Complete manuscript_20_11_10

Aim and Scope of the Series „Nothing endures but change“. Heraclitus the Ephesian (ca. 535–475 BC)

Institutions, defined as “the rules of the game”, are a key factor to the sustainable development of societies. They structure not only the multitude of human-human interactions of modern societies, but also most of the human-nature interactions. Poverty, famine, civil war, degradation of natural resources and even the collapse of ecosystems and societies often have institutional causes, likewise social and economic prosperity, sustainable use of resources and the resilience of socio-ecological systems. Agriculture, forestry and fisheries are those human activities where the interdependencies between human-human and human-nature interactions are perhaps most pronounced, and diverse institutions have been developed in history to govern them. Social and ecological conditions are, however, ever changing, which continuously challenge the existing institutional structure at a given point in time. Those changes may be long-term, like population growth or climate change, medium-term, such as new technologies or changing price relations, or short-term, like floods or bankruptcies, but all of them pose the question whether the rules of the game need to be adapted. Failures to adapt timely and effectively may come at a high social cost. Institutional change, however, face a principal dilemma: on the one hand, institutions need to be stable to structure expectations and effectively influence human behaviors; on the other hand, they need to be adaptive to respond to the ever changing circumstance mentioned above. Understanding stability and change as well as developing adaptive institutions and effective, efficient and fair mechanisms of change are, therefore, of central importance for societies and an ongoing research challenge for social scientists. If we want to improve the effectiveness, efficiency and adaptability of institutions, it stands to reason that we have to develop a good understanding of the causes, effects, processes and mechanism of stability and change. This is the aim of the series “Institutional Change in Agriculture and Natural Resources,” which attempts to answer the questions "How do processes and mechanism of institutional change actually work? What and who are the main determinants and actors driving, governing and influencing these processes? What are the economic, political, social and ecological consequences? How can adaptive institutions be designed and developed, and what governance structures are required to make them effective?” These are the questions at the heart of

vi

Aim and Scope of the Series

the series. The works published in this series seek to provide answers to these questions in different economic, social, political and historical contexts. Volker Beckmann and Konrad Hagedorn Humboldt-Universität zu Berlin

Preface This volume contains a selection of papers that were presented at the AsiaLink RECREATE seminar at the University of Economics, Ho Chi Minh City, Vietnam, in June 2007 and the international Asia-Link RECREATE conference at Nanjing Agriculture University, China, in October 2008. The seminar, the conference and the publication of this book are part of AsiaLink’s “Restructuring Higher EduCation in Resource and Environmental Economics in East Asian Transition Economies” project activities, which have been funded by the European Union under its Asia-Link programme. The primary objectives of the Asia-Link RECREATE project have been the • strengthening of staff capacity for education and research in resource and environmental economic at Nanjing Agricultural University (NAU) and at the University of Economics, Ho Chi Minh City (UEH) and • development and revision of educational curricula in environmental and resource economics at the undergraduate and graduate levels at the respective universities in Vietnam and China. The majority of the authors included herein are from partner institutions of the Asia-Link RECREATE project, including Wageningen University, The Netherlands; Nanjing Agricultural University, China; University of Economics, Ho Chi Minh City, Vietnam; Humboldt University Berlin, Germany; and the International Institute of Social Studies, The Hague, The Netherlands, but also from other researchers in Southeast Asia working on environmental and natural resource economic issues in the region. The selected papers provide an overview, albeit not a complete one, regarding the environmental and natural resource problems Southeast Asian countries are facing from an economic perspective. The papers are written in such a way so that they can serve as supporting case study material for classes in environmental and natural resource economics at the graduate level. But we also hope that environmental and natural resource economists with an interest in Southeast Asia will find the various chapters of interest as well as stimulating for their own research. In completing the book we received support from a number of persons. First of all, we would like to thank the EU programme officers, Le Chi Hieu and Do Quang Huy, of the delegation of the European Commission to Vietnam for their support to the project. The Asia-Link RECREATE project steering committee – Konrad Hagedorn, Ekko van Ierland, Arie Kuyvenhoven, Pham Van Nang, Hans Opschoor, and Futian Qu – deserves

viii

Preface

our gratitude for supporting and critically reviewing the process of the project and supporting the seminar and conference of which this book is one result. We would also like to thank the participants of the seminar, the conference, and the authors for their contributions. And we are grateful to Christopher Hank for his superb language editing job. Finally, we hope the spirit of the Asia-Link RECREATE project will be reflected in the chapters of this book and readers will enjoy the volume as much as we enjoyed editing it. Volker Beckmann Nguyen Huu Dung Xiaoping Shi Max Spoor Justus Wesseler

Table of Content Preface ........................................................................................................vii Table of Content ......................................................................................... ix Chapter 1 Economic Transition and Natural Resource Management in East and Southeast Asia: An Introduction and Overview ................. 1 Volker Beckmann, Nguyen Huu Dung, Xiaoping Shi, Max Spoor, and Justus Wesseler 1

Introduction....................................................................................... 1

2

Part I - Transition and Sustainability: The Southeast Asian Perspective ........................................................................................ 2

3

Part II – Sustainable Land Management: Land Tenure and Investment ......................................................................................... 3

4

Part III – Sustainable Land Management: Farmland Conversion ........................................................................................ 4

5

Part IV – Agricultural Intensification: Input Use Efficiency and Sustainability.............................................................................. 5

6

Part V – Agricultural Intensification: Pesticide Use and IPM.......... 7

7

Part VI – Natural Resource Endowment and Trade ......................... 9

8

Part VII – Natural Resource Risk and Coping Strategies............... 10

Part I Transition and Sustainability: The Southeast Asian Perspective ... 11 Chapter 2 Climate Change and Sustainable Development ................................. 12 Hans Opschoor 1

Introduction..................................................................................... 12

2

Climate change and sustainable development: where are we now? ................................................................................................ 13

2.1

Climate change: facts and expectations ...................................... 13

2.2

Climate change and future development .................................... 14

2.3

A case study: Vietnam ................................................................ 16

x

Table of Content

3

Economic impacts and costs – a global-level analysis ................... 17

3.1

Damage costs .............................................................................. 17

3.2

Mitigation costs........................................................................... 18

3.3

Adaptation costs.......................................................................... 19

4

Towards climate-resilient and sustainable development in developing economies..................................................................... 20

4.1

Adaptation and climate-resilient development ........................... 20

4.2

Climate and sustainable development: mitigation...................... 23

5

Conclusions and recommendations ................................................ 26

References ............................................................................................... 27 Part II Sustainable Land Management: Land Tenure and Investment...... 31 Chapter 3 The Land Rental Market and its Effect on Agricultural Production in Rural China................................................................... 33 Xiaobing Wang, Thomas Glauben, and Yanjie Zhang 1

Introduction..................................................................................... 33

2

Theoretical framework and econometric models............................ 35

3

Data description and variables ........................................................ 37

4

Empirical results ............................................................................. 40

4.1

Multivariate analysis of the incidence of land rental activity..... 40

4.2

The results of production functions ............................................ 42

4.3

The level and determinants of technical efficiency .................... 44

5

Conclusion ...................................................................................... 44

References ............................................................................................... 45 Chapter 4 Characteristics of the Rural Land Rental Market in China: A Village Level Study ............................................................................... 47 Stephan Piotrowski and Christian Böber 1

Introduction..................................................................................... 47

2

Data ................................................................................................. 48

3

Methodology ................................................................................... 51

Tables of Content

xi

4

Descriptive results........................................................................... 54

5

Econometric results......................................................................... 59

5.1

Village-level production function............................................... 59

5.2

Allocative efficiency in the land rental market........................... 61

6

Conclusions..................................................................................... 63

Acknowledgements ................................................................................. 64 References ............................................................................................... 64 Chapter 5 Impact of Soil and Water Conservation Investments on Agricultural Development in Western China..................................... 67 Nico Heerink, Rui Li, Shuyi Feng, Kaiyu Lu, and Xiaobin Bao 1

Introduction..................................................................................... 67

2

Trends in erosion and SWC investments........................................ 69

3

Results for erosion functions .......................................................... 76

4

Results for agricultural production function................................... 83

5

Conclusion ...................................................................................... 89

References ............................................................................................... 92 Part III Sustainable Land Management: Land Conversion .......................... 95 Chapter 6 What Is the Optimal Rate of China’s Conversion of Farmland? Statistical Experience from the Past 15 Years ................................... 97 Rong Tan and Futian Qu 1

Introduction..................................................................................... 97

2

Optimal degree of farmland conversion ......................................... 98

3

Methods......................................................................................... 100

4

Marginal ecological revenue of farmland..................................... 103

5

Estimation and results ................................................................... 105

5.1

Data ........................................................................................... 105

5.2

Estimation of marginal revenue in the two sectors................... 106

5.3

Estimation of the MR and MC curves ...................................... 109

xii

Table of Content

5.4

Calculation of the expense loss, excessive loss I and excessive loss II ........................................................................ 110

5.5

Analysis of the results............................................................... 112

6

Conclusions................................................................................... 114

Acknowledgements ............................................................................... 115 References ............................................................................................. 115 Chapter 7 Effects of the Public Domain Property Rights on Collective Farmland Prices: A Chinese Case Study .......................................... 117 Guancheng Guo 1

Introduction................................................................................... 117

2

Theory of public domain property rights...................................... 118

3

Analytical model ........................................................................... 119

4

Structure of collective farmland’s property rights in China ......... 121

4.1

Ownership ................................................................................. 121

4.2

Contractual management rights ................................................ 122

5

Public domain I of collective farmland in China.......................... 123

5.1

Function attributes of collective farmland in China ................. 123

5.2

Connections between function attributes, property rights and collective farmland prices ......................................................... 125

5.3

Function attributes and public domain I of collective farmland in China...................................................................... 126

6

Public domain II of collective farmland in China ....................... 126

6.1

Non-performance of institutions ............................................... 126

6.2

Organizational imbalances........................................................ 128

7

Effect of the public domain on collective farmland prices in China ............................................................................................. 129

8

Empirical data ............................................................................... 131

8.1

Brief introduction of the selected county.................................. 131

8.2

Expropriation price of collective farmland (P1)........................ 131

8.3

Market price appraisal of collective farmland (P2)................... 133

8.4

Comparison of two kinds of prices ........................................... 135

Tables of Content

9

xiii

Conclusion and discussion............................................................ 136

9.1

Conclusion ................................................................................ 136

9.2

Discussion ................................................................................. 137

Acknowledgements ............................................................................... 137 References ............................................................................................. 137 Part IV Agricultural Intensification: Input Use Efficiency and Sustainability ....................................................................................... 139 Chapter 8 Technical Efficiency of Shrimp Farms in the Mekong Delta, Vietnam ................................................................................................ 141 Tihomir Ancev, Md Abdus Samad Azad, Do Thi Den, and Michael Harris 1

Introduction................................................................................... 141

2

Literature review ........................................................................... 144

3

Theoretical framework.................................................................. 145

4

Data and method ........................................................................... 147

4.1

Socioeconomic characteristics of shrimp farmers .................... 148

4.2

Economic characteristics of shrimp farms................................ 149

4.3

Method ...................................................................................... 151

5

Results ........................................................................................... 153

6

Summary and conclusion.............................................................. 156

Acknowledgements ............................................................................... 157 References ............................................................................................. 157 Chapter 9 Understanding Environmental and Social Efficiencies in Indonesian Rice Production ............................................................... 161 Joko Mariyono, Budy P. Resosudarmo, Tom Kompas, and Quentin Grafton 1

Introduction................................................................................... 161

2

Brief literature review ................................................................... 163

3


xiv

Table of Content

3.1

Environmental and technical efficiencies ................................. 165

3.2

Chemical waste and environmental costs ................................. 168

3.3

Private and social efficiency ..................................................... 169

4

Implementation ............................................................................. 171

4.1

Environmental efficiency.......................................................... 172

4.2

Social efficiency........................................................................ 172

5

Data ............................................................................................... 173

6

Results and discussion .................................................................. 174

6.1

Environmental efficiency.......................................................... 176

6.2

Chemical waste and environmental costs ................................. 177

6.3

Social efficiency........................................................................ 179

7

Conclusion .................................................................................... 181

References ............................................................................................. 182 Chapter 10 Rural Household Energy Consumption and Choice: A Case Study of Nanjing, Jiangsu Province, China...................................... 187 Lina Shi, Xiaoping Shi, Nico Heerink, and Shuyi Feng 1

Introduction................................................................................... 187

2

Analytical framework ................................................................... 189

3

Description of the research site..................................................... 193

4

Model and results .......................................................................... 196

4.1

Biomass use model .................................................................. 196

4.2

Energy choice model................................................................. 197

5

Empirical results ........................................................................... 200

6

Conclusions................................................................................... 204

Acknowledgements ............................................................................... 205 References ............................................................................................. 206

Tables of Content

xv

Chapter 11 Community-Based Aquaculture for Poverty Reduction: Institutional and Technical Options for Sustainable Resource Use......................................................................................................... 209 Christine Werthmann and Chi Mai Thi Truc 1

Introduction................................................................................... 209

2

Materials and methods .................................................................. 210

3

Results ........................................................................................... 212

3.1

Local arrangements ................................................................... 212

3.2

Water management ................................................................... 212

3.3

Technical experiences ............................................................... 214

4

Discussion ..................................................................................... 216

5

Conclusions................................................................................... 219

References ............................................................................................. 220 Chapter 12 Shifting Livelihood Strategies of Small Cotton Farmers in Southern Xinjiang ............................................................................... 221 Max Spoor, Xiaoping Shi, and Chunling Pu 1

Introduction................................................................................... 221

2

Income inequality, poverty and agriculture .................................. 225

3

Income dependency from land and cotton.................................... 230

4

Diversification and policy-induced change .................................. 235

5

Conclusions................................................................................... 237

Acknowledgements ............................................................................... 239 References ............................................................................................. 239 Part IV Agricultural Intensification: Pesticide Use and IPM...................... 241 Chapter 13 Reaping Bitter Fruit? Farmers’ Health and Pesticide Use in the Mekong Delta, Vietnam ...................................................................... 243 Nguyen Huu Dung, Max Spoor, and Lorenzo Pellegrini 1

Introduction................................................................................... 243

xvi

Table of Content

2

Farmers and pesticides .................................................................. 246

2.1

Frequency of pesticide application ........................................... 246

2.2

Farmers’ perceptions of the effects of pesticide use on personal health .......................................................................... 249

2.3

Evidence of health impairment ................................................. 250

3

Determinants of acute health impairments ................................... 253

3.1

Headache................................................................................... 254

3.2

Skin effects................................................................................ 255

3.3

Eye effects................................................................................. 255

4

Private health costs associated with pesticide use ........................ 256

4.1

Model specification................................................................... 256

4.2

Estimated pesticide-related health costs ................................... 258

5

Conclusion .................................................................................... 261

References ............................................................................................. 261 Chapter 14 Training and Visit (T&V) Extension vs. Farmer Field School: The Indonesian Experience ................................................................ 269 Budy P. Resosudarmo and Satoshi Yamazaki 1

Introduction................................................................................... 269

2

Overview of rice production in Indonesia .................................... 271

3

BIMAS: The Indonesian T&V program....................................... 275

3.1

Historical perspective................................................................ 275

3.2

Accomplishments of BIMAS.................................................... 279

3.3

Issues with BIMAS ................................................................... 280

4

Farmer Field School: The IPM program....................................... 281

4.1

Historical perspective................................................................ 281

4.2

Accomplishments of FFS.......................................................... 284

4.3

Issues with FFS ......................................................................... 285

5

Discussion ..................................................................................... 287

6

Conclusion .................................................................................... 288

Acknowledgements ............................................................................... 290

Tables of Content

xvii

References ............................................................................................. 290 Chapter 15 Determining Factors of IPM Adoption: Empirical Evidence from Longan Growers in Northern Thailand .................................. 297 Chapika Sangkapitux, Pornsiri Suebpongsang, Sakdamneon Nonkiti and Andreas Neef 1

Introduction................................................................................... 297

2

Supports for and barriers to Integrated Pest Management adoption: a review......................................................................... 299

2.1

Off-farm income ....................................................................... 299

2.2

Labor organization .................................................................... 300

2.3

Knowledge of pesticide effects on human health ..................... 300

3

Pest management in longan farming systems in northern Thailand......................................................................................... 301

4

Sample selection ........................................................................... 302

5

Model specification....................................................................... 302

5.1

Defining IPM adoption ............................................................. 302

5.2

Factors hypothesized to determine IPM adoption: ................... 303

5.3

Economic decision model of adoption: Count model............... 304

6

Empirical results ........................................................................... 305

6.1

Descriptive results..................................................................... 305

6.2

Regression results ..................................................................... 310

7

Discussion and conclusion............................................................ 312

Acknowledgement................................................................................. 313 References ............................................................................................. 313 Chapter 16 Transaction Cost Analysis of Hired Labor Use in Pest Management: An Empirical Study of Fruit Tree Farming in Thailand ............................................................................................... 317 Evi Irawan, Volker Beckmann, and Justus Wesseler 1Introduction ......................................................................................... 317 2


xviii

Table of Content

2.1

Asset specificity ........................................................................ 319

2.2

Frequency.................................................................................. 320

2.3

Uncertainty................................................................................ 321

3

Methods......................................................................................... 322

3.1

Data source................................................................................ 322

3.2

Data of analysis ...................................................................... 322

4

Results ........................................................................................... 325

4.1

Hired farm labor characteristics and wage rate ........................ 325

4.2

Task attributes and farm labor organization ............................. 326

4.3

Estimated probit model ............................................................. 329

4.4

Durian farming model............................................................... 330

4.5

Tangerine farming model.......................................................... 331

5

Conclusion .................................................................................... 332

References ............................................................................................. 333 Appendix ............................................................................................... 334 Part VI Natural Resource Endowment and Trade....................................... 337 Chapter 17 The Feasibility of Founding OREC and China’s Countermeasures................................................................................. 339 Zhonghui Wang and Yujie Chan 1

Introduction................................................................................... 339

2

The reasons for creating OREC .................................................... 340

2.1

The experience of OPEC .......................................................... 340

2.2

The current situation of the international rice market .............. 341

2.3

The world food crisis and its causes ......................................... 341

3

The feasibility of the establishment of OREC .............................. 343

3.1

The five countries planning on joining OREC are the prime producers and exporters of rice................................................. 343

3.2

The five countries are for the establishment of OREC............. 343

4

The barriers to establishing OREC ............................................... 344

Tables of Content

xix

4.1

Because of the world food crisis, many countries could not agree on the establishment of OREC........................................ 344

4.2

Fierce opposition to the establishment of OREC have been from rice importing countries. .................................................. 344

4.3

The power of OREC would be weak after its establishment.... 345

5

Impacts of the establishment of OREC on China and China’s countermeasures............................................................................ 346

5.1

Impacts of the establishment of OREC on China..................... 346

5.2

China’s countermeasures against the founding of OREC ........ 348

References: ............................................................................................ 349 Chapter 18 Competitiveness Analysis for Major Agricultural Product in the Mekong Delta: The Case of Tien Giang Province............................ 351 Nguyen Trong Hoai 1

Introduction................................................................................... 351

2

Literature review ........................................................................... 352

2.1

Competitiveness theory: from comparative advantage to competitive advantage .............................................................. 352

2.2

Measurements of competitiveness in agriculture ..................... 353

2.3

Related research ........................................................................ 354

3

How to measure competitiveness.................................................. 355

4

Research design for focus group interview................................... 357

5

Product analysis: rice .................................................................... 357

5.1

Measurement of DRC ............................................................... 357

5.2

Results from focus-group interviews........................................ 357

5.3

Results from expert interviews ................................................. 358

6

Product analysis: Citrus ................................................................ 359

6.1

Measurement of competitiveness.............................................. 359

6.2


6.3


7 7.1

Product analysis: shrimp ............................................................... 361 Measurement of competitiveness.............................................. 361

xx

Table of Content

7.2


7.3


8

Conclusion .................................................................................... 363

Acknowledgement................................................................................. 363 References ............................................................................................. 363 Part VII Natural Resource Risks and Coping Strategies .............................. 365 Chapter 19 Analysis of Risk and Coping Strategies of Fishing Communities in the Philippines when facing Natural Calamities.......................... 367 Maria Rebecca Campos 1

Introduction................................................................................... 367

2

Objectives...................................................................................... 369

3

Methodology and analytical approach .......................................... 369

3.1

Risk and management assessment framework ......................... 369

3.2

Quantification and evaluation of risk........................................ 371

3.3

Sources of risk........................................................................... 372

3.4

Data limitations for assessing risk ............................................ 374

3.5

Secondary data analysis ............................................................ 375

3.6

Quantification of risk due to natural calamities........................ 375

3.7

Estimation of aquaculture yield losses due to natural calamities. ................................................................................. 375

3.8

Identification of risk management and coping strategies ......... 376

3.9

Field verification and validation activities for primary data collection................................................................................... 376

4

Results and discussion .................................................................. 377

4.1

Production and profitability ...................................................... 377

4.2

Average loss due to natural risks .............................................. 378

4.3

Coping and adaptive practices .................................................. 379

4.4

Proactive practices of QUEDANCOR borrowers to mitigate impacts of natural calamities .................................................... 380

4.5

Assistance provided by institutions .......................................... 382

Tables of Content

4.6

xxi

QUEDANCOR calamity bridge funds ..................................... 383

5

Conclusions................................................................................... 383

6

Recommendations......................................................................... 384

Acknowledgements ............................................................................... 388 References ............................................................................................. 388 Contributors ............................................................................................ 389

Chapter 1 Economic Transition and Natural Resource Management in East and Southeast Asia: An Introduction and Overview Volker Beckmann1, Nguyen Huu Dung2, Xiaoping Shi3, Max Spoor4, and Justus Wesseler5 1 Introduction Economic and institutional reforms in East and Southeast Asia have caused impressive economic growth and improved the livelihoods of millions of people. In several regions, however, this growth has been obtained at the expense of land quality or to the detriment of other natural resources. As a consequence, the sustainability of future growth is threatened. Efforts aimed at promoting sustainable resource use in rural East and Southeast Asia are being and will be confronted with and influenced by two major changes. First, the continuous transition towards a market-oriented economy implies that massive, centralized state regulation will decrease and that economic, decentralized and informal resource management institutions are likely to increase in importance. Second, domestic economic liberalization, international trade liberalization and globalization will greatly affect domestic agricultural prices and, hence, the use of natural resources. Based on these observations, three key questions emerge: first, can agricultural output growth justify alterations in natural resource quality for, among others, air, water, land, or natural forests? Second, do changes in natural resource quality threaten sustainable agricultural development and, third, if so what are appropriate policy responses? To provide answers to these three questions, the international dimensions of 1

2 3 4 5

Department of Agricultural Economics, Faculty of Agriculture and Horticulture, Humboldt Universität zu Berlin, Germany/ Department of Environmental Management, Chair of Environmental Economics, Brandenburg University of Technology (BTU), Cottbus, Germany University of Economics, Ho Chi Minh City, Vietnam Nanjing Agricultural University, Nanjing, China Institute of Social Studies, The Hague/Erasmus University, Rotterdam, The Netherlands, and Barcelona Institute of International Studies, Spain Environmental Economics and Natural Resources Group, Wageningen University, Wageningen, The Netherlands

Beckmann, V., N.H. Dung, X. Shi, M. Spoor, J. Wesseler (eds.) (2010). Economic Transition and Natural Resource Management in East- and Southeast Asia. Aachen: Shaker Publisher, pp. 1-10.

2

Beckmann, Dung, Shi, Spoor, and Wesseler

changes in natural resource quality and the integration of the East and Southeast Asian economies into the world economy have to be taken into account. Answering these questions completely would be a tall order. Our objective, then, is more modest. The following papers address some aspects of these challenging questions, but the volume as a whole is not aimed at fully answering them. We hope rather to provide some answers while challenging readers to further think about economic transition and natural resource management. The volume includes a selection of papers presented at the Asia-Link RECREATE symposium in Ho Chi Minh City, Vietnam and the Asia-Link RECREATE conference in Nanjing, China. The eighteen chapters are organized into seven parts, with some parts including more chapters than others, illustrating where research emphasis has been placed in the past. Those parts having fewer chapters also illustrate where additional research is needed. Some readers may miss chapters explicitly addressing issues related to forestry, fisheries and biodiversity conservation. This does not imply that these issues are not important, but is rather a result of the history of this volume; the authors are mainly participants from the Asia-Link RECREATE project, including partners located at agricultural universities or faculties, thus explaining the bias. Nevertheless, we are convinced that the papers provide sufficient insights for the problems at stake. 2 Part I - Transition and Sustainability: The Southeast Asian Perspective Hans Opschoor opens Part I by addressing the challenges posed for Southeast Asia and, in particular, Vietnam caused by climate change. In “Climate Change and the Economics of Adaptation and Mitigation: A Developmental Perspective,” Opschoor illustrates the challenges and implications for sustainable development brought on by climate change. While he does not favor either adaptation or mitigation strategies exclusively, he strongly recommends a path leading toward climateresilient economies. The major argument for following a combined strategy is long-run sustainable development, including short-term measures such as poverty eradication and long-term measures such as sustainable patterns of production and consumption as well as the protection of natural resources. While – given the nature of the contribution – specific policy recommendations are not provided, subsequent chapters address a number of the issues raised by Opschoor.

Introduction and Overview

3

3 Part II – Sustainable Land Management: Land Tenure and Investment Sustainable land management is strongly related to systems of property rights which provide incentives for short- and long-term investments to secure and improve soil fertility and agricultural production. Although many East and Southeast Asian transition countries still rely on state and collective ownership of land, most of them have introduced important land tenure reforms such as individual, tradable, but time-limited land use rights. Part II contains two papers focusing on the relationship between the emerging land rental market and agricultural production and one paper examining the effects of public and private investments in soil and water conservation on soil erosion and agricultural production. All three papers focus their attention on China. The first paper, by Xiaobing Wang, Thomas Glauben and Yanjie Zhang, explores the relationship between farmers’ participation in the land rental market and farm production and efficiency. Based on farm-level panel data gathered in two Chinese provinces and encompassing the period from 1995-2002, the authors first examine factors influencing farmers’ decisions to rent-in agricultural land and, then, estimate the effect of the share of rented-in land on agricultural production and efficiency by applying a translog stochastic production function approach. Their results reveal that land rental market participation has had a mixed impact on agricultural production, but uniformly improved the efficiency of production. This result largely confirms economists’ expectations: well functioning land markets are essential for achieving technical and allocative efficiency, but they do not necessarily increase overall agricultural production. The functioning of the land rental market in China is further investigated by Stephan Piotrowski and Christian Böber. Based on village-level data obtained in 2007 in one Chinese county in North China, they present detailed information on land rental market organization. Only about 5% of the households in the sample villages were engaged in land rental, and the transfer of land-use rights occurred mainly within the villages. Although land rental markets were comparatively small and inactive, further analysis by Piotrowski and Böber shows that land rental prices in this area were significantly influenced by the village-level marginal value product from agriculture. The authors conclude from this example that land rental markets in China do function comparatively well, despite high administrative burdens and small village markets. Nevertheless, they also identify room for improvement: land trusts or mediators may be able to reduce transaction costs on land markets and extend the borders of land markets beyond families, clans and villages.

4


In the third article of this part, Nico Heerink, Rui Li, Shuyi Feng, Kaiyu Lu and Xiaobing Bao study public and private investments in soil and water conservation (SWC) in Western China, a region which has been and continues to be severely affected by wind and water erosion, threatening agricultural production and ecological systems. Utilizing data from the Chinese national soil erosion surveys covering the period 1985-2002, Heerink et al. estimate the effects of SWC investments on soil erosion and, subsequently, the effects of soil erosion on agricultural GDP. Finally, they calculate the return on investment in SWC. The results show that, among others things, investments made by local governments have had significantly negative impacts on the extent of soil erosion, while private investments by farm households exhibit significantly negative effects on the severity of soil erosion. The severity of erosion subsequently turned out to have significantly negatively affected agricultural GDP in North-western China, though not, however, in South-western China. Overall, the authors conclude that SWC investments made by local governments and households in North-western China were profitable, but less profitable than other public investments in agriculture. Heerink et al. do acknowledge, however, that their analysis only takes into account the effects of SWC investments on agricultural production, disregarding its effects on the quality of natural resources and the functioning of ecological systems. Future research should take these effects into account. 4 Part III – Sustainable Land Management: Farmland Conversion Farmland conversion for urban development is a highly controversial issue worldwide, be it for environmental or food security reasons. The two papers constituting Part III focus on farmland conversion in China, where strong economic growth, rapid urbanization and the dominance of state ownership over collective ownership has resulted in a significant reduction of farmland resources. Farmland preservation has, thus, over time become an important policy objective in China. The two papers explore reasons for the excessive farmland conversion and raise questions of whether more or less governmental invention is needed and how property rights over agricultural land should be assigned. Rong Tan and Futian Qu analyze farmland conversion in China between 1989 and 2003 from an efficiency point of view. Theoretically, optimal farmland conversion should occur up to the amount of land at which the marginal benefits of farmland conversion equals its marginal social costs, where the latter is determined by the marginal agricultural revenue forgone and the marginal loss in external social and ecological benefits generated by agricultural land. In practice, however, not only are the external social and ecological benefits of farmland ignored, but the price of farmland is


5

also further suppressed by governmental invention. Based on provincelevel data and benefit transfers, Tan and Qu estimate the social optimal amount of farmland conversion as well as the amount of land that should have been converted when taking into account only marginal agricultural revenue forgone. They arrive at astonishing results. According to their calculations, 52.9% of the farmland conversion in China between 1989 and 2003 resulted from governmental invention in suppressing the price of farmland, while only 28.6% resulted from ignoring the external costs of farmland conversion. Thus, Tan and Qu argue that a policy aiming at reducing farmland conversion in China should focus first on “getting prices right” by reducing governmental intervention and improving the functioning of the land market. Guancheng Guo explores in more detail how farmland prices are related to the property rights systems in China. Based on the theory of public domain property rights, the author develops an analytical model distinguishing between private property rights and property rights in public domains I and II. Category I in the public domain results from high transaction costs that prevent individuals from privatizing all attributes of a commodity or resource, whereas public domain II is determined by institutional constraints that do not allow property rights to be privatized. The overall extent of public domains I and II, then, determines the price of a good or resource traded on a market. By using pricing data for farmland conversion from Jiangdu, the author shows that public domain II in China reduces the market price of farmland by 67.31%. In order to “get prices right”, Guo argues for property rights changes that reduce the distorting impact of public domain II and also enable farm households to reduce public domain I. Thus, similar to Tan and Qu, Guo also concludes that effective markets and stronger agricultural property rights would reduce farmland conversion in China considerably, thus increasing overall welfare. 5 Part IV – Agricultural Intensification: Input Use Efficiency and Sustainability Sustainable land management to a large extent depends on the intensity of agricultural production. In particular, the sustainability of natural resource use and land use has been challenged by intensification of agriculture production. Part IV includes five papers addressing input use efficiency and sustainability. While at the social level sustainability requires input use efficiency, at the farm or private level input use efficiency will not necessarily imply sustainability, if externalities have not sufficiently been internalized. The paper by Tihomir Ancev, Md Abdus Samad Azad, Do Thi Den, and Michael Harris investigates the technical efficiency of shrimp farms in the

6


Mekong Delta in Vietnam. Shrimp farming has significantly increased in Vietnam, but is also characterized by high financial risks and has been accused of causing destruction of mangrove forests and environmental pollution. Investigating the technical efficiency of shrimp farms can provide information for improving the sustainability of shrimp farming from the farmers’ as well as the societal point of view. The authors differentiate in their analysis between intensive and extensive shrimp farming systems and apply a stochastic production frontier approach for estimating technical efficiency among 193 shrimp farms. They find intensive shrimp farmers to be more technically efficient than extensive ones, explaining this as being a result of differences in managerial skills. The results indicate that there is substantial scope for improving production among existing shrimp farms. According to the authors, this is relatively good news, as production can be increased without expanding the area used. Similarly Joko Mariyono, Budy P. Resosudarmo, Tom Kompas, and Quentin Grafton investigate the environmental and social efficiencies in Indonesian rice production, using a stochastic production frontier approach as well. They find the environmental efficiency across regions in Indonesia to be low at 0.64. As expected, the efficiency of chemical inputs (fertilizer and pesticides) is higher when environmental costs are not included. Again, the relatively low efficiency levels found imply that there is scope for reducing the environmental impact of rice production via better allocation of inputs. The authors conclude that more flexibility with respect to land use might currently be the best solution for improving the efficiency of chemical use and, thereby, the relative impact of rice production on the environment. This may not necessarily result, however, in less use of chemicals. Consequently, reducing the overall environmental costs of rice production still remains a challenge. Lina Shi, Xiaoping Shi, Nico Heerink and Shuyi Feng investigate the use of biomass as part of the energy consumption of 258 farm households belonging to Nanjing City of Jiangsu province in China. Even though the households can be considered to belong to the economically well developed areas in South China, biomass (firewood, straw and stalks) as an energy source still comprises about 57% of total energy consumption. Wealthier and younger households consume on average less biomass, indicating a transition from more traditional energy sources to electricity, gas, and coal. Although it has not been explicitly investigated, the authors suggest that less use of traditional biomass will increase human health due to decreased indoor pollution. New technologies for converting biomass in energy, such as gasification, can further increase the production of clean energy and reduce environmental costs. As sources of biomass are still


7

abundant, promoting such cleaner technologies looks promising, but also needs further investigation. Christine Werthmann and Chi Mai Thi Truc investigate the intensification of rice-fish cultivation in the Mekong Delta of Cambodia and Vietnam being carried out and managed by groups of farmers using communal water bodies. According to the authors, the advantage of group management is a reduction in average production costs. Governance problems common to group-managed projects have also been found, however, such as difficulties in collecting payments, excluding non-group members from access to the resources used and produced, and lack of support from higher-level institutions. The reported productivity for participating farmers is consequently modest – hardly more than catches from wild populations – while spill-over benefits for non-group members have been reported, but not yet quantified. Non-group members seem to free-ride on the fish stock inputs provided by group members. The case study illustrates typical problems found in managing common pool resources in agricultural situations where inputs are provided by a defined group, but benefits are shared non-voluntarily between group and nongroup members. Max Spoor, Xiaoping Shi and Chunling Pu investigate the possibilities for intensifying agricultural production by substituting cotton with highvalue crops such as vegetables, fruits and nuts as part of a shifting livelihood strategy among small cotton farmers in Southern Xinjiang, China. Rural net income in Xinjiang is low by Chinese standards, with an average for the region at 2,737.3 Yuan per capita in 2006, with the major source of income coming from cotton, ranging between 65.6 and 73.1 percent. The intensification of agricultural production to shift households out of poverty is policy induced, but the authors note that the success of the strategy depends, on the one hand, on water availability and, on the other hand, on the possibilities for marketing high-value crops. The authors remain skeptical about the possibilities for marketing such high-value corps, as other prefectures in China are also promoting diversification of agriculture production via similar strategies. A particular problem in the study area is caused by the high use of pesticides in cotton, affecting pesticide residue levels of fruits and nuts intercropped in cotton fields. 6 Part V – Agricultural Intensification: Pesticide Use and IPM While the papers presented in Part IV address input use efficiency by considering different kinds of inputs, the papers from Part V address one specific input that has received a lot of attention – the use of pesticides – and one specific response towards increasing the efficiency of that input: integrated pest management (IPM).

8


Nguyen Huu Dung, Max Spoor and Lorenzo Pellegrini investigate the impact of pesticide use on farmers’ health in the Mekong Delta of Vietnam. While the authors observe a decline in pesticide use per crop, annual pesticide use per hectare has increased, due to intensified production. Although the introduction of IPM has increased awareness about the negative health impacts of pesticide use among farmers, nevertheless pesticide use still results in acute poisoning symptoms among farmers and still causes substantial private health costs. The authors calculate marginal health costs to be on the order of about 50VND per additional gram of pesticide active ingredient (a.i.) for farmers participating in the IPM programme, while marginal health costs in the control group of about 58VND were significantly higher by about 16%, illustrating the health-cost benefits of the IPM programme. Budy P. Reosudarmo and Satoshi Yamazaki investigate the implications of the farmer field school (FFS) approach on pesticide use and agricultural production and contrast the approach with the previously used Training and Visit (T&V) Extension Program or the Massive Guidance (BIMAS) Program in Indonesia. The authors provide a detailed review of rice intensification programs in Indonesia since the beginning of the 1960’s. While rice yields increased substantially during the period when the T&V approach was in force, the approach has been criticized for its high costs, collusion between the Department of Agriculture and chemical companies and the promotion of high pesticide use, resulting in pest resistance and high health costs. These problems gave support for the introduction of IPM, using the FFS approach. Yet the success of this approach has also been disputed, based again on high costs and, additionally, low adoption rates. Nevertheless, the programme did result in reduced pesticide use, though not in increased absolute yields. In their conclusions, the authors propose the T&V-approach for boosting yields in areas where the level of development is low, while later switching to an FFS approach for maintaining yield increases. The FFS approach has been widely used in Thailand. Two papers in this part assess the adoption of IPM among fruit tree farmers in Thailand. Chapika Sangkapitux, Pornsiri Suebpongsang, Sakdamneon Nonkiti, and Andreas Neef assess factors determining the adoption of IPM among longan growers in Lamphun Province in Northern Thailand. They find strong support for the effect of knowledge about pesticide use on the adoption of IPM, though the number of IPM training courses shows no significant effect. At the same time, off-farm income exhibits a pronouncedly significant negative effect on IPM adoption. This is an important observation when considering the integration of farm households into the overall economy and the increase in off-farm activities that often


9

go along with it. Evi Irawan, Volker Beckmann, and Justus Wesseler investigate the use of hired labor in pest management among durian and tangerine farmers in Thailand. They use a transaction cost approach to test the effect of pest management tasks on the likelihood of hired-labor farm employment performing such tasks. According to theory, hired farm labor is assigned to tasks that are easier to monitor, are applied repetitively, and do not require specific skills; hence, the adoption of IPM is more likely under farm labor organizations comprised largely of family, rather than hired, labor. The empirical results support these theoretical predictions and are especially stronger for the case of durian farmers. Both studies support the argument that farm labor organization is an important factor in explaining the adoption of IPM. 7 Part VI – Natural Resource Endowment and Trade The two papers in Part 6 of this volume investigate the role of natural resource endowment and trade. Zhonghui Wang and Chan Yujie explore the possibilities for founding an Organization of the Rice Exporting Countries (OREC) as a response to the rise of food prices in 2008 and possible countermeasures by China. This idea received some attention during the food crisis, and these authors are the first to have assessed its implications. The idea of OREC – initiated by the Thai Government – had been to include Myanmar, Laos, Vietnam, and Cambodia. The main idea was to regulate the supply of rice, address food crises and, thereby, protect the member countries’ own interests. The authors draw their conclusions about the need and potential success of OREC by looking at the history and current situation of the Organization of the Petroleum Exporting Countries (OPEC). Based upon its superb natural resource endowments for rice production, the five member nations of OREC would possibly have been able to control about 90% of world rice export, providing them with interesting opportunities for controlling world trade thereof. China, as a net rice exporter, would mainly have benefited if OREC were able to have controlled rice prices at a higher level. The authors assess that the power of OREC would have been weak after its establishment, as the control of rice production, which is decentralized and carried out by small-scale producers, would be difficult. While the idea of establishing OREC had already evaporated a few days after its announcement, the debate on controlling food supplies via government intervention is ongoing. Nguyen Trong Hoai assesses the competitiveness of major agricultural products in the Mekong Delta of Vietnam, measured in terms of domestic resource costs. Not surprisingly, rice and shrimp were identified as the two commodities to be internationally competitive, owing to natural resource endowments supporting their production. The author warns, however, that

10


further integration of Vietnam into international trade will expose producers to stronger international competition. As the paper by Ancev et al. from Part IV on the technical efficiency of shrimp production illustrates, there is scope for improving competitiveness in this sector. Hoai recommends strengthening the efficiency of shrimp and rice production by supporting vertical integration and technology transfer as well as providing better access to information for farmers. 8 Part VII – Natural Resource Risk and Coping Strategies In this last part, Maria Rebecca Campos analyses the risk and coping strategies of fishing communities when facing natural calamities: which are a regular phenomenon in places like the Philippines, where they are caused by typhoons. Not surprisingly, fishing communities there are joined by governmental organizations in response to this regular threat. One of the responses of the government has been supporting fishing communities via financing mechanisms, providing opportunities for hedging the risks caused by regularly occurring natural calamities. While the idea works in principle, implementation is more difficult. Campos points out a number of problems the Philippine government has met while implementing its scheme. In particular, she concludes, investing in better weather forecasting as well as more locally targeted financing schemes and improvement of local production technologies adapted to typhoons may improve the effectiveness of such schemes.

Part I Transition and Sustainability: The Southeast Asian Perspective

Chapter 2 Climate Change and Sustainable Development Hans Opschoor1 Abstract: Climate change is likely to affect development in many developing countries. The repercussions of this are discussed in general terms, with illustrations related to Vietnam. The emphasis is on economic aspects, in the perspective of 'sustainable development'. Even though most of the climate change problems are due to emissions outside the developing countries, curbing emissions in the latter countries as well is necessary if safe limits to warming are to be adhered to. Mitigation measures in developing countries would need financial compensation and support from the traditional industrialised countries. Economic aspects of climate change underlying global strategy development relate to: (i) the damage due to impacts of climate change, (ii) the costs of mitigating climate change and (iii) the costs of adaptation. For each of these estimates are summarised; these estimates are surrounded by a host of uncertainties but scenario analysis suggest that deep mitigation globally is economically viable and desirable. The two-way links between sustainable development and adaptation on the one hand, and mitigation on the other, are explored in a qualitative way. The chapter aims at underpinning that climate responses and development can reinforce one another. Keywords: Climate Change Economics, Sustainable Development, Mitigation, Adaptation, Vietnam

1 Introduction Climate change is a serious and urgent issue with great developmental significance. The scientific consensus is that human activities have contributed to it significantly and that the changes taking place are far more rapid and dangerous than thought earlier (IPCC 2007a; Stern 2006). While climate change results from activities all over the globe, with contributions to it being rather unevenly spread, it leads to very different impacts in 1

Dr. J.B. Opschoor, emeritus professor, Department Economics of Sustainable Development, Institute of Social Studies (ISS), The Hague, Netherlands; emeritus professor, Department of Spatial and Environmental Economics, Vrije Universiteit, Amsterdam, Netherlands.


Climate Change and Sustainable Development

13

different countries. It seems clear now that the worst impacts will fall on developing countries. Thus, there is a need to consider the links and feedbacks between climate change (and policies to address it) and development. On the one hand, development paths vary in the ways in which they affect climate; on the other, different climate policies will have different impacts on development trajectories. The policy challenge in this is obvious. International negotiations should, by the end of 2009, lead to a new strategy to deal with the challenge of climate change beyond 2012. The Bali Action Plan (UNFCCC 2007d) outlining this process holds that development will have to be climate-resilient and sustainable. In this chapter we will look at what this means for developing countries: from a largely global perspective, and focusing on economic aspects. We begin by sketching what climate change might mean for development (section 2) and, in particular, what climate change-related costs might be in general (section 3), to then move on (in section 4) to the questions of whether and how the main response strategies relate to notions like sustainable economic development at the national level for adaptation and for mitigation. In section 6 some conclusions are drawn. 2 Climate change and sustainable development: where are we now? 2.1 Climate change: facts and expectations Human activities since 1750 have caused the mean temperature of the earth’s atmosphere to rise by 0.70 C (IPCC 2007a). This has already begun to have noticeable climatological impacts (e.g., in the form of heat waves, frequency of extreme events, and recession of glaciers); further temperature increases contain the potential for much larger and even catastrophic impacts. If warming exceeds the threshold of 20 C, catastrophic changes are highly probable (Stern 2006) during the 21st century - much larger than those observed so far, more abrupt and painful, and irreversible. Events like ice-cap melting and conveyor belts may still be insufficiently understood, but they are tied to to impacts of a magnitude that, taking a precautionary approach, are to be avoided if possible. While climate change results from activities all over the globe, it may lead to very different impacts in different countries, depending on local/regional environmental conditions and on differences in vulnerability to climate change - independent of the contributions to climate change of these countries. Climate change is likely to undermine the sustainability of livelihoods as well as resource bases for development. The worst impacts will fall on developing countries and, in general, the poor will be exposed relatively more than other social layers (IPCC 2007 b). Climate change will adversely affect economic growth, both through the structural changes it

14

Opschoor

brings about and through its impacts on frequencies of extreme events. Slower growth is expected to cause an increase in poverty and child mortality. Stern (2006) estimates that climate change could cause an additional 145- 220 million people to live below the US$2/day poverty line in South Asia and sub-Saharan Africa alone, and could cause significant additional child deaths there. Responses to climate change (IPCC 2007 b and c) include coping with the impacts of it and suffer through the associated damages or reacting to them by altering behavior, institutions, structures and even development paths in such a way so as to reduce and curb damage (adaptation). A more fundamental response would be for the world economy to reduce its emission of greenhouse gases and alter its patterns of land use in such a way so as to prevent and curb warming itself, and to enhance sinks for greenhouse gases (mitigation). 2.2 Climate change and future development About 50 % of climate change is the result of adding more CO2 to atmospheric concentrations. The clearing of tropical forests and other forms of land use change account for some 25% of the increase in world carbon dioxide levels, and some 75 % of it is due to fossil energy use. This chapter is restricted largely to adaptation (which, of course, is insensitive to where carbon comes from), but when dealing with emissions the focus is on fossil energy use. The 2007 World Energy Outlook (IEA 2007) anticipates fast growth of global energy needs, from over 23 Gt towards some 43 Gt in 2030, if a “business as usual”(BAU) perspective were to prevail, with developing countries contributing 74% to the increase in global primary energy use in this scenario (Table 1). The different regions of the world contribute to emissions in very different ways and at quite divergent levels. Historically, almost all of the greenhouse gas emissions have originated in the North. But, even if we look at current emissions, the differences between countries are striking. Per capita, 2004 CO2 emissions were 4.5t; in OECD countries this was 11.5t and in developing countries 2.4t. In high-income developing countries this was 13.3t, but in low-income countries it was only .9t and in the so-called Least Developed Countries (LDCs) this figure even dropped to .2t (UNDP 2007: 69). Given these findings, it is all the more problematic that developing countries, and especially the poor in these countries, will be particularly vulnerable to climate change and will suffer the most from any global warming that occurs in the decades to come (IPCC 2007 b, c; CDP 2007).

15


Table 1: Primary energy demand and CO2 emissions for “business as usual”, 2000-2030

OECD

Primary Energy Demand (Btonnes oil eq)

Emissions (Gt CO2)

2000

2005

2030

2000

2005

2030

~5.1

~5.6

~6.9

~12

~13.5

~16,5

~1.0

~1.1

~1.8

~3

~4.4

~5.3

~3.6

4.7

~9

~8.5

~8.9

~20.2

~9.7

11.4

17.7

~23.5

~26.8

~42

USA EU TransEc Russia DevCies China India World

Sources: estimated (roughly) from World Energy Outlook (2007, 2005) (Ref scenario).

Developing countries often suffer already from the adverse impacts of current climatic conditions, in terms of trends (e.g. desertification) and shocks (e.g. floods). Stern (2006) mentions that health and agricultural incomes are most likely to suffer from climate change, and this may go up tremendously with climate change increasing during this century; moreover, severe climate change could induce mass migration and lead to conflict. Disasters in low-income countries may lead to losses at some 5% of GDP. Unchecked, climate change will become a major obstacle to poverty reduction (Banuri and Opschoor 2007). IEA estimates that realizing the 20 target would require energy-related emissions to be cut to around 23 Gt in 2030, which is 19 Gt less than in the reference scenario. Most of this emissions reduction would have to take place in the industrialized countries,2 but it is inevitable that developing 2

The UN Framework Convention on Climate Change (1992), following the Rio Declaration (1992) in which principles for sustainable development were laid down, recognizes as the key principle the so-called ”Common but differentiated responsibilities and capabilities” of developed and developing countries to deal with environmental issues in general and with climate change in particular. Under this, developed countries have a prime responsibility in the area of mitigation and to assist developing countries (financially, technologically, etc) in their responses to climate change (that will largely be in the area of adaptation). The Bali Action Plan (UNFCCC 2007d) includes consideration of mitigation actions by developing country Parties (art.1.b-ii) and emphasizes the need for international cooperation to support adaptation (art. 1.c-i).

16

Opschoor

countries, especially countries like China, India, and South Africa, would also need to contribute. According to recent analysis, the South’s cumulative carbon emissions are already large enough to jeopardize both climate stability and the South’s economic growth (Wheeler and Ummel 2007). 2.3 A case study: Vietnam To make the above more concrete, let us look at effects of climate change on the society and economy of one of the key countries in the RECREATEproject: Vietnam. The World Bank lists Vietnam as one of the five countries likely to be most adversely affected by rising seas (along with Bangladesh, the Bahamas, Egypt, Surinam). The following information (retrieved through vietnamnews.vnagency.com.vn) illustrates this: • In Vietnam, average sea levels would increase by 35cm by 2050, 50cm by 2070 and possibly by 1 m in 2100 (in the absence of major global mitigation efforts and catastrophic effects). About 70 per cent of Vietnam’s population live in disaster-prone zones; IPCC (2007: 59) expects that a 1 m sea level rise (SLR) in Vietnam would lead to flooding of up to 25,000 km2 of delta land; the country may lose 12% of its land due to rising sea levels. Vietnam is facing losses likely to run as high as US$17 billion per year in the event of a 1m SLR. Vietnam might then lose 70-80 per cent of the Cuu Long (Mekong) Delta, leading to the loss of 12-15 million tons of rice a year, with 20 million people made homeless by 2050. A 1 m rise might cut 10 per cent off the country’s GDP. • Mangrove forests are at risk due to rising sea levels, pollution, and farming (as well as other) activities. Since 1943, the area of mangrove forests had been cut in half by 2006. In good conditions they help reduce the devastation of tsunamis with 50-70 percent. Flood-controlling dykes built from concrete may be at risk of being breached during typhoons, but where salt-marsh forests are in good condition, soil dykes can provide good protection. A programme to recover mangrove forests throughout Vietnam will cost the country as much as VND1.9 trillion (US$120 million) by 2015. • The occurrence of extreme weather events like floods and heat waves as well as tropical storms and typhoons will increase. Storms, floods and drought will be more severe and thus their impact on life, production and development will be worse. Damage inflicted by 2006 typhoons cost VND18.7 trillion (US$1.1 billion) in damages. Natural disasters in 2007 killed 435 people and cost the nation VND11.6 trillion ($725 million), or


17

1 per cent of the GDP. Drought has dried up many water resources, and low levels in the Hong (Red) River – the biggest river in northern Vietnam – have caused severe water shortages for farmers over the last five years. • Moreover, there can be severe impacts upland. Drought has heavily affected Vietnam’s agriculture and forestry production during the last 10 years, especially in the centre and Tay Nguyen (The Central Highland, with drought having cost VND 1,400 billion (US$87.5 million) in the two years 1997-1998. To mitigate the effects of climate change, carbon emissions need to be reduced and trees planted to soak up carbon dioxide in the atmosphere. The ministry is aiming to increase the proportion of forest coverage from 37 per cent in 2005 to more than 42 per cent by the end of 2010. There will also be greater use of renewable energy sources, such as solar, wind, and hydroelectric power. 3 Economic impacts and costs – a global-level analysis The costs of climate change can be broken down into three categories: the costs of impacts or damage, the costs of adaptation to reduce damage costs and the costs of mitigation (World Bank 2006). These costs are interrelated, in that the sooner global warming is arrested and, if possible, reversed, the lower the damage costs or the eventual costs of adaptation should be. 3.1 Damage costs Damage costs are losses due to direct and indirect impacts of climate change. Direct impacts of climate change include the loss of life, livelihoods, assets, agricultural output, infrastructure, and, more generally, the depression of sources of economic growth and development. These could nullify the pro-poor potential of macroeconomic policies, trade and private sector investments (Richards 2003). More indirectly, climate change is predicted to alter the sectoral origins of growth, including the ability of the poor to engage in non-farm sectors, as well as to increase inequality. Even with 2 - 3°C of warming, poor countries would suffer significant costs. For example, Tol estimates a cost to Africa of 4.1% of GDP for 2.5°C warming, very close to Nordhaus and Boyer’s estimate of 3.9%. For warming beyond 2 - 3° C, the models disagree on the size of this cost, ranging from a very small fraction of global GDP to 10% or more (see UNFCCC 2007a). Obviously, GDP reductions do not capture the full

18

Opschoor

range (and extent) of societal costs. Stern (2006) has attempted to estimate welfare losses (rather than GDP reductions) from a very long-term perspective (200 years), avoiding the suppression effect on future costs/benefits by discounting, and applying equity-based weighting of costs and benefits. On that basis, damage due to BAU emissions give rise to average reductions in welfare equivalent to per capita consumption at 5 – 20% (Stern 2006: 164).3 3.2 Mitigation costs Damages could be reduced by mitigation. The estimation of mitigation costs is, however, surrounded by uncertainties and unknowns and, among other things, is highly dependent on the levels and pace of carbon concentration stabilization as well as on where emissions are to be cut. Different projections come up with widely varying estimates for the likely costs by 2100: from $1,800 trillion to less than $400 trillion (1990 US dollars) for stabilization at the 20 C line (IPCC 2001). Stern estimates that stabilizing at ~ 30 C would cost the equivalent to about 1% of global GDP (-1% to +3.5%) by 2050 (Stern 2006: 233, 234). Their meta-analysis of macroeconomic model estimates generates an estimate of stabilization costs (~ 30 C) of 1% of GDP (+/- 3%) over the next 50-100 years (Stern 2006: 240ff). These estimates suggest that mitigation is globally affordable (i.e., the figures are small in comparison to global product or income levels, though in absolute figures they would amount to hundreds of billions of dollars annually), at least to the 30 level, and that - when compared with the damage estimates in case of absence of action (see above) - mitigation would pass any broad-minded cost-benefit test.4 Table 2 summarizes some estimates of the macroeconomic consequences of climate change. The upshot is that estimates of mitigation costs are far less than the 5-20% damage costs (at 30 C with no action taken), but may grow significantly if the world is to adhere to the 20 target.

3

4

Critics have argued that the choice of discount rate was arbitrary and did not match real and prevalent time preference. More fundamental critiques, e.g. Baer (2007) and Spash (2007), state that cost-benefit analysis should not be applied on technical grounds (and there is some methodological truth in that) or is theoretically inappropriate, as it is derived from a totally flawed ethical paradigm. The figures reflect averages over very long periods of time, arrived at under rather special assumptions. Also, these costs will not be distributed evenly.

19


Table 2: Economic repercussions of CC-mitigation in 2030 and 2050, associated with capping temperature rise due to CC at 20 or 30 C Year

2030

2050

Atmospheric GWP ($PPP) warming foregone (%) by year indicated (IPCC) ~20 C < 3%

< . 12%

~30 C

< . 10%

~20 C ~30 C

21002200

~30 C

. 6% ( .2% – 2.5%) < 5. 5% 1. 3% ( χ² 0.00 Note: Standard errors in parentheses. * (**) [***] Significant at the 5% (1%) [0.1%] levels of error probability. Source: Own data (2007).

(20.184) ***

(0.031) (11.040) (0.661) (4.340) (23.991) (21.238) (21.040) (20.237) (24.186) (20.811) (26.575) (20.795) (20.904)

***

*** *** * ** **

The dependent variable in the OLS model is the reported average land rental price, which is explained by the estimated marginal value product (mvp). Other variables which could influence the rental price are whether the village keeps reserve land (reserve), the percentage of HHs renting in a village (rent_within) and the average plot size (psize). Reserve land seems to imply land of low quality, while the percentage of HHs renting could be

Characteristics of the Rural Land Rental Market in China

63

an indicator of high demand for land and, thus, high rental fees. Larger plots are expected to increase the value of land. The results in Table 3 confirm that rental prices are mainly based on the MVP and locationspecific factors. The variables rent and psize have the expected signs, but are not significant. The absolute value of the allocative inefficiency score (AI) is on average 0.78. Concerning the factors which determine the AI, as defined by Vranken and Mathijs (2001), the available data is inconclusive. The score is significantly lower (at the 5% level) in villages with a very active land rental market (≥10% of HHs), which is in line with the expectation that a more active market leads to more competitive prices. The significance disappears, however, when including township dummy variables and other control variables. This indicates that the gap between the productive value of land and rental prices is influenced by other factors which have not been covered in the survey. Furthermore, the AI is very sensitive to the estimated partial production elasticity of land, which may be faulty, since few variables could be used in the village-level production function. 6 Conclusions This study has identified a number of reasons why the rural land rental market has not proliferated well under certain conditions in China. The low extent of market-mediated land transfers seems to call for institutions to encourage and facilitate such transfers by reducing transaction costs for farm households. As one suggestion in this direction, it has been proposed to establish a “land bank” as a national policy bank with the specific purpose of encouraging land transfers and, thereby, finally ensuring “farmers’ enthusiasm for planting grain crops” (China Daily 2005). In such a land bank, farmers willing to rent out their land could deposit their land use rights at a high interest rate, while those willing to expand production could obtain land use rights at a low lending rate. Such a framework could decrease the wedge between supply and demand prices for land use rights and build up farmers’ trust in the land rental market. Zhang et al. (2004) found land trusts, a variant of the proposed land bank, to be in fact active in Zhejiang Province. Such land trusts are described as local institutions acting as brokers between lessors and lessees, disseminating information on supply and demand for land, helping to negotiate and write contracts and mediating disputes. An institutionalized land bank does not exist in Quzhou, but village leaders confirmed that households that wish to rent-in land from outside their village often depend on some mediating person (who could be a friend or relative) who has connections to another household willing to rent out. Apart from the administrative barriers, lack of market information can

64

Piotrowski and Böber

therefore be seen as one reason for the low level of inter-village land transfers. The low extent of exchange of production factors between villages is also reflected by the strong autocorrelation of many of the agroeconomic variables assessed in this study – regions with high economic development exist side by side with those of lower development. Nonetheless, further analysis showed that the MVP is a valid explanatory variable for land rental prices. For this analysis, a Heckman two-step regression model, which accounts for the selection bias caused by those villages with a non-existing land market, was estimated with the average land rental price as the dependent variable. In essence, the results show that, despite distortions due to administrative restrictions, land rental prices in the study area are based on the value of land and can thus be called market prices. However, due to the fact that transfers are mainly confined to members of the same village collective, and often to members of the same family or clan, the land rental market currently does not live up to its potential. Acknowledgements This research was conducted within the International Research Training Group (IRTG) “Sustainable Resource Use in North China”, funded by the German Research Foundation (DFG) and the Ministry of Education of the People’s Republic of China. Their support is gratefully acknowledged.

References Anselin, L. (1988). Spatial Econometrics: Methods and Models. Dordrecht, The Netherlands: Kluwer. Anselin, L., Bera, A.K., Florax, R., Yoon, M.J. (1996). Simple Diagnostic Tests for Spatial Dependence. Regional Science and Urban Economics 26(1): 77-104. Bareth, G. (2003). Möglichkeiten und Grenzen der regionalen agrarumweltrelevanten Modellierung unter Nutzung von GIS in China am Beispiel der Nordchinesischen Tiefebene. Habilitation (postdoctoral lecture qualification). Stuttgart, Germany: University of Hohenheim. Bowlus, A.J., Sicular, T. (2003). Moving Toward Markets? Labor Allocation in Rural China. Journal of Development Economics 71(2): 561-583. Chen, J., Yu, Z., Ouyang, J., van Mensvoort, M.E.F. (2006). Factors Affecting Soil Quality Changes in the North China Plain: A Case Study of Quzhou County. Agricultural Systems 91(3): 171-188. Chen, K., Brown, C. (2001). Addressing Shortcomings in the HRS: Empirical Analysis of the Two-Farmland System in Shandong Province. China Economic Review 12(4): 280-292.

Characteristics of the Rural Land Rental Market in China

65

Chen, Z., Huffman, W.E. (2006). County-level Agricultural Production Efficiency in China: A Spatial Analysis. In: China's Agricultural Development: Challenges and Prospects. Aldershot, UK: Ashgate. China Daily (2005). Ensuring Grain Security by Setting Up “Land Bank”, May 17, 2005. Cliff, A. D., Ord, J.K. (1973). Spatial Autocorrelation. London, UK: Pion. Fleisher, B.M., Liu, Y. (1992). Economies of Scale, Plot Size, Human Capital, and Productivity in Chinese Agriculture. Quarterly Review of Economics and Finance 32(3): 112-123. Hu, K., Huang, Y., Li, H., Li, B., Chen, D., White, R.E. (2005). Spatial Variability of Shallow Groundwater Level, Electrical Conductivity and Nitrate Concentration, and Risk Assessment of Nitrate Contamination in North China Plain. Environment International 31(6): 896 - 903. Hung, P. V., MacAulay, G., Marsh, S. P. (2007). The Economics of Land Fragmentation in the North of Vietnam. The Australian Journal of Agricultural and Resource Economics 51(2): 195-211. Jacoby, H. G., Li, G., Rozelle, S. (2002). Hazards of Expropriation: Tenure Insecurity and Investment in rural China. American Economic Review 92(5): 1420-1447. Keil, A. (2004). The Socio-economic Impact of ENSO-related Drought on Farm Households in Central Sulawesi, Indonesia. Göttingen, Germany: University of Goettingen. Kong, X., Zhang, F., Xu, Y., Qi, W. (2003). Arable Land Change and its Driving Forces in Intensive Agricultural Region: The Case of Quzhou County in Hebai Province (sic!). Resources Science 25(3): 57-63. Kuiper, M., van Tongeren, F. (2005). Growing Together or Growing Apart? A Village Level Study of the Impact of the Doha Round on Rural China. Policy Research Working Paper, The World Bank, Washington D.C., USA. Kung, J.K. (2000). Common Property Rights and Land Reallocations in Rural China: Evidence from a Village Survey. World Development 28(4): 701-719. Kung, J. K. (1995). Equal Entitlement Versus Tenure Security under a Regime of Collective property rights: Peasants' Preference for Institutions in Post-reform Chinese Agriculture. Journal of Comparative Economics 21(1): 82-111. Liu, C., X. Yao, Lavely, W. (1996). China Administrative Regions GIS Data: 1:1M, County Level, 1990. Published and disseminated by the Center for International Earth Science Information Network (CIESIN). ftp://ftpserver.ciesin.org/pub/data/China/adm_bnd/CT SAR90.bnd90/. November 28, 2007. Lohmar, B., Somwaru, A., Wiebe, K. (2002). The Ongoing Reform of Land Tenure Policies in China. Agricultural Outlook 294: 15-18. Lohmar, B., Zhang, Z., Somwaru, A. (2001). Land Rental Market Development and Agricultural Production in China. Annual Meeting of the American Agricultural Economics Association, August 5-8, Chicago, IL, USA. Minot, N., Baulch, B., Epprecht, M. (2006). Poverty and Inequality in Vietnam: Spatial Patterns and Geographic Determinants. Research Report, International Food Policy Research Institute, Washington, D.C., USA.

66

Piotrowski and Böber

Moran, P. (1948). The Interpretation of Statistical Maps. Journal of the Royal Statistical Society, Series B 10: 243-251. Pisati, M. (2001). sg162: Tools for Spatial Data Analysis. Stata Technical Bulletin 60: 21-37. Quzhou County Statistical Bureau (Various years). Quzhou Statistical Yearbook. Quzhou, Hebei, China. Vranken, L., Mathijs, E. (2001). The Allocative Efficiency of Land Rental Markets in Transition Agriculture. American Agricultural Economics Association Annual Meeting, August 5-8, 2001, Chicago, IL, USA. Wan, G. H., Cheng, E. (2001). Effects of Land Fragmentation and Returns to Scale in the Chinese Farming Sector. Applied Economics 33(2): 183 -194. Wang, W. (2005). Land Use Rights: Legal Perspectives and Pitfalls for Land Reform. In: Ho, P. (ed.), Developmental Dilemmas: Land Reform and Institutional Change in China. London, UK: Routledge.

Chapter 5 Impact of Soil and Water Conservation Investments on Agricultural Development in Western China Nico Heerink1,2, Rui Li3, Shuyi Feng2, Kaiyu Lu4, and Xiaobin Bao5 Abstract: Measures to control soil erosion are an important component of China’s Western region development program. This study uses a unique set of annual provincial data on soil and water conservation (SWC) investments during the period 1989-2005 to obtain empirical estimates of the impact of such investments on the extent and severity of erosion as well as agricultural gross domestic product (GDP) in Western China. We find that SWC investments made by local governments have a significant negative impact on the area affected by erosion, whereas SWC investments made by farm households have a significant negative effect on the severity of erosion during this period. In recent years, however, both local government investments and private investments have significantly affected reductions in the severity of erosion, but not in the share of land affected by erosion in Western China. In its turn, severity of erosion has a significant negative impact on agricultural GDP in North-western China (though not in South-western China). Estimation of the impact of the extent of erosion on agricultural GDP provides mixed results. Based on these results we derive that one RMB invested in SWC by local governments in North-western China increased agricultural GDP in 2002 by 1.16-1.38 RMB, while SWC investments made by farm households in the same region increased agricultural GDP by only 0.20-0.24 RMB. Keywords: Soil and water conservation, Erosion, Public investment, Agricultural development, West China

1 Introduction China’s Western region covers 71.5 percent of mainland China’s land area. A large proportion of China’s mountains and desert land is located within the Western region. It houses only 28 percent of China’s population, but 1 2 3 4 5

Development Economics Group, Wageningen University, Netherlands China Center for Land Policy Research, Nanjing Agricultural University, China. Institute of Soil and Water Conservation, Chinese Academy of Sciences, China Institute of Agricultural Economics and Development, Chinese Academy of Agricultural Sciences, China. Rural Development Institute, Chinese Academy of Social Sciences, China


68

Heerink, Li, Feng, Lu, and Bao

accounts for more than 70% of the total number of poor in China. The region has benefited less from the economic reforms that have been implemented since the end of the 1970s than the coastal and central regions have. As a result, per capita GDP in Western China was only 55.4 percent of per capita GDP in the rest of China in 2007 (NBS 2008). In 2000, the Chinese government initiated a policy to boost the development of Western China. This program, also known as the ‘Open Up the West Program’ aims at promoting development through infrastructure development, human resource development, and protection of natural resources and the environment (Lai 2002; Goodman 2004; Lu and Neilson 2004). Advancing the development of the Western region was reiterated as a major policy goal in China’s 11th five-year plan, covering the period 2006-2010 (NDRC 2006). Besides promoting economic development and reducing regional inequality, Chinese leaders hope that the Western China development program can play a vital role in restoring China’s ecological balance, especially in controlling soil erosion and desertification (Lai 2002). Despite its relatively vast size, agricultural land in the Western region is not very fertile. Soil erosion due to grazing activities, rapid deforestation in the recent past, and other improper land use activities is a major problem. Eroded areas in the Western region were estimated to make up 83.3 percent of China’s total eroded land in 2001-02. The strength and frequency of sandstorms have increased significantly in recent times. Water resources in the Northwest are in scarce supply and depend highly on climate variability. The lack of fertile soil combined with the uncertainties of precipitation makes it a difficult region for widespread agricultural production activities (Glantz et al. 2001). Investments in soil and water conservation (SWC) in Western China and elsewhere can have important socio-economic benefits. They may maintain or improve the productivity of agricultural land in areas where conservation measures (such as land terracing, check dams, afforestation or grass planting) are undertaken. They may also improve agricultural and nonagricultural productivity and environmental quality elsewhere, by reducing flood damages, sediments in waterways and reservoirs, water pollution, and so on. And they may have important indirect effects on rural gross domestic product (GDP), poverty alleviation, employment, migration, and so on. But SWC may also be costly, either directly through its investment requirements or (for techniques like diversion ditches or terracing) indirectly through foregone production (Lutz et al. 1994). A crucial question is, therefore, whether the socio-economic benefits of SWC investments outweigh their costs.

Impact of Soil and Water Conservation Investments

69

The purpose of this study is to examine the impact of SWC investments aimed at controlling soil erosion on agricultural GDP growth in Western China. To this end, we use a unique set of annual provincial data on SWC investments during the period 1989 - 2005 that was made available to us by the Chinese Ministry of Water Resources. The methodology that we use builds upon the methodology developed by Fan et al. (2002, 2004) for evaluating the benefits of various types of public investments, adjusted in such a way that it could be applied to examine the economic benefits of SWC investments by Heerink et al. (2009). The study is structured as follows. In section 2 we present recent trends in erosion and SWC investments in Western China, based on the annual provincial data on SWC investments that were made available to us and provincial soil erosion data that have recently become available. In section 3 we present the empirical results of estimating erosion functions from the provincial data set, using available SWC investment data for the periods 1987-2002 and 1997-2002, respectively. The results provide insights into the impact of SWC investments on the areas affected by erosion and the severity of erosion within those areas. In the next step, both the land area affected by erosion and the severity of erosion are used as explanatory variables in an agricultural production function. Section 4 presents the estimation results of the production function. The regression results for the erosion functions and the agricultural production function can be used to calculate marginal returns to agricultural GDP of investments in SWC and compare these with the returns of other types of public investments. In section 5 we derive such marginal returns from our estimation results and present some concluding remarks. 2 Trends in erosion and SWC investments Using results from the national soil erosion surveys, we have been able to investigate trends in soil erosion area in Western China since the mid1980s. The national soil erosion surveys have been carried out three times until now: 1985-86, 1995-96 and 2001-02. In these surveys, the size and severity of erosion is estimated from satellite data supplemented with (land-use) maps. Data from the 2001-02 survey have recently become available, making it possible to compare the trend from the mid-1990s to the beginning of the present century with the trend from the mid-1980s to the mid-1990s. Table 1 shows the shares of the land affected by erosion for all provinces in Western China. The provinces in North-western China are listed in the upper part of the table, while the provinces in South-western China are in the lower part. The averages for North-western China, South-western China, and Western China as a whole are shown in the last two columns.

70


The table shows that erosion is a much larger problem in North-western China, affecting 56-58 percent of the land, than in South-western China, where it affects around 21 percent of the land. Moreover, the area affected by erosion in North-western China (and as a result also in Western China as a whole) has increased over time, while in South-western China there was a small decline between the mid-90s and 2001-02. Erosion affects more than 50 percent of the land in all provinces in North-western China, except Qinghai. It affects more than one-third of the land in three provinces in South-western China, namely Guizhou, Yunnan and Sichuan. Table1: Areas affected by erosion in Western China (% of total land) 1985-86 1995-96 2001-02

Inner Mongolia 69.5 65.1 67.3 Guangxi

Shaanxi

Ningxia

Gansu

Qinghai

Xinjiang

64.2 62.6 61.5

75.0 71.2 71.6

57.1 64.6 64.0

25.5 25.4 28.4

Guizhou

Yunnan

Sichuan

Tibet

57.1 63.2 62.9 SW China2 20.6 21.3 21.0

NW China1 55.7 57.4 58.3 West China3 42.4 43.7 44.1

1985-86 4.7 43.5 37.5 32.6 9.4 1995-96 4.4 41.6 37.2 36.8 9.4 2001-02 4.4 41.4 36.2 34.2 10.3 Notes: 1 North-western (NW) China consists of Inner Mongolia, Shaanxi, Ningxia, Gansu, Qinghai and Xinjiang. 2 South-western (SW) China consists of Guangxi, Guizhou, Yunnan, Sichuan (which includes Chongqing) and Tibet. 3 Western China includes North-western and South-western China.

Table 2: Areas affected by water erosion in Western China (% of total land) 1985-86 1995-96 2001-02


Shaanxi

Ningxia

Gansu

Qinghai

Xinjiang

58.5 57.4 56.2

44.2 40.4 42.1

25.8 29.5 28.9

5.6 7.4 7.4

Guizhou

Yunnan

Sichuan

Tibet

37.5 37.2 36.2

32.6 35.8 33.1

5.2 5.2 5.8

6.8 7.0 7.2 SW China2 18.6 19.2 18.7

1985-86 4.7 43.5 1995-96 4.4 41.6 2001-02 4.4 41.4 Notes: 1, 2, 3 See notes below Table 1.


Information on the land affected by the two major types of erosion, water erosion and wind erosion is provided in Tables 2 and 3. Water erosion is a bigger problem in South-western China, where it affects around 19 percent of the land, than in North-western China, where 13-14 percent of the land is

71


affected. But there are large differences between the individual provinces within each region. In North-western China, water erosion is a major problem in Shaanxi and Ningxia, with more than 40 percent of the land affected and, to a lesser extent, in Gansu. In South-western China, it is a major problem in Guizhou, Yunnan and Sichuan, where more than 33 percent of the land is affected by water erosion. In both regions, water erosion worsened between 1985-86 and 1995-96, but has improved since then. The latter finding suggests that measures taken to control water erosion seem to have been successful in reversing the negative trend in Western China since the mid-90s. Table 3: Areas affected by wind erosion in Western China (% of total land) 1985-86 1995-96 2001-02


Shaanxi

Ningxia

Gansu

Qinghai

Xinjiang

5.6 5.2 5.2

30.8 30.8 29.6

31.2 35.1 35.2

19.9 18.0 21.0

Guizhou

Yunnan

Sichuan

Tibet

0.0 0.0 0.0

0.0 1.1 1.1

4.2 4.2 4.5

50.3 56.1 55.7 SW China2 2.0 2.2 2.3



Wind erosion is almost entirely a problem in North-western China only. More than 50 percent of the land is affected by wind erosion in Inner Mongolia and Xinjiang, while 20-35 percent of the land is affected in Gansu, Ningxia and Qinghai. Only in Shaanxi it is a relatively minor problem, as compared to the water erosion problem in that province. The averages for North-western (and West) China as a whole show that the measures taken to control wind erosion have not (yet) been successful. In fact, average annual increases in the size of the land affected by wind erosion are slightly higher for the 6-year period between 1995-96 and 2001-02 than for the preceding 10-year period between 1985-86 and 199596. A closer look at the data for the individual provinces reveals that the acceleration of wind erosion since the mid-90s has been almost entirely caused by the worsening of the problem in Inner Mongolia and Qinghai, two provinces with large shares of grassland. The land affected by erosion has been classified according to the severity of erosion (light / medium / intense / more intense / severe6) in all three soil 6

See Li et al. (2006) for the precise definitions of light, medium, intense, more intense and severe erosion.

72


erosion surveys. We used this information to derive an index of the severity of erosion by multiplying the land in each land class by the following weights: • light erosion: 0.25 • middle erosion: 0.5 • intense erosion: 1 • more intense erosion: 2 • severe erosion: 4 Dividing the resulting value by the total land area gives the value of the severity index (SI). The resulting values are presented in Tables 4-6. No results are presented for wind erosion and total erosion in Shaanxi and Gansu, because the results show implausibly large shifts in the SI between 1985-86 and 1995-96 for these two provinces.7 Table 4: Values of erosion severity index (SI) in Western China 1985-86 1995-96 2001-02

1985-86 1995-96 2001-02 Notes:

Inner Mongolia 0.97 1.19 1.16

Shaanxi4

Ningxia

Gansu4

0.65 0.61 0.59

Qinghai

Xinjiang

0.67 0.87 0.88

1.13 1.18 1.18 SW China2 0.54 0.50 0.47

Guangxi

Guizhou

Yunnan

Sichuan

Tibet

0.59 0.35 0.30

0.54 0.44 0.44

0.39 0.40 0.39

0.82 0.60 0.57

0.27 0.48 0.44

NW China1,5 1.01 1.14 1.13 West China3,5 0.91 1.00 0.99

1, 2, 3 4 5

See notes below Table 1. Data for Shaanxi and Gansu not presented because of measurement problems. Excluding Shaanxi and Gansu.

The results presented for total erosion in Table 4 show that the SI increased from the mid-80s to the mid-90s, but has slightly decreased since then. Inner Mongolia and Xinjiang have values that are higher than the average for the whole of Western China, implying that the more severe types of erosion dominate in these two provinces. All the other provinces have SI values that are below the average of Western China.

7

The values of SI for wind erosion increased from 0.56 to 2.04 in Gansu and from 0.88 to 1.86 in Shaanxi during this period. As a consequence, the value of SI for total erosion increased from 0.66 to 1.45 in Gansu. Because wind erosion is a relatively minor problem in Shaanxi, the value of SI for total erosion was not much affected in that province.

73


Table 5: Values of severity index (SI) for water erosion in Western China 1985-86 1995-96 2001-02


Shaanxi

Ningxia

Gansu

Qinghai

Xinjiang

1.38 1.05 1.03

0.69 0.47 0.58

0.78 0.74 0.75

0.57 0.50 0.52

Guizhou

Yunnan

Sichuan

Tibet

0.39 0.40 0.39

0.82 0.60 0.58

0.26 0.30 0.29

0.25 0.32 0.32 SW China2 0.57 0.47 0.46



The value of the SI for water erosion declined considerably between the mid-80s and mid-90s in North-western China (from 0.70 to 0.61) as well as South-western China (from 0.57 to 0.47), as can be seen from Table 5. Even though soil and water conservation programs were not successful in reversing the trend in the total land area affected by water erosion (see Table2), they do seem to have been successful in reducing the most severe types of water erosion in Western China during that decade. The only major exceptions are Xinjiang and Tibet, where the value of the SI increased between the mid-80s and mid-90s. No major changes between 1995-96 and 2001-02 could be observed in the values of the SI for water erosion in North-western and South-western China, as well as in most provinces within these regions, with the increase from 0.47 to 0.58 in Ningxia being a major exception. Table 6: Values of severity index (SI) for wind erosion in Western China 1985-86 1995-96 2001-02

Inner Mongolia 1.09 1.38 1.33

Shaanxi4

Guangxi

Guizhou

Ningxia

Gansu4

0.58 0.80 0.60 Yunnan

1985-86 n.a. n.a. n.a. 1995-96 0.31 n.a. n.a. 2001-02 n.a. n.a. n.a. Notes: 1, 2, 3, 4, 5 See notes below Tables 1 and 4.

Qinghai

Xinjiang

0.70 1.03 1.00

1.25 1.28 1.29 SW China2 0.64 0.55 0.54

Sichuan

Tibet

n.a. 0.40 0.40

0.28 0.72 0.62

NW China1,5 1.13 1.29 1.27 West China3,5 1.11 1.27 1.25

The value of the SI for wind erosion is considerably higher than the SI for water erosion in North-western China, as is evident from Table 6. In other

74


words, the areas affected by wind erosion suffer relatively more from it than the areas affected by water erosion. Among the provinces that have reliable data on the severity of wind erosion, the SI is highest for Inner Mongolia and Xinjiang. The SI for wind erosion increased considerably between the mid-80s and mid 90s in North-western China, but has hardly or not increased since then. Annual provincial data on SWC investments during the period 19892005 were collected for this study by the Institute of Soil and Water Conservation (ISWC) of the Chinese Academy of Sciences (CAS) and the Ministry of Water Resources (MWR) in close collaboration with the Administrative Bureau of the Upper and Middle Reaches of the Yellow River. These data are sub-divided into investments made by the central government (i.e. the Ministry of Water Resources), local government (province, prefecture and county) and farmers8, respectively. There are several gaps in these data, however, which makes it difficult to sketch an overview picture of the developments in SWC investments for Western China as a whole. Yet, the available data do allow us to make some limited analyses. Out of the 11 provinces in Western China, full time-series on central and local government investments in SWC for the whole period 1989-2005 and data on private SWC investments for the period 2001-05 are available for 5 provinces, namely Shaanxi, Gansu, Ningxia, Sichuan and Yunnan. The resulting trends (in constant prices of the year 2000)9 are shown in Figure 1. The figure shows that central government investments greatly exceeded local government investments in most years, except 1996-97. There was a rapid increase in central government investments and, to a lesser extent, also in local government investments in the years 1998 and 1999. Since 2002, central government investments and private investments show large declines. Their levels in 2005 were only 64 and 55 percent of their levels in 2002, respectively. Local government investments, on the other hand, remained more or less stable during that period.

8 9

Data on SWC investments by farmers include time input as well as financial outlays for such investments. The implicit GDP deflator for each province is used to convert the investment data from current prices into constant prices.

75

Impact of Soil and Water Conservation Investments 1000 900 800

Million RMB

700 600

Central Gov., 5 Provinces

500

Local Gov., 5 Provinces

400 300

Private, 5 Provinces

200 100 2005

2003

2001

1999

1997

1995

1993

1991

1989

0

Year

Figure 1: Soil and water conservation investments, 1989-2005 (5 provinces) Some provinces have information for the whole period 1989-2005 about only one or two sources of investment: Information on central government investments in SWC is available for Qinghai and Guizhou, information on local government SWC investments is available for Guangxi and Guizhou, and information on private SWC investments is available for Inner Mongolia and Qinghai. Adding this information to the data used for drawing Figure 1 results in Figure 2. Notice that it is no longer possible to compare the levels of the three different types of investments, because they are obtained for different sets of provinces now. But because the data on each type of investment are calculated for a consistent set of provinces, it is possible to examine the trends in each type of investment individually. 1200

Million RMB

1000 800

Central Gov., 7 Provinces

600

Local Gov., 7 Provinces

400

Private, 7 Provinces 200

2005

2003

2001

1999

1997

1995

1993

1991

1989

0

Year

Figure 2: Soil and water conservation investments, 1989-2005 (7 provinces)

76


The trends shown in the figure closely resemble the ones of Figure 1. The decline in private investments during the period 2002-05, however, is more pronounced (with the level in 2005 being only 44 percent of the 2002 level), while the decline in central government investments is a bit smaller (with the 2005 level being 70 percent of the 2002 level). 3 Results for erosion functions We use annual provincial data on public and private investments in SWC and data from the three national soil erosion surveys to obtain empirical estimates of the impact of SWC investments on soil erosion. We specify two equations, one for the size of the land affected by erosion and one for the severity of erosion, as both dimensions are likely to affect agricultural production. The specification of the regression equations is as follows: CEROSHit = c10 + c11*log(Cit) + c12*log(Lit) + c13*log(Fit) + c14*DWi*log(Cit) + c15*DWi*log(Lit) + c16*DWi*log(Fit) + c17*CSOWNit + c18*DPt + ε1it

(1)

CEROSIit = c20 + c21*log(Cit) + c22*log(Lit) + c23*log(Fit) + c24*DWi*log(Cit) + c25*DWi*log(Lit) + c26*DWi*log(Fit) + c27*CSOWNit + c28*DPt + ε2it

(2)

where: CEROSHit = Average annual percentage change in the share of the land affected by erosion for province i during period t CEROSIit = Average annual change in the value of the severity index (SI) of erosion for province i during period t Cit = Average annual investments in SWC by the central government (Ministry of Water Resources) in province i during period t (in 10,000 RMB) per square km of eroded land at the start of the period, in constant prices of the year 2000 Lit = Average annual investments in SWC by the local (province, prefecture and county) government in province i during period t (in 10,000 RMB) per square km of eroded land at the start of the period, in constant prices of the year 2000 Fit = Average annual investments in SWC by farm households in province i during period t (in 10,000 RMB) per square km of eroded land at the start of the period, in constant prices of the year 2000


77

DWi = Dummy variable that equals one if province i is located in Western China, and equals zero otherwise. CSOWNit = Average annual percentage change in sown areas to crops in province i during period t DPt = Dummy variable for first period (= 1 for first period; = 0 for second period) ε1it, ε2it = Random disturbance terms with standard properties Soil erosion depends on investments in SWC made by the (central and local) government and farm households, human-induced factors (such as land clearance, planting of erosive crops or use of inappropriate technologies), biophysical factors (such as soil type, vegetative cover, share of hilly and mountainous land, average slope of the land), and climatic factors (such as quantity and intensity of rainfall). To avoid the need for measuring biophysical and climatic factors, for which provincial level data are generally not available, we examine the change in land affected by erosion over a specific period. We assume that biophysical and climatic factors remain constant over time and, therefore, do not affect changes in the areas affected by erosion. We consequently model the change in land affected by erosion as a function of SWC investments and human-induced factors during that period only. SWC investments are expected to have a negative effect on the share of land affected by erosion. Likewise, it may be expected that the severity of erosion is reduced when more investments in SWC are made. The three national soil erosion surveys give us information on changes that took place during two periods, the period 1985-86/1995-96 and the period 1995-96/2001-02, for all provinces. Unfortunately, we do not have complete information on the public and private investments in SWC for these two periods for all provinces in Western China. Yearly information on SWC investments by the central government, local government as well as farm households is available for all years from 1989 to 2002 for only 3 provinces in Western China: Shaanxi, Ningxia and Sichuan. For the same three provinces, plus Inner Mongolia, complete annual information is available for all years between the second and the third survey. To obtain empirical estimates of the impact of SWC investments on the level and severity of soil erosion, we therefore decided to run regression for all provinces in mainland China for which SWC investment data are available for the whole 1989-2002 period and derive estimates of the impact for the provinces in Western China by including a dummy variable (DW) in the regressions.

78


We take the period 1987-1996 as the first period and the period 19972002 as the second period. Because the length of the two periods is different, we examine average annual changes in soil erosion instead of changes over the whole period. For the first period, we only have investment data for the period 1989-1996 for the observations in our sample. Average annual investment data for that period are assumed to be representative for average annual investments during the whole period. Furthermore, the method of data collection on SWC investments has changed considerably since 2001, which means that we cannot compare the data before 2001 with those collected since then. We therefore calculate average annual SWC investment data for the years 1997-2000 and assume that these data are representative for the whole period 1997-2002. SWC investments from different sources are expected to have different types of impacts. For example, the impact of local government investments may differ from those of central government investments, because local governments tend to invest in relatively small-scale projects that focus on local circumstances and the local needs for SWC. On the other hand, central government investments tend to concentrate on large-scale projects that are intended to benefit the whole country and usually take a longer period to have an impact. Moreover, central government investments tend to pay more attention to poorer and relatively sparsely populated areas (see Li et al. 2006). It may therefore be relevant to examine the impact of local and central government investments separately, even though local governments obtain a large share of their funds for making SWC investments from the central government. Compared to central and local governments, farm households are probably most familiar with the local circumstances and the SWC investment needs, but they are less likely to take off-site effects of erosion into account in making their SWC investment decisions. It will thus be interesting to compare the impact of farm household investments on changes in the extent and severity of soil erosion with those of public investments. On the other hand, public investments in SWC often require a certain amount of input from farm households, thereby reducing the independence of farm households in decision making. Cross-products of each of three types of SWC investment and the dummy variable for Western China are added to the equation to test whether the impact of SWC investments differs significantly for the provinces in Western China in our sample. The impact of human-induced factors on soil erosion is proxied by change in area sown with crops.10 This variable is intended to combine the 10

Alternatively, we also used the average annual growth rate in agricultural GDP instead of the change in area sown with crops to represent the impact of humaninduced factors. The conclusions drawn from both analyses are similar.


79

impact of extensification (land clearance) as well as agricultural intensification on soil erosion. The impact of extensification may be distinguished from that of intensification by using changes in the size of the cultivated land area as an explanatory variable. Unfortunately, problems with the reliability of cultivated land data preclude this option.11 Finally, a dummy variable is added to the equation to control for possible unobserved systematic differences between the two periods. We expect that SWC investments will have a negative impact on the share of land affected by erosion (CEROSH) and on the severity of erosion (CEROSI). The coefficients for the cross-products of SWC investments and the dummy variable for Western China may have either sign. The land area sown with crops can have both a positive and a negative effect on the extent and severity of erosion, as some land use types reduce erosion while others worsen it. We have 25 observations in total, 11 for the first period and 14 for the second period. The regression results (using ordinary least squares) are summarized in Table 7. 12 Column (1) in Table 7 shows the regression results for the change in the share of land affected by erosion (CEROSH). The results indicate that local government SWC investments have a significant negative effect on the share of land affected by erosion, central government SWC investments have a strongly significant positive effect, and farm household investments do not have a statistically significant effect. These results suggest that local government investments are more efficient in fighting erosion than central government and private investments. The estimated coefficients for the cross-products are not significantly different from zero, which suggests that the impact of SWC investments does not significantly differ between Western China and the rest of China. The significant positive impact of central government investments is surprising. When interpreting this result, it should be taken into account that the estimated coefficient only reflects the impact on erosion within the same period and, therefore, represents short-term effects. Some typical central government SWC investments, such as tree planting or check dams, usually take longer than 3-5 years (the average length of the periods used in our regression analysis) to materialize. Moreover, climatic and biophysical factors may not be constant over time, as we assume in our analysis, but may contribute to an increase in soil erosion in the relatively poor and 11

12

Official data on cultivated land area in China’s provinces show that a major change took place in 1996 (the year when the Agricultural Census was held) and that no changes have taken place since then. Results of the White test for heteroskedasticity show that the null hypothesis of homoskedasticity is not rejected (at a 5 percent testing level) for all four equations reported in the table.

80


sparsely populated areas in which the central government concentrates its SWC investments. No suitable province-level data are available at the moment to examine these possible explanations in more detail. Table 7: Regression results for erosion share and severity of erosion, 1987-2002 Explanatory Variables Central government investment

Erosion share (1) 1.27*** (3.73)

Local government investment

-1.05** (-2.87)

Farm household investment DW * Central government investment DW * Local government investment DW * Farm household investment Area sown with crops

-0.07 (-0.67) -0.70 (-0.59) 0.81 (0.63) -0.03 (-0.09) -0.06 (-0.33) -0.22 (-0.69) 0.48

First Period Adjusted R-squared

(2) 1.31*** (4.81)

Water erosion severity (3) (4) 0.0022 0.0051 (0.60) (1.56)

-1.04*** (-3.14)

0.0063 (1.62)

0.0041 (1.05)

-0.07 (-0.49) -0.91 (-1.00) 1.00 (1.01) 0.00 (0.01) -

-0.0047** (-2.81) -0.0135 (-1.06) 0.0078 (0.57) 0.0062 (1.74) 0.0039 (1.43) -0.0097 (-1.60) 0.30

-0.0042** (-2.40) -0.0186 (-1.72) 0.0161 (1.37) 0.0038 (1.14) -

0.54

0.21

Notes: t-statistics between brackets. *** significant at one percent testing level. ** significant at five percent testing level. * significant at ten percent testing level.

The finding that farm household investments in SWC do not significantly affect the extent of erosion is at variance with the conclusion drawn by Lutz et al. (1994) for Central America and the Caribbean that farmers invest in soil conservation when such investments are profitable. It suggests that the independence of farm households in China in making decisions on SWC investments is limited. Additional research on farm household decision making is needed to test this hypothesis. Human-induced factors, as measured by the area sown with crops, do not have a significant effect on the annual change in the share of land affected by erosion. In other words, the combined impact of extensification and agricultural intensification is not significantly different from zero. Assuming that bringing (in China usually marginal) land into cultivation contributes to soil erosion, this result means that agricultural intensification takes place through planting less-erosive crops or using technologies that


81

contribute to SWC. Finally, the estimated coefficient for the first period dummy variable does not differ significantly from zero. In column (2), regression results are shown when the latter two variables are dropped from the equation. The adjusted R-squared is considerably higher as compared to the results presented in column (1). But the conclusions that emerge with respect to the impact of SWC investments are similar to those drawn for the first equation. Columns (3) and (4) show the regression results for the annual average change in the SI of erosion (CEROSI). Because the available information on severity of erosion shows some very implausible trends for land affected by wind erosion and hence for total erosion in Gansu and Shaanxi (see section 2 above), we present the results for severity of water erosion only.13 Farm household investments in SWC are found to have a significant negative impact on the severity of soil erosion. The impact of central and local government SWC investments, on the other hand, does not differ significantly from zero. The estimated coefficients for the cross-products are again not significantly different from zero, re-affirming that the impact of SWC investments made during the period 1987-2002 did not significantly differ between Western China and the rest of China. Finally, the estimated coefficients for human-induced factors, as measured by the area sown with crops, and the dummy variable for the first period are also not statistically significant. When the latter two variables are dropped from the equation, as was done for the results reported in column (4), farm household investments are still found to have a significant negative effect on the annual change in the SI of water erosion. Information on SWC investments for a larger number of provinces is available over the period 2001-2005. Due to improvements in data collection methods over time, it may be assumed that the quality of this information is relatively better. In the second part of this section we use this recent information to estimate the impact of SWC investments on erosion in order to test the robustness of the results that we obtained so far. Because the third round of the national soil erosion survey gathered information for 2001/02, we can only use the SWC investment data for the years 2001 and 2002 for our analysis. On the other hand, the dependent variables CEROSH and CEROSI measure the annual changes in erosion between the last two rounds of the soil erosion survey, i.e. between 1997 and 2002. We assume that the annual investments in SWC in 2001 and 2002 in each province are strongly related to the levels in 1997-2000 and can, therefore, be used as approximations of the annual investments in the whole period 1997-2002. The models that we estimate for CEROSH and 13

Similar conclusions emerge, however, when severity of total erosion is specified as the dependent variable.

82


CEROSI are similar to the ones for 1989-2002. But, as the models are estimated for only one period, the dummy variable for the first period is left out. Table 8: Regression results for erosion share and severity of erosion, 1997-2002 Explanatory variables

Erosion share

Water erosion severity

(1) 0.40 (1.53)

(2) 0.39 (1.56)

(3) -0.0015 (-0.89)

(4) -0.0014 (-0.83)

-1.67*** (-3.90)

-1.65*** (-4.03)

-0.0001 (-0.05)

-0.0005 (-0.17)

0.26 (1.32)

0.25 (1.33)

-0.0026* (-2.07)

-0.0025* (-2.03)

DW * Central government investment

-1.07 (-1.02)

-1.32 (-1.70)

0.0023 (0.35)

0.0055 (1.10)

DW * Local government investment

1.65** (2.69)

1.70** (2.97)

-0.0083* (-2.14)

-0.0090** (-2.44)

DW * Farm household investment

-0.40 (-0.48)

-0.18 (-0.32)

0.0051 (0.98)

0.0024 (0.65)

-0.09 (-0.37) Adjusted R-squared 0.46 Notes: t-statistics between brackets. *** significant at one percent testing level. ** significant at five percent testing level. * significant at ten percent testing level.

-

0.0012 (0.75) 0.40

-

Central government investment Local government investment Farm household investment

Area sown with crops

0.50

0.42

The estimation results are presented in Table 8. Again it is found that, out of the three types of SWC investment, only local government investments in SWC have a significant negative impact on the share of land affected by erosion. But, in contrast to the findings for the entire 1987-2002 period, we now find that the cross-product of this variable with the dummy variable for Western China has a significant positive effect on the share of land affected by erosion. The estimated coefficient for the cross-product is almost equal to minus the coefficient of local government SWC investments. In other words, these findings indicate that local government SWC had a negative impact on the share of land affected by erosion in China, but not in Western China, in the period 1987-2002. In Western China, local government SWC investments did not have a significant impact during that period. The impact of the area sown with crops does not differ significantly from zero. Deleting the area sown with crops from the erosion share equation does not change the conclusion that local


83

government SWC investments have a statistically significant impact on the area affected by soil erosion, but not in Western China (see column (2)). The last two columns present the results for the factors explaining the annual average change in the SI of erosion during the period 1997-2002. We find that SWC investments made by local governments in Western China have a significant negative impact on the SI of erosion. In other words, in Western China both local government and private investments in SWC negatively affected the SI for erosion during the period 1997-2002. The size of the area sown with crops again does not have a statistically significant impact. Deleting this variable from the equation improves the adjusted R-squared but does not change the conclusions drawn on the importance of the different sources of SWC investments in and outside of Western China. 4 Results for agricultural production function To estimate the impact of the share of land affected by erosion and the severity of erosion on agricultural GDP in Western China, we use the following Cobb-Douglas production function14: log(AGDPit) = d0 + d1*log(LANDit) + d2*log(LABit) + d3*log(CAPit) + d4*log(IRit/ LANDit) + d5*log(FERit) + d6*log(RAINit) + d7*EROSHit + d8*DNW*EROSHit + d9*DSW*EROSHit + d10*EROSIit + d11*DNW*EROSIit + d12*DSW*EROSIit + d13*DNWi + d14*DSWi + d15*DCi + d16*Y86t + d17*Y96t + ε3it (3) where: AGDPit = Gross domestic product of the primary sector of province i in year t, in constant prices of the year 2000 (in million RMB) LANDit = Arable land area in province i in year t (in million hectares) LABit = Agricultural labor in province i in year t (in 10,000 persons) CAPit = Capital stock of agriculture in province i in year t (in million RMB) IRit = Irrigated area in province i in year t (in million hectares) 14

More flexible functional forms such as the translog production function impose fewer restrictions on estimated parameters. However, theoretical consistency of such a function cannot be imposed globally (see Sauer et al. 2006). Moreover, many estimated coefficients are not statistically significant due to multicollinearity problems among various interaction variables, and the derivation of marginal effects becomes much more complex. For these reasons, we select the Cobb-Douglas function for our analysis.

84


FERit = Chemical fertilizers consumption in province i in year t (in 10,000 tons of pure nutrients) RAINit = Precipitation in the capital of province i in year t (in mm) EROSHit = Share of the land affected by erosion in province i in year t EROSIit = Value of the severity index (SI) of erosion in province i in year t DNWi = Dummy variable for provinces in North-western China (= 1 if province is located in North-western China; = 0 otherwise) DSWi = Dummy variable for provinces in South-western China (= 1 if province is located in South-western China; = 0 otherwise) DCi = Dummy variable for provinces in Central China (= 1 if province is located in Central China; = 0 otherwise) Y86t = Dummy variable for 1986 (= 1 if observation is for 1986; = 0 otherwise) Y96t = Dummy variable for 1996 (= 1 if observation is for 1996; = 0 otherwise) ε3it = Random disturbance term with standard properties This specification combines conventional production factors (land, labor and capital) with two other major determinants of agricultural productivity in China (irrigation and chemical fertilizer use), one major agro-climatic indicator (annual precipitation) and two indicators of soil quality (share of land affected by erosion and a measure of the severity of erosion). Crossproducts of the two erosion indicators and dummy variables for Northwestern China and Northeast China are added to the equation to test whether the impact of erosion on agricultural production differs between these two regions and the rest of China. As discussed in section 2, there exist major differences in the types of erosion and their severity between the provinces in North-western China and South-western China. For that reason, separate dummy variables for these two regions are included in the analysis. The information on erosion is available for only three years (1986, 1996 and 2002), so all information on all the other variables was collected for the same three years.15 No information on the capital stock in agriculture was available for the province of Tibet, which was therefore excluded from the 15

For simplicity, we assume that the data of the first national erosion survey (covering 1985-86) refer to 1986, and the results of the other two surveys (covering 1995-96 and 2001-02) refer to 1996 and 2002.


85

analysis. In addition Beijing, Tianjin and Shanghai are not included, because the agricultural sector in these three provinces is very small. The total number of observations is therefore 3 (years) x 25 (provinces) = 75.16 Data on capital stock in agriculture are not available from official statistics. Available information on gross capital formation and depreciation rates of fixed assets were used to construct such capital stock data in a previous IFPRI study of the impact of public investments in rural China (see Fan et al. 2002, 2004).17 In order to be consistent with economic theory, the coefficients for the three conventional production factors and irrigation and fertilizer use should be positive and less than one (i.e. have diminishing marginal productivities). Likewise, we expect annual precipitation to have a positive impact on agricultural production, with diminishing marginal returns. In water-scarce areas, located mainly in Northern China, water is the limiting factor in agricultural production. The impact of precipitation on output in such areas will be much larger than in areas with abundant precipitation, where agricultural production is limited by other factors that promote plant growth. Because the signs of the coefficients of the conventional inputs and precipitation are dictated by (economic and agronomic) theory, one-sided significance tests are used to test these signs. In our model, agricultural production depends not only on conventional agricultural inputs and annual precipitation, but also on the size of the land that is affected by erosion and on the severity of erosion. The latter two variables are both expected to have a negative impact on agricultural GDP. The larger the share of the land affected by erosion, the smaller agricultural GDP is expected to be. Furthermore, it may be expected that the negative impact on agricultural output is much stronger for land affected by severe erosion than for land affected by slight erosion. Both soil erosion indicators in our model are measured for the total land area in a province. Unfortunately, no separate information is available on the size or severity of erosion on cultivated land. We therefore use the two soil erosion indicators for the total land area as proxies for the share of arable land affected by erosion and the severity of erosion on arable land, respectively. Severity of water erosion is used as a proxy of the severity of erosion, as was also done in section 3. Exploratory analysis indicated that semi-log specifications for the two soil erosion variables provide better results, in terms of R-squared and t-values, than double-log specifications. Because semi-log specifications are consistent with economic theory (i.e. result in 16 17

The data for Sichuan province include Chongqing, and the data for Guangdong province include Hainan in our data set. Capital stock data have been constructed until the year 2000. We use the agricultural capital stock data for 2000 here as an approximation of the stock in 2002.

86


negative first derivatives and positive second derivatives) and greatly simplify the calculation of the marginal effects, we use semi-log instead of double-log specifications for these two variables.18 Again, one-sided significance tests are used to test for the (negative) signs of the coefficients. The coefficients of the cross-products of these two variables and the two regional dummy variables, however, can be positive as well as negative. So, two-sided tests will be applied for the cross-products. Dummy variables for provinces located in North-western China, Southwestern China or Central China are included as separate variables to control for regional differences in agro-climatic and other unobserved factors affecting agricultural production. Likewise, dummy variables are included for the years 1986 and 1996 to test for systematic differences in productivity between the three years in the data set. Although the coefficients of these five dummy variables can in principle be positive as well as negative, we expect that total factor productivity will be highest in East China and in the year 2002, meaning that all five coefficients will have negative values. The regression results are summarized in Table 9.19 All conventional agricultural inputs except capital are found to have a significant positive impact on agricultural GDP. The estimated coefficients are all smaller than one, implying diminishing marginal productivities. Given the large differences in agro-climatic conditions and the persistent water shortages in the northern parts of China, it is also no surprise to find that annual precipitation has a significant positive impact on agricultural production. Of the two erosion variables, only the severity of erosion in Northwestern China is found to have a significant negative impact on agricultural production. The other estimated coefficients for the erosion variables (including the other cross-products) are not significantly different from zero. A possible explanation is that erosion is measured as a share of all the land in a province, not just the land used for agriculture. Moreover, the relationship between erosion and agricultural production is not straightforward. High erosion rates may have little impact on agricultural productivity when soils in a region are deep and contain a high percentage of organic matter. Conversely, areas with shallow soils or unfavorable subsoils can be very sensitive to even limited rates of erosion (Lutz et al., 1994). Due to lack of suitable data, we cannot control for soil depth and soil quality in our analysis. A third possible explanation is the fact that provinces with large shares of eroded land tend to be located in areas with low rainfall, especially in North-western China (see section 2), causing a 18 19

See also Walpole et al. (1996) for a justification of this approach. Results of the White test for heteroskedasticity show that the null hypothesis of homoskedasticity is not rejected (at a 5 percent testing level). .

87


relatively high negative correlation (-0.72) between annual precipitation and the share of land affected by erosion in our data set. Table 9: Regression results for agricultural production function Explanatory variables Sown land

(1)

(2)

(3)

(4)

0.37*** 0.33** 0.38*** 0.33** (3.21) (2.32) (3.62) (2.43) Labor 0.33*** 0.36*** 0.33*** 0.36*** (3.23) (3.70) (3.34) (3.67) Capital -0.01 0.00 (-0.26) (0.02) Irrigation 0.13** 0.17** 0.16*** 0.18*** (1.87) (2.38) (2.44) (2.71) Fertilizer 0.28*** 0.26** 0.22*** 0.25** (2.58) (2.11) (2.43) (2.31) Precipitation 0.20*** 0.18*** (3.22) (3.03) Erosion share -0.30 -0.68** -0.34 -0.68*** (-1.07) (-2.32) (-1.23) (-2.40) DNW* Erosion share -0.11 0.28 0.24 0.30 (-0.24) (0.48) (0.73) (0.68) DSW* Erosion share 0.65 0.84* 0.61 0.84** (1.38) (1.86) (1.34) (2.02) Erosion severity 0.06 0.09 -0.01 0.08 (0.31) (0.40) (-0.06) (0.44) DNW*Erosion severity -0.62** -0.53 -0.45* -0.52* (-2.09) (-1.53) (-1.77) (-1.74) DSW*Erosion severity 0.23 0.17 0.38 0.18 (0.48) (0.42) (0.86) (0.55) North-western China 0.37 0.02 (1.14) (0.06) -0.61*** -0.72*** -0.62*** South-western China -0.65*** (-2.80) (-3.54) (-3.23) (-4.54) Central China -0.37*** -0.33*** -0.37*** -0.33*** (-5.08) (-4.35) (-5.37) (-4.47) Year 1986 -0.40*** -0.43*** -0.44*** -0.43*** (-3.42) (-3.45) (-4.98) (-4.23) Year 1996 -0.20*** -0.19*** -0.20*** -0.19*** (-3.68) (-2.93) (-3.91) (-3.04) Constant 0.13 0.14 -1.20* 0.14 (0.20) (0.20) (-1.90) (0.20) Adjusted R-squared 0.962 0.956 0.963 0.958 Notes: t-statistics between brackets. *** significant at one percent testing level. **:significant at five percent testing level. * significant at ten percent testing level. White’s heteroskedasticity consistent covariance estimator is used for estimating the results presented in columns (2) and (4).

To examine the impact of the latter factor, we also ran a regression with annual precipitation excluded from the list of explanatory variables. The

88


results are shown in column (2) of Table 9. 20 The results for the conventional input factors are comparable to those presented in the first column. But the major difference is the significance of the estimated coefficient for the share of land affected by erosion (at a five percent testing level) and the cross-product of that variable with the dummy variable for South-western China (at a ten percent testing level). The coefficient of the former variable is negative, while the coefficient of the cross-product is positive and slightly larger in (absolute) size. This finding suggests that the share of land affected by erosion has a negative impact on agriculture production in mainland China, with the exception of the provinces located in South-western China. The coefficient of the crossproduct of the severity index of soil erosion and the dummy variable for North-western China, however, is no longer statistically significant (if tested two-sided). The estimated coefficients for the five dummy variables are all negative and significantly different from zero in both equations, with the exception of the dummy variable for North-western China. The larger (absolute) coefficient estimated for the 1986 dummy than for the 1996 dummy indicates that, as expected, total factor productivity in agriculture was lowest in the year 1986.21 The results for the regional dummies indicate that total factor productivity in North-western China and Central China is significantly lower than that in East China.22 For North-western China, however, total factor productivity is not significantly different from what it is in East China. This finding suggests that the relatively low levels of agricultural production in North-western China can to a large extent be explained from the relatively large share of land affected by erosion and the severity of that erosion (and from differences in the level of inputs in the other explanatory variables). Finally, in columns (3) and (4) of Table 9 we show the results of dropping the only insignificant conventional input variable, capital stock, 20

21

22

Results of the White test for heteroskedasticity show that the null hypothesis of homoskedasticity should be rejected (at a 5 percent testing level) for this equation and for the equation shown in column (4) in Table 9. White’s heteroskedasticity consistent covariance estimator is therefore used for estimating the equations. Application of the Wald test for equality of the two coefficients for the 1986 and 1996 dummy variables shows that the null hypothesis of equality should be rejected (at a 10 percent testing level) in both equations. The estimated coefficient for South-western China is larger in (absolute) size than the coefficient estimated for Central China, but the difference is not statistically significant. Application of the Wald test for equality of the two coefficients for North-western China and Central China dummy variables shows that the null hypothesis of equality cannot be rejected (at a 10 percent testing level) in both equations.


89

and the dummy variable for North-western China from the equation. This causes some minor changes in the magnitudes of the estimated elasticities for the other conventional input variables. But the t-values of the three significant erosion values – the share of land affected by erosion, the crossproduct of that variable and the dummy variable for South-western China, and the cross-product of severity of erosion and the dummy variable for North-western China – become larger in absolute value, thereby providing additional support for the conclusions drawn above on the impact of erosion on agricultural production in Western China. 5 Conclusion In recent years, the Chinese Government has invested considerable amounts of money in soil and water conservation (SWC) as an important component of the ‘Open up the West’ program, but systematic assessments of their impact on soil erosion and agricultural growth have been lacking until now. This study has used a unique set of annual provincial data on SWC investments during the period 1989-2005 to estimate the impact of such investments on the extent and severity of erosion as well as agricultural gross domestic product (GDP) in Western China. The SWC investment data that we used have been sub-divided by source (central government, local government and farm households), making it possible to analyze the efficiency of different types of SWC investments in reducing erosion and stimulating agricultural growth. We first made an analysis of recent trends in soil erosion and SWC investments in China. Using data from three national soil erosion surveys, carried out in 1985-86, 1995-96 and 2001-02, respectively, we found that the share of the land affected by erosion has increased from 42.4 to 44.1 percent between the first and last surveys. The severity of erosion also increased between the first two surveys, but slightly declined between the second and third surveys. Erosion is a bigger problem in North-western China, both in terms of the area affected by it and its severity, than in South-western China. Wind erosion is almost entirely concentrated in North-western China, where it affects more than one-half of the land in Inner-Mongolia and Xinjiang. The share of the land affected by it increased from 43.3 to 44.6 percent in North-western China between the first and last surveys. Its severity increased between the first and second surveys, but has slightly declined since then. Water erosion affects a larger share of the land in South-western China than in North-western China (18.7 vs. 13.7 percent in 2001-02, respectively), but its severity is somewhat larger in Northwestern China. The trends in water erosion are more favorable than those in wind erosion. The share of the land affected by water erosion showed a

90


declining trend since 1995-96 in both regions, while its severity showed an important decline in both regions between 1985-86 and 1995-96. The data on SWC investments reveal several gaps, making it difficult to sketch an overview picture of the developments for Western China as a whole. Yet, it is possible to draw some conclusion for those provinces that have complete data on all three sources of investment. We found that central government investments greatly exceeded local government investments in most years between 1989 and 2005, except 1996-97. There was a rapid increase in central government investments and, to a lesser extent, also in local government investments in the years 1998 and 1999. Since 2002, central government investments and private investments have shown large declines. Local government investments, on the other hand, remained more or less stable during the latter period. Using the annual SWC investment data and the data from the national erosion surveys, a model of the short-term impact of annual SWC investments on changes in the extent and severity of soil erosion during the periods 1987-1996 and 1997-2002 has been estimated. We found that SWC investments made by local governments have a significant negative impact on the extent of erosion, whereas SWC investments made by farm households have a significant negative effect on the severity of erosion. In recent years, however, both local government investments and private investments have significantly affected reductions in the severity of erosion but not in the share of land affected by erosion in Western China. Central government investments in SWC have not been found to cause significant reductions in soil erosion. Such investments often take longer to materialize than the period (3-5 years) that we have examined in our analysis. Moreover, climatic and biophysical factors may contribute to an increase in soil erosion in the relatively poor and sparsely populated areas in which the central government concentrates its SWC investments. This may explain why we did not find a negative impact of central government investment on the extent and severity of erosion in Western China in our data set. The same data sets have been combined with provincial data on agricultural GDP and other socio-economic indicators made available by the International Food Policy Research Institute (IFPRI) into a provincial panel data set for the years 1986, 1996 and 2002, used to estimate an agricultural production function. We found that severity of erosion has a significant negative impact on agricultural GDP in North-western China (but not in South-western China), while the impact of the share of land affected by erosion is not significantly different from zero. The latter finding may have been caused by measurement problems in the indicator that we used, by omitted variables that would be needed to control for soil depth and land quality, or by the relatively high negative correlation


91

between annual precipitation and the share of land affected by erosion. When annual precipitation was omitted from the equation, the share of land affected by erosion was also found to have a significant negative impact on agricultural GDP in North-western China (but not in South-western China). Combining the estimation results for all equations, we were able to derive the marginal effects of public investments in SWC on agricultural GDP in Western China (see Appendix 2 in Heerink et al., 2009 for the mathematical details). To this end, we used the erosion function regression results for the period 2001-2002, because the quality of the SWC investment data is likely to be better as compared to the period 1989-2000 and because these data are available for a larger number of provinces. We took the regression results for the changes in the share of land affected by erosion and the severity of erosion presented in columns (2) and (4) of Table 8 and the production function results presented in columns (1) and (4) of Table 9. To derive the marginal effect on agricultural GDP, we combined these estimation results with available data for the year 2002 on agricultural GDP, share of land affected by erosion and local government SWC investments for the 7 provinces in Western China that were used for estimating the erosion functions. We found that one RMB invested in SWC by local governments in North-western China in 2002 increases agricultural GDP with 1.16 - 1.38 RMB 23, depending on which version of the agricultural production function was used. One RMB invested by farm households in Northwestern China, on the other hand, increases agricultural GDP with 0.200.24 RMB. As can be seen from Table 8, both local government investments and private investments affect agricultural GDP by reducing the severity of erosion, not the share of land affected by erosion, with local government investments having a larger impact (as can be seen from the estimated coefficients in that table). The marginal effects of SWC investments in South-western China are not significantly different from zero, because erosion was not found to have a significant impact on agricultural production in that region (see section 4). To examine the sensitivity of our result, we applied the formula for calculating the marginal effect to each of the 7 observations for Western China and examined the spread of the resulting marginal effects. The standard deviation of the marginal effects for local government investment equals 1.90 if the production function in column (1) of Table 9 is used, giving a 95% confidence interval of [-0.02, 2.79]. If the production function in column (4) of the same table is used, the standard deviation equals 1.59 and the 95% confidence interval equals [-0.02, 2.34]. For 23

All marginal effects are expressed in RMB in constant prices of the year 2000.

92


private farm household investments, the 95% confidence intervals equal [0.72, 1.21] and [-0.61, 1.01], respectively. Comparing our results with the marginal impact of different types of public investment in Western China presented in Fan et al. (2004: Table 5), we find that local government investments in SWC have a higher return to (provincial) agricultural GDP than public investments in electricity (0.88 RMB), but lower returns than public investments in agricultural R&D (10.19 RMB), telephones (2.49 RMB), education (2.33 RMB), irrigation (2.13 RMB) and roads (1.73 RMB). The estimated return to farm household SWC investments in North-western China is smaller than all the estimates presented by Fan et al. (2004). We can therefore conclude that SWC investments made by local governments and farm households in North-western China make a non-negligible but, compared to other public investments, relatively small contribution to agricultural GDP. It should be remembered, however, that our analysis has only focused on the economic impact of such investments. The main rationale for making SWC investments is restoration of the ecological balance in Western China. Our, rather crude, analysis in section 3 points to some limited successes in this respect. A more profound analysis of the impact of SWC investments on the quality of the natural resource base in Western China would require the availability of a more complete data base on such investments as well as relevant natural resource indicators at the provincial level and at lower levels of analysis, like counties and households. References Fan, S., Zhang, L., Zhang, X. (2002). Growth, Inequality, and Poverty in Rural China – The Role of Public Investments. Research Report 125. Washington, D.C.: International Food Policy Research Institute (IFPRI). Fan, S., Zhang, L., Zhang, X. (2004) Reforms, Investment and Poverty in Rural China. Economic Development and Cultural Change 52: 395-421. Glantz, M.H., Ye, Q., Ge, Q. (2001). China's Western Region Development Strategy and the Urgent Need to Address Creeping Environmental Problems. Aridlands Newsletter 49, May/June 2001 Available at: http://ag.arizona.edu/OALS/ALN/aln49/glantz.html#concerns

Goodman, D.S.G. (2004). The Campaign to “Open Up the West”: National, ProvincialLevel and Local Perspectives. China Quarterly 178: 317-334. Heerink, N., Bao, X., Li, R., Lu, K., Feng, S. (2009). Soil and Water Conservation Investments and Rural Development in China. China Economic Review 20: 288-302. Lai, H.H. (2002). China’s Western Development Program: its Rationale, Implementation, and Prospects. Modern China 28: 432–466. Li, R., Heerink, N., Bao, X. (2006). Final Report on SWC Data Collection and Analysis. Report submitted to Asian Development Bank under ADB TA4404.


93

Lutz, E., Pagiola, S., Reiche, T. (1994). The Costs and Benefits of Soil Conservation: the Farmer's Viewpoint. World Bank Research Observer 9: 273-295. National Bureau of Statistics (NBS) (2008). China Statistical Yearbook 2008. Beijing: NBS. National Development and Reform Committee (NDRC) (2006). The Outline of the Eleventh Five-Year Plan for National Economic & Social Development of the People’ Republic of China – Profile’. Beijing: NDRC. Available at: http://en.ndrc.gov.cn/hot/t20060529_71334.htm Sauer, J., Frohberg, K., Hockmann, H. (2006). Stochastic Efficiency Measurement: the Curse of Theoretical Consistency. Journal of Applied Economics 9: 139-165. Walpole, S., Sinden, J., Yapp, T. (1996). Land Quality as an Input to Production: the Case of Land Degradation and Agricultural Output. Economic Analysis and Policy 26: 185-207.

Part III Sustainable Land Management: Land Conversion

Chapter 6 What Is the Optimal Rate of China’s Conversion of Farmland? Statistical Experience from the Past 15 Years Rong Tan1 and Futian Qu2 Abstract: Farmland conversion, i.e. when farmland is occupied for nonagricultural use, not only supports rapid economic growth in China, but also brings with it some negative effects. The tradeoff between these aspects of conversion has already attracted the attentions of researchers and the government. In order to find out the optimal amount of China’s farmland conversion, this paper first defines the expense loss, excessive loss I and excessive loss II of farmland conversion. It then uses the C-D production function to set up a two-sector production function model and employs it with statistical panel data from 1989 to 2003 to estimate the optimal amount of farmland conversion in China. The results show that the expense loss is 18.5%, excessive loss I is 28.6%, and excessive loss II is 52.9%, which account for the total of converted farmland. As advice for policy makers, this paper draws the conclusion that the current optimal degree of farmland conversion in China would best be set at 47.1%, which is the sum of expense and excessive loss I, because excessive loss I is still too difficult to be grasped and separated from the sum in reality. Keywords: Farmland Conversion, Marginal Efficiency, Expense Loss, Excessive Loss

1 Introduction With an increasing population, rapid economic growth and urbanization, much land has been converted from agricultural to non-agricultural use in China. According to statistics, about 4,701,500 ha of cultivated land has been converted for non-agricultural use from 1978 to 2006. Being influenced by Marxism, China’s land was administratively allocated for long periods of use, although land was considered as an important production factor. The market system of urban land use rights only appeared after the end of 1987. Since then land has its price and land use rights can be sold for capital accumulation and also used to attract 1 2

College of Public Administration, Zhejiang University, Hangzhou, China China Land Policy Center, Nanjing Agricultural University, Nanjing, China


98

Tan and Qu

investment by the government. Therefore, in the following years, all local governments became very enthusiastic about selling land. Because of the limitations of the already existing land in urban area, local governments then diverted their attention to farmland in the rural areas. But the conversion of farmland leads to a series of negative effects for the sustainability of the economy, the stability of the society and the preservation of ecological services. What’s more, farmland conversion has often been taking place near the towns and cities in the coastal and eastern areas of China, where the land is usually of high quality and having a high cropping intensity (Yang 2000); this has been a cause of quality degradation of farmland at the national level in recent years, which has been worrying the central government regarding the nation’s food security. There is now continual debate in China about whether to convert more farmland resources for non-agricultural use to support rapid economic growth or to limit conversion to protect the farmland for food security. It is really a dilemma for the government: on one side the central government implements the so-called strictest land law in the world to protect the farmland, whereas on the other side the local governments continuously set up different kinds of industrial parks. So the answer to this debate is significantly important for the sustainable development of China’s economy. Will the debate last forever? In fact, the two sides have both been mentioned, but an important issue has not yet been highlight, i.e. the optimal amount of conversion. If the optimal degree is ascertained, then the tradeoff between them will be balanced. So the following questions arise: What is the optimal amount? Is today’s farmland conversion in China the optimal one? What should we do if it is not optimal? In order to solve these problems, we first define a series of concepts: expense loss, excessive loss I and excessive loss II for farmland conversion. We then employ the C-D production function with statistical panel data to build a demand and supply model in both agricultural and non-agricultural sectors to calculate the optimal degree and the three different kinds of loss for farmland nowadays in China. Then, based on the study recommendations regarding farmland conversion policy are discussed in the last part of the paper. 2 Optimal degree of farmland conversion Essentially, farmland conversion is the reallocation of land resources between agricultural and non-agricultural production sectors, which is not a new issue in economics. The key requirement for the optimal allocation of land resources in the two sectors is to make sure that the marginal revenue of land input in the two sectors is equal. So we begin the discussion with this standard.

What Is the Optimal Rate of China’s Conversion of Farmland?

99

In China the farmland conversion process begins with farmland acquisition by the government, which changes the ownership of farmland from rural-collective-owned to state-owned; then the government conveys land use rights to land users in different ways in the primary land market, such as transferring without charge (huabo), negotiating (xieyi), bidding (zhaobiao), auction (paimai), bulletin (guapai). Thirdly, the land users change the land use and complete the process of the conversion. Three kinds of prices exist during the conversion process, which are shown in Figure 1. P MC′

A MR

MC

P3 P2 P1

O

Q3

Q2

Q1

Q

Figure 1: Land prices and farmland conversion P3 is the equilibrium price when the marginal ecological revenue of the farmland resource (MC´) is considered. P2 is the equilibrium price for only the market marginal revenue, without the ecological being considered. P1 is the market price when the trade is influenced by some kinds of external factors, e.g. government intervention. The auction and the suspending price can be seen as P2, while other prices in the conversion process belong to P1. Nowadays, no price can represent P3 because of the difficulty of measuring the ecological value of natural resources in the traditional market system. The equilibrium quantity of farmland conversion with for price P3, P2 and P1 is Q3, Q2 and Q1 respectively. Equilibrium price P1 is a special price that only occurs in China’s land market. It results from government intervention. So far, the quantity of farmland conversion can be subdivided into three types, i.e. expense loss, excessive loss I and excessive loss II. Expense loss is the needed quantity of farmland to be converted into nonagricultural use to support economic growth under a pure market system, which covers the cost of all externalities in the allocation process, i.e. OQ3

100

Tan and Qu

in Figure 1. Excessive loss I refers to the conversion quantity being greater than the expense loss due to traditional market failure, where the ecological and other external values are not considered in the market trade decision. It underestimates prices and causes more farmland loss, which is shown as Q3Q2 in Figure 1. Excessive loss II is another type of excessive farmland conversion due to land prices being distorted in the market because of governmental intervention and it is shown as Q2Q1 in Figure 1. Therefore, China’s optimal farmland conversion should be OQ3 according to Figure 1. If the government grasps the optimal amount, then the dilemma between economic growth and farmland protection can be changed into a win-win situation at Q3. In order to obtain that amount, calculation of expense loss, excessive loss I and excessive loss II is needed. The following section of this paper will attempt to calculate all the three losses by setting up a model to simulate the demand and supply of the conversion. 3 Methods In order to calculate the optimal amount, the basic way is to measure the marginal revenue in both agricultural and non-agricultural sectors. So the main purpose of the model should be to estimate the marginal revenue of the land resource in both sectors. Production function is an important method to estimate the cause-and-effect relation between input factors and total output value (Lin 1992; Fan 1991). An average production function has been used here to measure marginal revenue, for our purpose is not to calculate the inefficiency of the production, but rather the input factor marginal revenue during a certain period. The C-D production function is employed to simulate the production of the two sectors, formulated as follows: α

β

Yagr = A × K agr × Lagr × Land agr χ

γ

(1)

δ

Ynonagr = B × K nonagr × Lnonagr × Land nonagr

ε

(2)

Y denotes the total revenue of the sector. K is the capital input. L represents the labor input. Land means the input of the land resource. The subscript agr means the agricultural sector, and the subscript nonagr is the non-agricultural sector. So the marginal revenue of the land resource in two sectors can be shown thus: α

β

MRagr = A × γ × K agr × Lagr × Land agr

γ −1

(3)

What Is the Optimal Rate of China’s Conversion of Farmland? χ

δ

MR nonagr = B × ε × K nonagr × Lnonagr × Land nonagr

101

ε −1

(4)

In theory, when the MRagr and MRnonagr are calculated, the optimal allocation of the land resource in the two sectors can be computed based on the total amount of the land resource. There is a rule of thumb for pricing in microeconomics3 (Pindyck and Rubinfeld 2001): P=

MC 1 1+

(5)

MR 1 1+

(6)

εd

P=

εs

Formula (5) describes the relationship between the product price and the producer’s marginal cost. P is the price of the product. MC is the marginal cost of the production, while εd is the demand elasticity, which is negative. Formula (6) is derived from the deduction of formula (5), which reflects the price of the product and the marginal revenue of consumers. MR is the marginal revenue of consumers, while εs is the supply elasticity, which is positive. If we consider the farmland resource as an output of the agricultural sector, and an input of the non-agricultural sector, the process of farmland conversion can be seen as a production process. The marginal revenue of the non-agricultural sector is the marginal revenue of the farmland conversion, and the marginal revenue of the agricultural sector is the marginal cost of the farmland conversion. Thus the relationship between land prices and marginal revenue and costs can be measured by formula (5) and (6). Then, with the results of equations (3) and (4), the marginal revenue and cost curves shown in Figure 1 can be estimated. In detail, first, presume the style of the demand and supply curves of farmland conversion to be like the following equations: Q D = a × P − b , b >0

(7)

Q S = c × P d , d >0

（8)

The reason for assuming these equations this way is to make the elasticity of demand and supply constant, i.e. b and d respectively, which will 3

The detailed deduction: MR= P+Q(∆P/∆Q)=P+P(Q/P)(∆P /∆Q)=P+P(1/Ed),if MR=MC，then P+P(1/Ed)=MC, that is P=MC/（1+(1/Ed)）

102

Tan and Qu

facilitate simulation of the demand and supply curves via formulas (5) and (6). In fact, China’s economy is in a rapid growth phase, which is causing demand for land resources to steadily increase (Qu and Chen, 2001). If the sample period is not too long, and the demand for land for economic growth is not influenced by natural restrictions or disasters, the assumption can be accepted. From formulas (5) and (6) and from equations (7) and (8) the following equations can be deduced: 1 log QD = log a + b × log(1 + ) − b × log MR d

1 log QS = log c − d × log(1 − ) + d × log MC b

(9) （10)

Consequently, based upon equations (9) and (10), the marginal revenue and cost curves in Figure 1 can be estimated by the following model: log Q D = C1 + C 2 × log MR

(11)

log Q s = C 3 + C 4 × log MC

(12)

After getting the estimation results for C1, C2, C3 and C4 with the equation MR = MC, the optimal quantity of farmland conversion Q can be calculated by equations (11) and (12). Or, from the estimation results of C1, C2, C3 and C4, and with equations (9) and (10), the numerical value of a, b, c and d can be calculated. Then, with MR = MC the optimal quantity of conversion Q also can be calculated.  (C × C 4 − C 2 × C 3 )  Q = exp  1  C4 − C2  

(13)

or

Q=a

d d +b

×c

b d +b

b×d − d  ×   b×d +b 

b× d b+d

(14)


103

Furthermore, if the MC includes the marginal ecological revenue of farmland resources, the computed result Q should be the same quantity as Q3 in Figure 1. Q is the expense loss of the farmland conversion, which is also the optimal amount of farmland conversion. If the MC does not include the marginal ecological revenue of farmland resources, the computed result Q should be the same quantity as Q2 in Figure 1; meanwhile, Q is the sum of expense loss and excessive loss I, and then the excessive loss II can be calculated by the total conversion quantity minus the expense and excessive loss I. So far, the model has been built to measure the expense loss, excessive loss I, and excessive loss II of the farmland conversion. But the measurement of the marginal ecological revenue of farmland is still missing, which will be discussed in section 4.

4 Marginal ecological revenue of farmland There are some values of farmland resources that are ignored in the market system. For example, farmland has the functions of water preservation, erosion control, biological diversity, and even aesthetic enjoyment from landscape views. These kinds of value are so-called external values, which are benefits for the whole society. External value is a well defined concept, and some research has already taken it into account during the allocation of farmland (Zhang 2001; Zhu 2002; Qian 2003). But there are seldom papers that are using a quantitative method to measure ecological value for allocation. So far, Costanza (1997) can be seen as a complete and detailed study which provides most essential values for ecosystem services and the natural capital of such resources. Thus, we will refer to some results from Costanza (1997), combined with real conditions in China to calculate marginal ecological revenue. Some results from Costanza（ 1997） on the ecological value of farmland are shown in Table 1.

104

Tan and Qu

Table 1: Ecological value of different agricultural resources according to Costanza Type of land

Forest

Grassland

Water

Cropland*

232

8,498

92

969（ average） Ecological value

2,007（ tropical） 302 （ temperate）

Source： Costanza (1997). Unit: 1994 US$ha-1yr-1. * Include the cultivated land and the garden plot.

The method for calculating the ecological value of farmland in China is as follows. First, calculate the acreage of different types of farmland in different provinces. The grassland and water data can be gathered directly in government reports, forestry should be subdivided into different zones, and cropland is the sum of cultivated land and garden plots. Second, based on the data in Table 1, calculate the average ecological value of different provinces in different years, weighted by their farmland acreage. Last, change prices into 2003 constant prices. Results are shown in Table 2.

Table 2: Ecological value of farmland resources in different provinces in China 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 Beijing

0.44 0.46 0.52 0.57 0.68 0.86 1.01 1.38 1.18 1.21 1.22 1.26

Tianjin

1.75

Hebei

0.46 0.48 0.51 0.56 0.66 0.82 0.93 1.06 1.03 1.02

1

Shanxi

0.33 0.33 0.36 0.38 0.44 0.55 0.64 0.71 0.72 0.71

0.7

Neimeng

0.22 0.26 0.27

Liaoning

0.53 0.57 0.66 0.71 0.81 0.97 1.13 1.24 1.21 1.18 1.16 1.16 1.16 1.15 1.17

Jilin

0.43 0.45 0.47 0.51 0.58

Heilongjiang

0.47

0.44

0.5

0.59 0.73 0.87 0.95 0.97 0.97 0.94 0.93 0.93 0.93 0.93

Shanghai

1.29 1.37 1.52

1.7

2.07 2.61

Jiangsu

1.67 1.61 1.84

1.8

2.33 2.86 3.44 4.18 3.82

Zhejiang

0.52 0.53 0.54 0.59 0.81 1.38 1.13 1.24

Anhui

0.98 1.02 1.06 1.12 1.27 1.73 1.72 1.85 1.83

Fujian

0.45 0.45 0.45 0.48 0.54

Jiangxi

0.61 0.62 0.64 0.69 0.77 1.42 1.09 1.24 1.16 1.15 1.13 1.13 1.13 1.13 1.14

Shandong Henan

0.89 0.95 0.99 1.03 1.16 1.42 1.67 1.88 1.88 1.87 1.85 1.86 1.89 1.88 0.71 0.71 0.71 0.75 0.86 0.98 1.14

Hubei

0.81 0.85 0.92 1.19

Hunan

0.61 0.61 0.64 0.73 0.87 1.64 1.38 1.46 1.52 1.53 1.53 1.56 1.54 1.53 1.57

1.8

0.4

1.3

1.28 1.28

1.99 2.22 2.72 3.36 3.85 4.21 4.33 4.31 4.26 4.24 4.29 4.28 4.32

0.3

0.34 0.42

1.1

0.7

1.2

0.5

1

0.99 1.01

0.73 0.73 0.72 0.73

0.54 0.56 0.56 0.56 0.56 0.57 0.57 0.58

0.81 0.89 3.1

0.99

0.9

0.89 0.88 0.86 0.87 0.87 0.88

5.95 3.48 3.48 3.53 3.62 3.62 3.64 3.64 1.2

3.8

3.75 3.75 3.78 3.75 3.79

1.18 1.17 1.18 1.18 1.16 1.19 1.8

1.76 1.77 1.78 1.76 1.79

0.79 0.88 0.85 0.85 0.84 0.86 0.85 0.85 0.85

1.3

1.3

1.9

1.27 1.23 1.22 1.23 1.23 1.25

1.76 1.85 2.02 2.09 2.05 2.01 1.99

2

1.99 2.03

105


Table 2: Continuation Guangdong

0.54 0.52 0.54 0.59 0.74 2.01 1.02 1.13 1.11 1.09 1.07 1.09 1.08 1.07 1.07

Guangxi

0.33 0.33 0.34 0.36 0.43 1.69 0.71 0.77 0.76 0.73 0.72 0.72 0.72 0.71 0.72

Hainan

0.3

Sichuan

0.36 0.37 0.38 0.41 0.48 0.89 0.64 0.69 0.74 0.74 0.73 0.73 0.74 0.74 0.75

Guizhou

0.23 0.23 0.24 0.26 0.31

Yunnan

0.26 0.26 0.27

Tibet

0.32 0.33 0.57 0.63 0.71 1.02 1.09 1.17 1.23 1.24 1.24 1.24 1.24 1.24 1.25

Shaanxi

0.27 0.27 0.28 0.31 0.34 0.44 0.52 0.58

Gansu

0.27 0.29 0.31 0.35

Qinghai

0.3

Ningxia

0.23 0.25 0.29 0.32 0.42 0.53 0.62 0.66 0.69 0.69 0.68 0.68 0.69 0.68 0.69

Xinjiang

0.34 0.41 0.44

0.29 0.31 0.34 0.46 1.48 0.78

0.4

0.3

0.7

0.45

1.1 0.5

0.8

0.78 0.79 0.78 0.77 0.78

0.51 0.51 0.51

0.5

0.51 0.51 0.51

0.38 1.79 0.56 0.62 0.63 0.64 0.64 0.63 0.62 0.62 0.63

0.4

0.5

0.6

0.6

0.59 0.58 0.57 0.58 0.57 0.58

0.59 0.68 0.67 0.66 0.65 0.68 0.68 0.69

0.49 0.52 0.69 0.88 1.11 1.31 0.5

0.82

1.3

1.31

1.3

1.29 1.33 1.36 1.39

0.64 0.81 1.08 1.21 1.24 1.25 1.21 1.21 1.25 1.25 1.25

Source： Calculated by the authors. Unit: 104yuan ha-1 yr-1 (2003 constant prices).

The ecological value of farmland in Table 2 is an average value, and it is not equal to the marginal value in theory. But the ecological value is a nonmarket value, and the relationship between the gross value and the quantity of the farmland is a linear equation, which means the gross value is equal to quantity of the farmland times the value per unit. The differential coefficient of the gross value (marginal revenue) is exactly the value of per unit quantity (a constant), so the average value can be seen as the marginal revenue. This can also be explained more intuitively: the loss of the ecological value of the last unit of the converted farmland is exactly the average ecological value of each unit.

5 Estimation and results 5.1 Data The farmland data used to estimate the model comes from the Statistical Data of the Land Administration of China (1989-1995); the Land Yearbook of China (1994-1997), compiled by the former National Land Management Bureau; and Annals of The Land Resources of China (1999-2003), compiled by the Ministry of Land and Resources of China. The data of the other variables come from the Statistical Yearbook of China (1990-2004). In order to obtain comparable data for the agricultural and nonagricultural sectors, an index for each variable was chosen in terms of whether the subdivided data of each industry of the index can be found. The gross revenue of the agricultural sector, Yagr, is the primary industry data that composes one part of the total GDP, while the other part of the GDP is the gross revenue of the non-agricultural sector, Ynonagr. Capital investment K is the sum of new basic construction investments, new

106

Tan and Qu

renewal and reconstruction investments, and fixed capital investments by urban and rural collective communities each year4. The labor data is the data for employed persons in each sector. The land resource input data in the agricultural sector, i.e. Landagr, includes the acreage of forestland, grassland, cultivated land, garden plots, and water area. The land input data in the non-agricultural sector, i.e. Landnonagr, is the acreage of built-up land. The conversion data for farmland, QD and QS, are the data for the cultivated land that is converted into buil-up land each year. Because farmland not only includes cultivated land, so this index underestimates the quantity of the farmland conversion5. But for the limitation of the data and the cultivated land to be occupied is the main proportion of the total conversion data, this index is acceptable. All price data have been changed into 2003 constant prices. The provincial panel data of 30 provinces, municipalities, and autonomic regions from 1989 to 2003 have been used to estimate the model, which excludes the data of Hong Kong, Macao, and Taiwan. Chongqing’s data is also excluded because of its incompleteness. 5.2 Estimation of marginal revenue in the two sectors Because of the different situations between different provinces in China, and the lopsided policy regarding quotas that permits farmland conversion in each area from the central government (Tan et al. 2009), this study divides the whole country into three regions, according to the national standard, as follows: (a) Eastern China: municipalities of Beijing, Tianjin, Shanghai, provinces of Hebei, Liaoning, Jiangsu, Zhejiang, Fujian, Shandong, Guangdong, Guangxi and Hainan; (b) Central China: provinces of Shanxi, Jilin, Heilongjiang, Anhui, Jiangxi, Henan, Hubei, Hunan, the autonomous region of Neimeng; (c) Western China: provinces of Sichuan, Guizhou, Yunnan, Shaanxi, Gansu, Qinghai, and the autonomous regions of Tibet, Ningxia, and Xinjiang.

4

5

Total investment includes four aspects: basic construction, renewal and reconstruction, real estate development, and other fixed capital investments. The data used in this paper underestimates investment in production. But for the aim of the model the MR and MC of the land resource have been found, so the estimation will not be influenced by the error regarding investment, for the error is considered within the constant term. The underestimation of the farmland conversion data leads to the underestimation of the slope of the MC and overestimation of the slope of the MR curves, so it overestimates expense loss and excessive loss I, which causes the underestimation of excessive loss II.

107


Table 3: Results of the regression of formulas (1) and (2) Equation (1) Eastern China (t test) Central China (t test) Western China (t test) Equation (2) Eastern China (t test) Central China (t test) Western China (t test)

A 2.491 (68.65) 2.122 (60.23) 2.013 (58.56) B 0.497 (4.93) 0.423 (5.12) 0.391 (4.93)

α 0.003 (20.89) 0.002 (21.34) 0.004 (19.24) χ 0.042 (7.30) 0.038 (8.30) 0.032 (6.30)

β 0.034 (28.94) 0.028 (22.85) 0.030 (25.54) δ 0.140 (9.01) 0.128 (8.35) 0.114 (7.25)

γ

R2

DW

Omit6

0.999

2.271

Omit

0.999

2.121

Omit

0.999

2.215

ε 0.028 (5.28) 0.021 (4.78) 0.023 (6.12)

R2

DW

0.999

1.787

0.998

1.841

0.992

1.812

The estimation strategy first estimates equations (1) and (2), and then calculates the marginal revenue of the two sectors with equations (3) and (4). To estimate (1) and (2), first change the equation into a logarithmic style, then employ the general least squares method with the first order component of autocorrelation, i.e. AR(1), which indicates an autoregressive component. During the estimation of equation (1), in order to get different coefficients for each province a cross-section weight for each province is used to estimate the coefficients of AR(1) and Landagr, this is done because the contribution of farmland in different provinces is different. During the estimation of equation (2), only the AR(1) is estimated by the cross-section weights, because built-up land is not influenced by terrain and climate in any way that is relevant here. In addition, White HeteroskedasticityConsistent Standard Errors & Covariance is employed to reduce the heteroskedasticity. The regression results are shown in Table 3; each coefficient is significant at the 1% level. Based on the results of equations (1) and (2), the marginal revenue of the land resource in the two sectors can be calculated with equations (3) and (4), which are shown in Table 4 and Table 57. 6 7

The data here are corresponding to 30 provinces, and they are omitted for restriction of the table. Here assume that the provinces in each region have the same production function but with different constant terms to control for the differences of each province. A study (Tan 2005) has shown that the Gini coefficient of the GDP averaged by the amount of the farmland conversion in the three regions are from 0.1 to 0.3, which means that there is not much disparity between farmland conversion and economic growth in each region. So offering them the same production function has no great affect on the calculation of the marginal revenue.

108

Tan and Qu

Table 4: Marginal revenue of land resources in the agricultural sector of each province in China 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 Beijing

122

123

123

124

124

127

139

170

175

178

180

180

180

180

180

Tianjin

192

193

192

194

193

185

197

210

232

246

266

262

262

252

253

Hebei

528

542

556

570

638

660

720

785

819

842

851

845

838

819

817

Shanxi

75

75

75

75

76

78

82

90

94

96

96

96

96

89

89

Neimeng

21

20

20

20

20

20

21

23

23

23

23

23

23

23

23

Liaoning

248

280

313

346

348

367

391

412

435

447

454

445

441

440

441

Jilin

163

168

173

179

181

187

210

222

230

230

235

233

230

230

233

Heilongjiang

51

51

52

52

52

53

59

64

67

68

68

63

62

63

64

Shanghai

381

396

410

424

436

462

496

395

435

467

501

505

507

691

643

Jiangsu

1166 1221 1276 1330 1251 1311 1401 1401 1506 1595 1686 1655 1643 1762 1763

Zhejiang

596

597

599

600

606

621

676

727

749

762

764

754

754

759

750

Anhui

508

497

485

474

470

462

458

509

538

552

560

548

542

532

531

Fujian

417

427

437

447

453

480

535

560

586

587

594

590

588

576

573

Jiangxi

240

244

249

253

255

262

281

296

307

312

315

311

309

313

314

Shandong Henan

1045 1081 1116 1152 1158 1173 1240 1334 1426 1462 1485 1464 1461 1380 1383 790

809

829

849

858

871

944 1028 1095 1115 1134 1130 1125 1090 1088

Hubei

345

338

331

324

325

332

367

384

404

413

416

398

393

394

394

Hunan

321

324

327

330

332

349

379

403

421

429

429

418

413

404

403

Guangdong

578

581

585

588

594

611

612

647

665

675

678

674

665

665

664

Guangxi

186

190

195

200

203

212

225

231

246

248

238

230

226

224

223

Hainan

415

415

415

415

422

460

494

495

512

519

528

524

524

527

528

Sichuan

124

139

154

170

172

178

195

220

247

234

230

227

225

218

218

Guizhou

72.8 76.6 80.5 84.3 87.3 85.8 99.7 106.3 110.5 111.7 111.2 110.4 109.4 103 102.9

Yunnan

88.1 90.5 92.9 95.2 97.8 102.8 110.4 117.8 123.5 125.4 126.3 125.1 125 119.2 119.2

Tibet

11.4 12.1 12.8 13.5 13.7 14.4 14.4 14.7 15.4 15.8 17.1 17.3 17.5 18.2 18.2

Shaanxi

80

80

80

80

79.8 81.8 87.2 92.6 95.1 96.7 96.2 95.2

95

95.5

Gansu

41.4 41.4 41.3 41.3 42.4 43.8 44.8 50.1

Qinghai

20.5 20.5 20.5 20.5 16.7 17.2 17.8 20.5 20.9 21.1 21.3 21.1

21

22

22

Ningxia

0.6

0.6

0.6

0.6

0.8

0.8

0.8

Xinjiang

22

22.5

23

23.4 23.1 23.2 24.2 26.2

0.7

0.7

0.8

0.8 -1

57

95

59.3 59.6 59.1 58.9 56.6 56.2

0.8

0.9

27

27.4 27.7 26.8 26.8 26.8 26.9

-1

0.9

0.8

Source： Calculated by the authors. Unit: Yuan ha yr (2003 constant prices).

109


Table 5: Marginal revenue of land resources in the non-agricultural sector of each province in China 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 Beijing

6.5

6.62 6.73 7.17 7.41 7.84 8.17 7.82

Tianjin

3.81 3.98 4.14 4.08 4.39 4.89 5.58 5.88 5.35 4.78 4.09 4.02 4.06 4.94 4.94

Hebei

1.46 1.47 1.49 1.62 1.67 1.83 1.96

Shanxi

0.76 0.78 0.81 0.89 0.94

Neimeng

0.4

0.4

Liaoning

1.9

1.95 1.99 5.25 2.21 2.42 2.57 2.29 2.29 2.29 2.18 2.18 2.23 2.68 2.64

Jilin

0.62 0.64 0.66 0.65

0.7

Heilongjiang

0.75 0.79 0.83

0.99 1.12

Shanghai

13.73 14.13 14.52 14.47 15.15 16.6 17.99 16.38 16.13 15.7 14.47 13.97 14.04 16.18 15.88

Jiangsu

2.73 2.84 2.95

Zhejiang

5.93

6

0.4

1

2.1

8 2.09

7.87 7.77 7.57 7.39 7.61 7.77 2

1.86 1.84 1.85 2.31 2.34

1.06 1.13 1.16 1.14 1.11 1.01 1.02

1.2

1.24

1.06 0.44 0.48 0.52 0.49 0.49 0.49 0.47 0.46 0.47 0.57 0.58

0.9 3.1

0.77 0.84 0.87 0.85 0.82 0.76 0.76 0.76 0.91 1.4

1.54 1.55 1.44 1.32

1.34 1.79 1.77

3.17 3.69 4.03 4.25 3.77 3.31 2.88 2.82 2.83 4.01

6.08 6.14 6.18 6.48 7.01 7.02

6.63 6.31 6.19 6.18 6.17 6.21

0.7

Fujian

2.06 2.41 2.75 3.08 3.53 4.07 5.69 5.57 5.37 5.14 4.82 4.68 4.69 5.25 5.22

Jiangxi

0.81 0.85 0.89 0.93 1.01 1.48 1.59 1.67 1.61 1.49 1.32 1.26 1.26 1.42 1.45

Shandong

1.58 1.77 1.96 2.05 2.14

Henan

0.79 0.88 0.97 1.01 1.14 1.32 1.47 1.58 1.59 1.56 1.46 1.45 1.49 1.83 1.84

Hubei

1.1

1.22 1.35

Hunan

1.08

1.3

1.52 1.48 1.56

4

4.15

4.3

1.5 4.3

0.85 1.15 1.16 1.07 0.99 0.89 0.87 0.85 1.11 0.97

2.2

2.48 2.75 2.64 2.48

2.2

2.15 2.15 2.82 2.87

1.64 1.83 2.33 2.48 2.39 2.24 2.01 1.98 1.6

2

2.37 2.35

1.75 1.96 1.91 1.83 1.68 1.64 1.67 1.86 1.87

4.56 5.06 5.66 5.56 5.31 5.09 4.77 4.65 4.78 4.71 4.69

Guangxi

0.85 0.92

1

Hainan

1.41 1.46

1.5

1.51 1.52 1.52

1.6

1.56 1.53 1.48 1.39 1.39 1.41 1.28

Sichuan

0.39 0.43 0.47 0.47 0.48 0.52

0.6

0.67 0.69

Guizhou

1.31 1.37 1.42 1.53 1.51 1.64 1.63 1.68 1.67 1.67 1.61 1.67

Yunnan

0.81 0.89 0.97 1.07 1.13 1.19 1.32

1.4

1.48 1.44 1.42 1.37 1.35 1.63 1.62

Tibet

2.2

1.8

1.58 1.49 1.21 1.29 1.32

Shaanxi

0.79 0.86 0.94

Gansu

0.53 0.51 0.49 0.62 0.67

2.39 1.58

1

0.7

6.9

4.6

Anhui

Guangdong

0.73 0.65 0.66

1.3

0.9

1.7 1

1.08 1.44 1.72

1.8

1.74 1.72 1.88

1.68 1.62 1.54 1.52 1.49 1.56 1.58 0.8

1.3

0.85 0.84 0.84 0.96 0.97 1.7

1.85 1.87 2.2

2.15

1.06 1.14 1.13 1.31 1.35 1.33 1.28 1.29 1.31 1.57

1.6

0.7

0.71 0.83 0.87 0.87 0.83 0.83 0.81 0.97 0.97

Qinghai

0.49 0.55 0.61 0.83 1.35 1.43 1.53 0.98

Ningxia

1.24 1.39 0.93 0.97 0.75 0.88

Xinjiang

0.37 0.39 0.41 0.44 0.37 0.33 0.38 0.45 0.41 0.38 0.34 0.34 0.37 0.49 0.49

1

1

1.07 0.96

1.01 0.97 0.98 0.9

1

0.93 0.98

0.85 0.84 0.85 1.21 1.21

Source： Calculated by the authors. Unit: 104 yuan ha-1 yr-1 (2003 constant prices).

5.3 Estimation of the MR and MC curves With the data in Table 2, Table 4, Table 5 and the quantity of farmland conversion, MR and MC curves are estimated using equations (11) and (12). The regression uses the fixed effects general least squares method, with AR (1) and cross-section weighted. A conversion of the coordinates of the MR is needed to make MR and MC lie on the same coordinate system

110

Tan and Qu

(reference frame) as shown in Figure 1. The results of the estimation are in Table 6.

Table 6: Results of the regression of formulas (11) and (12)

MR

MC′

MC

Equation（ 11） Eastern China (t test) Central China (t test) Western China (t test) Equation（ 12） Eastern China (t test) Central China (t test) Western China (t test) Eastern China (t test) Central China (t test) Western China (t test)

C1 Omit* Omit Omit C3 Omit Omit Omit Omit Omit Omit

C2 -0.490 (-5.14) -0.475 (-6.34) -0.468 (-4.64) C4 0.154 (2.75) 0.148 (2.86) 0.142 (3.12) 0.721 (3.82) 0.721 (3.56) 0.721 (3.78)

R2

DW

0.984

2.034

0.982

2.052

0.991

2.037

R2

DW

0.987

2.097

0.992

2.086

0.983

2.043

0.986

2.028

0.985

2.025

0.988

2.031

Note: *The data here correspond to 30 provinces; they are omitted for restriction of the table.

All of the coefficients are significant at the 1% level. From the results, the marginal revenue and cost curves of farmland conversion in China from 1989 to 2003 have been estimated. Based on the value of C1, C2, C3, and C4, the value of a, b, c, and d can be calculated, giving the demand and supply curves of the farmland conversion in China from 1989 to 2003. 5.4 Calculation of the expense loss, excessive loss I and excessive loss II Based on the results of the above calculations, using equations (13) or (14) the expense loss, excessive loss I8 and II9 for the sample period were calculated. The results are shown in Table 7. 8

The computing of excessive loss I depends on the calculation of the ecological value of the farmland. The ecological value of the farmland in this paper has been derived from the results of Costanza (1997), so the accuracy directly depends on the Costanza’s (1997) research. Prof. Costanza has himself mentioned that the appraisal of the ecological value of natural resources still needs to be perfected. So with development of the methods to appraise the ecological value of natural resources, the accuracy of the expense loss for farmland conversion can be improved.


111

Table 7: Different rates of loss from farmland conversion in China from 1989 to 2003 (units are in ha)

China Eastern China Beijing Tianjin Hebei Liaoning Shanghai Jiangsu Zhejiang Fujian Shandong Guangdong Guangxi Hainan Central China Shanxi Neimeng Jilin Heilongjiang Anhui Jiangxi Henan Hubei Hunan Western China Sichuan Guizhou Yunnan Tibet Shaanxi Gansu Qinghai Ningxia Xinjiang

Expense loss proportion

Optimal conversion (MC¹=MR) (1)

Hypo-optimal conversion (MC=MR) (2)

Real conversion

Excessive loss I proportion [(2)-(1)]/(3)

Excessive loss II proportion [(3)-(2)]/(3)

395201.3

1007259

2137467

(1)/(3) 18.5%

28.6%

52.9%

232888.5

521694.6

1140274

20.42%

25.33%

54.25%

10563.82 2998.735 28620.15 10074.56 11585.16 37866.69 58169.63 18394.71 22076.68 23645.9 7587.888 1304.618

40086.04 15136.04 48236.37 35179.03 51459.28 72430.66 84662.02 33231.68 56739.57 58952.87 22889.67 2691.4

51811.81 20864.65 124325.5 76523.76 84355.33 207222.3 157000.1 60424.66 189313.4 111733 44924.69 11775.28

20.39% 14.37% 23.02% 13.17% 13.73% 18.27% 37.05% 30.44% 11.66% 21.16% 16.89% 11.08%

56.98% 58.17% 15.78% 32.81% 47.27% 16.68% 16.87% 24.55% 18.31% 31.60% 34.06% 11.78%

22.63% 27.46% 61.20% 54.03% 39.00% 65.05% 46.08% 45.00% 70.03% 47.24% 49.05% 77.14%

107644.6

284908.5

638030.6

16.87%

27.78%

55.35%

11073.96 6089.703 5295.434 12732.59 18947.36 12405.77 22521.84 8367.554 10210.43

34333.33 26766.53 12617.14 49289.79 39486.44 17789.31 36713.67 42583.32 25329.01

57626.08 41658.78 30061.22 78052.39 117218.9 44592.43 125919.4 89369.17 53532.11

19.22% 14.62% 17.62% 16.31% 16.16% 27.82% 17.89% 9.36% 19.07%

40.36% 49.63% 24.36% 46.84% 17.52% 12.07% 11.27% 38.29% 28.24%

40.42% 35.75% 58.03% 36.85% 66.31% 60.11% 70.84% 52.35% 52.68%

54668.13

200655.8

359162

15.22%

40.65%

44.13%

13445.77 5733.071 13942.06 1305.495 8896.624 3398.73 1877.637 1707.901 4360.84

49553.85 21546.3 36183.4 7154.233 33154.36 15850.23 5320.954 7724.995 24167.51

116109.6 38779.61 62445.39 7243.453 52434.91 23831.07 6161.115 12090.21 40066.71

11.58% 14.78% 22.33% 18.02% 16.97% 14.26% 30.48% 14.13% 10.88%

31.10% 40.78% 35.62% 80.75% 46.26% 52.25% 55.89% 49.77% 49.43%

57.32% 44.44% 42.06% 1.23% 36.77% 33.49% 13.64% 36.11% 39.68%

(3)

Source： Calculated by the authors.

9

The regression data is panel data, with the results being the average of the highest level of the data. That means the results denotes the average value at the national level, with no difference between each province having been observed. So the value of the loss is more accurate at the highest level in Table.7, and the accuracy of the results decreases from national level to provinces level. If a more accurate value for each province is needed, then panel data for each province is required.

112

Tan and Qu

Table 7 shows that there has been a serious excessive loss II of farmland during the process of farmland conversion in China from 1989 to 2003, and its proportion, accounting within the real total conversion for 52.9%, is 2.86 times the expense loss. The proportion of excessive loss I is 28.6%, which is 1.55 times the expense loss. From the comparison of eastern China, central China and western China, the amount of farmland conversion decreases from east to west, on a scale of 3.17: 1.78: 1. But the proportion of excessive losses I and II in the three regions are not so much different, with the proportion of excessive loss I in each region being 25.33%, 27.78% and 40.65%, respectively, and the proportion of excessive loss II in each region being 54.25%, 55.35% and 44.13%, respectively. This means that the inefficiency of land allocation between the two sectors in the three regions is similar. 5.5 Analysis of the results 5.5.1 Optimal amount of farmland conversion Expense loss, as described above, is defined as the optimal amount of farmland conversion, which includes ecological and social values in the decision making process of the market system, i.e. the social optimal conversion quantity. However, to ensure a farmland conversion equal to the expense loss can still only be seen as an assumption in theory, because of the difficulty of including the externality of the farmland resource in the traditional market. Although the theoretical optimal amount of farmland conversion is difficult to be achieved in practice, a hypo-optimal conversion can be discussed. In order to be implemented in practice, the hypo-optimal amount should be no more than excessive loss I plus expense loss. On one hand, the amount can be calculated by referring to the private optimal conversion in reality; on the other hand, because the sum of the expense loss and excessive loss I is only 47.1%, choosing the hypo-optimal conversion also can reduce the excessive loss of farmland enormously. So, in the current situation the optimal amount of farmland conversion should be no more than the sum of expense loss and excessive loss I. If the hypo-optimal amount of farmland conversion is followed, the quantity of farmland conversion in China should be reduced by about 52.9% in comparison to 1989 to 2003. Will this baffle the growth of the economy? First, 52.9% is the newly converted farmland in that period. It is not to reduce by 52.9% of the total built-up land, and the total quantity of land in the non-agricultural sector will not be reduced. Second, let’s refer the experiences of other countries or areas. For instance, in Japan, after 1973 economic growth was supported by the knowledge and technology


113

sectors, and the speed of farmland conversion was slowed down, while the economy still maintained a steady increase (Guo 1993). In Taiwan, when industrialization was finished, the economy changed into a technologyintensive style and then farmland conversion decreased each year, while the growth rate was still maintained at 8% each year (Qu and Chen, 2001). In mainland China, there were 2,046 developing zones which were removed in 2003, and the removed rate of the total zones is 36% (China land and resource news, 2003), but the GDP in that year still remained at a high level. All of the above examples show that economic growth is not only supported by farmland conversion but that, if the Chinese government stops unreasonable and illegal farmland conversion projects, it will not hinder stable economic growth. 5.5.2 Reasons for excessive loss Nowadays, the proportion of excessive loss I is lower than that of excessive loss II, i.e. 28.6% versus 52.9%. As excessive loss I is caused by market failure and excessive loss II is caused by government failure, this means the main reason behind current excessive loss of farmland is government failure. This result is in accord with the real farmland conversion process in China. The government has been distorting land prices during the conversion process, causing excessive demand for land. As mentioned above, there are several different ways of transferring land use rights: transfer without charge (huabo), negotiating (xieyi), bidding (zhaobiao), auction (paimai), and bulletin (guapai). The costs associated with each type of transfer are different; only the auction and bulletin prices are similar to the final market price. Although the land market system and commercialized land system has been implemented since 1987, and the marketing rate increases every year, the proportion of auction and bulletin was only 24.6% of total converted farmland acreage in 2003 (Annals of The Land Resources of China, 2003). So it can be concluded that the intervention of the government is the main reason for the excessive loss due to farmland conversion in present day China. To reduce the intervention of the government and let the market work is one of the most important ways to reduce this excessive loss. 5.5.3 Comparison with other studies on the trade of quota for conversion Chen et al. (2004) studied the marginal revenue of the built-up land for the three regions of China and found that the contribution to economic growth from the built-up land input in eastern China is higher than that of central and western China. Consequently, they suggested that the quotas for farmland conversion should be transferred from the western and middle regions to the eastern regions to increase the revenue from land resources.

114

Tan and Qu

However, Chen et al. (2004) could have been mistaken, because they only focused on the marginal revenue of built-up land, whereas study of the farmland conversion process should not only observe the revenue from built-up land, but also should the revenue from farmland. So their conclusions may be not correct regarding the transferring of quotas between the three regions. Tan et al. (2005) studied the disparity of the contributions from farmland conversion to economic growth between the three regions. It found that the disparities within the whole country are caused by the disparities of the provinces inside each region, rather than disparities between the three regions. So, in order to reduce the level of disparity of the whole country, the transfer of quotas for farmland conversion should be restricted to within each region and should not be transferred between the regions. So the conclusion of Tan et al. (2005) is opposed to offering the eastern region more quotas for farmland conversion. From the research presented in this paper, we can see that the marginal revenue from farmland conversion within the eastern region is surely higher than that of the middle and western regions (Table 5), but the inefficiency of land resource allocation in the three regions is similar, i.e. the proportion of excessive loss in each region is similar for each region. So if more quotas for farmland conversion are transferred to the eastern region, the revenue from built-up land will likely increase, but the total revenue of the society, including the revenue from farmland, will decrease due to the excessive loss that will have occurred in every province in China. This means that the marginal revenue would already have been lower than the marginal cost in every province, i.e. any further land input in the non-agricultural sector would reduce the gross revenue of the society. And because of marginal costs of the middle and western regions are lower than that of the eastern region (Table 4), transfer of quotas from the western and central regions to the eastern region will cause a worse tradeoff: decrease of the gross revenue of the whole country. So it seems inappropriate to transfer quotas in the ways suggested in the current literature.

6 Conclusions Firstly, to find an optimal amount of farmland conversion is the only way to harmonize the contradiction between farmland protection and economic growth theoretically. The optimal farmland conversion rate should be ascertained by the marginal revenues from land in the agricultural and nonagricultural sectors, i.e. the expense loss due to farmland conversion, which has been defined by this paper.


115

Secondly, the optimal amount of farmland resource conversion should be 47.1% of the real conversion quantity that took place from 1989 to 2003, which is the sum of the expense loss and excessive loss I. But expense loss is difficult to measure and implement in reality, and excessive loss I is much smaller than excessive loss II. So, in order to currently reduce the excessive utilization of the farmland resource, more attention should be paid to how to reduce excessive loss II. Thirdly, because excessive loss I is much lower than excessive loss II, one of the most important measures for the government to reduce excessive loss is to reduce the intervention of the government itself and improve the functioning of the land market system. Lastly, as the proportion of excessive loss due to farmland conversion is similar under the economic growth conditions in each region, so is the efficiency of the land utilization also similar. If there were to be a transfer of quotas for farmland conversion from the west to the east of China, although the revenue from built-up land would improve, there would in fact be a decrease in the gross revenue of the whole country, and the efficiency of land resource utilization would also decrease. So it is not reasonable to transfer quotas for farmland conversion from the west to the east of China.

Acknowledgements The authors would like to express their thanks for the grants from National Nature Science Foundation of China within the project for Outstanding Youth Researchers (70425002) and the project for Youth Researchers (70903057). The comments and help from our colleagues of Nanjing Agricultural University and Zhejiang University are very much appreciated. Of course any remaining mistakes belong to the authors.

References Chen J., Qu, F., Chen, W. (2004). The Dissimilarity in Space of the Non-agricultural Efficiency of Farmland, and its Inspiration to the Policy Adjustment of Use of Land. Management World 8: 37-42. Costanza, R., et al. (1997). The Value of the World’s Ecosystem Services and Natural Capital. Nature 387: 253-260. Fan, S. (1991). Effects of Technological Change and Institutional Reform on Production Growth in Chinese Agriculture. American Journal of Agricultural Economic 73: 266-275. Guo, Z. (1993). The Takeoff and Options of Japan. Shanghai Peoples Press. Hong Y., Li X. (2000). Cultivated Land and Food Supply in China. Land Use Policy 17: 73-78.

116

Tan and Qu

Lin J.Y. (1992). Rural Reforms and Agricultural Growth in China. The American Economic Review 82: 34-51. Pindyck R.S., Rubinfeld D. (2001). Microeconomics (Fifth Edition). Tsinghua University Press. Qian Z. (2003). China Farmland Protection. Management World 10: 60-70. Qu F., Chen J. (2001). Comparison of the Farmland Conversion between the Mainland and Taiwan Area in Rapid Economy Growth Phase. China Land Science 6: 59. Tan, R., Beckmann, V., Van den Berg, L., Qu, F. (2009). Governing Farmland Conversion: Comparing China with the Netherlands and Germany. Land Use Policy 26: 961-974. Tan R., Qu, F., Guo, Z. (2005). On the Disparity of Farmland Conversion to Regional Economic Performance in China. Resources and Environment in the Yangtze Basin 14: 277-281. Zhang H., Jia S. (2001). An Analysis on Adjustment Mechanism of Non-agricultural Utilization of Cultivated Land. Economic Research 15: 50-54. Zhu P., Qu F. (2002). Economic Analysis on Cultivated Land Allocation Between Agricultural and Non-agricultural Sectors. China Land Science 5: 14-17

Chapter 7 Effects of the Public Domain Property Rights on Collective Farmland Prices: A Chinese Case Study Guancheng Guo1 Abstract: Measuring every attribute of commodities is not costless, which makes it very expensive to perfectly delineate property rights. And because of this, not all attributes are priced. Using unpriced attributes is equivalent to placing the attributes in the public domain. The author intends to analyze the effect of the public domain property rights on collective farmland’s price through a Chinese case study. In China, the collective farmland is not only a means of livelihood but also a form of security for peasants. The farmland has many functions, including social security, social stabilization, ecological environmental and other social functions. Because of limitation on the knowledge of collective farmland’s function attributes and difficulties of measuring these attributes accurately, some property rights remain in the public domain. In the course of utilization and transaction of the farmland, peasants have to face a large amount of transaction cost, which also makes some valued properties into the public domain. Both of these two aspects may result in a reduction of collective farmland’s price. By one Chinese case study, the author calculates the expropriation price of collective farmland in the selected county, and appraises its market price by hypothetical development method. Compared with the market price of the collective farmland, the expropriation price is decreased 67.31 percent on the average level because of the public domain. So the effect of the public domain on collective farmland’s price is very remarkable. Keywords: Public domain; Property rights; Transaction cost; Collective farmland’s price

1 Introduction China is in the course of transforming its centrally planned economy into a market one, taking increasingly rapid steps towards industrialization and urbanization, whereby much collective land, most of which is farmland, is being expropriated into state-owned land,. According to the results of the Research on Alteration of Land Utilization in China, made public by the Ministry of Land and Resources of the P.R.C., up to October 31, 2004 the 1

China Center for Land Policy, Nanjing Agricultural University, China


118

Guo

total cultivated land in China was 0.122 billion ha, which had been reduced by 8 million ha as compared with the total area for October 31, 1996 (Wei 2006). This reduction poses many problems, including food selfsufficiency and security, environmental and social problems. Many researchers have seen urbanization, industrialization, capital investment and changes in property rights as reasons for the farmland reduction (Huang, Zhu and Qiao 1998; Li 1999; Zhu, Pu and Peng 2001; Qu, Feng and Zhu 2004). This paper tries to analyze the problem from the perspective of the public domain property rights. The author’s viewpoint is that collective farmland prices are being depressed due to public domain property rights, which results in decreasing amounts of farmland. The paper is organized as follows: section 1 gives a brief introduction to the theory of public domain property rights. Model analysis is done in section 2. Section 3 explains the structure of collective farmland’s property rights in China. Public domains I and II are respectively analyzed in sections 4 and section 5 with regard to the situation in China. Section 6 does some analysis on the effects of the public domain on collective farmland prices in China. To test the paper’s hypothesis, empirical data are given in section 7. Finally, section 8 concludes the paper.

2 Theory of public domain property rights In classical economics, a price system can be looked at as being voluntarily operated without any cost. However, the economists of property rights insist that the nature of exchange in markets touches not only on physical commodities or services, but rather on bundles of property rights, and the value of goods is determined by the exchanged property rights. The same resource or asset will have different market prices if it is linked to different property rights or if it has varying property rights boundaries. Property rights are an instrument of society and derive their significance from the fact that they help people form the expectations which can reasonably be held in dealings with others (Demsetz 1967). In order that the rights to an asset be complete, or be perfectly delineated, both its owner and other individuals potentially interested in the asset must possess full knowledge of all its valued properties. With full knowledge, the transfer of rights to an asset can be readily effected (Barzel 1997). If the property rights are fully clear and perfectly executed without any obstruction or any extra costs, the actors involved can freely negotiate according to the Coase Theorem, up to the Pareto optimum. It’s unfortunate, however, that the arrangement of property rights does not amount to a free lunch. Both delineation and exertion of property rights bear costs: transaction costs. In the real world, transaction costs exist wherever exchange happens, as does friction in physics. So, in such a real world with positive transaction costs, an owner

Public Domain Property Rights and Collective Farmland Prices

119

of property rights will have no incentive to specify or make use of these rights when transaction costs are higher than expected benefits. With regard to this situation, Barzel (1997) had a penetrating saying, “Given that property rights are never perfectly delineated, some valued properties will always be in the public domain.” According to his analysis of property rights, measuring every attribute of commodities is not costless because of the complexity of transactions, which makes it very expensive to perfectly delineate property rights. And because of this, not all attributes are priced. Consequently, unpriced attributes tend to be excessively used and inadequately provided. Using unpriced attributes is equivalent to placing them in the public domain (Barzel 1997).

3 Analytical model According to the theory of property rights, some aspects of property rights remain in the public domain owing to transaction costs. There are two origins of transaction costs. One is the technology factor. The complexity of a commodity’s attributes may result in such high transaction costs during the definition or exertion of property rights that people have to place some those property rights into the public domain, which is called public domain I. The other is the institutional factor, including irrational regulation or restriction of an institution, or a lack of organization. The irrationality factor can create such high transaction costs that property rights actors have to give up their rights and place them into the public domain, which we call public domain II. It is convenient to use such property rights in the public domain for those who can take advantage of them, which would then however be an infraction of the owner’s property rights and leading to a decrease of the transaction value of the good in question. On the assumption that property rights can be quantified, the following models may be built. PQ=PA+PB

(1)

PB =F(C1)

(2)

C1=f(T,θ)

(3)

In the above formulae, PQ represents the total quantity of property rights, in which PA is the part that can be defined and executed, while PB cannot. So PB has to be kept in the public domain because of its high transaction costs. Formula (2) represents the function relationship between PB and transaction cost (C1): the higher C1 is, the greater PB is. Here, the transaction cost is determined by the technology factor. In formula (3), θ is the remaining factors. In general, transaction costs may be reduced with the development

120

Guo

of technology. However, according to the theory of property rights, the public domain I will always exist because of the incomplete definition of property rights due to transaction costs. In Figure 1, PA2 is one part of PA on account of the institution factor, resulting in public domain II of property rights. PA1 is the part that can be really delineated and executed. It is transaction cost (C2) that leads to PA2, which is related to the non-performance institution of property rights, lack of organization, and so on. The function between PA2 and C2 can be described as: PA2=G(C2)

(4)

C2=g(I, O,ε)

(5)

In the formula (5), I stands for the non-performance institution, O for lack of organization, and θ for the remaining factors. Because, PA =PA1+PA2

(6)

PA1 = PQ-PB-PA2= PQ-F(C1) -G(C2)

(7)

Then,

From formula (2) and formula (3), PB = F(f(T,θ))

(8)

From formula (4) and formula (5), PA2=G(g(I, O,ε))

(9)

Assuming, H(T,θ)= F(f(T,θ))

(10)

L(I, O,ε)= G(g(I, O,ε))

(11)

It follows that, PA1 = PQ-H(T,θ)-L(I, O,ε)

(12)

Formula (12) shows that the property rights which can be fully delineated and really executed are reflected by many factors, including technology, property rights institutions, organization, etc. If we want to enlarge our


121

exchanging property rights and obtain greater income flow, a good way is to improve our technology, institutions and organization. Public domainⅡ

Public domainⅠ

PA PB PA1

PA2

Figure 1: Transaction costs and the public domain property rights Source: adapted from Barzel (1997)

4 Structure of collective farmland’s property rights in China 4.1 Ownership In the early days after the foundation of the People’s Republic of China, the central government carried out the Land Reform in rural areas. It abolished the ownership right of non-cultivating landowners over their land, replacing it with a peasants’ land ownership system. After the mid-1950s, the Chinese government promoted establishing cooperative groups and primary unions in rural areas, when the land was still owned privately by members of the groups and unions (Wang 2006). But after 1955, setting up agricultural cooperatives was hastened, and advanced cooperatives prevailed throughout the country by the end of 1956. The land was owned collectively by the cooperatives, which arranged the members’ work and owned the means of production and output. Then the system of ownership on three levels—the communes, production brigades, and production team—was formed. Since the beginning of the 1980s, China has been reforming rural areas and adopted the Household Contract Responsibility System (HCRS), whereby households obtained land management rights by signing land contracts, while ownership remains with the collectives. Article 10 of the Constitution of China prescribes: “Land in the rural and suburban areas is owned by collectives except for those portions that belong to the state in accordance with the law; house sites and private plots of cropland and hilly land are also owned by collectives.”

122

Guo

Although this Article affirms that rural land is owned by collectives, it does not clarify the specific owners of collective land, which has to be solved by legislation. The National People’s Congress enacted the General Principles of the Civil Law in 1988, which declared: “Collectively owned land should be owned collectively by the village peasants in accordance with the law, and they should be managed and administered by the village agricultural production cooperatives, other agricultural collectives, or villagers’ committees. Land already under the ownership of township(town) peasants’ economic collectives may be collectively owned by the economic collectives.”

Article 10 of the Land Administration Law, which was revised in 1998, prescribed: “The land owned by the peasants’ collective is owned by the peasants’ collective of the village according to law, and managed and administered by the village collective economic organization or the villagers’ committee; what is already owned by two or more rural collective economic organizations of the peasants’ collective can be managed and administered by each of these rural collective economic organizations or villagers’ groups; what is already owned by the peasants’ collective of the township(town) is managed and administered by the rural collective economic organization of the township (town).”

From the above description, there are 3 types of categories for the ownership of collective farmland: (1) The land is owned by the peasants of villages and managed and administered by the village economic organization or villagers’ committee; (2) the land is owned by the peasants of the township(town); (3) the land is owned by the peasants of two or more rural collective economic organization or villagers’ groups. 4.2 Contractual management rights Farmland contractual management rights are enjoyed by members of collective economic organizations towards collectively-owned or stateowned farmland that can be used collectively. This right was created with the adoption of HCRS. In 1988, the General Principles of Civil Law considered farmland contractual management rights as a form of property rights. The Rural Land Contract Law, enacted in 2002, further clarifies the principles, procedures, and terms of the farmland contract. In Article 20 it is stated that, “The term of cultivated land contract shall be thirty years. The term of grassland contract shall be thirty to fifty years. The term of forestland contract shall be thirty


123

to seventy years; the forestland contract for special types of trees may be extended with the permission of the forest administration of the State Council.”

However, this law does not affirm the right of farmland contractual management and hence adversely affects the stability and transferability of contractual management right, while hindering the circulation of farmland as property. Without an adequate system of property rights, the land contractual management right can hardly become a long-term stable property right, and the contracted households face much difficulty defending themselves against unlawful interference and damages. Another thing that should be noticed is that collective farmland contractual management rights are owned by every member of the collective. The Rural Land Contract Law has this statement: “Each member of a rural collective economic organization has the right to contract rural land of the same organization, and no one can deprive or unlawfully restrict this right.” So it can be deduced that the collective farmland contractual management right is naturally obtained by every peasant if she or he is a member of the collective. In fact, the right is kept in close touch with membership rights of the collective, which is important in HCRS.

5 Public domain I of collective farmland in China 5.1 Function attributes of collective farmland in China In China, collective farmland serves more than just the function of an important factor of production that generates a return on its use (Kung 2000). Where rural markets remain to be developed, farmland also provides a source of food security and insurance for households in the event they fail to obtain income from other sources. In addition to the production function, collective farmland has many other functions as follows: (1) Social security function Urban citizens are covered by social security rights in China, but not peasants. Thus, farmland is the last life guarantee for the peasants. When peasants are too old to cultivate their lands, their farmland can be inherited by their children who will be responsible for them in their old age. If a member of the collective does not have any children, he or she will get life security from the collective, whose own source is income from the collective land. China has a large rural labor force, and today much of it is also engaged in some kind of part-time jobs in urban areas. However, these rural laborers are short of education or enough training, which adds much risk for their easily finding employment. These laborers could, however,

124

Guo

return to their collective farmland if they can’t survive in non-agricultural fields. Therefore, the farmland has a further function of employment security. Even in the south of Jiangsu province, where the non-agricultural income of peasants is very high compared to other areas of China, 89.7 percent of the peasants won’t abandon the property rights on their collective farmland (Liang 2000). (2) Social stabilization function There is an old Chinese saying: “People depend on food, food on agriculture, and agriculture on farmland.” This shows that farmland is considered to be very important for the society. What’s more, China has a population of more than 1.3 billion, so the farmland has a paramount meaning for food security. If the collective farmland sustains great losses, the country’s food security will fall into a serious situation and will become a challenge to social stability. In 1995, Lester Brown shocked the Chinese government with his prediction that the People’s Republic of China would face critical food shortages in the future (Peter 2001). (3) Ecological-environmental function Land, a basic factor of the natural ecological environment, has been in existence since before the origin of human beings. Collective farmland has many ecological-environmental functions, such as avoiding soil erosion, regeneration of water sources, improvement of weather, purification of air, and so on. In the course of the urbanization of China, many collective farms are being transferred to state-owned construction land, which is reducing the land area that is beneficial for our ecological environment. (4) Other social functions From the viewpoint of Resource and Environment Economics, farmland resources possess the non-use values of option value, bequest value and existence value. Option value indicates that people will pay for future choices regarding farmland use, which can be used for themselves and for their offspring as well as others. Bequest value denotes that present owners receive satisfaction by voluntarily preserving and bequeathing their farmland to their offspring. It implies that people wish their children to achieve some benefits from the land, for example, sightseeing, education, culture, entertainment, etc. Existence value means the value put on


125

attempting to insure the presence of a farmland resource because of its inherent value as such, which is independent of land use. 5.2 Connections between function attributes, property rights and collective farmland prices Barzel (1997) has pointed out that the structure of a set of rights should be designed to allocate ownership of individual attributes among actors in such a way that the actors who have a comparative advantage in managing those attributes that are susceptible to the common-property problem will obtain rights over them. According to this rule, peasants have a comparative advantage over the production function, the relevant property rights of which are utilization for agriculture, so peasants should own these property rights. As for the social security function, the state would have the advantage here, because the state has a responsibility to set up social security for every citizen of the state. When peasants lose their farmland, it means that they lose its social security function, so we can set up a development right on farmland to incarnate the security function. Development rights on farmland indicate the right of transfer from agricultural use to construction use, which should be allowed by the state. It is obvious that the social stabilization function, ecological-environmental function and other social functions are not aimed at one farmland user or one person, but rather benefit the whole society. This means that these functions have the characteristics of public goods, so that they cannot be offered by private or market approaches, but can only be provided by the public sector—the state. Therefore, the corresponding property rights are public ones. In Figure 2, OMNH represents revenue curve, and the part of OM is to reflect income from the production function in agriculture, under which the area of OV1M reveals the value of economic income, so V1 is the price of farmland’s income from agricultural use. Then, the part of MN refers to the social security function, under which the area of MNV2V1 reveals the value of development rights; V2 is the price of farmland’s development rights plus V1, which in China is called the immature land price after its use transfer. Finally, part of NH is drawn for the social stabilization, ecological-environmental and other social functions, under which the area NHV3V2 means the value of public property rights, so V3 is the price of farmland’s public property rights plus V2, which in China is recognized as the full price of farmland.

126

Guo H

ecological environment function

public property rights

other social functions social stabilization function N

social security function M

Production function

O

development right

V1

farmland property rights for agricultural use V2

V3

Figure 2: Connections between function attributes, property rights and prices for collective farmland Source: Own compilation.

5.3 Function attributes and public domain I of collective farmland in China According to Barzel’s analysis of property rights, transacted commodities have many attributes, and the rights to different attributes of a given commodity or to different attributes of a transaction are not all equally well defined. In China, the people don’t have enough consciousness about collective farmland’s function attributes, especially the social stabilization, ecological-environmental and other social functions. In fact, it is very difficult to calculate the value of these functions (Xiao, Qian and Qu 2005). Therefore, the value of farmland can only be identified within the scope of the people’s ability to recognize its particular values. And this becomes an opportunity for those who want to snatch more property rights in the public domain.

6 Public domain II of collective farmland in China 6.1 Non-performance of institutions According to the analysis models, the inconsequence of institutions, including irrational regulations or restrictions, can create such high a transaction costs that property rights actors have to give up their rights, which thus may be put into the public domain. In China, the inconsequence of institution can be analyzed from two aspects, as follows.


127

(1) Instability of property rights over collective farmland The above description about the structure of collective farmland property rights in China shows that the farmland contractual management right is not actually a real right. According to the HCRS, every member of a collective has a right to contract farmland of the same organization, which means fairness for everyone of the collective. However, it causes serious problems when a new born baby or a new bride of a collective member needs to be given the membership right to contract collective farmland, as the collective farmland should be distributed again. In fact, collective farmland is reallocated every three or five years in accordance with changes of population in many parts of China. Surely, this affects the stability of collective farmland property rights, and peasants should hand over their contractual collective farmland when the tenure of a farmland contract ends or the population changes, then these farmlands would be reallocated. For the peasants, therefore, it is almost certain that they will lose collective farmland which they have cultivated for several years, creating a situation which will not ensure the land’s users stable expectations and a positive attitude regarding land investment. It is concluded that property rights over collective farmland are not stable because community boundaries are in a state of uncertainty. (2) Insecurity of property rights of collective farmland According to the institutional structure governing collective farmland in China, there are plural subjects of the ownership of the collective farmland, which means ambiguity of property rights and that these rights may become infracted. On the one hand, it is difficult to establish a land registration system, which brings significant troubles to the cadastre administration. Because of the ambiguity of ownership, it is hard to determine who should register as the owner of the collective farmland. On the other hand, when villagers find their collective farmland’s rights have been infracted or are inconsequently disposed of by the villagers’ committees, they want to bring suits to the courts and demand the protection of their rights, but the courts refuse to handle such cases on the grounds that the villagers have no standing under the present law (Wang 2006). Both of these institutional characteristics show the insecurity of property rights over the collective farmland. Nevertheless, the more serious threat is from actual expropriation of collective farmland. According to Article 43 of the Land Management Law, “any enterprise or individual must apply for state-owned land when they need land for construction.” This generally means that the land users or developers can’t apply for

128

Guo

collective land to construct projects or other things. What’s more, article 63 of this Law prescribes “the usage right of rural collectively-owned land shall not be remised, transferred or leased for the purpose of nonagricultural construction, except for usage right transfer due to bankruptcy and mergence, etc., according to law by enterprises that conform to the general planning of land utilization and acquired land used for construction.” This regulation totally prohibits transfer of collective land used for non-agriculture construction. Consequently, under the present framework of Chinese law, the only legitimate way for collective farmland to enter the market is when it is expropriated by the state and can then be circulated in the market. This easily leads to abuse of collective farmland expropriation which may be beyond the scope of public interest. In the course of industrialization and urbanization in China, a large amount of farmland has been needed for construction and projects, and too much collective farmland has been expropriated: whether the aim has been for the public interest or not. Some officials of local governments expropriate farmland in the name of the public interest, but do not employ public bidding or auctions, thus seeking their own personal interests. At the same time, it is the villagers’ committees that actually manage collective farmland in China. Because of corruption and lack of supervision, the villagers’ committees often dispose of farmland and change farmland contracts arbitrarily, thus harming villagers’ security on their contractual farmlands. Given the villagers’ weak understanding of law, collective farmland usually becomes owned actually by the head or a few cadres of the village. These village’s cadres may rent or sell farmland for their private interests without authorization, or misappropriate state compensation for farmland expropriation, which also harms villagers’ interests. 6.2 Organizational imbalances Under the present frame of Chinese institutions governing farmland, peasants or their collectives are placed in a disadvantageous position because of their weak organization. In the course of collective farmland expropriation, for instance, the peasant may have not rights to participate in negotiatation of compensation issues. Though the peasants are the majority in the overall population of China, they exist politically at such a low level of organization that they have almost no voice in negotiations or in the course of legislation and policy making. Because of this, the peasants face great costs in terms of time and effort when they want to exchange or protect their property rights over collective farmland. For them, it’s difficult to search for information, negotiate, make decisions, monitor, or attend to other tasks. For example, if one peasant finds that an individual or


129

organization has infracted her or his farmland property rights and wants to exercise or maintain them, then he or she may have to search for information and evidence, engage a lawyer, pay court fees in advance and so on. It is possible for court fees to be so high for the peasant that he or she has to give up the property rights in question, which can then be snatched up by those who have comparative advantage because the rights are now in the public domain.

7 Effect of the public domain on collective farmland prices in China From Figure 1 it can be seen that the greater the share of public domain rights, the fewer individual property rights can be really delineated and exercised; as a result, the value or price an actor can obtain decreases. Because of the limitation on the available knowledge of farmland’s function attributes and difficulties in measuring these attributes, some property rights remain in the public domain, which reduces the overall exchange price of farmland. Lacking stability and security of property rights, the households who use collective farmland do not have enough incentive to make long-term investments on the land. For example, such peasants will apply more chemical fertilizer than organic fertilizer, as they only see a small probability of being able to continually farm their currently assigned plots. As we know, the stability and security of property rights can give people a long-term perspective that tends towards ensuring sustainable use of goods or resources. But under present conditions of instability and insecurity the peasant will use the farmland with an attitude of relative indifference, which is not beneficial for farmland quality and will reduce its value. Today, farmland expropriation is a forcible action of government in China, and in order to develop their local economies local governments have great incentive to expropriate masses of collective farmland at very low compensation. In this process, many public resources can’t be operated according to the market, but can only be disposed of by government power, which distorts the price of the collective farmland. For their own personal interests, some officials of local governments engage in rent-seeking activities in the course of collective farmland expropriation, which depresses the price of transfer of the collective farmland’s ownership because some land developers can get the land at low price if they give bribes to the officials . Some village cadres, working as agents of the collective farmland’s property rights, may also engage in rent-seeking activities as the officials do, resulting in a reduction of the expropriation price. Finally, weak organization also makes peasants’ costs for transferring some of their property rights so great that they reduce the price that the peasant can acquire.

130

Guo

The effect of the public domain on collective farmland can also be explained by economic models. In figure 3, the curve AC describes the demand of the land market, and the curves MEC, MPC and MSC respectively show marginal expropriation cost, marginal private cost and marginal society cost. When collective farmland is expropriated by the public domain, the equilibrium at point E3 can’t be realized for the cost of delineation and neither can the equilibrium at point E2 for the cost of exertion, so the final equilibrium is at point E1. As a result, the land’s owner can only obtain price P1, not P2 or P3. For a whole society, P3 should be the optimum price, because gross surplus (including consumer surplus and producer surplus) is maximized under this price, which can be represented by the area of triangle OAE3, and the quantity of expropriated farmland is Q3 in this case. However, there are many externalities related to social stabilization, ecological environment and other social functions which are not measured by the market mechanism and are part of public domain I, so the land’s potential buyer is just willing to pay P2. In this case, the society’s gross surplus is reduced by the area of triangle ME2E3 than under P3, and collective farmland will be expropriated at Q2, which is more than the socially optimum quantity. Because of public domain II, P1 is the only price peasants can get. Compared with P2, the gross surplus continues to be decreased by the area of trapezoid MNE1E2. What’s more, the quantity of expropriated farmland is much more than the socially optimum (Q2). Therefore, the public domain leads not to only lower prices, but also decreased social welfare and more expropriated farmland. P

N MSC

A

M

P3 P2

E3

MPC E2

P1

E1 O

Q3

Q2

Q1

C

MEC

Q

Figure 3: Effect of the public domain on collective land prices


131

8 Empirical data 8.1 Brief introduction of the selected county Empirical data used in this paper is from Jiangdu, a county of Jiangsu province, located in the eastern part of China. In 2003, it had a population of 1,072,154, with 23 towns and a land area of 1330.16 km2, of which 76,460.26 hm2 was farmland. Jiangdu is close to Shanghai and belongs to a developed district in China. Its gross domestic production was up to 12,093 billion yuan (RMB) in 2002, which was an increase of 12.93% compared to 2001. The author participated in a project seeking to appraise the value of collective farmlands of Jiangdu in 2003 and attained the data presented below. 8.2

Expropriation price of collective farmland (P1)

In China, the expropriation price of collective farmland is calculated by adding up all expenses occurring in the course of the farmland expropriation, consisting of the following items (Table 1): (1) Land compensation fee (X1) According to the regulations of the government of Jiangsu province, land compensation fee is 21 yuan(RMB) per m2 in Jiangdu, which is ten times the average annual output of the expropriated farmland over three years prior to expropriation. (2) Crop compensation fee (X2) This fee is to compensate for those crops which have not been harvested by the peasants. In Jiangdu, the rate of this fee is 1.87 yuan (RMB) per m2. (3) Land attachment compensation fee (X3) This fee is to compensate costs for peasants when they renew or demolish buildings and trees on the expropriated land. In Jiangdu, the present rate for this fee is 1.87 yuan (RMB) per m2. (4) Settling fee (X4) In order to not worsen the living standard of peasants’ whose contractual lands are expropriated by government, a setting fee is to be paid, which

132

Guo

varies among the towns because cultivated farmland area per peasant is different. (5) Tax of occupying cultivated land (X5) Based on the regulation of the state, the tax is 3 yuan (RMB) per m2 in Jiangdu. (6) Reclamation fee of cultivated land (X6) This fee is to protect cultivated land and seeks to ensure food security. In Jiangdu, the rate for this fee is 7 yuan (RMB) per m2. (7) Management fee of land expropriation (X7) In Jiangdu, this fee is 2 yuan (RMB) per m2. (8) Agricultural development fund (X8) The rate for this cost is 3 yuan (RMB) per m2 in Jiangdu. (9) Special fund of land reclamation (X9) The fund is calculated as 10 percent of the tax for occupying cultivated land, so it is 0.3 yuan(RMB) per m2 in Jiangdu. (10) Land management fee (X10) According to the regulations, the rate for this fee is computed as 3 percent of the sum of above fees.

133


Table 1: Calculation of expropriation prices of collective farmland in Jiangdu Town X1 X2 X3 X4 X5 X6 X7 X8 X9 Fanchuan 21 1.87 6 10.2 3 7 2 3 0.3 Yong’an 21 1.87 6 9.96 3 7 2 3 0.3 Zhenwu 21 1.87 6 10.32 3 7 2 3 0.3 Guocun 21 1.87 6 7.5 3 7 2 3 0.3 Tangtou 21 1.87 6 9.66 3 7 2 3 0.3 Xiaoji 21 1.87 6 10.08 3 7 2 3 0.3 Dinggou 21 1.87 6 9.12 3 7 2 3 0.3 Huicun 21 1.87 6 7.32 3 7 2 3 0.3 Yiling 21 1.87 6 10.62 3 7 2 3 0.3 Daqiao 21 1.87 6 9.78 3 7 2 3 0.3 Putou 21 1.87 6 8.28 3 7 2 3 0.3 Wuqiao 21 1.87 6 9 3 7 2 3 0.3 Huadang 21 1.87 6 8.4 3 7 2 3 0.3 Jiangdu 21 1.87 6 11.28 3 7 2 3 0.3 Sima 21 1.87 6 7.8 3 7 2 3 0.3 Gaoxu 21 1.87 6 8.46 3 7 2 3 0.3 Wubao 21 1.87 6 8.4 3 7 2 3 0.3 Wujian 21 1.87 6 9.96 3 7 2 3 0.3 Zhouxi 21 1.87 6 8.22 3 7 2 3 0.3 3 7 2 3 0.3 Dinghuo 21 1.87 6 11.16 Shaobai 21 1.87 6 7.92 3 7 2 3 0.3 Shuanggou 21 1.87 6 8.64 3 7 2 3 0.3 Shaoguan 21 1.87 6 7.98 3 7 2 3 0.3 Note: The unit is yuan (RMB) per m2 and the appraisal time is 30th, June, 2003.

X10 1.63 1.62 1.63 1.55 1.61 1.63 1.60 1.54 1.64 1.62 1.57 1.60 1.58 1.66 1.56 1.58 1.58 1.62 1.57 1.66 1.56 1.58 1.56

P1 56.00 55.75 56.12 53.22 55.44 55.88 54.89 53.03 56.43 55.57 54.02 54.77 54.15 57.11 53.53 54.21 54.15 55.75 53.96 56.99 53.65 54.39 53.71

8.3 Market price appraisal of collective farmland (P2) The market price of collective farmland can be appraised by the hypothetical development method. Collective farmland can’t be immediately used for building or other construction because it lacks basic requirements and is called “immature land” in China. After some land development, it is turned into “mature land”, which can then be used for construction. Therefore, the price of immature land can be evaluated by the price of mature land minus all costs entailed in the course of land development, including land development fees, management fees, investment interest, taxes, land development profits, land increment income, and so on. (1) Price of mature land (V) In Jiangdu, this price can be obtained based on urban land appraisal, as listed in table 2.

134

Guo

(2) Land development fee (C1) The land development fee is an input to make collective land into mature land, mainly through basic foundational construction and development of the plot. The rate for this fee is different among the towns of Jiangdu and is shown in table 2. (3) Investment interest (C2) Since all cost are incurred at the same time as the land price appraisal, interest can be viewed as zero. (4) Tax (C3) In Jiangdu, this is calculated as 2 percent of the land development fee. (5) Land development profit (C4) The rate of profit is surmised according to the various urban rates in Jiangdu, then the profit can be computed. (6) Land increment income (C5) In Jiangdu, the rate of this income is 10 percent. Given the above costs, the price of immature land (collective farmland) can be appraised by the hypothetical development method (table 2).

135


Table 2: Market price of collective farmland appraised by the hypothetical development method Town V C1 C3 C4 C5 Fanchuan 368 125 2.5 10 33.45 Yong’an 267 85 1.7 5.1 24.27 Zhenwu 368 125 2.5 10 33.45 Guocun 312 105 2.1 7.35 28.36 Tangtou 267 85 1.7 5.1 24.27 Xiaoji 368 125 2.5 10 33.45 Dinggou 368 125 2.5 10 33.45 Huicun 267 85 1.7 5.1 24.27 Yiling 368 125 2.5 10 33.45 Daqiao 368 125 2.5 10 33.45 Putou 267 85 1.7 5.1 24.27 Wuqiao 267 85 1.7 5.1 24.27 Huadang 267 85 1.7 5.1 24.27 Jiangdu 368 125 2.5 10 33.45 Sima 312 105 2.1 7.35 28.36 Gaoxu 267 85 1.7 5.1 24.27 Wubao 267 85 1.7 5.1 24.27 Wujian 267 85 1.7 5.1 24.27 Zhouxi 267 85 1.7 5.1 24.27 Dinghuo 312 105 2.1 7.35 28.36 Shaobai 368 125 2.5 10 33.45 Shuanggou 312 105 2.1 7.35 28.36 Shaoguan 267 85 1.7 5.1 24.27 Note: The unit is yuan (RMB) per m2 and the appraisal time is 30th, June, 2003.

P2 197.05 150.93 197.05 169.19 150.93 197.05 197.05 150.93 197.05 197.05 150.93 150.93 150.93 197.05 169.19 150.93 150.93 150.93 150.93 169.19 197.05 169.19 150.93

8.4 Comparison of two kinds of prices Actually, P2 in figure 3 is equal to the market price of collective farmland appraised by the hypothetical development method. From table 3, it can be seen that the ratio of the expropriation price to the market price of collective farmland is between 27.23 percent and 36.94 percent in Jiangdu, whose average level is 32.69 percent. Compared with the market price for collective farmland, the expropriation price is decreased by 67.31 percent on average because of the public domain. Thus we can see a remarkable effect of the public domain on collective farmland prices.

136

Guo

Table 3: Comparing the expropriation price (P1) with the market price (P2) of collective farmland Town Fanchuan Yong’an Zhenwu Guocun Tangtou Xiaoji Dinggou Huicun Yiling Daqiao Putou Wuqiao Huadang Jiangdu Sima Gaoxu Wubao Wujian Zhouxi Dinghuo Shaobai Shuanggou Shaoguan Average

P1 (yuan per m2) 56.00 55.75 56.12 53.22 55.44 55.88 54.89 53.03 56.43 55.57 54.02 54.77 54.15 57.11 53.53 54.21 54.15 55.75 53.96 56.99 53.65 54.39 53.71 54.90

P2 (yuan per m2) 197.05 150.93 197.05 169.19 150.93 197.05 197.05 150.93 197.05 197.05 150.93 150.93 150.93 197.05 169.19 150.93 150.93 150.93 150.93 169.19 197.05 169.19 150.93 170.14

P1/P2 (%) 28.42 36.94 28.48 31.46 36.74 28.36 27.86 35.14 28.64 28.20 35.79 36.29 35.88 28.98 31.64 35.92 35.88 36.94 35.75 33.68 27.23 32.15 35.59 32.69

9 Conclusion and discussion 9.1 Conclusion In China, collective farmland is not only a means of livelihood, but also a form of security for peasants. Farmland has many functions, including social security, social stabilization, ecological-environmental and other social functions. Because of the difficulty of measuring these functions accurately, part of farmland’s property rights fall into the public domain. In the course of utilization of and transacting with their farmland, peasants have to face a large transaction costs, which often leads to turning some valued properties over to the public domain. Both of these two aspects result in a reduction of collective farmland prices. According to the above analysis of the structure of collective farmland property rights in China, it is neither stable nor secure and makes the farmland user scared of making long-term investments and indifferent about engaging in sustainable use, as well as providing opportunities for local government officials and village cadres to engage in rent seeking or abuse of authority. Within the framework of the present law, peasants have to pay high costs for obtaining


137

rights and maintaining their interests. All of these aspects may lead to reduction of the price of collective farmland, which has been validated by the data of Jiangdu county of Jiangsu province, China. 9.2 Discussion For a long time, the social and ecological functions of collective farmland—especially cultivated land, grass land and forest land—have not been taken into consideration (Chen, Liu and Wang 1998; Shan, Yu and Ye 2000; Zhou, Zeng and Wang 2002). This has certainly had a negative impact on the property rights and, consequently, prices associated with such land. With the development of industrialization, urbanization, economic growth and an increasing population, the rapid reduction of collective farmland, especially cultivated land, will be a huge challenge in China. Perhaps the challenge is mainly to the security of food selfsufficiency, the environment and social conflict. Given the existing, and seemingly deliberate, institutional ambiguity, local governments have great incentives to expropriate collective farmland. It has been estimated that the share of gains taken by local governments from the expropriation and sale of collective farmland amounts to 60-70 percent, while the amount allocated to the village collective is 25-30 percent, and the amount to farm households is between 5 and 10 percent (Guo 2001). So, of all the actors involved in farmland expropriation, the ones who are dependent on farmland for a living turn out to be the least compensated. It is clear that the institution of property rights over collective farmland should be reformed to reveal its market price, protect peasants’ rights and interests, and also to aid the economic growth of China.

Acknowledgements The author would like to thank Dr. Volker Beckmann and Dr. Max Spoor for their helpful comments and advice. Special thanks also to those who helped with comments and discussion at the international conference on Economic Transition and Sustainable Agricultural Development in East and Southeast Asia, held in Nanjing, China, October 29-30, 2008, funded by European Union’s Asia-Link Programme. All errors and omissions are entirely the responsibility of the author, who can be reached at [email protected].

References Chen, F, Liu, W., Wang, T. et al. (1998). A Study of Evaluation of Agricultural Land Price in Korla. Journal of Natural Resources 13(2),162-167.

138

Guo

Demsetz, H. (1967). Towards a Theory of Property Rights. American Economic Review 57(2): 347-359. Huang, N.S., Zhu, Z.Y., Qiao, Y.L. et al.(1998). Analysis on characteristic of cultivated land area change in the course of urbanization in Pearl River Delta. Tropical Geography 18(4), 296-301. Kung, J.K. (2000). Common Property Rights and Land Reallocations in Rural China: Evidence from a Village Survey. World Development 28(4): 701-719. Li, X.B (1999). Cultivated land area change in the latest 20 years and its policy revelation in China. Journal of Natural Resources 14(4),329-333. Liang, H. (2000). Study on the social security function of rural household land in the south of Jiangsu Province. Chinese Journal of Population Science (5),32-39. Ho, P. (2001). Who Owns China’s Land? Policies, Property Rights and Deliberate Institutional Ambiguity. The China Quarterly 166(2): 394-421. Qu, F., Feng, S., Zu, P. (2004). Institution Arrangement, Price Mechanism and Farmland from Agriculture to Non-agriculture. Economics 4(1), 229-248. Shan, S., Yu, J., Ye, X. et al. (2000). Study on Appraisal Methods of Agricultural Land. Resources Science 22(1),45-49. Wang, L. (2006). Rural Land Ownership Reform in China’s Property Law. Frontiers of Law in China 1(3), 311-328. Wie, G. (2006). Problems and countermeasure faced with land intensive use. Shandong Land Resources 22(2) ：57-62. Xiao, Y., Qian, Z., Qu, F. (2005). Transaction cost, public domain property rights and infraction of peasants’ interest in the expropriation of agricultural land. Agricultural Economic Problems 9, 58-63. Guo, X. (2001). Land Expropriation and Rural Conflicts in China. The China Quarterly 166(2), 422-439. Barzel, Y. (1997). Economic Analysis of Property Rights. Second Edition. Cambridge: Cambridge University Press. Zhou, X., Zeng, L., Wang, J. (2002). Rating-Revenue Integrated Appraisal: A Synthetical Approach for Cultivated Land Appraisal in china. Resources Science 24(4),35-42. Zhu, Z.H., Pu, L.J., Peng, B.Z. et al. (2001). Cultivated land quantity change and protection countermeasure in Yangzi River Delta—A case of Wujiang city. Resources and Environment in Yangze Basin 10(4),316-322.

Part IV Agricultural Intensification: Input Use Efficiency and Sustainability

Chapter 8 Technical Efficiency of Shrimp Farms in the Mekong Delta, Vietnam Tihomir Ancev1, Md Abdus Samad Azad2, Do Thi Den3, and Michael Harris1 Abstract: Shrimp production in aquaculture systems has been playing an important economic role in Vietnam in recent years. The area for aquaculture production has been continuously expanding, especially in the provinces of the Mekong Delta. Despite this increase in production, there is some evidence suggesting that many shrimp farms are experiencing significant financial losses. Profitability can, among other things, be affected by the economic efficiency of shrimp farming operations. This chapter examines the effect of technical efficiency on the profitability and financial viability of shrimp farming, based on a sample of intensive and extensive shrimp farms surveyed in the Bac Lieu province, in the Mekong Delta of Vietnam. Technical efficiency was estimated using the stochastic production frontier approach. The results suggest that less technically efficient farms are prone to more substantial financial losses. Experience in shrimp farming and the educational level of farmers are some of the factors that were found to affect technical efficiency. Keywords: Mekong delta, shrimp farming, technical efficiency.

1 Introduction Over the last twenty years various types of aquaculture operations, including shrimp culture, mangrove farming, shrimp-mangrove farming, crab culture, crab-mangrove farming, fish farming and alternate rice-fish farming have been rapidly developing throughout South–East Asia, particularly in Vietnam. Shrimp production systems have been playing an important economic role, contributing to poverty alleviation, increased employment and foreign currency earning (Estelles et al. 2002). 1 2 3

Senior Lecturer, Agricultural and Resource Economics Group, The University of Sydney, Sydney, NSW, Australia. PhD Candidate, Agricultural and Resource Economics Group, The University of Sydney, Sydney, NSW, Australia. Researcher, Mekong Delta Development Research Institute, Can Tho University, Can Tho, Vietnam.


142

Ancev, Azad, Den, and Harris

In Vietnam, the rapid expansion of shrimp farming has resulted in a significant increase of production, ranking the country seventh in total global shrimp production (Kautsky et al. 2000). In recent years the area devoted to aquaculture in Vietnam has increased from 309,760 ha in 1991 to 865,414 ha in 2003. Of this acreage, shrimp farming covered the largest area in 2003, with 573,388 ha of shrimp production in brackish water, and 19,044 ha of other forms of shrimp production (MOFi and WB 2005). Export earnings from fish, shrimp and other seafood products in 2003 were about US $2.2 billion, with more than half of these earnings being derived from shrimp exports (MOFi and WB 2005). In the Mekong Delta, shrimp culture continues to be popular with farmers mainly because of the high potential financial return. In the last few decades the area for aquaculture has been continuously expanding throughout the provinces of the Mekong Delta and, most of all, in the provinces near the Ca Mau peninsula (Soc Trang, Bac Lieu, Ca Mau) (Figure 1).

Figure 1: Map of the provinces near the Ca Mau Peninsula, Vietnam The Mekong Delta is considered to be an especially significant aquaculture region, since it accounts for more than 60 % of the annual export value of aquatic products from Vietnam (Thanh et al. 2005). According to Sinh (2006), fisheries and aquaculture contributed 8.1% and 29.2 % to the agricultural sector’s share of the national GDP of Vietnam and to the regional GDP of the Mekong Delta in 2003, respectively. Despite the significance of shrimp farming for the national economy and the economy of the Mekong delta, recent evidence indicates that some 30% of all shrimp farms are experiencing financial losses (Sinh 2006). What factors may contribute to this unusually high incidence of loss, which is occurring in the face of the growing popularity of shrimp farming in the Mekong Delta, is one of the key questions facing the shrimp industry. The

Technical Efficiency of Shrimp Farms in the Mekong Delta, Vietnam

143

anecdotal evidence from the field suggests that many farmers are rushing into investing in intensive aquaculture operations, in the hope of quickly gaining wealth, but many of them fail after only a year or two, pressed by significant losses. A hypothesis that this chapter sets out to test is that many of these farmers lack experience, managerial skills, and commitment to the operation, and that these factors have a strong influence in determining whether the shrimp growing operation will succeed or fail financially. Testing of this hypothesis is pursued by estimating a stochastic production frontier for a sample of shrimp growing farms in Bac Lieu province of the Mekong delta and determining technical efficiency scores of each farm in the sample, which are then correlated with reported net revenues of the surveyed shrimp farming operations. Technical efficiency, as a component of economic efficiency, reflects the ability of a production unit to obtain maximum output from a given set of inputs and available technology (Farrell 1957). The other component of economic efficiency is allocative efficiency, which reflects the ability of a firm to use the inputs in optimum proportions, given their relative prices. While technical efficiency indicates the firm’s ability to use the best available practices and technology in the most effective way, allocative efficiency takes into account market signals—prices—and measures the firm’s ability to make optimal decisions on product mix and resource allocation given those prices. Combining technical and allocative efficiency provides a measure of total economic efficiency. In addition to the economic efficiency measurement, scale efficiency measures the optimality of the firm’s size. A production unit that is technically, allocatively and scale efficient, is operating at a level that maximizes profits. This chapter focuses on the first of these three conditions. The financial viability of any production enterprise depends on its longterm profitability. As technical efficiency is one of the key determinants of profitability, the major objective of this study is to establish a link between the technical efficiency scores of sampled shrimp farms in Bac Lieu province, their profitability, and consequently their financial viability. An additional objective is to estimate a model of technical inefficiency for this sample, and to determine the economic and socio-economic factors that may influence this technical efficiency. Since there are substantial differences in technology, productivity, profitability and riskiness between the extensive and intensive shrimp farming systems practiced in the Mekong Delta; an additional objective of this study is to compare the technical efficiency and financial performance of these two shrimp farming systems, as well as to compare some of the other characteristics of these two types of farms, and of the farmers that operate them, that may influence their performance.

144


These objectives were pursued by examining data collected via a survey of 193 shrimp farms in three districts of the Bac Lieu province. The districts were chosen based on two important criteria: (a) that the district has a history of conversion from rice farming to shrimp culture, and (b) that the district has experienced some level of intensification in shrimp farming. While there has been substantial recent published work on the economics of aquaculture in the context of Southern and South-Eastern Asia, some of which is reviewed below, research findings that specifically link the technical efficiency and financial viability of shrimp growing operations in Vietnam have not yet been reported. The main contribution of the current chapter lies in addressing that gap by uncovering some of the story behind the highly variable profitability of shrimp farms in the Mekong Delta and attributing financial losses to particular characteristics of farmers, such as for example experience in shrimp growing.

2 Literature review In recent years, many studies have reported estimates of technical efficiency (TE) in aquaculture of various species in the broader region of South-East Asia. Several of these studies used stochastic production frontier as a method for estimating TE for aquaculture operations. Chiang et al. (2004) conducted a study in Taiwan which specified a stochastic production frontier function to estimate milkfish farm output and efficiency. Estimates of maximum potential yield indicate that about 80% of all milkfish farms reach a technical efficiency level of 0.8. The average technical efficiency of milkfish farms was estimated at 0.82. Another study conducted in the Philippines aimed to determine farmlevel technical efficiency of tilapia growout pond operations by estimating a stochastic production frontier function. The study found the technical efficiency of tilapia growout ponds to be 83 % (Dey et al. 2000). Technical efficiency of carp production in peninsular Malaysia was studied by Iinuma et al. (1999), also using the stochastic production frontier. The mean technical efficiency of the sample carp farms was estimated to be 42 %, with the intensive farms being more technically efficient than extensive ones (Iinuma et al. 1999). A broader, comparative study was carried out for selected Asian countries by Dey et al. (2005). It presented the estimated levels and determinants of farm-level TE in freshwater pond polyculture systems in China, India, Thailand and Vietnam. The levels of country-specific TE were estimated for different production intensity levels by estimating stochastic production frontier functions, including a model of technical inefficiency.


145

Several other studies looked specifically at shrimp production. In a study conducted in India, Kumar et al. (2004) determined the TE of shrimp farming to be 69%. The average productivity of shrimp farms in Vietnam was investigated in a recent survey (Estelles et al. 2002) where it was reported that average shrimp production in extensive farming is 150 kg/ha/year, in semi-intensive farming 600–1,800 kg/ha/year and 6,000– 10,000 kg/ha/year in the case of intensive farming systems. This finding can be compared with the estimated average annual shrimp productivity in some major shrimp producing countries in the region, like Thailand (3,116 kg/ha), Malaysia (1,500 kg/ha), China (800 kg/ha), Philippines (770 kg/ha) and India (635 kg/ha) (Kumar et al. 2004). The current chapter builds on this published work, adding to it by providing specific focus on the Mekong Delta in Vietnam. In addition, the chapter contributes to the literature by establishing a link between poor financial performance and the technical efficiency of shrimp farms.

3 Theoretical framework Farrell (1957) was the first to establish the possibility of estimating frontier production functions, which have been applied widely in the last three decades, especially following the studies of Aigner et al. (1977) and Greene (1980). Frontier functions include stochastic frontier production functions, frontier cost functions and frontier profit functions. The stochastic frontier production function was independently proposed by Aigner et al. (1977) and Meeusen and van den Broeck (1977). Its main application has been in estimating and calculating TE of various production processes. The method allows output to be specified as a function of controllable factors of production, random noise and a technical inefficiency (TI) term. The technical inefficiency error term has two components: one to account for random effects (e.g. measurement errors in the output variable, weather conditions, diseases, and the combined effects of unobserved/uncontrollable inputs on production) and another to account for pure technical inefficiency in production (Coelli et al. 1998)). The stochastic frontier production function can be specified as: Yi = f (Xi; β ) exp (Vi – Ui)

i = 1, 2…N

(1)

where Yi is the output of the ith farm; Xi is a vector of inputs used by the ith farm; β is a parameter vector; Vi is a random variable representing pure random error, assumed to be normally, independently and identically distributed with a mean of zero and variance of σv2. Ui is a random variable independent of Vi that represents technical inefficiency in production and is

146


assumed to follow a truncated (at zero) normal distribution with mean, µi, and variance, σu2, such that, µi = Zi δ

(2)

Zi is a p x 1 vector of farm-specific variables that may cause inefficiency and δ is a 1 x p parameter vector. The stochastic production frontier at a technically efficient farm represents the maximum attainable output (Yi*) such that: Yi* = f (Xi; β) exp (Vi )

(3)

This can be used to measure the technical efficiency of all other farms, relative to this efficient farm. The technical efficiency of the ith farm (TEi) is given by: TEi =

Yi = exp (– Ui) Yi *

(4)

where TE may be defined as the capacity of a producer to produce a maximum output from a certain amount of input and using the available technology. Estimation of the stochastic production frontier function may be viewed as a variance decomposition model, which can be expressed as: σ 2 = σ 2v + σ u2

(5)

γ = σ u2 /( σ 2v + σ u2 )

(6)

where γ is the variance ratio parameter which relates the variability of Ui to total variability ( σ 2 ). The value of γ lies in the range 0 ≤ γ ≤ 1 (Battese and Corra 1977). If the value of γ equals zero, it indicates that the difference between an individual unit’s performance relative to the efficient performance (a point on the frontier) is entirely due to statistical noise. On the other hand, a value of one indicates that the difference can be entirely attributed to technical inefficiency (Coelli 1995). The relationship between technical efficiency and profitability has been investigated in the theory of productivity analysis since its early days (Farrell 1957; Nerlove 1965). Theoretically, technical efficiency is a necessary but not a sufficient condition for overall economic efficiency and, hence, maximum profitability. The additional conditions of allocative and scale efficiency are necessary for profit maximization, as technical efficiency alone cannot guarantee profit maximization, but it is an important condition for it (Cherchye et al. 2000). While the other components of overall efficiency (allocative and scale) can potentially


147

compensate for low technical efficiency, it is likely that a technically inefficient production enterprise will eventually experience low or negative profitability, which will endanger its financial viability.

4 Data and method The data used in this study were obtained from a survey conducted in the Bac Lieu province in 2004. The survey was administered by interviewing the heads of 212 households from three districts of the province: Gia Rai, Dong Hai, and Vinh Loi. The respondents in each of the districts were chosen by random sampling from a list of residents in the districts. The design, sampling procedure and survey implementation were conducted by the Mekong Delta Development Research Institute (MDDRI). Data on 193 households were retained and are analyzed in this chapter, while the data on the remaining 19 households were of poor quality and were discarded from the sample. Data were collected on a number of characteristics, including basic household information (household size; age, education, and occupation of the household head), land use and assets (number and size of land parcels, machinery, animals), and on main farming practices (parcel level farming, inputs, yields, costs and prices). Collected data on shrimp farming were further classified according to the type of the system. In Vietnam, as in other South-Eastern Asian countries, shrimp farming can be classified in one of the four types of farming systems: extensive, improved extensive, semi-intensive and intensive (Rashid and Chen 2002). The key difference between these farming systems is the intensity of input use. While the extensive system uses very little purchased inputs and relies on available natural resources (sea water, land etc.), the intensive shrimp farming system employs more labor, and relies heavily on purchased inputs such as seed and feed. The difference in quantity of input use between these two farming systems is given in Table 3. Since the technology and the intensity of input use is quite different between these shrimp farming systems, the data were classified into two categories: one called “intensive”, which included semi-intensive and intensive farming systems, and another called “extensive”, which included extensive and improved extensive farming systems. The farms of the latter type were dominant in the sample, with 163 surveyed households falling into the “extensive” category, with only 30 falling into the “intensive”. This proportion of extensive and intensive farms is representative of the actual situation in the region, where typically one intensive or semi-intensive farm can be found for every 4 extensive farms (Shang et al. 1998).

148


4.1 Socioeconomic characteristics of shrimp farmers Data were collected on the following socioeconomics characteristics of the surveyed respondents: household size; farm size; age, education and experience in growing shrimp of the household head. In general, there was not much difference in the average number of completed years of education of the head of household operating one of the two types of shrimp farms, but the operators of intensive farms reported a slightly higher average level of education (6.8 years), compared to operators of extensive farms (5.7 years) (Table 1). It can be hypothesized that, since intensive farming requires more technical knowledge, people with relatively lower levels of education would be hesitant to adopt this system and, hence, the finding that intensive shrimp farming operations tend to be operated by relatively better educated farmers. Likewise, organization of labor may affect the level of conversion from extensive into intensive shrimp farming systems. It has been reported that the adoption rate of a technology is lower in farms where the decision maker has higher opportunity cost of labor as a result of better education, and where relatively cheap labor can be employed (Beckmann and Wesseler, 2003). The household size of most of the intensive shrimp farmers was found to be smaller than that of extensive shrimp farmers. Data in Table 1 show that 10% of the intensive shrimp farming households had a family size of less than two members, while only 3% of the extensive shrimp farming household fell into this category. On the other hand, 34% of the extensive shrimp farming households had more than six members, while only 27% of the intensive shrimp farming families were in this category. This finding is consistent with the tendency of the operators of an intensive shrimp system to be less dependent on family labor compared to the extensive system, and the former’s strong tendency to hire labor. The average age of both intensive and extensive shrimp farmers was found to be around 50 years. Average farm area for intensive and extensive shrimp farming was found to be 1.2 ha and 1.74 ha respectively. However, 50% of intensive farms and 29% of extensive farms reported having farms of less than one hectare. The average number of years of experience in shrimp farming was reported to be 3.8 and 5.3 years, respectively, for intensive and extensive shrimp farmers, whereas 63% of intensive operations and 46% of extensive operations had less than four years shrimp farming experience (Table 1). This indicates that intensive shrimp operations are both smaller in area than the extensive operations and are operated by less experienced farmers.

149


Table 1: Socio-economic profile of the surveyed shrimp farmers in the Bac Lieu province, Mekong Delta, Vietnam, 2004 Item

Intensive (n=30)

Extensive (n=163)

All (n=193)

Head of household education (years of schooling completed) 1-5 years 6-9 years 10-12 years College

47 43 10 00

63 30 06 01

61 32 07 01

Average head of household education (years of schooling completed)

6.8

5.7

5.9

Household size (number of household members) 6 members

% of respondents

Percentage (%) of Respondents 10 13 50 27

03 10 53 34

04 11 52 33

Average household size (number of household members)

4.7

4.9

4.9

Head of household age (number of years)

Percentage (%) of Respondents

60 years

03 20 37 27 13

06 18 34 22 20

05 18 35 23 19

49.5

49.7

49.6

Average age ( number of years) Distribution of farm size

Percentage (%) of farms

2.0 ha

50 33 17

29 37 34

32 37 31

Average farm size (ha)

1.2

1.7

1.7

Experience in shrimp farming

Percentage (%) of Respondents

4 years

63 30 07

46 36 18

49 35 16

Average shrimp farming experience (number of years)

3.8

5.3

5.1

4.2 Economic characteristics of shrimp farms Table 2 presents key economic data for the two shrimp farming types: intensive and extensive systems.

150


Table 2: Productivity, income and costs of the surveyed shrimp farms in the Bac Lieu province, Mekong Delta, Vietnam, 2004 Intensive Average shrimp production (kg/ha)

Extensive

1570

203

134,205

17,885

Average net revenue (1000VND/ha)

51,532

6,719

Average cash cost (1000VND/ha)

74,494

8,018

Average gross revenue (1000VND/ha)

The findings indicate that average net revenue reported for the intensive farming system is much higher than that reported for the extensive shrimp farming. The reported figures were about 60 million VND and 10 million VND per hectare, respectively.4 Shrimp yield was also found to be nearly eight times greater in the case of intensive farming. However, it was found that the production costs for intensive farming is nine times greater than the costs involved in extensive shrimp farming. In terms of input use, Table 3 shows that intensive operations use much greater quantities of all relevant inputs compared to extensive farms. This translates into substantially higher cost for inputs incurred by intensive farms. While intensive shrimp farming is on average more profitable when compared to extensive shrimp farming, over one third of farms in both categories were unprofitable (Table 4). This is consistent with some previous findings, as reported by Sinh (2006). Very high positive net revenues (greater than 150 million VND) were reported by 14% of the respondents for intensive shrimp farming. In the mid-range, 33% of the intensive shrimp farms and 61% of extensive farms reported positive net revenues of less than 50 million VND. At the very extreme end of financial loss, 7% of intensive farms, but only 3% of extensive farms, reported a financial loss of more than 20 million VND (Table 4). The data in Table 4 suggest, on one hand, that there is significant opportunity for generating higher profits by engaging in intensive shrimp farming, while, on the other hand, there is an increased possibility of heavy loss which can occur in intensive shrimp farming, reflecting the greater riskiness of this type of operation.

4

1000 VND (Vietnamese Dong) = 0.063 USD, 0.047 EUR, 0.081 AUD, in December 2004, at the time when the survey was administered.

151


Table 3: Input use and costs of the surveyed shrimp farms in the Bac Lieu province, Mekong Delta, Vietnam, 2004 Intensive

Extensive

Seed quantity (1000 postlarvae) Seed cost (1000VND)

152 10,704

109 4,690

Feed quantity (kilograms) Feed cost (1000VND) Fuel quantity (liters) Fuel cost (1000VND) Hired labor quantity (days) Hired labor cost (1000VND)

2420 40,622 1224 6,987 83 3,677

06 346 153 922 24 959

Family labor quantity (days) Family labor cost (1000VND)

273 8,179

98 2929

16,714

1,727

Other costs

Table 4: Distribution of net revenues from shrimp farming by type of farming system of the surveyed farms in Bac Lieu province, Mekong Delta, Vietnam, 2004 Net revenue (1000 VND/household)

Intensive

Extensive

(n=30)

(n=163)

Percentage (%) of Respondents Negative

37

37

> -20000

07

03

>=-20000 to -15000

00

02

>-15000 to -10000

13

05

>-10001 to -5000

17

10

=50000 to 100000

16

01

>=100000 to 150000

00

01

>=150000 to 200000

07

00

> 200000

07

00

4.3 Method To determine the technical efficiency (TE) of surveyed shrimp operations, a two-step procedure was followed. In the first step, a stochastic frontier

152


model was specified to estimate TE for each shrimp farming operation. In the second step, TE scores were used as the dependent variable and were regressed on variables representing individual characteristics of the surveyed households. For the purpose of the analysis, stochastic production frontier functions were fitted through the survey data. There are several functional forms that have been used in the literature to measure the physical relationship between inputs and outputs. One of the most common forms is the CobbDouglas function (Nerlove 1965). The Cobb-Douglas model was specified as: Log(Yi) = β0 + β1 log(Si) + β2 log(Fi) + β3 log(Li) + Vi - Ui,

(7)

where Yi is the reported shrimp production of farm i (kg/ha/year), Si is seed (1000 postlarvae /ha), F is feed (kg/ha), L is labor (days/ha), and Vi and Ui are error terms as described previously. This model was run separately for the intensive and for the extensive shrimp farms. In effect, this meant that the data were grouped into two separate samples. There were several reasons for this. One was that, in terms of technology, the intensive and extensive systems are so different that estimating from the data in a single sample would greatly distort the results. This suspicion was confirmed by the preliminary testing conducted. While initially a model with an intercept and slope shift parameters using the whole data sample was estimated (not reported here), it only provided estimates for the coefficients on individual inputs used in extensive and intensive farms, but it did not allow for separate measures of technical efficiency for the two farming systems to be estimated. This meant that the estimated stochastic production frontier was a surrogate of the actual frontiers for the two farming systems and was not representing the relevant reference points for either one. Another reason for estimating separate stochastic production frontiers for intensive and extensive shrimp operations was that this enabled estimation of the differences in technical efficiency between the intensive and extensive systems. In addition, the data quality for the labor input was very different across the systems. While data on both hired and household labor for the extensive systems were of acceptable quality (no outliers or obviously implausible values), household labor data for the intensive operations were plagued with problems. Given that the sample size for the intensive operations was fairly small in the first place (only 30 intensive farms), there was no possibility to discard the observations with erroneous household labor records. Hence, a decision was made to only include hired labor as a dependent variable for the stochastic production frontier estimation for intensive farms, while total labor was used in the estimation


153

of the frontier for extensive farms. To establish a link between the estimated technical efficiency scores and the profitability of the farms, a correlation coefficient between the technical efficiency score and the reported net income from shrimp farming was estimated using spreadsheet software. Subsequent to the estimation of a technical efficiency model for extensive and intensive shrimp farms, a technical inefficiency model for each of the two types of farming systems was specified as follows: Ui = α0 + α1 log(Ai) + α2 log(Ei) + α3 log(Edi),

(8)

where Ui is the mean farm-specific technical inefficiency, Ai is area under shrimp farming (ha), Ei is experience of the head of household (years), and Edi is the level of attained education of the household head (years of schooling completed). These variables were included in the above model because they were theoretically expected to have an effect on the level of technical efficiency and because their significance was indicated in the initial stages of estimating the model. The maximum likelihood estimates (MLE) of the parameters specified in the models above were obtained using FRONTIER version 4.1 (Coelli 1996).

5 Results Parameter estimates for the stochastic production function model and the technical inefficiency model for the sample of intensive shrimp farms are given in Table 5. The results show that “feed” is the most influential input variable affecting shrimp production. This is expected and intuitively justified. The input variables “labor” and “seed” are not significant, even though they are close to being significant at the 10% level, suggesting that they are also important determinants of shrimp productivity in intensive farming systems in the Mekong Delta. The estimated technical efficiency (TE) of the sampled intensive shrimp farms was 71.3 %, which indicates relatively high efficiency in using inputs in this type of operation. The coefficient estimates from the technical inefficiency (TI) model for intensive farms are presented in the bottom part of Table 5. All estimated parameters in this model are statistically significant. The variable “shrimp area” is significant and positive, with the interpretation that larger intensive shrimp operations tend to be less efficient. The reason for this may be that operators of larger farms are often unable to manage their shrimp operations properly, due to the managerial complexity of intensive operations. Attempting to manage this complex system at a larger scale results in less efficient use of inputs and, hence, a low technical efficiency

154


score. The effect of the variable “experience” is significant and negative, suggesting that operations operated by more experienced farmers tend to be more efficient. This is a significant but expected finding, having in mind the managerial skills required for operating an intensive shrimp farming operation, which are enhanced through experience.

Table 5: Maximum-likelihood estimates of the stochastic CobbDouglas production frontier function and technical inefficiency model for the surveyed intensive shrimp farms in the Bac Lieu province, MD, Vietnam Coefficient

Standard error

T-ratio

Constant

1.256

1.120

1.05

Log (Seed)

0.506

0.340

1.49

Log (Feed)

0.464***

0.106

4.38

0.064

0.049

1.30

Stochastic frontier

Log (Hired labor) Log- likelihood value

-37.21

Mean technical efficiency index

71.29%

Technical inefficiency model Constant

-14.02*

7.146

-1.96

Total shrimp area (ha)

2.238**

1.094

2.04

Shrimp experience (no. of years)

-0.439*

0.225

-1.96

Education (no. of years)

1.084**

0.501

2.16

2.787***

0.765

3.645

0.752***

0.134

5.61

Variance parameters

σ2 γ

Note: *significant at 10%, **significant at 5%; ***significant at 1%.

Results from the technical inefficiency model also reveal that farmers that have higher levels of education operate less technically efficient shrimp farms than less-educated farmers. At the intuitive level, this is a surprising result, but similar results were reported in previous work (Tham and Fleming 2004; Kalirajan and Shand 1985). One explanation may be that better-educated farmers have a greater tendency to be engaged in off-farm employment; empirical evidence in support of this hypothesis has been reported elsewhere (van de Walle and Cratty 2004). In such circumstances, it may be expected that relatively better educated shrimp farmers spend


155

more time in off-farm employment, which limits the amount of time that they can devote to managing their farm. To establish whether such a hypothesis may hold for this data set, the correlation coefficient between education level and off-farm employment was estimated. Its value was found to be insignificantly different from zero for both extensive and intensive shrimp farms, which renders the hypothesis invalid. This indicates that reasons other than off-farm employment might be influential in explaining relatively lower technical efficiency of farms operated by more educated farmers. The results for the estimated stochastic frontier production function and the technical inefficiency model for the extensive shrimp farms are presented in Table 6.

Table 6:

Maximum-likelihood estimates of the stochastic CobbDouglas production frontier function and technical inefficiency model for the surveyed extensive shrimp farms in the Bac Lieu province, MD, Vietnam. Coefficient

Standard error

T-ratio

4.680***

0.542

8.63

Log (Seed)

0.089

0.094

0.95

Log (Feed)

0.035**

0.015

2.25

Log (Total labor)

0.204**

0.089

2.30

Stochastic frontier Constant

Log- likelihood value

-236.17

Mean technical efficiency index

49.58%

Technical inefficiency model Constant Total shrimp area (ha) Shrimp experience (no. of years) Education (no. of years)

-6.748*

4.056

-1.66

-0.791

0.490

-1.62

-0.574**

0.293

-1.96

0.618**

0.278

2.22

9.383**

3.892

2.41

0.965***

0.019

51.28

Variance parameters

σ2 γ

Note: *significant at 10%, **significant at 5%; ***significant at 1%.

For the production function estimates, both variables “feed” and “labor” were found positive and significant. The estimate of the coefficient on “seed” is not significant, even though it is close to being significant at 10%,

156


indicating that improving shrimp culture by adequately supplying postlarvae is an important factor in improving productivity of shrimp farms. The TE of the sampled extensive shrimp farms was estimated at 49.6 %, which is quite low, but not unexpected considering that a key characteristic of extensive shrimp farming is that it does not use many inputs, and the ones that are used are only provided in minimal amounts. These findings can be compared to the findings reported elsewhere. Iinuma et al. (1999) also found low levels of technical efficiency in their study, with an estimated mean technical efficiency for carp farms of 42%. In contrast, Rashid and Chen (2002) estimated the technical efficiency for the improved extensive shrimp farms at 61%, which was considerably higher than the average technical efficiency score for extensive farms (which include improved extensive) in the current study. The coefficient estimates from the technical inefficiency (TI) model for extensive shrimp farms are presented in the bottom part of Table 6. The results indicate that the area of the extensive shrimp farms is not significant in explaining TI. This is expected, since extensive farms are relatively large and the variability of the farm size in the sample was relatively low. However, both “experience” and “education” are significant and with effects that are consistent with the technical inefficiency model for intensive farms. Extensive shrimp farms that are operated by more experienced farmers tend to be more efficient, while more educated farmers tend to operate less efficient extensive shrimp farm operations. Possible explanations for this are similar to those offered for intensive shrimp farms.

6 Summary and conclusion In this chapter, data from a survey of shrimp growers in the Bac Lieu province in Vietnam’s Mekong Delta were used to estimate the technical efficiency of shrimp farming operations. This was carried out within the framework of stochastic production functions. The main aim of the study was to test whether low technical efficiency coincides with negative financial results. For this purpose, correlation coefficients between the estimated technical efficiency scores and reported net income were estimated as 0.55 for extensive farms and 0.28 for intensive farms. These values indicate a relatively strong positive relationship between the technical efficiency of the surveyed shrimp farms and their profitability. In addition, to discern the effect that alternative farming systems have on technical efficiency, the sampled shrimp farms were grouped into two groups: extensive and intensive farms. The mean level of technical efficiency for the sample of intensive shrimp farms was estimated to be 71.3 %. The input of feed was found to be most influential in determining the productivity of this type of farm, while seed and labor were found less


157

influential. The mean level of technical efficiency for extensive farms was estimated at 49.6 %, which is fairly low. The key determinants of productivity in the extensive farms were identified to be feed and labor. The technical inefficiency of shrimp farming from the sampled farms may be explained by several factors. One of them is the lack of experience of the shrimp growers. Shrimp farming in Vietnam is still a relatively new activity, and it may be expected that its technical efficiency will improve with time, as operators become more experienced. Shrimp farming continues to be one of the more profitable and, thus, popular farming activities in Vietnam. Profitability has however been variable, especially for more intensive shrimp farming systems, which can be both highly profitable but are also susceptible to substantial losses. It is possible that some farmers, after having observed the success of their longstanding shrimp producing peers, invest in intensive shrimp operations with a hope for quick profit. However, they are immediately confronted with the high managerial requirements for this type of farming, for which they lack experience, and are under continuing pressure to meet often heavy repayments on borrowed funds. It is suspected that these factors are at least partly contributing to the high rate of economic failure of newly established intensive shrimp farming operations, which has been consistently the case in the Mekong Delta in recent years. The results from this study, however, do suggest that the technical efficiency of shrimp farms in Vietnam is still not at par with similar operations in South-East Asia, and this may be one of the other reasons for the instability of profits. More efficient use of inputs, such as feed, seed and labor may in the future be the key for lifting productivity and hence profitability of shrimp farming in Vietnam. In addition, such improved use of inputs and general level of management may lift the environmental performance of the shrimp farms, which has been a perennial concern.

Acknowledgements The authors would like to thank the editors of this volume for valuable comments and suggestions on this chapter. Assistance by Deepa Pradhan in preparing this chapter is also greatly appreciated. The authors, however, take responsibility for all remaining errors. The research work reported here was generously supported by an AusAID CARD Project 025/05 VIE.

References Aigner, D., Lovell, C.A.K., Schmidt, P. (1977). Formulation and Estimation of Stochastic Frontier Production Function Models. Journal of Econometrics 6 (1): 21-37.

158


Battese, G.E., Coelli, T.J. (1995). A Model for Technical Inefficiency Effects in a Stochastic Frontier Production Function for Panel Data. Empirical Economics 20:325-332. Battese, G.E., Corra, G.S. (1977). Estimation of a Frontier Model: with Application to the Pastoral Zone of Eastern Australia. Australian Journal of Agricultural Economics 21: 167-179. Beckmann, V., Wesseler, J. (2003). How Labour Organization May Affect Technology Adoption: an Analytical Framework Analysis the Case of Integrated Pest Management. Environment and Development Economics 8: 437-450. Cherchye, L., Kuosmanen, T. , Post, T. (2000). What is the Economic Meaning of FDH? A reply to Thrall. Journal of Productivity Analysis 13: 263-267. Chiang, F., Sun, C., Yu, J. (2004). Technical Efficiency Analysis of Milkfish (Chanos chanos) Production in Taiwan: An Application of the Stochastic Frontier Production Function. Aquaculture 230: 99-116. Coelli, T.J. (1996). A Guide to FRONTIER Version 4.1: a Computer Program for Stochastic Frontier Production and Cost Function Estimation. CEPA Working Paper 96/7, Department of Econometrics, University of New England, Armidale NSW Australia. Coelli, T.J. (1995). Recent Developments in Frontier Modelling and Efficiency Measurement. Australian Journal of Agricultural Economics 39 (3): 219-245. Coelli, T., Prasada Rao, D.S., Batese, G. E. (1998). An Introduction to Efficiency and Productivity Analysis. Kluwer Academic Publishers, Boston. Dey, M.M., Paraguas, F.J., Bimbao, G.B., Regaspi, P.B. (2000). Technical Efficiency of Tilapia Growout Pond Operations in the Philippines. Aquaculture Economics and Management 4: 33-47. Dey, M.M., Paraguas, F.J., Srichantuk, N., Xinhua, Y., Bhatta, R., Dung, L.T.C. (2005). Technical Efficiency of Freshwater Pond Polyculture Production in Selected Asian Countries: Estimation and Implication. Aquaculture Economics and Management 9: 39-63. Estelles, P., Jensen, H., Sanchez, L. (2002). Sustainable Development in the Mekong Delta. Centre for Environmental Studies, University of Aarhus. Farrell, M.J. (1957). The Measurement of Productive Efficiency. Journal of Royal Statistical Society 120(3): 253-290. Fleming E., Farrell, T., Villano, R., Fleming P. (2006). Is Farm Benchmarking the New Acceptable Face of Comparative Analysis? Australasian Agribusiness Review, 14, paper 12. Greene, W.H. (1980). Maximum Likelihood Estimation of Econometric Frontier Functions. Journal of Econometrics 13: 27-56. Iinuma, M., Sharma, R., Leung, P. (1999). Technical Efficiency of Carp Culture in Peninsula Malaysia: an Application to Stochastic Production Frontier and Technical Inefficiency Model. Aquaculture 175: 199-213. Kalirajan, K.P., Shand, R.T. (1985). Type of Education and Agricultural Productivity: a Quantitative Analysis of Tamil Rice Farming. Journal of Developing Studies 21: 222-243.


159

Kautsky, N., Ronnback, P., Tedengren, M., Troell, M. (2000). Ecosystem Perspectives on Management of Disease in Shrimp Pond Farming. Aquaculture 191: 145161. Kumar, A., Birthal, P.S. (2004). Technical Efficiency in Shrimp Farming in India: Estimation and Implications. Indian Journal of Agricultural Economics 59(3): 413-420. Meeusen, W., van den Broeck, J. (1977). Efficiency Estimation from Cobb Douglas Production Function with Composed Error. International Economic Review 18: 435-444. Ministry of Fisheries and the World Bank (MOFi and WB) (2005). Vietnam Fisheries and Aquaculture Sector Study. Final Report, Accessed December 2007, available at: http://siteresources.worldbank.org/INTVIETNAM/Resources/ vn_fisheries- report-final.pdf Nerlove, M. (1965). Estimation and Identification of Cobb-Douglas Production Functions. Princeton: Princeton University Press. Rashid, M.H., Chen, J.R. (2002). Technical Efficiency of Shrimp Farmers in Bangladesh. Bangladesh Journal of Agricultural Economics 25 (2): 15-31. Shang, Y.C., Leung, P., Ling, B.H. (1998). Comparative Economics of Shrimp Farming in Asia. Aquaculture 164: 183-200. Sinh, L.X. (2006). Major Considerations on the Fishery Sector in the Mekong Delta of Vietnam. Presentation to the CARD workshop, July 15-20, 2006, Can Tho Tham, M., Fleming, E. (2004). An Analysis of Scope Economies and Specialization Efficiencies among Thai Shrimp and Rice Smallholders. Working paper 200410, School of Economics, University of New England. Thanh, P.N., Son, V.N. (2005). River Pen Culture of Giant Freshwater Prawn Macrobrachium rosenbergii (De Man) in Southern Vietnam. Aquaculture Research 36: 284-291. van de Walle, D., Cratty, D. (2004). Is the Emerging Non-farm Market Economy the Route out of Poverty in Vietnam? The Economics of Transition 12 (2): 237– 274.

Chapter 9 Understanding Environmental and Social Efficiencies in Indonesian Rice Production Joko Mariyono1, Budy P. Resosudarmo2, Tom Kompas3, and Quentin Grafton4 Abstract: Intensive agricultural productions worldwide are associated with significant use of inputs that are detrimental to the environment. Indonesia is no exception. Its use of agrochemicals in rice production has increased considerably since the early 1970s, particularly in irrigated rice production areas, and has been claimed to cause significant environmental destruction. This study aims to examine rice agriculture efficiencies related to the use of agrochemicals at the farm level in Indonesia. The measures cover environmental and social efficiencies. The results indicate that in general rice production in Indonesia has a low environmental efficiency, leading to chemical wastes. Rice farmers in Java tend to be more efficient than those in other areas. Large-scale farms tend to waste more than small-scale farms. Not taking into account the environmental costs of chemical pollution exacerbates the overuse of chemicals and reduces the utilisation of other inputs. Limited land availability and the need to absorb abundant rural labour have been the main constraints on Indonesian rice farmers aiming to be socially efficient. Keywords: Environmentally detrimental inputs, Agrochemical waste, Production economics, Agricultural economics

1 Introduction The Green Revolution that started in the early 1960s certainly brought with it both blessings and curses. On one hand, it was a blessing since intensive 1

2

3

4

Research Associate for Socio-economics, AVRDC – The World Vegetable Centre, Taiwan. Associate Professor, The Arndt-Corden Division of Economics, Crawford School of Economics and Government, The Australian National University, Canberra, Australia. Professor, Crawford School of Economics and Government, The Australian National University, Australia Professor, Crawford School of Economics and Government, The Australian National University, Canberra, Australia


162

Mariyono, Resosudarmo, Kompas, and Grafton

use of high-yielding varieties and agrochemicals significantly improved the world’s food production, avoiding the world-wide famine anticipated in 1970s–1980s. Yet, at the same time, it was also a curse since the intensive use of agrochemicals has caused serious environmental and health problems worldwide. Evidence indicates that agricultural practices using intensive agrochemical inputs indeed lead to non-point-source pollution, thus inducing high external costs. For example, Houndekon and de Groote (1998) report that the external cost of controlling migratory locust pests during 1992–1996 in Nigeria was around US$ 417 thousand, the value of livestock poisoned by insecticides. In Thailand, Jungbluth (1999) reports that the external costs of agrochemical use in 1992 reached about US$ 43 million, the market value of chemical-contaminated vegetables and fruits. In terms of health costs, each Pilipino farmer must spend approximately an extra US$ 24 for recovering from health problems associated with each kg of pesticide application (Rola and Pingali 1993). In the1980s, concern over the impact of agrochemical use on the environment intensified, and many scientists published works concerning the sustainability of intensive agriculture, examples being Barbier (1989) and Conway and Barbier (1990). Nowadays, demand for a clean and healthy environment is greater than it has ever been. Various organisations have lobbied for the development of environmental regulations and policies to improve the quality of environmental conditions, including more environmentally friendly agricultural practices. Indonesia has also been involved with the Green Revolution since the early 1970s. At that time, the Indonesian government decided to develop a food-production intensification program, mainly in terms of rice, as its top priority. This program included the large-scale adoption of high-yielding modern seed varieties, development of irrigation systems, expansion of food-crop producing areas, increased use of chemical fertilisers and pesticides, expansion of agricultural extension services, establishment of farmer cooperatives and input subsidies, and stabilisation of national foodcrop prices (Resosudarmo and Thorbecke 1998). During the 1970s and 1980s, this food intensification program caused food-crop production to grow at an annual rate of approximately 3.7 percent (BPS 1973–1991). A major miracle occurred in rice production. Pushing the average annual growth rate of rice production to approximately 4.7 percent, the rice intensification program transformed Indonesia from the world’s largest importer of rice, importing approximately two million tons per year by the end of the 1970s, to self-sufficiency in 1983 (Resosudarmo and Thorbecke 1998). Note, however, that since the mid 1990s Indonesia has again had to import rice to fulfil national demand.

Environmental and Social Efficiencies in Indonesian Rice Production

163

Despite the remarkable success of the food intensification program, the excessive use of agrochemicals, particularly in rice production, caused serious environmental problems, such as acute and chronic human pesticide poisoning, animal poisoning and contaminated agricultural products, destruction of beneficial natural parasites and pest predators, and pesticide resistance in pests (Pimentel, Aquay, Biltonen, Rice, Silva, Nelson, Lipner, Giordano, Horowitz, and D’Amore 1992; Antle and Pingali 1994; Resosudarmo 2001). Research on how efficient rice production is in Indonesia when using agrochemical inputs has been relatively limited (Roche 1994). As yet no research has been conducted to identify efficiency differences in using agrochemical inputs among regions or among different types of rice farmers. But acquiring such knowledge is important in order to be able to develop policies targeting certain regions or types of farmer to improve their efficiency. Using a stochastic production frontier model, this paper aims, first, to examine the environmental efficiency of rice farmers in Indonesia in using agrochemicals. In general, environmental efficiency is regarded as the ratio of minimum amount of feasible environmentally detrimental inputs to the actual amount of those inputs used. Second, we intend to estimate the environmental costs of chemical waste in rice production due to inefficient use of chemical inputs. Third, the paper examines the social efficiency (or inefficiency) of rice farmers due to their inefficient use of chemicals inputs. Social efficiency, in general, measures how far producers are from an optimal situation where they reach maximum profits at which external costs due to inefficiency of chemical use have been internalised into production costs. Here, chemical inputs are considered to be environmentally detrimental materials. The outline of this paper is as follows. The introduction is followed by a brief literature review section, summarising the existing literature related to measuring efficiency in agricultural production. Then a theoretical framework section is followed by an implementation section which describes the practical application of the theoretical framework regarding the case of rice production in Indonesia. The data section presents the source and details of the available data. A results and discussion section follows and, finally, conclusions.

2 Brief literature review Although there have been many studies worldwide on agricultural efficiency, most focus on technical efficiency (Battese and Tessema 1993; Kumbhakar 1994; Bravo-Ureta and Evenson 1994; Ahmad and BravoUreta 1996; Tadesse and Krishnamoorthy 1997; Amaza and Olayemi

164


2002), which in general is concerned with an efficient allocation among inputs utilised (Widodo 1989).5 Since the early 1990’s, concern over intensive agriculture that uses environmentally detrimental inputs has been raised by several authors, for example Byerlee (1992), Ali and Byerlee (2002), Apel, Paudyal, and Richter, (2002), Pujara and Khanal (2002), Singh (2002), Schumann (2002) and Toryanikova, Karaseva, Lutsenko, and Nishonov (2002). However, there is still only limited discussion regarding agricultural efficiency that takes environmental impacts into account, that is, environmental efficiency in agriculture. In general, there are two approaches to estimating environmental efficiency. The first uses the method of data envelopment analysis to estimate a deterministic production frontier. Environmental efficiency is then derived from the production frontier. The second approach uses an econometric method to estimate a stochastic production frontier, which is then used to derive environmental efficiency (Reinhard, Lovell, and Thijssen 1999 and 2000). In deriving environmental efficiency using an estimated production frontier, either approach typically uses a flexible transcendental logarithmic (translog) function, which needs to be monotonic, guaranteeing that there is a unique solution in which the observable output can be feasibly produced with a minimum level of inputs at the frontier. 6 If the input is environmentally detrimental, the ratio of the minimal level to the observable level will represent the rate of environmental efficiency. An example of work in this area is by Hadri and Whittaker (1999), who analysed the relationship between technical efficiency of dairy farms in England and environmental pollution related to agrochemicals that are potential environmental contaminants. The study used a stochastic production frontier to estimate technical efficiency, which is dependent on some farmer characteristics and the use of chemical inputs. The important outcome of this study was in showing that more efficient farms use more chemical inputs. Another example is the work of Reinhard, Lovell, and Thijssen. (2002), who used a stochastic frontier method to estimate the environmental efficiency of Dutch dairy farms and examine the sources of variation in environmental inefficiency. A two-step estimation method was used to account for some producer characteristics assumed to have impacts on environmental efficiency. In the first step, environmental efficiency was calculated from a production frontier. In the second step, environmental 5 6

Other literature also refers to this as allocative efficiency. Monotonic production function means that the producer is assumed to operate the firm under conditions of production increasing at a decreasing rate.


165

efficiency was regressed with the producer characteristics hypothesised to have strong relationships to environmental performance. A more recent study on environmental efficiency was conducted by Gang and Felmingham (2004). Using a stochastic frontier method they calculated the potential reduction of environmentally detrimental material. The potential reduction can be regarded as chemical waste. As mentioned before, the work in this area with regard to Indonesia is rather limited. One of the very few works is that conducted by Roche (1994) on rice production. He shows that the use of Nitrogenous fertilisers determines technical efficiency and argues that Javanese farmers tend to overuse fertilisers, in contrast to those off Java. However, Roche did not calculate the environmental efficiency of rice production in Indonesia. Based on the above literature review, this paper analyses the environmental efficiency in Indonesian rice agriculture, with particular attention to the use of agrochemicals, utilising a stochastic production frontier technique.

3 Theoretical framework This section discusses in great detail the concepts of environmental, technical and social efficiencies. 3.1 Environmental and technical efficiencies In modern agricultural practices, including rice farming, the use of chemical inputs is common. When these chemical inputs are not perfectly absorbed by the plant, and so to some extent are discharged into the environment, it is defined as pollution or, to be precise, a non-point source of pollution, which is considered to be environmentally damaging (Cacho 1999; Grafton, Adamowicz, Dupont, Nelson, Hill, and Renzetti 2004). Based on this phenomenon, the initial definition of environmental efficiency is the ratio of minimum attainable environmentally detrimental input use to actual use, given the actual level of output and other inputs of the existing technology (Reinhard, Lovell, and Thijssen 2002). With the production frontier concept, the above definition of environmental efficiency should be modified to include technical efficiency, that is, efficient use of all inputs, including environmentally detrimental inputs. And so, the modified definition of environmental efficiency is the ratio of minimum feasible environmentally detrimental input to actual use, given the actual level of output and efficiency level of other inputs. Figure 1 illustrates the difference between the initial and the modified definition of environmental efficiency. Consider a production frontier

166


technology f ( X , Z ) , where X is an environmentally detrimental input and Z is a usual input. Let point A be the current actual situation where a producer utilises input as much as X act and Z act to produce Y act . This producer is certainly not efficient and, thus, not on the frontier. Based on the production frontier, Y act can actually be attained with X min and Z min . If one uses the initial definition of environmental efficiency, then the level of environmental efficiency of this producer is the ratio of X to X act . The producer should then acquire better knowledge in order to reduce input X from X act to X to produce the potential output, that is, from point A to C. But this move is not technically efficient (Huang and Liu 1994). The most efficient move is if the producer can apply a better practice to move from A to B. Hence, environmental efficiency should be the ratio of X min to X act .

Z

A

C •

Zact

B •

Zmin

Y = f (X , Z)

O

X

Xmin

Xact

X

Figure 1: Environmental efficiency: input oriented Let us define ψ as the ratio of X min to X act , ξ as the ratio of Z min to Z act , and ϕ or the technical efficiency as the ratio of Y act to Y pot . The actual level of output can be represented by a frontier function: Y act = f (ψ ⋅ X , ξ ⋅ Z )

(1)

where X = X act and Z = Z act . Furthermore, based on the definition of technical efficiency: Y act = ϕ ⋅ Y pot

(2)


167

where Y pot is the potential output, if the producer is efficient, having the proper knowledge for utilising X and Z . For simplicity, let the kernel deterministic production frontier function be a Cobb-Douglas function where Y pot = a ⋅ X α ⋅ Z β

(3)

Combining (1), (2) and (3) results in ϕ ⋅ a ⋅ X α ⋅ Z β = a ⋅ (ψ ⋅ X )α ⋅ (ξ ⋅ Z ) β

(4)

By assuming ξ = ψ , ψ can be solved as 1

ψ =ϕ

α +β

(5)

Let us consider relaxing the assumption of ξ = ψ and define the true situation as ξ = δ ⋅ψ . The first case is when 0 < δ < 1 . In this case, 1

ψ ' =ψ ⋅δ ψ "= ψ ⋅δ

β (α + β ) 1 β (α + β )

where ψ ' < ψ . The second case is when δ > 1 . In this case, where ψ ' ' > ψ . Note that the upper bound of the second 1 (max)

condition is when ξ = δ ⋅ψ = 1 , and so ψ " = ϕ . By definition, ψ "(max) ≤ 1 . In the real world, it is likely that the differences between ψ ' < ψ and ψ ' < ψ as well as ψ ' < ψ will be small, within a two-decimal number, so it is accurate enough to utilise equation (5) to calculate environmental efficiency. From equation (5), it can be seen that the level of environmental efficiency can be indirectly estimated in two steps. First, estimate technical efficiency and technology parameters using the production frontier technique. Second, measure environmental efficiency utilising the estimated technical efficiency rate and the output elasticity with respect to inputs. There are two conditions that make environmental efficiency exactly the same as technical efficiency. The first is when the firm is operated at full technical efficiency (ϕ = 1) and the second when production exhibits constant returns to scale, that is, (α + β = 1) . It is important to note that, with a Cobb-Douglas production technology, output elasticity with respect to all inputs is constant. This is rather restricted. Output elasticity is expected to vary among producers, and it could be the case that producers with high technical efficiency have more output elasticity, or vice versa. A more flexible production technology is therefore preferable. α

168


3.2 Chemical waste and environmental costs Let us define AW as the amount of chemical waste in agricultural production. AW hence can be defined as AW = (1 −ψ ) ⋅ X

(6)

Estimating the monetary value or the environmental cost for society of this AW has been a source of debate in the economic literature. The difficulty in putting monetary value on such environmental waste is generally due to there being no market for pollution. A significant body of environmental economic literature has nevertheless been devoted to dealing with this issue (Hanley, MacMillan, Wright, and Bullock 1998; Carson 2000; Grafton, Adamowicz, Dupont, Nelson, Hill, and Renzetti 2004).

Figure 2: Valuation of chemical waste using a modified “Effect on Production” approach This paper modifies one of the approaches that has been developed in this literature, namely the “effect on production” method (Garrod and Willis 1999), which suggests that the existence of additional pollution will affect production such that the level of output will be different from production without the existing level of pollution. The difference of monetary value of output represents the environmental cost. We have modified this method so that the definition of environmental cost is the value of potential output, calculated using the frontier function with the actual use of chemicals


169

minus the actual output, that is, the cost of not using the chemical input efficiently. From Figure 2, this definition can be calculated as

{(

) (

EC = P f X act , Z − f X min , Z

)}

(7)

where EC is the environmental cost and P is the prevailing price of output. 3.3 Private and social efficiency The general definitions for private and social efficiencies are as follows. Private efficiency is when producers are maximising their profit without taking into account the externalities caused by their pollution, whereas social efficiency is where producers take these externalities into account. Let the profit identity be Π ≡ PY ⋅ f (•) − PX ⋅ X − PZ ⋅ Z

(8)

where Π is profit, f (•) is a production frontier technology that produces single-output Y using environmentally detrimental input X and usual input Z , PY is the prevailing market price of output, PX and PZ are prevailing market prices of input X and input Z , respectively, t is time index, β is the coefficient of technology, and ε represents the composite disturbance terms. Note that the production technology should be strictly concave to guarantee profit maximisation. The first-order conditions that are necessary for profit maximisation are PY

∂f (• ) = PX ∂X

(9a)

PY

∂f (•) = PZ ∂Z

(9b)

In terms of elasticity, the conditions can be expressed as VY ⋅θ X = W X

(10a)

VY ⋅θ Z = WZ

(10b)

where θ X and θ Z are output elasticity with respect to inputs X and Z , respectively, W X = PX ⋅ X and WZ = PZ ⋅ Z are costs of inputs X and Z , respectively, and VY = PY f (•) is the value of product. Let us denote VY ⋅θ = VY ⋅θ X + VY ⋅θ Z and W = W X + WZ ; then the first-order conditions that are necessary for profit maximisation can be arranged as

170

Mariyono, Resosudarmo, Kompas, and Grafton θX = SX θ

(11a)

θZ = SZ θ

(11b)

WX W and S Z = Z . The notation of θ X θ and θ Z θ are called W W normalised output elasticity with respect to inputs X and Z , respectively.

where S X =

A firm is said to be privately efficient if and only if normalised output elasticity with respect to each input is equal to the share of costs for each respective input. The conditions will hold if Ω X + ΩZ = 0

where Ω X =

(12)

θX θ − S X and Ω Z = Z − S Z . This represents a marginal rate of θ θ

technical substitution ( MRTS ) between both inputs that is equal to the price ratio of both inputs (Kumbhakar and Lovell, 2000). When the conditions do not hold, representing deviation of MRTS from the price ratio, this implies that the use of one input must be excessive relative to other inputs. The closer the value of Ω X + Ω Z is to zero, the higher the private efficiency. Pretty and Waibel (2005) suggest that environmental costs resulting from chemical use should be taken into account to obtain social efficiency. Social efficiency can be determined by internalising the environmental costs associated with an environmentally detrimental input, X , that is θX = Sˆ X θ

(13a)

θZ ˆ = SZ θ

(13b)

W X + EC ˆ WZ , SZ = and EC is environmental cost associated W + EC W + EC with the use of X . A firm is said to be socially efficient if and only if

where Sˆ X =

normalised output elasticity with respect to each input is equal to the share of social costs for each respective input. The conditions will hold if Φ X + ΦZ = 0

where Φ X =

(14)

θX ˆ θ − S X and Φ Z = Z − Sˆ Z . The closer the value of Φ X + Φ Z is θ θ

to zero, the greater the social efficiency.

171


The different allocation of inputs satisfying equations (11a) and (11b) on one side and equations (13a) and (13b) on the other represents the problem when a producer does not internalise environmental costs, even if this producer operates on the frontier.

4 Implementation Consider a farm uses land A, capital K , labour L , material M , and chemical input X to produce a single output, Y, where technology is characterised by production function Y = F ( A, K , L, M , X , t ) , with F' > 0 and F' ' < 0 . Millan and Aldaz (1998) state that the inclusion of t in time-series econometric models of production is to measure technological change over time. This paper employs a translog production frontier function introduced by Christensen, Jorgenson, and Lau (1973) as follows: ln Yit = β 0 + β1 ⋅ ln Ait + β 2 ⋅ ln K it + β 3 ⋅ ln Lit + β 4 ⋅ ln M it + β 5 ⋅ ln X it

+ 0.5 ⋅ β 6 ⋅ ln Ait ⋅ ln Ait + 0.5 ⋅ β 7 ⋅ ln K it ⋅ ln K it + 0.5 ⋅ β 8 ⋅ ln Lit ⋅ ln Lit + 0.5 ⋅ β 9 ⋅ ln M it ⋅ ln M it + 0.5 ⋅ β10 ⋅ ln X it ⋅ ln X it + β11 ⋅ ln Ait ⋅ ln K it + β12 ⋅ ln Ait ⋅ ln Lit + β13 ⋅ ln Ait ⋅ ln M it + β14 ⋅ ln Ait ⋅ ln X it + β15 ⋅ ln K it ⋅ ln Lit + β16 ⋅ ln K it ⋅ ln M it + β17 ⋅ ln K it ⋅ ln X it + β18 ⋅ ln Lit ⋅ ln M it + β19 ⋅ ln Lit ⋅ ln X it + β 20 ⋅ ln M it ⋅ ln X it + β 21 ⋅ t ⋅ ln Ait + β 22 ⋅ t ⋅ ln K it + β 23 ⋅ t ⋅ ln Lit + β 24 ⋅ t ⋅ ln M it + β 25 ⋅ t ⋅ ln X it + β 26 ⋅ t + β 27 ⋅ t 2 + vit + u it

(15)

In the above model, the disturbance term is assumed to have two components. One component, u it , is assumed to have a strictly nonnegative distribution and the other component, vit , is assumed to have a symmetric distribution. In the econometrics literature, u it , is often referred to as the inefficiency term and vit is often referred to as the idiosyncratic error. Stochastic production frontier models for panel data permit two different parameterisations of the inefficiency term u it : a time-invariant model and a time-varying decay model (Battese and Coelli, 1992). In the time-invariant model, u it = u i ; while in the time-varying decay model, u it = u i exp{η (t − T )}. Consider σ u2 and σ v2 as the variances of the parameters u it and vit and define σ 2 = σ u2 + σ v2 . According to Jondrow, Lovell, Materov, and Schmidt

172

Mariyono, Resosudarmo, Kompas, and Grafton 2

(1982), γ = σ u

σ2

represents the total variation of actual output deviating

from the frontier. Hence, following Battese and Coelli (1992), this paper will examine the significance of σ 2 and γ to measure the significance differences between the average production function and the frontier. 4.1

Environmental efficiency

FRONTIER 4.1, a computer program created by Coelli (1996), has been used to estimate the parameters of the stochastic production frontier function in the equation (15). Environmental efficiency is then calculated using the formula 1

θ ψ =ϕ ∑

k k

(16)

where ϕ = exp{− u i } is technical efficiency, θ k is output elasticity with respect to input k, for k = 1,2,…,5. The estimation of environmental efficiency using this formula is expected to overcome the problem when a producer uses zero level of an environmentally detrimental input.7 4.2 Social efficiency Analysis is needed to find the proportionality of input use, which can be specified as θk − Sk = Ωk θ

(17)

where k = land, capital, labour, materials and agrochemicals. The firms are privately efficient when Ω for all inputs is not statistically different from zero. A non-zero Ω represents private inefficiency in the use of respective input. If Ω k is positive, the use of input k is excessive relative to other inputs. Likewise, the social efficiency is analysed as θk ˆ − Sk = Φk θ

(18)

When Φ for all inputs is not statistically different from zero, the firms are socially efficient. A non-zero Φ represents social inefficiency in the use of respective input. Overall private and social efficiencies are shown by the 7

At the micro level of agricultural practices, it is likely that a producer does not use fertilisers and/or pesticides.

173


values of ∑k Ω k and to the price ratio.

∑

k

Φ k , which show that MRTS of inputs is not equal

5 Data This study uses a data set established from a longitudinal survey being conducted by the Indonesian Centre for Agricultural, Socioeconomic and Policy Studies (CASEPS) of the Ministry of Agriculture. The data set is unbalanced panel data consisting of 358 farm operations in Indonesia during 1994, 1999 and 2004, with the sample having been collected from five regions: Lampung, West and East Java, West Nusa Tenggara, North Sulawesi and South Sulawesi. Several villages were randomly selected in each region and farmers cultivating rice sampled randomly. Once farmers have been selected, they become respondents of the survey and will be interviewed every five years. The total number of observations used is 817. Table 1 presents detailed observations from each year in each region, most of which are from 1994 and 1999. The reasons for the significant reduction of observations in 2004 are that some farmers were no longer cultivating rice, others have died and the family did not continue cultivating rice, and rice farmers in North Sulawesi were no longer interviewed. Seeing that not all the same farmers were interviewed in these three years, the data has become an unbalanced panel.

Table 1: Number of observations Region Lampung West and East Java Nusa Tenggara North Sulawesi South Sulawesi Total

1994

1999

Year 2004

74 36 126 21 84 341

79 33 121 21 50 304

54 19 63 0 36 172

Total 207 88 310 42 170 817

The actual number of variables observed in the data collection obtained by interviewing sampled farmers varies widely. This is because the survey accommodates variations in which farming is very spatially and temporally specific. For example, certain fertilisers are not used in one place and always used in another place. In some regions, it is usual that there is voluntary labour during early planting and harvesting seasons, but this not the case in others. In addition, some farmers are able to separate expenses of rice agriculture in some detail, but others are not. For the purpose of this study, however, the data is aggregated to avoid problems of missing data.

174


Table 2 shows the disaggregated variables utilised in this paper and their definitions.

Table 2: Variable definitions Variable Rice production (Y) Area (A) Labour (L)

Description Un-husked production Total rice-sown area Total labour comprises family, voluntary and hired labour, used for six stages of farming Capital (K) Capital consists of tractors and animals mainly used in land tillage Material (M) Total material used in rice production comprises seed, water irrigation, and green manure Chemicals (X) Chemical fertilisers and pesticides. Fertilisers consist of Urea, Triple Super Phosphate (TSP), Ammonium Sulphate (ZA) and Potassium Chloride (KCl). Pesticides comprise solid and liquid formulations. Note: * Monetary value is at 1993 constant price.

Unit Kilogram Hectare working day tractor/animal working day monetary term* monetary term*

Summary statistics of the key variables are presented in Table 3. It can be seen that, in terms of land, Java farmers, on average, are smaller than those off Java.

Table 3: Summary statistics of key variables by region Lampung Production

2,477 (3,989)

Area

0.58 (0.62)

West and East Java 1,341 (1,318)

West Nusa Tenggara 2,482 (2,365)

0.27 0.80 (0.23) (0.80) Capital 1.03 (2.99) 2.67 7.34 (3.41) (17.01) Labour 61.65 42.39 60.74 (67.57) (44.12) (47.79) Material 26,676 23,712 77,028 (32,297) (30,368) (76,000) Chemical 62,884 44,294 158,399 (79,700) (45,151) (168,134) Note: Numbers in parentheses are standard deviation.

North Sulawesi 1,284 (1,645)

South Sulawesi 2,445 (2,574)

0.56 (0.66)

0.66 (0.53)

2.54 (4.40)

2.81 (9.06)

30.56 (24.45) 32,634 (36,157) 37,013 (43,612)

68.02 (62.59) 64,968 (90,548) 81,588 (84,620)

6 Results and discussion The test for functional form of production technology shows that translog production technology with non-Hicks neutral technological change fits


175

with the farm-level data of Indonesian rice agriculture. The magnitudes and signs of all estimated coefficients of the translog production frontier in the two models are given in Table 4.

Table 4: Parameter estimates of stochastic production frontier function TFP Area (A) Capital (K) Labour (L) Material (M) Chemicals (X) 0.5 A*A 0.5 K*K 0.5 L*L 0.5 M*M 0.5 X*X A*K A*L A*M A*X K*L K*M K*X L*M L*X M*X T*A T*K T*L T*M T*X T T2

σ2 γ Log-likelihood LR-test

Coefficient 8.05a 1.23a 0.15b -0.06n -0.11n 0.08n 0.05c 0.01a -0.01n 0.01c 0.01a -0.01n -0.02n -0.06b 0.02a 0.002n -0.01b -0.00005n 0.01n -0.004n -0.01b 0.02n 0.03a 0.03n -0.05n 0.09a -0.34n 0.3a

z-ratio 7.10 3.33 2.81 -0.19 -0.98 1.38 1.61 3.49 -0.62 1.72 6.97 -1.20 -0.45 -2.06 3.11 0.44 -2.18 -0.08 0.56 -0.88 -2.10 0.44 4.81 0.82 -1.32 7.32 -0.67 5.48

1.10a 0.88a -645.56 137.47

5.92 38.48

Note: Dependent variable is rice production (kg); all variables are in log; a = significant at 1% level; b = significant at 5% level; c = significant at 10% level; and n = not significant.

176


Commonly, in estimating a translog production technology, some coefficients are significant and others not (Ahmad and Bravo-Ureta 1996). As the insignificant coefficients are not different from zero, in this paper significant coefficients will be used for calculating output elasticity with respect to each input. Mean output elasticity is calculated at the average level of input uses during the period. The mean output elasticity is shown in Table 5.

Table 5: Mean output elasticity 1994 Land 0.72 Capital 0.03 Labour 0.00 Material 0.10 Chemicals 0.00 Scale elasticity 0.86 Note: 0.00 means the number is trivial.

1999

2004

Total

0.70 0.05 0.00 0.10 0.09 0.94

0.74 0.03 0.00 0.12 0.19 1.09

0.72 0.04 0.00 0.11 0.09 0.96

From Table 5, it can be observed that the mean output elasticity of land is the highest. The second highest is material inputs. This result is uncommon. Trewin, Weiguo, Erwidodo, and Bahri (1995) estimated using a Cobb-Douglas technology and found that output elasticity with respect to land is superior compared with other inputs. Sumaryanto, Wahida, and Siregar (2003) provide support that output elasticity with respect to land is around 0.8. By using translog and quadratic forms, Villano and Fleming (2006) also found that output elasticity of land is the highest in rice farming in the Philippines. In China, the output elasticity of land is 0.9 (Yao and Liu, 1998). In the case of Vietnamese rice production, however, the highest output elasticity is not with respect to land but with respect to material inputs (Che, Kompas, and Vousden 2006). 6.1 Environmental efficiency The results of calculating environmental efficiency are given in Table 6. On average, the level of environmental efficiency for Indonesia is 0.64. This number is considered to be low. Note that the most environmentally efficient level is 1. This result, nevertheless, corresponds to Roche’s (1994) finding of low N-uptake efficiency in rice production.

177


Table 6: Average environmental efficiency across regions over time Year 1994 1999 Mean 0.55 0.59 Lampung S.D 0.24 0.25 West and East Mean 0.74 0.81 Java S.D 0.15 0.09 West Nusa Mean 0.62 0.66 Tenggara S.D 0.16 0.18 North Mean 0.51 0.57 Sulawesi S.D 0.23 0.20 Mean 0.57 0.65 South Sulawesi S.D 0.15 0.17 Mean 0.60 0.65 By year S.D 0.19 0.20 Note: S.D = standard deviation; n.a. = not available. Region

2004 0.67 0.22 0.74 0.15 0.71 0.14

By region 0.60 0.24 0.77 0.13 0.65 0.17 0.54 0.21 0.63 0.16 0.64 0.19

n.a. 0.75 0.10 0.71 0.17

Among regions in Indonesia, on average, rice farmers in (West and East) Java are the most environmentally efficient in using chemical inputs in cultivating their rice plants. The least environmentally efficient, on average, are rice farmers in North Sulawesi, followed by those in Lampung. What is also interesting is that, while the level of environmental efficiency in Java remained relatively constant between 1994 and 2004, in other regions environmental efficiencies have been improving. As expected from equation (15), the correlation between technical and environmental efficiency is close to unity: both move in the same direction. This means that farmers who are able to use other inputs efficiently are also able to utilise chemical inputs efficiently. Those who cannot do so discharge the excess chemicals into the environment. Table 7 shows the correlations between environmental efficiency and farm size, revealing that the larger the rice farm, the higher its environmental efficiency.

Table 7: Correlations between farm size and several indicators Technical Environmental Chemical Environmental Efficiency Efficiency Waste Cost Farm size 0.21a 0.19a 0.36a 0.45a Note: a = significant at 1%; b = significant at 5%; n = not significant.

Use of Chemicals 0.14a

6.2 Chemical waste and environmental costs In this paper, we have emphasised that chemical waste is a consequence of environmentally inefficient farms. As stated in equation (6), the amount of chemical waste resulting from each producer will be dependent on the level of environmental efficiency and the level of use of environmentally

178


detrimental inputs. Consequently, the ranks of farms based on environmental and technical efficiency will be different from those based on level of chemical waste. We have regressed the amount of chemical waste with dummy regions to identify in which regions on average rice farmers are wasting more chemicals. The results of this regression are in Table 8, where it can be seen that the highest level of chemical waste results from rice farms in South Sulawesi. It is important to note, however, that rice farmers in South Sulawesi are not the least environmentally efficient. One reason for this could be farm size. The correlation shows that, the greater the size of rice farms, the higher the levels of chemical waste (Table 7). On average, the size of rice farms in South Sulawesi is relatively larger than in other regions, except for West Nusa Tenggara.

Table 8: Regression of chemical waste with dummy regions Year 1994 1999 2004 Overall 879.25n 691.79n 1,335.01n 907.36n Constant (=Java) 0.42 0.72 1.39 0.94 n b n 2,323.85 2,665.12 1,061.34 2,143.99c Lampung 0.90 2.32 0.95 1.86 3,309.59a 1,773.16c 3,248.80a 3,949.54c West Nusa Tenggara 1.65 3.04 1.61 2.97 n c 3,871.12 2,810.03 3,218.74c North Sulawesi (dropped) 1.11 1.82 1.90 2,567.22b 5,870.74a 4,139.24a 4,306.02c South Sulawesi 1.71 2.07 4.93 3.48 2 R 0.011 0.030 0.182 0.017 F-stat 0.44 2.32 12.47 3.52 No. Obs. 341 304 172 817 Note: Numbers below the coefficients are their t-ratios; a = significant at 1%; b = significant at 5%; c = significant at 10 %; n = not significant. Regions

Recall that we have previously argued that the larger the farm size, the more environmentally efficient the farm. But now, it can be seen that the larger the farm, the greater the chemical waste. The main reason for the latter is that, although large farms tend to be more environmentally efficient, they also use much greater quantities of chemicals per ha than smaller ones. Table 9 provides a summary of the costs of this chemical waste across the five regions observed. It can be seen that the environmental costs caused by rice farms in West Nusa Tenggara and South Sulawesi have been significantly greater than those in (West and East) Java, while other regions are not statistically different to Java.


179

Table 9: Regression of environmental cost with dummy regions Regions Coefficient Constant (=Java) 15,161n Lampung 34,483n West Nusa Tenggara 139,335c North Sulawesi -13,895n South Sulawesi 202,958b 2 R 0.01 F-stat 2.92 Note: b = significant at 5%; c = significant at 10 %; n = not significant.

t-ratio 0.23 0.43 1.84 -0.12 2.47

6.3 Social efficiency Table 10 shows the private and social efficiency of factor inputs in Indonesian rice production. Recall that privately or socially efficient with regard to a certain input is when the indicator of private or social efficiencies, respectively, equals zero. Observing the levels of social efficiency in any year for any inputs, it can be seen that Indonesian rice farmers have not utilised inputs efficiently. The source of this social inefficiency is not so much because they do not take into account the environmental costs of using excess chemicals, but rather because they are not able to allocate their inputs efficiently, that is, due to their poor performance regarding their private efficiency. Table 10 indicates that most inefficient is the use of land. The cause is that the observed cost of land as represented by land taxes is very low in rural areas; meanwhile, the output elasticity with respect to land is very high. In other words, where rice production is concerned, the value of the marginal product of land is much greater than the “price” of land. First, this could mean that land has in general been underpriced in Indonesia. Second, availability of land has been the constraint for rice farmers, preventing them from becoming more efficient. In Java, fertile paddy land exists, but the amount is decreasing over time, mostly due to land conversion (Ashari 2003; Firman 1997; Mariyono 2006; Mariyono, Harini, and Agustin 2007). Off Java, there is relatively limited available land suitable for rice plantation. If more land becomes available and the price is relatively low, farmers should expand their land for rice production to improve both their private and social efficiencies as well as to enhance their profits. The gaps for other inputs are negative, meaning that the uses of capital, labour, materials and agrochemicals are relatively excessive. Least excessive are the use of capital in 1994 and 1999 and the use of materials in 2004. Considering how expensive capital is for farmers, it can be understood that the priority of rice farmers is to utilise capital as efficiently as possible. Rice farmers have always been able to improve their material

180


input efficiency. Increasing knowledge and experience could be the main reason for this.

Table 10: Private and social efficiencies

Year

Inputs

Z

Land Capital Labour Materials Chemicals

Private Efficiency

Social Efficiency

θk − Sk = Ωk θ

θk ˆ − Sk = Φ k θ

mean 0.79a -0.02a -0.26a -0.16a -0.35a ∑ Ω = 1.59

S.D 0.06 0.08 0.20 0.20 0.20

mean 0.79a -0.02a -0.26a -0.16a -0.35a ∑ Φ = 1.59

S.D 0.06 0.08 0.20 0.20 0.20

1999


0.73a -0.05a -0.47a -0.11a -0.10a ∑ Ω = 1.46

0.01 0.12 0.27 0.22 0.17

0.73a -0.05a -0.47a -0.11a -0.11a ∑ Φ = 1.46

0.01 0.12 0.27 0.21 0.17

2004


0.56a -0.17a -0.40a -0.01a 0.03a ∑ Ω = 1.18

0.14 0.20 0.25 0.12 0.13

0.59a -0.14a -0.32a 0.02a -0.16a ∑ Φ = 1.23

0.09 0.16 0.22 0.10 0.22

Note: a = significantly different from zero tested at 1% significance level.

Overuse of agrochemicals has been largely due to the invention during the Green Revolution of rice varieties responsive to fertilisers. The Green Revolution also induced excessive use of pesticides as the main technique to control pests (Barbier 1989; Resosudarmo 2001). The impact of this on Indonesian farmers’ behaviour prevailed until 1994, with the overuse of chemicals having been mostly in Java’s intensified rice areas, and not off Java, where most rice farms were not covered by the intensification program. This finding is in line with a study by Roche (1994) indicating excessive use of fertilisers in Javanese rice farming. After 1994, there is an impressive improvement in allocation of agrochemicals. This is understandable, because the use of agrochemicals has been rationalised and pesticide use consequently reduced. Pesticides are only used when serious pest infestation exists, and timely and correct dosage of fertilisers is applied (Rolling and van de Fliert 1994; Winarto 2004; Resosudarmo and Yamazaki 2006).


181

An interesting situation can be seen in 2004. When the environmental costs of excessive use of chemicals are taken into account, it can be seen that, in general, Indonesian farmers overuse chemicals and should utilise more material inputs. The obverse happens if the environmental costs are ignored. This case shows the importance of considering the environmental costs of chemical pollution in rice agriculture. Labour, in general, is the input that has been overused the most. Labour is relatively abundant in rural areas, while non-farm activities are relatively limited. Additionally, institutional and cultural norms in rural areas have pushed farming sectors to utilise more labour than they need. When the economic crisis hit Indonesia in 1997–98, there was a massive flow of unskilled labour from the urban sector to rural areas (Mellor, Falcon, Taylor, Arifin, Said, and Pasandaran 2003), and the farming sector was pushed to absorb this extra labour.

7 Conclusion This paper has proposed a framework to measure environmental and social efficiencies in using chemicals that are detrimental to the environment, including estimating the environmental costs of pollution. With the case of Indonesian rice farming, this paper has also shown how to utilise this proposed framework. It is certainly an open discussion whether or not the framework proposed in this paper to measure environmental and social efficiencies in using environmentally detrimental chemicals in the agricultural sector is a proper one. There are several weaknesses in implementing the framework to calculate environmental and social efficiency for the Indonesian rice sector, largely due to limited data availability. The two main weaknesses are as follows. First the available data only cover several Indonesian provinces. West Sumatra, Central Java and Bali, all important rice-producing regions in Indonesia, are not in the data set. Hence, it is difficult to argue that the result of this paper is able to represent the whole situation in Indonesia. Second, observations within the area covered in the data set are too few. Hence, it is impossible to develop a frontier analysis for each region that could lead to a more interesting analysis. Despite these weaknesses, several conclusions can be drawn from the analysis in this paper. First, in general, the level of rice farmers’ environmental efficiency in using chemicals is low. Farmers in Java tend to be more efficient than those in other areas. The least efficient among areas covered in this paper are rice farmers in North Sulawesi. However, it is important to note that, while the level of environmental efficiency among Javanese rice farmers has been relatively stable throughout the period of the analysis in this paper, that of rice farmers off Java has been improving.

182


Second, inefficiency in using chemicals results in a significant level of chemical waste and so becomes a non-point source of pollution. In general, rice farmers in South Sulawesi and West Nusa Tenggara waste more chemicals than those in other regions. One reason is most likely due to the relatively larger rice farms in those areas compared to other regions in Java, North Sulawesi and Lampung. There is evidence that the larger the farm, the more chemicals it wastes. Third, in achieving social as well as private efficiency, availability of land for rice farming is the main constraint. Fertile land in Java has decreased over time, due to land conversion for housing and modern sectors. There is abundant land off Java, but it is mostly unsuitable for rice farming. Fourth, a further constraint in achieving better social as well as private efficiency is imposed by rural institutions and norms that require agricultural sectors to absorb abundant labour in rural areas. Non-farm activities should be developed while, at the same time, the quality of rural workers should be improved to be able to work in non-farm sectors, so reducing the amount of labour in the agricultural sector. Fifth, ignoring environmental costs does exaggerate the overuse of chemical inputs and reduces the utilisation of other inputs. Nevertheless, so far, ignoring the environmental cost of pollution has not been the main reason that Indonesian rice farmers are not efficient. Rather, limited land availability and the need to absorb abundant rural labour are the main causes.

References Ahmad, M., Bravo-Ureta, B. (1996). Technical Efficiency Measures for Dairy Farms Using Panel Data: A Comparison of Alternative Model Specifications. Journal of Productivity Analysis 7: 399–416. Ali, M., Byerlee, D. (2002). Productivity Growth and Resource Degradation in Pakistans Punjab: A Decomposition Analysis. Economic Development and Cultural Change 50(4): 839-864. Amaza, P.S., Olayemi, J.K. (2002). Analysis of Technical Inefficiency in Crop Production in Gombe State, Nigeria. Applied Economics Letters 9: 51-54. Antle, J.M., Pingali, P.L. (1994). Pesticides, Productivity, and Farmer Health: A Philippine Case Study. American Journal of Agricultural Economics 76: 418430. Apel, H., Paudyal, M.S., Richter, O. (2002). Environmental Risk Assessment of Pesticide Use Based on the Modelling of the Environmental Fate of Pesticides in Soil. In Herrmann, A., Schumann, S. (eds.), Proc. Int. Workshop on Environmental Risk Assessment of Pesticides and Integrated Pesticide Management in Developing Countries 6-9 Nov. 2001 Kathmandu, Nepal, Landschaftsökologie und Umweltforschung 38, Braunschweig, 13-23.


183

Ashari. (2003). Tinjauan Tentang Alih Fungsi Lahan Sawah ke Non Sawah dan Dampaknya di Pulau Jawa. Forum Penelitian Agro Ekonomi 21(2): 83-98. Barbier, E.B. (1989). Cash Crops, Food Crops, and Sustainability: The Case of Indonesia. World Development 17(6): 879-895. Battese, G.E., Coelli, T.J. (1992). Frontier Production Functions, Technical Efficiency and Panel Data: With Application to Paddy Farmers in India. Journal of Productivity Analysis 3: 153-169. Battese, G.E. and Tessema, G.A. (1993). Estimation of Stochastic Frontier Production Functions with Time-varying Parameters and Technical Efficiency Using Panel Data from Indian Villages. Agricultural Economics 9: 313-33. Bravo-Ureta, B.E., Evenson, R.E. (1994). Efficiency in Agricultural Production: The Case of Peasant Farmers in Eastern Paraguay. Agricultural Economics 10: 2737. Byerlee, D. (1992). Technical Change, Productivity, and Sustainability in Irrigated Cropping Systems of South Asia: Emerging Issues in the Post–Green Revolution Era. Journal of International Development 4(5): 477-496. Cacho, O.J. (1999). Dynamic Models, Externalities and Sustainability in Agriculture. Working Paper Series in Agricultural and Resource Economics 99-4, April 1999. The University of New England, Armidale. Carson, R.T. (2000). Contingent Valuation: A Users Guide. Environmental Science and Technology 34: 1413-1418. Che, T.N., Kompas, T., Vousden, N. (2006). Market Reform, Incentives and Economic Development in Vietnamese Rice Production. Comparative Economic Studies 48: 277-301. Christensen, L.R., Jorgenson, D.W. and Lau, L.J. (1973). Transcendental Logarithmic Production Frontier, Review of Economics and Statistics 55(1): 29-45. Coelli, T. (1996). A Guide to FRONTIER Version 4.1: A Computer Program for Stochastic Frontier Production and Cost Function Estimation. Centre for efficiency and Productivity Analysis University of New England, Armidale. Conway, G.R., Barbier, E.B. (1990). After Green Revolution: Sustainable Agriculture for Development. London: Earth Scan Publication. Firman, T. (1997). Land Conversion and Urban Development in the Northern Region of West Java, Indonesia. Urban Studies 34(7): 1027-1046. Gang, L., Felmingham, B. (2004). Environmental Efficiency of the Australian Irrigation Industry in Treating Salt Emission. Australian Economic Papers 43(4): 475490. Garrod, G., Willis, K.G. (1999). Economic Valuation of the Environment: Methods and Case Studies. Cheltenham: Edward Elgar. Grafton, R.Q., Adamowicz, W., Dupont, D., Nelson, H., Hill, R.J., Renzetti, S. (2004). The Economics of the Environment and Natural Resources. Carlton: Blackwell Publishing. Hadri, K., Whittaker, J. (1999). Efficiency, Environmental Contaminants and Farm Size: Testing for Links Using Stochastic Production Frontiers. Journal of Applied Economics 2(2): 337-356.

184


Hanley, N., MacMillan, D., Wright, R.E., Bullock, C. (1999). Contingent Valuation versus Choice Experiments: Estimating the Benefits of Environmentally Sensitive Areas in Scotland. Journal of Agricultural Economics 49(1): 1-15. Houndekon, V., de Groote, H. (1998). Health Costs and Externalities of Pesticides Application in Locust and Grasshopper Control in The Sahel. Paper prepared for the Annual conference of the American Agricultural Economics Association, 2-5 Aug. 1998, Utah. Huang, C. J., Liu, J-T. (1994). Estimation of a non-neutral stochastic frontier production function. Journal of Productivity Analysis 5: 171-180. Jondrow, J., Lovell, C.A.K., Materov, S., Schmidt, P. (1982). On The Estimation of Technical Efficiency in the Stochastic Frontier Production Function Model. Journal of Econometrics 19: 233-238. Jungbluth, F. (1996). Crop Protection Policy in Thailand: Economic and Political Factors Influencing Pesticide Use. A Publication of the Pesticide Policy Project 5. Kumbhakar, S.C. (1994). Efficiency Estimation in a Profit Maximising Model Using Flexible Production Function. Agricultural Economics 10: 143-152. Kumbhakar, S.C., Lovell, C. A. K. (2000). Stochastic Frontier Analysis. Cambridge: University Press. Mariyono, J. (2006). Agricultural Land Conversion in Jogjakarta: An Empirical Analysis. Jurnal Ekonomi dan Studi Pembangunan 7(1): 9-18. Mariyono, J., Harini, R., Agustin, N.K. (2007). Impact of Economic Development and Population Growth on Agricultural Land Conversion in Jogjakarta: A Dynamic Analysis. Jurnal Ekonomi Pembangunan 8(1): 50-62. Mellor, J.W., Falcon, W.P., Taylor, D.M., Arifin, B., Said, E.G., Pasandaran, E. (2003). Main Report of Agricultural Sector Review, Indonesia, Part I. CARANA Corporation, US Agency for International Development. Millan, J.A., Aldaz, N. (1998). Agricultural Productivity of the Spanish Regions: A Non-Parametric Malmquist Analysis. Applied Economics 30: 875-884. Pimentel, D., Aquay, H., Biltonen, M., Rice, P., Silva, M., Nelson, J., Lipner, V., Giordano, S., Horowitz, A., D’Amore, M. (1992). Environmental and Economic Costs of Pesticides Use. BioScience 42: 750-760. Pretty, J., Waibel, H. (2005). Paying The Price: The Full Cost of Pesticides. In Pretty, J.N. (ed.). The Pesticide Detox: Towards a More Sustainable Agriculture. London: Earthscan, 39-54. Pujara, D.S., Khanal, N.R. (2002). Use of Pesticides in Jaishidihi Sub-Catchment, Jhikhu Khola Watershed, Middle Mountain in Nepal. In Herrmann, A., Schumann, S. (eds.). Proc. Int. Workshop on Environmental Risk Assessment of Pesticides and Integrated Pesticide Management in Developing Countries 6-9 Nov. 2001 Kathmandu, Nepal. Landschaftsökologie und Umweltforschung 38, Braunschweig, 168-177. Reinhard, S., Lovell, C.A., Thijssen, G. (1999). Econometric Estimation of Technical and Environmental Efficiency: An Application to Dutch Dairy Farms. American Journal of Agricultural Economics 81: 44-60. Reinhard, S., Lovell, C.A., Thijssen, G. (2002). Analysis of Environmental Efficiency Variation. American Journal of Agricultural Economics 84(4): 1054-1065.


185

Resosudarmo, B.P. (2001). The Economic-Wide Impact of Integrated Food Crop Pest Management in Indonesia. EEPSEA Research Report Monograph. Economy and Environment Program for South East Asia, Ottawa. Resosudarmo, B.P., Thorbecke, E. (1998). Reducing the Number of Pesticide Related Illnesses The Impact on Household Incomes in Indonesia. Bulletin of Indonesian Economic Studies 34(2): 143-157. Resosudarmo, B.P., Yamazaki, S. (2006). Mass Guidance vs. Farmer Field School: The Indonesian Experience. Paper presented in the Economics of Trade and Development Seminar Series, Economics Division, Research School of Pacific and Asian Economics ANU College of Asia and the Pacific, August 01. Roche, F.C. (1994). The Technical and Price Efficiency of Fertiliser Use in Irrigated Rice Production. Bulletin of Indonesian Economic Studies 30(1): 59-83. Rola, A.C., Pingali, P.L. (1993). Pesticides, Rice Productivity, and Farmers Health: An Economic Assessment. IRRI and World Resource Institute, Washington. Rolling, N., van de Fliert, E. (1994). Transforming Extension for Sustainable Agriculture: The Case of Integrated Pest Management in Rice in Indonesia. Agriculture and Human Value 11(2/3): 96-108. Schumann, S. (2002). Fast Estimation of Pesticide Risk Potential on Groundwater Through the Use of a Dye Tracing Technique. In Herrmann, A., Schumann, S. (eds.). Proc. Int. Workshop on Environmental Risk Assessment of Pesticides and Integrated Pesticide Management in Developing Countries 6-9 Nov. 2001 Kathmandu, Nepal. Landschaftsökologie und Umweltforschung 38, Braunschweig, 178-189. Singh, A. (2002). Modeling Overland Flow and Sediment Discharge from Agricultural Lands on Himalayan Foot Hills under Monsoon Conditions. In Herrmann, A., Schumann, S. (eds.). Proc. Int. Workshop on Environmental Risk Assessment of Pesticides and Integrated Pesticide Management in Developing Countries 6-9 Nov. 2001 Kathmandu, Nepal. Landschaftsökologie und Umweltforschung 38, Braunschweig, , 224-235. Sumaryanto, W., Siregar, M. (2003). Determinan Efisiensi Teknis Usahatani Padi di Lahan Sawah Irigasi. Jurnal Agroekonomi 21(1): 72-96. Tadesse, B., Krishnamoorthy, S. (1997). Technical Efficiency in Paddy Farms of Tamil Nadu: An Analysis Based on Farm Size and Ecological Zone. Agricultural Economics 16: 185-192. Toryanikova, R.V., Karaseva, T.A., Lutsenko, O.A., Nishonov, E. (2002). The Dynamics of Herbicides in the Water Environment of Uzbekistan. In Herrmann, A., Schumann, S. (eds.). Proc. Int. Workshop on Environmental Risk Assessment of Pesticides and Integrated Pesticide Management in Developing Countries 6-9 Nov. 2001 Kathmandu, Nepal. Landschaftsökologie und Umweltforschung 38, Braunschweig, 250-255. Trewin, R., Weiguo, L., Erwidodo, Bahri, S. (1995). Analysis of the Technical Efficiency over Time of West Javanese Rice Farms. Australian Journal of Agricultural Economics 39(2): 143-163. Villano, R., Fleming, E. (2006). Technical Inefficiency and Production Risk in Rice Farming: Evidence from Central Luzon Philippines. Asian Economic Journal 20 (1): 29-46.

186


Widodo, S. (1989). Production Efficiency of Rice Farmers in Java Indonesia. Yogyakarta, Gadjah Mada Press. Winarto, Y.T. (2004). The Evolutionary Changes in Rice-Crop Farming: Integrated Pest Management in Indonesia, Cambodia, and Vietnam. Southeast Asian Studies 42(3): 241-272. Yao, S., Liu, Z. (1998). Determinants of Grain Production and Technical Efficiency in China. Journal of Agricultural Economics 49 (2): 171-184.

Chapter 10 Rural Household Energy Consumption and Choice: A Case Study of Nanjing, Jiangsu Province, China Lina Shi 1, Xiaoping Shi 1, Nico Heerink2, and Shuyi Feng1 Abstract: Rural development in China is characterised by highly diversified patterns, including the patterns of transition to modern energy use. In relatively developed areas, rural households generally show a high degree of diversification in terms of income sources, off-farm activities and agricultural production and, as a consequence, in their access to and use of biomass energy. We analyze the factors driving biomass energy use and energy transition for 24 villages in six townships of Nanjing city, Jiangsu province, China. The analysis shows that household features, particularly household size and age of the head, have a significant impact on the transition from biomass to modern energy types. Participation in off-farm activities also facilitates the energy transition, while differences in accessibility of modern energy does not play a significant role in our case study area. Income levels of farm households have a negative impact on biomass energy use and a positive, but gradually declining, impact on the energy transition. Our analysis indicates that rural energy transition is in progress, but it is far from being complete. Biomass is currently underused as an energy source. Given the concerns over energy shortages in the long-run, policies to promote clean-energy use in rural areas should give high priority to how to move from traditional to modern biomass energy use.

Keywords: Energy transition, Farm household, Income, Off-farm employment, Bio-mass use

1 Introduction Rural energy consumption is an important focus of China’s national energy strategy. China is characterized by a dual economy, with the major part of the population living in rural areas and heavily relying on biomass fuels such as firewood and straw for cooking, heating and other purposes. After three decades of economic development, and with increasing incomes and development of rural infrastructure in rural areas, energy consumption for 1 2

China Centre for Land Policy Research, Nanjing Agricultural University, Nanjing 210095, P.R. China Development Economics Group, Wageningen University, Hollandseweg 1, 6706 KN, Wageningen, The Netherlands


188

Shi, Shi , Heerink, and Feng

rural farm households is now in transition towards modern energy use. According to the “White Cover Book of Energy Use and Policy” (published in December 2007), China will mainly rely on domestic energy sources, trying to optimize their usage, and will give high priority to rural energy consumption and enhancing the capacity for rural energy supply. Energy transition in developing countries entails that more and more traditional biomass fuels will be replaced by modern energy in household consumption along with economic growth and urbanization. In the rural areas, energy transition may also mean that the increasing demand for energy from households will be met by a combination of modern/commercial energy and biomass. There is some evidence for limited substitution of kerosene for biomass in India (Leach 1987) and Mexico (Sheinbaum et al. 1996). Limited evidence from rural China offers mixed results. Demurger and Fournier (2006) found that in 10 villages of Beijing municipalities increased wealth leads to reduced use of fuel wood and increased use of modern energy (coal). Chen et al. (2006), using data from three villages of Jiangxi province in Southeast China, found the same to be true for the two relatively rich villages. In a remote village in Southeast China, Shi et al. (2009) found that increased income leads to an increase of fuel wood consumption, indicating that fuel wood is not an ‘inferior good’ at very low income levels. According to studies on energy consumption, generally, energy transitions in rural areas very much depend on the income levels of farm households, accessibility of biomass energy and modern energy, end-uses of energy, etc (Jiang and O’Neill 2004; Hao 2005). Energy transition is determined by a number of factors that are different across regions and across income levels. To meet the increasing demand for energy by rural households and to reduce its impact on the rural environment, it is important to know which factors play a role in the transition towards modern rural energy sources and how biomass-use efficiency can be improved. There are some studies on rural energy consumption in China, focusing on energy use trends (Zhou et al. 2009; Jiang and O’Neill 2004; Hao 2005). Rural households may exhibit important differences during energy transition due to income levels, demographic conditions, availability of biomass etc. Especially in relatively developed areas, rural households generally exhibit a high degree of diversification in terms of income sources, off-farm activities and agricultural production and, as a consequence, in their access to and use of biomass energy. We may expect that rural households that focus on agricultural production have fewer incentives to switch to modern energy sources, but may also want to take

Rural Household Energy Consumption and Choice

189

initiatives to reduce the side-effects of bio-mass energy use, for instance by using bio-gas. The objectives of this paper are, therefore, (i) to examine the recent patterns of rural energy consumption in a relatively developed area in China; (ii) to analyze the factors that determine biomass use in rural areas (with a focus on the use of crop residues); (iii) to examine the factors that influence the transition towards the use of modern energy sources. To this end, we collected empirical data for the year 2006 based on 258 farm households from three districts, six townships and 24 villages in the rural area around Nanjing city. We use these data to estimate a Tobit model that analyzes the determinants of biomass energy use in a rural area and an ordinal Logit model that examines the factors influencing the transition of energy use from traditional biomass towards modern energy for different household groups. The remainder of the paper is organized as follows. Section 2 briefly presents the analytical framework. Section 3 describes the research area selected and data set used in this study. Section 4 describes the model used in the paper. Section 5 presents and interprets the results of the model, while section 6 summarizes the major conclusions and discusses policy implications.

2 Analytical framework Energy transition differs across countries and regions. For example, Foley (1995) argues that substitution is not a major feature of the energy transition in rural households, and additional demand for energy is met by commercial energy, but biomass use remains. Limited research in China shows that in most of cases biomass is substituted by coal (Demurger and Fournier 2006; Chen et al. 2006; Jiang and O’Neill 2004). But some other energy sources are also gradually being introduced into the daily life of rural households. In rural China, electricity is the most widely used energy source for lighting but is also gradually being used for cooking and heating. In some areas, farm households have also started to use bottled gas for cooking (Chen et al. 2006; Jiang and O’Neill 2004). Hence, energy transition may mean that farm households start from using biomass energy only to gradually using more and more combinations of a number of modern energy sources and biomass at the same time. Many studies have demonstrated the existence of this tendency (e.g. Campbell et al. 2003; Kebede et al. 2002; Foster et al. 2000 and Soussan 1987). A number of researchers have focused their studies on the determinants of energy transition. The ‘Energy Ladder’ theory shows that income level is one of most important factors that drive energy substitution (Gupta and Kohlin 2006; Michael et al. 2003). Research in India (Rao and Reddy

190


2007; Gupta and Kohlin 2006; Reddy 1995) and in Africa (Piana 2003) also found that income is a very important factor in energy transition. Increased income level is not only related to increased household food consumption and heating demand, but is also related to access to some new energy sources. For example, farm households with increased income levels may switch to bottled gas, but some may not due to an initial investment needed for using bottled gas, and lack of such capital may hamper use of such energy forms. By using different sources of macro level data, Jiang and O’Neill (2004) suggest a general picture for the energy transition trends in rural China, namely that transition takes place in two stages. In the first stage, at relatively low income levels, demand for all fuels increases, but use of commercial energy increases faster than the use of biomass. It was also found in a study of one remote village of Jiangxi province that fuel wood consumption increased with increased income levels (Shi et al. 2009). In the second stage, at higher income levels, commercial energy use continues to grow, but the use of biomass declines as preferences for fuel types shifted. Other researchers have argued that they did not observe energy transition with increased income levels in rural areas, rather only in urban areas (Kelakar and Nathan 2004). Baland et al. (2007) found that, due to lack of other ‘clean’ sources of energy, rural households in Nepal have not followed the model assumed by the ‘Energy Ladder’ theory, with bio-fuel consumption having increased with increased income levels. Use of more modern energy sources also depends on the opportunity costs of female labour (Kelakar and Nathn 2004). When these costs do not change, increased income level does not necessarily lead to a switch to modern energy. Because saving time spent on bio-fuel collection by female labour does not change household income, increased energy expenditure is not compensated by time-saving and new energy forms are, consequently, not adopted. Therefore, giving more opportunity to female labour may encourage energy transition. In other parts of the world, for example in India (Rosenzweig and Foster 2003) and Kenya (Patel 1995), bio-fuel consumption actually increases, to a point, when income increases. A study done by Ouedraogo (2006) found that fuel wood is a kind of ‘transition good’, because at certain lower income levels the possibility for increasing fuel wood consumption is higher when income level increases. When income rises to a certain level, fuel wood consumption will start to decrease with further increase of income. However, the experience from China may differ from countries mentioned above. More and more farmer households in China are


191

participating in off-farm employment. Therefore, they not only obtain higher overall income through higher wage income from off-farm activities, but also increase the opportunity costs of labour for fuel wood collection and processing of agricultural residuals for cooking. Thus, the effects of increased income and labour opportunity costs may jointly lead to energy transition. Shi et al. (2009) found in a poor area that fuel wood consumption does not increase with increased income from off-farm employment and reduced labour availability in agricultural production (including fuel wood collection). It seems, then, that energy transition is not significant in rural China. However, off-farm employment may lead to energy transition in relatively developed areas of China, because average income levels there are much higher compared with poor areas, given the argument that a certain income level act as a ‘switching’ point for energy transition. Some studies (Masera et al. 2000) have argued that increased income does not lead to energy transition because modern energy consumption requires a cash income, and rural households do not have stable cash incomes in some regions. Therefore, in such areas, increased income levels do not lead to rural households changing according to the model of the ‘Energy Ladder’, as factors other than income are equally important, such as the stability of cash income. Policy related to energy use may have an important impact on energy transition. Subsidies to some energy sources may encourage their use. In some cases, policy has tried to encourage farm households to use more modern energy, but not always successfully. For example, in Kolkata (India), a government subsidy of LPG did not lead to increased LPG use and decreased traditional fuel wood consumption. The reason is that, although LPG is cheaper, its supply cannot be guaranteed (Gupta and Köhlin 2006). Apparently then, in addition to income, energy accessibility and supply stability are very important to energy transition. As we have just discussed, due to lack of stable supply of modern energy forms, farm households may need to stick to traditional energy use, even with increasing income levels. Campbell et al. (2003) examined electricity use to replace traditional energy use in Zimbabwe and found that people at different income levels converted to electricity use at the same speed, simply in response to increased supply of electricity in Zimbabwe. Lack of rural infrastructure, such as roads, may also hamper the use of modern energy (Chen et al. 2006). The features of various energy sources and their end-use may influence choice of energy for households, with the safety of an energy form and the capacity of fixing end-use being two factors found to be important in a

192


study by Masera et al. (2000). Difficulties in storage and use of a particular form of energy lead to its being less used (Leach 1987). The initial investment required for end-use of LPG may steer some farm households away from choosing to use it (Kebede et al. 2002). However, in some cases, the price of LPG investment is not effectively reducing its use because, for richer households, the cost for end-use LPG takes a small share of overall household expenditure (Gupta and Köhlin 2006). Burning traditional energy, such as fuel wood, may cause respiratory diseases. Awareness of the effects of indoor pollution on health, especially for women and children, may reduce use of traditional energy and increase use of more ‘clean’ modern energy. However, Masera (2000) has argued that the impact of indoor pollution on health often shows up in the long term, and in the short term households do not have incentives to reduce it. But perhaps because households may simultaneously use a number of different energy sources, it is difficult to distinguish the impact from each one. Some researchers also argue that subsidizing modern energy sources will encourage use of ‘healthy’ energy3 instead of ‘inferior’ energy in urban areas, but in rural areas the subsidized energy is still relatively expensive compared with traditional energy collected by women, with lower opportunity costs (Kelakar and Nathan 2004). Household size may affect energy consumption and energy transition. Larger households need more traditional energy for subsistence, though there may be economies of scale in energy consumption; thus, energy use per capita in a large family may be smaller than that in a small family. The study by Chambwera and Folmer (2007) confirmed that, to a degree, the share of electricity in total energy consumption increases with increased household size, but further increase of household size leads to a rising share of traditional energy sources. The education levels of household members may also be important in making energy choices. Rao and Reddy (2007) found that households with higher education levels will choose more LPG instead of fuel wood. The same is found by Chen et al. (2006), though they explain that higher education levels may not mean increased awareness of the need for environmental protection, rather just indicating that children in school and members participating in off-farm employment may reduce their fuel wood consumption at home. Consequently, the number of educated people in a family does not necessarily affect energy choice. The education level of the household head does not, for example, determine use of LPG in Brazil, but the education level of women in a family does matter for energy choice, because they are in charge of family energy use (Heltberg 2004). 3

‘Healthy energy’ means that burning of energy will not produce air pollution.

193


Meanwhile, Gupta and Köhlin (2006) found that the age of the household head does affect choice of LPG in India, because the young generation is not able to buy end-use of LPG, but older household heads in families can.

3 Description of the research site Data used in this study is from a strategic sample of 3 districts, 6 townships, 24 villages and 258 farm households in Nanjing city, capital of Jiangsu province, a relatively developed are in South China. Three districts and 6 townships were selected according to their different economic development levels, distances to the core area of Nanjing city, differences in agricultural development and natural conditions. Farm households were selected randomly in the villages. The questionnaire used in the interview included demographic features of households, such as employment of family members, income from different sources, agricultural production, expenditures (including energy use) for 2006. Table 1 gives basic data on the three districts. Jiangning district of Nanjing city has the largest area and Lishui district the smallest. The districts also show differences in distance to the core area of Nanjing city, with Jiangning district just bordering it. Lishui district has forest cover due to being in a mountainous area.

Table 1: Basic situation of three sampled districts Liuhe District

Jiangning District

Lishui District

Total areas (Square km)

1467

1573

1067

Population (thousands)

872.2

845.5

405.9

Share of rural population in total (%)

60.09

61.65

76.5

Arable land (thousands mu)

1,100

628

460

Total sowing areas (thousands mu)

91.46

67.74

67.46

Topography

Hilly

Hilly

Hilly and lower mountain

Distance to core area (km)

60

15

40

Source: Statistical yearbook of Nanjing city (2007) and own survey (2007).

Table 2 shows the income levels of the sampled farm households in 2006. Average income levels per capita in Jiangsu province and nationally are 5,813 and 3,587 yuan in 2006, respectively. Income per capita of sampled households is 27 percent higher than that of Jiangsu province and is more than double the national average. Because Nanjing is the provincial capital, income levels there are relatively higher than in the rest of Jiangsu province, which itself is a relatively developed area in China. On average, 74.3 percent of income of the farm households in our sample is from off-farm activities, and only 16.7 percent is from agricultural production. About 9 percent of income is from other sources,

194


for example subsidies from government, selling assets, etc. Farm households have great differences in total income and income from agricultural production, but show relatively smaller differences in off-farm income.

Table 2: Income composition of farm households

Income per capita (Yuan) Farm household income (Yuan) S.D. Share in total income level (%) Source: Own survey (2007).

Off-farm income 5113 20393 7709 74.3

Farm income 1424 4562 18900 16.7

Other income 829 2476 5966 9.0

Total income 7366 27431 33127 100

Farm households in our research area consume different types of energy, such as firewood, straw, coal, LPG and electricity. Table 3 shows the breakdown of average total energy use by fuel type. On average, total energy use in per capita terms is 337.5 standard coal equivalent (kgce4), and biomass (firewood and straw) is more than half of total energy consumption (58%). Electricity is the second largest energy type used, with 29 percent of total energy use, whereas coal and LPG are both less than 10 percent. The total energy consumption in our sample is quite close to the energy consumption levels of rural households in 1999 (Jiang and O’Neill 2004)5. But the proportion of biomass in our sample is smaller than in Jiang and O’Neill (2004), where biomass still is the main source of energy, accounting for 72% of total use. The proportion of electricity use in our sample is much higher than that of 5% in Jiang and O’Neill (2004). In Jiang and O’Neill (2004)’s data, coal and coal products consumption is about 21% of total energy use, while in our sample it is only 7.8%. Around 2000, the central government started a project to reduce the price of electricity and improve the quality of electricity in rural China, which may explain part of the difference between our sample and the data from Jiang and O’Neill (2004). Increases in the number of household appliances and gadgets also can explain the higher share of electricity compared to coal.

4 5

1 kgce = 11,825 kjoules = 11,216 BTU. The data used by Jiang and O’Neill (2004) is from the 1999 National Rural Household Survey, an annual survey conducted by the China Rural Socio-Economic Survey Division of the State Statistical Bureau, and they took around 80% of the surveyed households randomly.

195


Table 3: Rural energy consumption by energy types in 2006 Firewood Straw and stalks Coal LPG Electricity Total commercial energy Total Source: Own survey (2007).

Per capita (kgce) 109.7 82.5 26.4 19.8 99.0 145.2 337.4

% 32.5 24.5 7.8 5.9 29.3 43.0 100

Table 4 shows farm households using different types of energy in our research areas. Farm households using a single type of energy are rare; most of them use at least two types. More than 86 percent of farm households either use firewood or straw and stalks, and half of them use one type of biomass. All the households use electricity, and more than 90 percent use LPG. Coal is only used by half of the households. In our data, the tendency is the same as the data from Jiang and O’Neill (2004): biomass plus electricity is the most common combination.

Table 4: Percentage of farm household using different types of energy Biomass* Firewood

Straw and stalks 134

No. of households using 222 150 Percentage of 86.0 58.1 51.9 households use in total * Biomass use includes both firewood and straw and stalks. Source: Own survey (2007).

Coal

LPG

Electricity

150

236

258

58.1 91.5

100

Not all biomass is used as an energy source; especially straw and stalks, mainly from rice and wheat in our research areas, are left unused and thrown away as waste. Table 5 shows where straw and stalk are consumed. Straw and stalks have three main destinations: used as energy, burnt or back to the field as fertilizer. More than 22 percent of straw and stalks is burned as waste and generates air pollution. Therefore, there still is great potential for better use of these materials as an energy source, at least around Nanjing city.

196


Table 5: Straw and stalk use for different purposes Straw and Stalk 905999

Energy

Burning

Not Back to Feedstuff used field 10835 53125 235948

Other

Quantity (Jin) 354733 203339 48020 Share in total 100 39.2 22.4 1.2 5.9 26.0 5.3 (%) * The quantity of straw and stalks is calculated according to a standard coefficient of crops for straw and stalks. Source: Own survey (2007).

4 Model and results The main purposes of this study are to (i) analyze factors determining biomass energy use and (ii) reveal factors determining energy transition. Hence, our analysis has two parts. In the first part of this section, we are going to examine which factors determine the use of biomass energy by using a Tobit model. In the second part of the section, we then try to analyze which factors are important for farm households in switching their energy use from traditional biomass use to modern energy use by using an ordinal Logit model. 4.1 Biomass use model Biomass energy use by farm households is converted to standard coal equivalent (kgce). If a household uses biomass energy, then the amount of biomass will be larger than zero; otherwise it will be zero. Therefore, we choose to use a Tobit model to estimate the factors determining biomass use, specified as Yi = C0 +C1X1i +C2X2i +C3X3i +εi

(1)

where Yi = variable representing amount of biomass use by farm household i; X1i= (column vector of) demographic and human capital characteristics of farm household i; X2i= (column vector of) household resources and characteristics of farm household i; X3i= (column vector of) local institutional factors and village characteristics; C0, C1, C2, C3 = (row vectors of) coefficients to be estimated; εi = error term with standard properties.


197

4.2 Energy choice model To be able to examine the factors determining the choice of different combinations of energy sources by farm households, we also calculated energy use (in standard coal equivalent) for modern energy types. Most farm households in our research areas consume a combination of energy types, but the share of modern energy use is different across farm households. There are some farm households that only use modern energy sources. To examine the households that consumed different combinations of energy source, we applied an ordinal Logit model. Three categories of farm households were distinguished. The first group consists of those households who consumed more energy from traditional sources, where the share of traditional energy is more than half of total energy use. The second group of households consumed modern energy sources as over 50 but less than 100 percent of total energy. The third group is the households who only use modern energy (i.e. 100 percent). The ordinal Logit is specified as: ln f (Pj) = αj +βjXi

(2)

where f(Pj) is the distributional probability function of Pj, and Pj is the probability of farm households choosing energy use combinations j: j can be 1 (group 1), 2 (group 2) or 3 (group 3). Xi stands for the independent variables, which include household characteristics and resources, institutional factors and village characteristics; αj is the intercept when farm households choose the energy use combination j; and βj is the vector of coefficients of Xi. The marginal effects of independent variables on household energy use combinations do not equal the sum of coefficients of Xi; therefore, we need to calculate the marginal effects separately. Table 6 gives a summary of variable definitions and sample statistics that are used in the regression. Table 7 shows the expected sign of the explanatory variables in the two models. We will, first, try to explain which factors determine the use of biomass energy. The first variable listed in the table is the age of the household head. Older household heads normally would prefer to use traditional energy; therefore, households with an older household head will increase their biomass consumption. Younger household heads will use more modern energy, since they prefer to have clean indoor air and environment. Education of household heads is also important, as it may increase knowledge about the impact of indoor pollution on health and lead to less use of biomass.

198


Table 6: Variable definitions and sample statistics Independent variables Head_age Head_edu Homesize

Variable definitions

Mean

S.D.

Min.

Max.

52.83 5.98 3.32

10 4.05 1.35

29 0 1

80 15 8

1.51

0.47

1

5

1.58

0.93

0

4

Contracted_land Income_tot Income_off Income_agri Income_oth

Age of household head Number of years of schooling Farm household size Total population divided by number of laborers Number of laborers participating in off-farm activities Contracted land area Farm household total income Off-farm income of farm household Agricultural income of farm household Other source of income

5.12 27430 20393

3.39 33127 27709

0 97 0

24 284460 230000

4561 2475

18900 5966

-985 0

269460 50115

Rent_out Distance_lpg HenLiang ZhuZheng Tangshan Guli Honglan

Rented out land? (0 = no/1 = yes) Distance to nearest LPG station Henliang dummy variable Zhuzheng dummy variable Tangshan dummy variable Guli dummy variable Honglan dummy variable

0.42 4110 0.15 0.18 0.12 0.18 0.20

0.49 13707 0.36 0.39 0.33 0.39 0.40

0 50 0 0 0 0 0

1 100000 1 1 1 1 1

Denp_ratio Pop_off

Larger household may have higher energy consumption levels, but they may also have economies of scale. Otherwise, an increase of population will increase consumption of energy (including biomass energy). The number of dependents (children and elderly) in a household can have mixed effects on energy use. On the one hand, more dependents in a household implies a need for higher energy use to satisfy their food and other needs. On the other hand, more dependents also means that more time is needed to care for them and, therefore, less time is available for processing firewood and straw. A higher share of family members working off-farm also means that there is less time available for firewood collection and using straw and stalks. A large area of irrigated land contracted by a household from the village group or village committee is expected to lead to larger straw and stalk output, which will provide biomass energy to the households. Households with higher income may possibly consume more modern energy, because those households will be able to afford the initial investment for end-use of energy and have cash income for obtaining a regular supply. The squared term is added to the equation to check for nonlinearities in the impact of income. Local institutions such as land markets may affect the choice regarding participation in off-farm employment and, therefore, reduce the time

199


availability for firewood collection and processing of straw. Renting out of land, defined as a dummy variable in our model, is expected to reduce the use of biomass energy of the households. Distance to LPG will negatively affect use of LPG, because it may take much effort for households to buy it, given poor road conditions in the rural areas. Finally, five dummy variables for Henliang, Zhuzheng, Tangshan, Guli and Honglan townships are added to the model to control for unobserved factors that systematically differ between six townships. Lower transportation costs for households in these townships and villages, which are close to major roads, may translate into lower prices for modern energy, thereby encouraging households to use it more than traditional sources.

Table 7: Expected signs of independent variables Independent variables Head_age Head_edu Homesize

Tobit + +

Ordinal logit + -

Denp_ratio

+/-

+/-

Pop_off

-

+

Contracted_land Income_tot

+ -

+

Income_tot2

+/-

-

Rent_out

-

+

Distance_lpg

+

-

HenLiang ZhuZheng Tangshan Guli Honglan

+/+/+/+/+/-

+/+/+/+/+/-

The second part of Table 7 gives (by way of hypotheses to be tested) the expected signs of the explanatory variables in explaining the energy transition from traditional biomass to modern energy use. It suggests that young people may be more likely use modern energy, and consequently less traditional energy, given a concern over health impacts and indoor pollution. With higher education of the household head, they may know more about the impacts of indoor pollution and will use more modern energy. Therefore, education of household head may be important to push the transition of energy use. With smaller household size, people may start to use more modern energy, which is more convenient. Hence, the impact on household size of the ‘one-child’ policy may facilitate the transition of energy use in rural China.

200


The number of dependents (children and elderly) in a household may have mixed effects on energy transition, like with biomass consumption. On the one hand, a larger number of dependents in a household requires the consumption of more energy; therefore, these households may consume more biomass to meet their requirements. On the other hand, due to their having to take care of children and elderly, they may have less time available to collect and process biomass energy. Household members working off-farm activities will have more chance to bring more cash income back home; consequently, these households may increasingly seek to replace traditional energy and will be more likely to shift to more modern energy use. Households with relatively large areas of contracted land will have more sources for biomass energy, which may reduce the incentives to shift to modern energy. Income level is also important for energy transition. Given the accessibility of modern energy, higher incomes will give households greater opportunity to afford modern energy. Therefore, higher income levels may increase the shift towards modern energy use. The square term of income level will show to what extent income level accelerates energy transition. We may expect that, as things are at present, farm households will not completely stop using biomass energy. For the lower income level households, increased income levels will facilitate their energy transition, but for the higher income level households, further increases in income level do not further improve the energy transition. Local institutions, such as land rental markets, may enable famers to be more specialized in certain activities. Farm households that specialize in agricultural production may still consume more biomass, because they have enough readily available sources of it. But the households that work more at off-farm activities, may only seldom use biomass energy. More convenient modern energy services, for example LPG, may also help the transition of energy use. Finally, the dummy variables for different townships are to indicate that local situations and transportation costs may be different and, due to this, the energy transition in each location may also show some differences.

5 Empirical results The estimation results of the Tobit model are presented in Table 8. The results for demographic and human capital characteristics show that age of household head has a significant impact on biomass energy use, and households with older household heads are most likely to use more traditional energy. The education level of household heads does not have significant impact on biomass energy consumption, which may indicate that households give little concern to indoor pollution or its impacts on

201


health in the short-term, though maybe in the long-term. Large households have higher biomass consumption, which may be due to larger household size biomass use being more efficient. This result is the same as the finding of Jiang and O’Neill (2004) that household size is one of the key factors for determining biomass use. However, number of dependents (children and elderly) in a household does not significantly affect the amount of biomass use, which Jiang and O’Neill (2004) did not study.

Table 8: Regression result of biomass consumption Tobit model Dependent variable: biomass consumption Coef. t Head_age

8.30*

Head_edu

-8.63

1.85 -0.82 ***

Homesize

121.97

Denp_ratio Pop_off

135.38 -7.60

1.60 -0.14

Contracted_land

28.33**

2.24

Income_tot Income_tot

-0.007 2

**

2.04E-08 ***

Rent_out -256.6 Distance_lpg 0.002 HenLiang -190.22 ZhuZheng -198.71 Tangshan 132.32 Guli 34.03 Honglan 53.99 Constant -159.98 Log likelihood -1761.66 LR chi2 56.25*** Pseudo R2 0.016 * Denotes statistically significant at 10% level. ** Denotes statistically significant at 5% level. *** Denotes statistically significant at 1% level.

3.31

-2.08 1.46 -3.04 0.73 -1.38 -1.39 0.92 0.25 0.41 -0.45

202


Table 9. Regression results regarding energy transition

Coef.

Ordinal Logit Dependent variable: energy transition Z

Head_age

-0.04**

-2.36

Head_edu

0.002

0.04

Homesize

-0.70***

-4.32

Denp_ratio Pop_off

-0.24 0.41*

-0.72 1.94

Contracted_land

-0.06

Income_tot Income_tot

0.00004 2

-1.24 ***

-1.22e-10 **

***

Rent_out 0.71 Distance_lpg -0.000012 HenLiang -0.013 ZhuZheng 0.58 Tangshan -0.43 Guli 0.20 Honglan -0.23 _cut 1 -2.60 _cut 2 -1.01 Log likelihood -194.63 LR chi2 72.18** Pseudo R2 0.1564 * Denotes statistically significant at 10% level. ** Denotes statistically significant at 5% level. *** Denotes statistically significant at 1% level.

3.50 -2.67 2.26 -1.15 -0.02 1.12 -0.72 0.39 -0.45

The number of laborers in a family participating in off-farm activity does not significantly affect the amount of biomass use. And larger availability of biomass energy (cultivated land) leads to higher use of biomass. Indeed, as expected, households with higher income will consume less biomass energy. However, the squared income term is insignificant. Rising incomes are not likely to show an accelerated pattern of transition in energy use. Biomass energy is a kind of ‘inferior good’ in our research areas, and households will consume less with increased income levels. Biomass consumption is also affected significantly negatively by local land renting activities. Households that have decided to leave farming for off-farm activities will definitely use less biomass energy, but this may partly be so because they have lower biomass energy availability (due to their land having been rented out).


203

Distance to a major, modern energy service (LPG) does not show a significant impact on biomass use. Consumption of LPG is only around 5% of total energy use (Table 3) and, therefore, households only show small differences there. None of the dummy variables for the townships are significant. This may indicate that factors at township level affecting household energy consumption are negligible. The estimation results of the ordinal Logit model analysis are presented in Table 9. The estimated equations perform satisfactorily in terms of goodness of fit. Due to the marginal effects of independent variables on household energy use, their combination does not equal the sum of coefficients of the variables, so we calculated the marginal effects of independent variables on household energy use, as shown in Table 10, and will focus on explanation of these results. Age of household head has a significant positive impact on the use of biomass energy and a negative impact on the use of modern energy. It confirms the tendency of energy transition between different generations of household heads. However, the education level of household heads does not show significant effects on energy transition. Household size has a positive impact on the consumption of biomass energy and negative impacts on modern energy use. Smaller household size is one of the consequences of ‘the one-child’ policy in China. The number of dependents does not show significant effects on energy transition. The number of household members working in off-farm activities not only shows a significantly negative impact on biomass energy use, but also positively affects modern energy use. As the number of household members participating in off-farm activities increases, the opportunity costs of collecting or processing biomass energy grow, and households more involved in off-farm activities will consume less biomass energy. Availability of one type of biomass energy (contracted land) does not show a significant impact on energy transition. Household income level does play an important role in energy transition. A higher income level leads to more modern energy consumption. The square term of income level for the households is significantly negative for modern energy use, which indicates that the impact becomes less at higher income levels. Renting out land is positively related to the energy transition. Households less involved in farming activities will also adapt their living styles to more closely match those of urban people. Distance to LPG services does not show any impact on energy transition, and different townships do not show any differences in energy transition either.

204


Table 10: Marginal effects on energy transition (ordinal Logit model) Dependent variable: biomass consumption Group 1 Coef.

Z **

Head_age

0.009

Head_edu

-0.0004

Homesize

0.15

Denp_ratio

0.05

***

Pop_off

-0.09

Contracted_land

0.01

Income_tot 2

Income_tot

*

-8.68e-06 2.68e-11 **

***

***

Group 2 Coef.

Z **

2.37

-0.006

-0.04

0.0003

4.42

-0.096

0.72

-0.03 *

-1.95

0.06

1.24

-0.008

-3.51

***

5.45e-06

*** **

Group 3 Coef.

Z **

-2.26

-0.003

0.964

0.0002

-3.78

-0.06

-0.72

-0.02

0.04

***

-4.04 -0.72

*

0.06

0.03

-1.22

-0.005

3.18

-2.29

3.24e-06

1.91 -1.23 ***

3.27

**

-2.56

2.67

-1.68e-11

-2.52

-9.99e-12

-2.24

0.10

**

2.23

0.06

**

2.04

Rent_out

-0.16

Distance_lpg

2.67e-06

1.15

-1.67e-06

-1.14

-9.94e-07

-1.14

HenLiang

0.003

0.02

-0.002

-0.02

-0.001

-0.02

ZhuZheng

-0.14

-1.09

0.08

1.15

0.06

1.0

Tangshan

0.09

0.76

-0.06

-0.74

-0.03

-0.80

Guli

-0.04

-0.38

0.03

0.39

0.02

0.37

Honglan

0.05

0.46

-0.032

-0.46

-0.02

-0.47

* Denotes statistically significant at 10% level. ** Denotes statistically significant at 5% level. *** Denotes statistically significant at 1% level.

6 Conclusions In this paper we have analyzed factors driving choice of energy use and energy transition for 24 villages in six townships of Nanjing, Jiangsu province, China. The data shows that, on average, off-farm income is more than 74 percent of total income for these farm households, and income from agricultural production is only 17 percent of the total (Table 2). Nevertheless, biomass is still quite important, comprising 57% of total energy consumption. Electricity is the most popular and important modern energy, and all households use it to some degree. The second is coal, which accounts for 7.8% of total energy use. Most of the studied households use a combination of energy sources, with only a few of them using modern energy exclusively. At the same time, we also found that straw from rice of household fields was being not completely used as an energy source, indicating that the potential of biomass energy is not fully exploited. Our analysis of the factors explaining biomass use show that the age of household heads significantly impacts biomass use. Additionally, households with more members are more likely to use more biomass. Households specialized in agricultural production (not renting-out land or renting-in land) are more likely to consume more biomass energy. Income


205

is very important for biomass use, and a lower income level implies higher use of biomass. Biomass is now increasingly seen as a kind of ‘inferior good’ for farm households in our research areas. The analysis of factors driving farm households to switch from biomass use to modern energy (Tables 8, 9 and 10) shows that household features such as age of household head and household size have significantly impacted the transition from biomass to modern energy types. Younger household heads and smaller households are more likely to use more modern energy forms. Household members’ participation in off-farm activities also facilitates energy transition, as less time is available for collecting and processing firewood and straw. Availability of biomass and accessibility of modern energy do not, however, have significant impacts on energy transition, due to an over-supply of biomass and lack of variation in our sample regarding accessibility of modern energy. The differences among townships also exhibit no differences in terms of impact on energy transition. However, income levels of farm households do significantly impact energy transition. Lower income households would prefer to use more biomass energy and, with increased income levels, households will switch more to modern energy. But at certain income levels, households will reduce their rate of use of modern energy and may keep a certain share of biomass energy. Our analysis indicates that rural energy transition in (the studied region of) China is in progress, but it is far from being complete. Compared with the data from 1999, as analyzed by Jiang and O’Neill (2004), we found that biomass use still takes quite a high share in total energy use and has only decreased to a small extent. At the same time, biomass is not being well used in rural areas, given the concerns over energy shortages in the longrun. Therefore, policies to promote clean-energy use in rural areas should give high priority to how to move from traditional biomass to modern energy, such as implementation of the gasification of biomass and bio-gas programs. These will not only solve the problem of traditional biomass use, but also provide opportunities for better use of biomass energy in rural areas.

Acknowledgements The authors thank the participants of the international conference “Greening Asian Growth - Economic Transition and Sustainable Agricultural Development in East and Southeast Asia”, 29-30 October 2008, Nanjing, China. Financial support from the SinoDutch program of Strategic Alliances of the Royal Netherlands Academy of Sciences (KNAW) and the Chinese Ministry of Science and Technology (MOST) and the

206


National Natural Science Foundation of China (70773057 and 70833001) are gratefully acknowledged.

References Baland, J.M., Bardhan, P., Das, S. et al. (2007). The Environmental Impact of Poverty: Evidence from Firewood Collection in Rural Nepal. Mimeo. Campbell, B.M., Vermeulen, S.J., Mangono, J.J., Mabugu, R. (2003). The Energy Transition in Action: Urban Domestic Fuel Choices in a Changing Zimbabwe. Energy Policy 31(6): 553-562. Chambwera, M., Folmer, H. (2007). Fuel Switching in Harare: An Almost Ideal Demand System Approach. Energy Policy 35: 2538–2548. Chen, L., Heerink, N., van den Berg, M. (2006). Energy Consumption in Rural China: A Household Model for Three Villages in Jiangxi Province. Ecological Economics 58: 407-420. Démurger, S., Fournier, M. (2006). Rural Poverty and Fuelwood Consumption: Evidence from Labagoumen Township (China). Paper presented at International Seminar on Transition towards Sustainable Rural Resource Use in Rural China, held 22-24 October, 2006 in Kunming, China. Available at: http://www.gate.cnrs.fr/ perso/demurger/Demurger_Fournier_110706.pdf. Foley, G. (1995). Photovoltaic Applications in the Rural Areas of the Developing World, World Bank Technical Paper Number 304, Energy Series, Washington, DC. Foster, V., Tre, J-P., Wodon, Q. (2000). Energy Consumption and Income: An InvertedU at the Household Level. World Bank Poverty Group. Gupta, G., Köhlin, G. (2006). Preference for Domestic Fuel: Analysis with Socioeconomic Factors and Rankings in Kolkata, India. Ecological Economics 57(1): 107-121. Gundimeda, H., Köhlin, G. (2008). Fuel Demand Elasticities for Energy and Environmental Policies: Indian Sample Survey Evidence. Energy Economics 30(2): 517-546. Hao, X. (2005). Rural Energy Development in China, (keynote talk), NISP Dissemination Workshops 2005, Beijing, January, 14-16, available on: http://ehs.sph.berkeley.edu/hem/printpage.asp?id=29. Heltberg, R. (2004). Fuel Switching: Evidence from Eight Developing Countries. Energy Economics 26: 869– 887. Jiang, L., O’Neill, B.C. (2004). The Energy Transition in Rural China. International Journal of Global Energy Issues 21(1-2): 2-26. Kebede, B., Bekele, A., Kedir, E. (2002). Can the Urban Poor Afford Modern Energy? The Case of Ethiopia. Energy Policy 30: 1029-1045. Kelakar, G., Nathan, D. (2004). Gender Relations and the Energy Transition in Rural Asia. The report of the Collaborative Research Group on Gender and Energy (CRGGE). Available at www.energia.org/crgge.


207

Leach, G. (2087). Energy Transition in South Asia' in Transitions between Traditional and Commercial Energy in the Third World, Surrey Energy Economics Centre, Discussion Paper Series No 35, University of Surrey. Masera,O.R., Saatkamp, B., Kammen, D.M. (2000). From Linear Fuel Switching to Multiple Cooking Strategies: A Critique and Alternative to the Energy Ladder Model. World Development 28(12): 2083-2103. Michael, A., Köhlin, G., Persson, R. (2003). Fuel Wood Revisited: What has Changed in the Last Decade? CIFOR Occasional Paper No. 39. Ouedraogo, B. (2006). Household Energy Preferences for Cooking in Urban Ouagadougou, Burkina Faso. Energy Policy 34: 3787–3795. Patel, S.H., Pinkney, T.C., Jaeger, W.K. (1995). Smallholder Wood Production and Population Pressure in East Africa: Evidence of an Environmental Kuznets Curve. Land Economics 71(4): 516–531. Piana, G. (2003). Energy & Gender Issues: Reflections and Strategies on the Road from Johannesburg, A special focus on African and Asian Rural Areas, Rome, FAO. Rao, M.N., Reddy, B.S. (2007). Variations in Energy Use by Indian Households: An Analysis of Micro Level Data. Energy 32: 143-153. Rosenzweig, M., Foster, A. (2003). Economic Growth and the Rise of Forests. Quarterly Journal of Economics 118(2): 601-637. Sheinbaum C., Martínez, M., Rodríguez, R. (1996). Trends and Prospects in the Mexican Residential Energy Use. Energy 21(6): 493–504. Shi, X., Heerink, N., Qu, F. (2009). The Role of Off-farm Employment in the Rural Energy Consumption Transition – A Village-level Analysis in Jiangxi Province, China. China Economic Review 20(2): 350-359. Soussan, J. (1987). Fuel Transitions Within Households. In: Transitions Between Traditional And Commercial Energy. The Third World Surrey Energy Economics Centre, Discussion Paper Series No 35, University of Surrey, UK. Statistical Yearbook of Nanjing City (2007) Compiled by Nanjing Statistical Bureau, Nanjing Publishing House, (in Chinese). Wang, X., Feng, Z. (1996). Survey of Household Energy Consumption in China. Energy 21(7/8): 703-705. “White Cover Book of Energy Use and Policy” published in Dec., 2007. by State News Agency, China. (in Chinese) Zhou, Z., Wu, W., Wang, Z. (2009). Analysis of Changes in the Structure of Rural Household Energy Consumption in Northern China: A Case Study. Renewable and Sustainable Energy Reviews 13(1): 187-193.

Chapter 11 Community-Based Aquaculture for Poverty Reduction: Institutional and Technical Options for Sustainable Resource Use Christine Werthmann1 and Chi Mai Thi Truc2 Abstract: We present findings from Action Research conducted within a community-based aquaculture project in Vietnam. The project aims at increasing food security and income of rural households in the Mekong Delta of Vietnam, where people deprive their livelihoods mainly from natural resources. Within this project, farmers culture fish communally in large water bodies and agree on sharing arrangements concerning the investments and benefits. In comparison to other sites, profits have been rather low in the project sites presented here. This is due to several reasons, including the time of harvesting and marketing. Further, several aquaculture groups report conflicts within groups as well as with nonproject members that reduced the motivation to continue the project. The research on technical and institutional options for community-based aquaculture presented here, shows that institutional issues play a major role for the implementation of the approach. Keywords: Community-based aquaculture, Food security, Poverty, Mekong Delta, Vietnam

1 Introduction Poverty is still very high in the rural floodplain areas of the Vietnamese Mekong Delta. People there derive their livelihood from natural resource use, especially through fishing and farming. The Mekong Delta region accounts for almost half of Vietnam’s rice and fish production and is also an important source for foreign exports (Ratner 2003). Thus, greater access to natural resources and their sustainable management can play a significant role in poverty reduction in Vietnam. The underlying assumption of the The WorldFish Center /Research Institute for Aquaculture No.2 (RIA No. 2) cooperative approach is that seasonal waterbodies (flooded crop fields, ponds and reservoirs in irrigation schemes) can 1 2

PhD student: Philipps-University Marburg, CGIAR System-wide Initiative on Collective Action and Property Rights (Capri), WorldFish Center Researcher: Research Institute for Aquaculture No. 2, Ho Chi Minh City, Vietnam

Beckmann, V., D.H. Nguyen, X. Shi, M. Spoor, J. Wesseler (eds.) (2010). Economic Transition and Natural Resource Management in East- and Southeast Asia. Aachen: Shaker Publisher, pp. 209-220.

210

Werthmann and Mai Thi Truc

be communally managed by all stakeholders under equitable and sustainable sharing arrangements. Experiences from a ten-year WorldFish project (Community-Based Fisheries Management: CBFM-2) in Bangladesh showed that livelihoods of fishers and poor communities can be improved – with 23,000 direct beneficiaries in Bangladesh – through community-based fisheries management and that, simultaneously, environmental issues can be addressed and fisheries resources improved in terms of production, diversity and sustainability (WorldFish 2007). Due to the successful implementation of the approach in Bangladesh and also in Vietnam in 2005, the WorldFish initiated a five year interdisciplinary research project (2005-2010) in three water basins, namely the Indus-Ganges (Bangladesh), Niger River basin (Mali) as well as the Mekong Delta (Cambodia and Vietnam), called the "Consultative Group on International Agricultural Research (CGIAR) Challenge Program on Water and Food: Community-Based Fish Culture in Irrigation Systems and Seasonal Floodplains". The project's main activity is to integrate the culture of fish and other living aquatic resources into existing water use systems and, thus, to enhance the productivity of seasonally occurring floodwaters. This is intended to contribute to reducing poverty, generating employment and increasing income of all classes of rural society in floodplain and irrigated areas. Research on institutional arrangements and technical options for community-based aquaculture in different hamlets of four different floodprone areas in the Lower Mekong area of Vietnam is now being conducted in collaboration with RIA No. 2. Expected outputs of this research are, among others, the development and testing of technical options for integrating fish and other living aquatic resources into irrigation systems and seasonal floodplains as well as the identification of locally appropriate institutional options for benefit sharing.

2 Materials and methods In co-operation with local officials from the Departments of Agriculture and Rural Development (DARDs) as well as with office representatives from Extension and Commune Offices, nine sites in total were selected. These sites are representative of managed floodplains in the Mekong Delta region of Vietnam and are within the Long Xuyen Quadrangle, the Plain of Reeds, or between Mekong and the Bassac River. The selection was made through carefully designed surveys and formal discussions between related stakeholders. The first and main site selection criterion was that a suitable infrastructure for community-based approaches (dikes, suitable sized areas of paddy fields, etc.) must already be in place. However, also willingness

211

Community-Based Aquaculture for Poverty Reduction

for cooperation from local officials as well as farmers played a key role in the selection process. This paper presents data from four different hamlets in one large floodplain area (Long Xuyen Quadrangle), where community-based aquaculture is practiced and institutional arrangement have been tested since 2006. Table 1 gives an overview about the sites presented in this paper.

Table 1: Selected RIA No. 2 research sites Ecological region (Can Tho city)

District

Commune

Hamlet

No. of farmers involved

Thoi Dong

Thoi Trung

9

Thoi Lai

Truong phu B

Co Do Long Xuyen Quadrangle Vinh Thanh

28

D1

34

E2

31

Thanh Thang

A “farmer group” was established in each of the sites in order to implement fish culture activities during the flood season. Working as a group enables farmers to enclose their fields using dikes and share necessary guarding duties as well as costs. On an individual basis, fish culture is only possible immediately next to the homestead in order to avoid losses from robbery. Financial support for nets and fingerlings as well as technical training was offered by the project. However, farmers were responsible for planning their activities, establishing a financial saving system, organising collective work activities and ensuring the guarding of the cultured fish. The following species were grown: common carp, grass carp, red tilapia, bighead carp, silver carp, silver barb as well as featherback and snakehead. Together with technical staff from RIA No. 2, the project members came to an agreement about suitable fish species, structure and density as well as about fence and dike constructions. Thus, common carp, grass carp, red tilapia, bighead carp, silver carp, silver barb, were nursed at a density of 10 pcs./m2; snakehead and featherback were nursed in hapas suspended in either ponds or in the river at a density of 400 pcs./m2. The nursing period lasted for two to four weeks. Thereafter the fish was released into the rice fields at a density of 0.2-0.25 pcs./m2, when natural food was used, or only 2 pcs./m2 when supplementary feed was provided. The total culture period lasted four to five months

212


In addition to secondary data analysis and literature review, Participatory Rural Appraisal (PRA) was applied in the hamlets. PRA is a method that facilitates community participation in examining issues of resource management, problem solving and decision making (Chambers 1994; IFAD, ANGOC and IIR 2001). This allows both the researcher and the participant to better understand actual resource use patterns and other related issues, to identify primary and secondary stakeholders and to examine difficulties associated with the natural resource. The PRA undertaken within this research included the following tools: resource mapping, social mapping, seasonal calendars, and time lines, transect walks, wealth ranking and semi-structured interviews. Information collected through a socio-economic baseline survey which was undertaken in both hamlets is presented in this paper, too.

3 Results The following section will present findings from research in four hamlets. In all project sites, different meetings were organised with project members and RIA No. 2 staff, who offered the facilitation for discussions about cultivation plans and members’ duties. RIA No. 2 technicians offered their technical expertise and built capacity within the group. Plans were established about who would be responsible at given times for guarding the water resource and maintaining the dike. Additionally, benefit-sharing agreements were set up, whereby profits are only calculated after the harvest. 3.1 Local arrangements In all hamlets a number of different organisations were already in place before the project implementation. People are also used to work collectively in field preparation, maintenance of dikes/water storage/distribution systems, harvesting crops as well as in guarding aquatic resources. Three of the four hamlets even have had experience in collective fish culture beforehand. Although the degree of cooperation differs, the approach of collective activities is not totally new to these selected hamlets. The groups participating in the WorldFish/RIA No2 project were all established in 2006, except for the D1 group, which was established in 2005. 3.2 Water management All research hamlets use water from canals which derive their water from the Bassac River. These canals contain water throughout the whole year, with water levels changing seasonally. Water is used for multiple purposes


213

including irrigation, fishing, washing, doing the dishes, irrigation of gardens, transport (commercially as well as private), and sanitation, as well as providing a market place (floating markets) and waste disposal site. The flood season in the region starts at the beginning of September when the water level in the canal starts rising, reaching a peak level in October/November. The water is used for gravity irrigation during this time and pumped irrigation between December and August. Water in the hamlets is used according to daily needs, and sluices are opened according to private decisions for irrigation. However, some households reported that they synchronise field preparation, also choosing similar duration crop varieties and water release times. In E2, more than 100 small private sluices connect the paddy fields with the canal and are used for individual private water release. These small sluices were built by the individual farmers themselves, accessing the water whenever they need it for irrigation. There are no collective decisions about when and how to irrigate. In Truong Phu B, farmers report that water pumping into the fields is done on the basis of individual decisions, without any common considerations. However, before the next rice crop can be sewn, water needs to be pumped out of the fields, and this must be done at the same time by all in order to facilitate effective disease prevention. Thus, the government announces a certain time frame within which water must be pumped out entirely. This is a local regulation from the DARDs. In Vietnam, people hold legal private use rights to their lands, and transfer of land is regulated through the willing buyer and willing seller principle. Only water resources, like canals, rivers and reservoirs, are considered to be public goods and remain under the management of the government (The Socialist Republic of Vietnam 1993). However, the canals in the two hamlets face an open access situation year round, as anyone can use it for any purpose at any time. Meanwhile, access rights to the paddy fields change with the season. The fields are private land during the dry season, when borders between paddy fields can be easily observed and where small dikes mark the borders of private property. However, during the wet season, these obvious borders disappear due to flooding; thus, at this time flooded agricultural land becomes common property. The fields can then be accessed by anybody, including members of other communities, for fishing purposes. In principle, landowners do not consider this to be an open access situation or a change in property rights, because legally they still have the right to deny access to their land. However, as it is not common to deny anyone access to fields for fishing purposes, so in practice these areas can be classified as open access. Conflicts about water resources are not reported, even in times where there is a lack of fish for daily subsistence.

214


Rules concerning water use in the hamlets are based on national law and only refer to illegal fishing gear, including electric fishing gear as well as small mesh nets (Vietnamese Fisheries Law, MOFI 2005). A police officer, employed by the commune, is responsible for the monitoring of illegal fishing gear use. Illegal fishing activities are reported to be more common during the flood season, when illegal fishers are apprehended about once per week. Seasonally, there are no restrictions: fishing is allowed during the whole year and is permitted anywhere. Villagers are not involved in decision making about rules, and all laws concerning fisheries are made by the government. The national rules are considered to lessen conflict between the farmers and are considered to be important, because they try to “ensure that there will be enough fish for the next generation” (pers. comment farmers). 3.3 Technical experiences Together with technical staff from RIA No. 2, the project members came to an agreement about suitable fish species, structure and density as well as regarding fence and dike construction. The fish structure decided upon was based on the different needs of the several species concerning feeding, and achieving a balance according to available feeds was considered. The main fish introduced was the common carp, because it had previously proved to give a high economic return. Over 60% of fish introduced in the project sites is thus common carp. The decision to release the fish into the field only after nursing them in suitable ponds was a financial consideration. At the same time, water productivity was measured and the pH, temperature, transparency and natural food (plankton), were compared between the different sites, and natural fish abundance was monitored. During this period, all project farmers were trained in fish culture techniques by working with RIA No. 2 technicians. Figure 1 shows the fish production in kg per hectare in Thoi Trung and D1 in the year 2006. The average fish production of the two trial hamlets is approximately 200 kg/ha. This yield is approximately the same as for wild fish in floodplain paddy fields. It is lower, however, than the results of fish production in the Plain of Reed, where about 400kg/ha were produced.

215

Community-Based Aquaculture for Poverty Reduction Production (kg/ha) 300 250 200 150 100 50 0 Thoi Trung

Average production

D1

Figure 1: Fish Production in kg/ha in two hamlets Figure 2 shows the survival rate of the main species of cultured fish, namely common carp, silver carp, bighead carp and snakehead. The bighead carp as well as the common carp showed significantly higher survival rates in all sites than the other species, up to almost 80% in Thoi Trung (Halwart and Gupta 2004). Another important finding was that the grass carp seemed to be not suitable for any of the sites, although, previous research has found that grass carp actually is one of the more suitable species (alongside common, silver and bighead carp), so the reasons for their low performance here require further research. survival rate (%)

100 common carp silver carp bighead carp grass carp

10

snake head

1

Thoi Trung

Average

D1

Figure 2: Survival rate of fish cultured in two hamlets Figure 3 shows the ultimate financial outcome of the collective fish culture, with the figures confirming the initial assumption that farmers in the Thoi Trung project group would receive a higher income from their fish culture activities than farmers in other sites. They were able to earn about 800,000

216


VND/ha, while people in D1 group earned about 500,000 VND/ha and average income for this activity in the Long Xuyen Quadrangle was about 650,000 VND/ha. Net profit/ha

VND in thousand

1000 800 600 400 200 0 Thoi Trung

Average

D1

Figure 3: Net profit per hectare in VND in two hamlets One main problem concerning the income from the project is that harvest time for the cultured fish, during and at the end of the flood season, occurs at the same time as the catching of wild species; thus farmers face a highly competitive price situation. Different options to avoid low prices include creating a marketing network within the project group in order to strengthen their bargaining position or storing the fish until dry season and selling only when price levels rise. However, this would need large ponds, where fish could be stored, which are seldom available. Thus, the farmers plan to harvest at different times and start selling even earlier to avoid low prices. Another crucial factor for attaining a high productivity and income must be seen in the successful organisation of collective activities and the fulfilment of duties in the sites with a higher realised income. At present, the participants are, for example, responsible to protect the cultured fish from natural risks as well as from poaching. Reasons for the low income of D1 farmers can be found in the loss of cultured fish through a broken dike as well as losses due to poaching.

4 Discussion In Thoi Trung, nine farmers grouped together to engage in fish production within their paddy fields in 2006. Individual pond culture was already common in the hamlet, but a collective approach was new. The project group, encouraged by the commune officer and project staff, started the project of collective aquaculture production, because of a lack of wild fish. The other landowners that hold land within the project sites left their


217

property to the project farmers during the wet season. Instead of a third rice crop in wet season, farmers now cultured fish from August until harvest in November and December. Farmers were not satisfied with their results, as they expected to gain a higher overall profit. Reasons for the low income can be seen in a lack of co-operation and missing regulations, as members did not always fulfil their responsibilities. This led to conflicts between project members. Due to the financial disappointment, farmers decided not to continue the project the following year (2008). In 2006, however, 28 farmers in Truong Phu B grouped together and have now been working together for one year to raise fish. They started fish culture, stocking fish along with their normal third rice crop period in June. In November, water was pumped out for the fish harvest, which was then sold to middlemen or farmers in the hamlet within 10 days. Although being disappointed concerning the realised price for the fish at the end of the culture season, the group decided to continue the fish culture. For the next year, it is planned to stock fish without rice and plant only two rice crops. Further, the management system was adapted with a change of rules regarding the duties and interims harvesting of fish by group members. In D1, 34 farmers grouped together and have now been working together for 3 years in order to raise fish. The idea was introduced to them by the district economic division; thereafter, the farmers decided to organise the project themselves. In November, water was pumped out for the fish harvest, which was then sold to middlemen or farmers in the hamlet within 10 days. Notwithstanding a lower income than expected, the project management can be considered well functioning and people report being satisfied with their cooperation. However, farmers also reported difficulties within the group, particularly because of ambiguity regarding different roles. Thus, in 2006, RIA No. 2 staff facilitated a process of developing written rules and regulations. These were developed based on the farmers’ previous co-operation experiences, their mutual agreements, as well as their own decisions. This change led to a higher sense of satisfaction within the whole group. In E2, 120 ha of paddy fields were enclosed by a dike in order to culture fish during the wet season. In this area a total of 55 farmers hold private land titles, but only 31 were interested in participating in the joint project. Fish was stocked from August until harvest in November and December. Here, membership in the project groups is only possible for farmers that hold land in the respective areas. The farmers established links to a local bank that has provided loans to farmers who were not able to participate due to a lack of money for investment in fingerlings. Farmers supported each other in cases of a lack of financial capital. Besides the financial contribution, membership duties included labour participation in dike

218


construction, guarding and participation in other collective activities, like stocking and harvesting as well as marketing. After the first year, farmers decided not to continue the fish culture due to the low profits realised. Further, conflicts within the group led to a low satisfaction. None of the groups had written regulations or had official membership records at the beginning of the project. Meetings were held approximately once a month and all decisions were taken through majority approval. However, all groups established a formal administration/management group, including a leader, a vice leader as well as an accountant. The projects do have a strict book keeping system and an accountant in charge; all farmers are allowed to review the books at any time, and the processes are kept transparent. Although the projects were supported by commune authorities, farmers organised themselves for setting-up the project, establishing collective organisation with formal leadership, organising the collective work and in finding markets for fingerlings and fish. However, members in the groups and especially the leaders were not satisfied overall with their first year’s experience in collective fish culture. Truong Phu B and E2 group leaders explained that they would appreciate support from higher level institutions, especially at the commune level. So far, there is neither a connection to other community-based projects nor to the communal government. The farmers explained that they lack financial support, because some members had difficulties paying their share of investments and had to borrow money from other project farmers in order to be able to participate in the project. Furthermore, farmers report that they lack technical support in terms of training, information or learning networks. Farmers also explained that they are interested in establishing common distribution channels with other aquaculture groups in order to have a stronger bargaining position, but do not feel able to establish such networks themselves. Problems that groups faced focused mainly on guarding duties; most of the time, group discussions solved these conflicts. However, the E2 farmer leader also reports a high level of frustration within the management committee, because of recent free-riding within the group. There is only low attendance at the meetings and participation in collective work activities is highly unsatisfactory (pers. comment group leader E2). Another difficulty that arose is that access that was formerly open during the wet season now became restricted. Because of the cultured fish, the project areas were no longer available for hamlet members that usually fished there for wild fish. E2 farmers report that usually group members followed this rule. However, farmers holding property rights to land within the project area, but are not participating in the project, have been particularly likely to continue to fish in the area and some even increased


219

their fishing efforts. On the one hand, this must be considered as a serious impact of the project, which affects non-members. On the other hand, if non-members end up catching the cultured fish, frustration within the group increases and investment in the common project becomes less attractive, thus potentially weakening commitment to the project. The experiences of the groups clearly reveal that hamlets were able to organise community-based aquaculture with respect to management, accounting and guarding responsibilities. However, groups reported a lack of support in technical and financial areas. Additionally, motivation within the groups seems to be still weak, as results were not as satisfactory as expected and organisational hindrances make coordination of group activities difficult.

5 Conclusions • Community-based aquaculture should be considered as a sustainable measure towards reducing poverty in the rural areas of the Mekong Delta. The experience of the projects reported on here clearly shows that the different groups of farmers were able to organise community-based aquaculture. • Farmers in research hamlets were able to organise themselves in order to culture fish through common action and also reported increased income. In some cases farmers are now also cooperating in other fields, like irrigation and the use of insecticides, promising better overall water management in future. • Bighead carp, common carp as well as silver carp proved to be suitable species with high survival rates and, thus, gave a relatively high return on investment. In one of the project sites, project farmers were able to earn about 800,000 VND/ha. With the community-based approach, water productivity in the common waters was also increased. Farmers also took into consideration marketing issues and plan to change their harvest times, so that they do not compete with wild-fish sellers, in order to attain higher prices for their cultured fish. • Although there are existing certain written rules and regulations, not having an efficient way of monitoring and maintenance of the project sites could lead to low fish yield. These findings helped to highlight the need for focusing on institutional arrangements right from the beginning of projects in new project sites. • Collective action in fish culture as well as participation in general waterrelated policies are still low. So far, local users are lacking long-term support from higher level institutions in supporting their needs regarding

220


issues such as water quality management or flood protection. Rules and regulations as well as monitoring/sanctions systems are rarely monitored and are inefficient, usually because local users are not involved in either decision making or monitoring. • Institutional arrangements differ not only along socio-economic dimensions between the hamlets, but come as well as with seasonal changes. Findings from examining different technical and institutional options are that blueprint approaches to collective aquaculture can not be successful, because of the differing local governance schemes that exist, even in similar natural settings. This all leads to the conclusion that strategies for pro-poor community-based aquaculture must be adapted to the specific local conditions in each community. • Although many development projects in Vietnam focus on the enhancement of collective action and better adaptation to local conditions, there is still a lack of institutional support from other government levels, as experienced by farmers in some of the research hamlets. Financial, technical and institutional support, according to the principle of subsidiary, still needs to develop in order to facilitate community-based aquaculture in the region. However, collective action in the Mekong Delta is a promising approach for future development of the region.

References Chambers, R. (1994). The origins and practice of participatory rural appraisal. World Development 22(7): 953-969. IFAD, ANGOC and IIR (2001). Enhancing Ownership and Sustainability. A Resource Book on Participation. Philippines and India, International Fund for Agricultural Development (IFAD), Asian NGO Coalition for Agrarian Reform and Rural Development (ANGOC) and International Institute for Rural Reconstruction (IIRR). Halwart, M., Gupta, M. V. (2004) Culture of fish in rice fields. FAO, Rome and The WorldFish Center, Penang, Malaysia, pp.83. Available online at http://www.worldfishcenter.org/Pubs/CultureOfFish/CultureOfFish.htm. Ratner, B. D. (2003). The Politics of Regional Governance in the Mekong River Basin. Global Change 15(1): 59-86. The Socialist Republic of Vietnam (2005). Vietnamese Fisheries Law, MOFI. Socialist Republic of Vietnam (1993). Constitutional Provisions, Fundamental Laws and Regulations of Vietnam, GIOI Publisher, Hanoi WorldFish Center (2007). CBFM-2, http://www.cbfm-bd.org/html/project.html

Chapter 12 Shifting Livelihood Strategies of Small Cotton Farmers in Southern Xinjiang Max Spoor1, Xiaoping Shi2, and Chunling Pu3 Abstract: This chapter focuses on the livelihood strategies of small Uyghur farmers in Southwestern Xinjiang (West-China), arguing that the continuation of cotton as a main crop and cash earner does not represent a sustainable way out of poverty, which is widespread in the region. Two prefectures will be analyzed, namely Kashgar and Aksu (with case studies of two counties, Shache and Awaiti) in order to highlight the limitations for these small farmers, in particular because of land availability and access. Income distribution is indeed depending much on the initial land allocation, which -contrary to other provinces of China- has not changed much in the past decades. In spite of some expansion towards waste lands, with population growth average land holdings have become smaller and dependency cotton will not be sufficient for sustainable income growth. Recent state-led policies are discussed that promote crop diversification for small farmers, although these are also not without problems, in particular in terms of expected price development and marketing. Keywords: China, Xinjiang, Cotton, Small Farmers

1 Introduction This paper investigates current changes in livelihood strategies of small cotton farmers in Xinjiang with regard to the limitations they encounter in terms of (1) resources (in particular land and water4), (2) government policies (which are being transformed from ‘instructions’ on what to grow and to whom to sell the output, to the use of more indirect policies that stimulate changes in crop-mix to stimulate greater diversification of farm 1 2 3 4

Institute of Social Studies, The Hague/Erasmus University Rotterdam, The Netherlands, and Barcelona Institute of International Studies Nanjing Agricultural University, Nanjing, China College of Economics and Management, Xinjiang Agricultural University, Urumqi, China Although water is a fundamental resource in these very dry areas of China, in this paper we will focus on cultivated or arable land, which in southwestern Xinjiang means ‘land with water’, i.e. largely irrigated land.

Beckmann, V., N. H. Dung, X. Shi, M. Spoor, J. Wesseler (eds.) (2010). Economic Transition and Natural Resource Management in East- and Southeast Asia. Aachen: Shaker Publisher, pp. 221-240.

222

Spoor, Shi, and Pu

produce); and (3) rapidly emerging rural factor markets. Xinjiang province of China, also known as the Xinjiang Uyghur Autonomous Region (XUAR), is a large producer of cotton, which has been grown as a near mono-culture for a long period of time. In the area where most cotton is cultivated, in the southwestern parts of the province, which is limited in terms of water and arable land resources, the crop is produced by Uyghur minority farmers with tiny farms and by large scale Han-Chinese populated state (or ‘regiment’) farms, belonging to the Xinjiang Production and Construction Corps (XPCC) or Bin Tuan (Spoor and Shi 2009).5 Rural income differences are substantial in Xinjiang, in which the prefectures with the highest populations have the lowest incomes. As the dependence on agriculture (and in particular crop production) is still high, and the contribution of the rural non-farm economy to farm households’ incomes is still limited, we will focus here on the relation between income inequality and access to arable (or cultivated) land, which is also highly differentiated. While in other parts of China policies for the regular re-allocation of land have been implemented (for example after demographic changes at village level), in Xinjiang this has not been the case (see Ho 2009). This chapter focuses on the livelihood strategies of small Uyghur farmers, arguing that the continuation of cotton as a main crop and cash earner does not represent a sustainable way out of poverty, which is still widespread in the region, in particular in the most remote southwestern prefectures6 of Kashgar and Khotan (Chen et al. 2004; Heilig et al. 2005). There have much lower than average rural income levels in these areas, where most of the cotton in Xinjiang is produced. The relative importance of cotton depends also on the availability of resources (primarily water and land), distance to markets and rural infrastructure and the institutional environment (taxes, forced procurement and influence in crop-mix). This can be seen when comparing Awati and Shache counties (in Aksu and Kashgar prefectures, respectively), in particular focused on their highly unequal land availability and its effect on net rural incomes within and between the two counties. We argue that intra-prefecture and county income differences, down to the level of townships and even villages, can by and large be explained by the initial land allocation households received, which in the specific case of 5

6

We have actually been able to visit Bin Tuan Regiment Farm Nr. 1 in Xinjiang (with a population of around 21,000), but in this paper we will not enter in comparisons (such as we did in Spoor and Shi 2008), but rather concentrate on the livelihood strategies of small Uyghur farmers. In China, provinces are divided into prefectures, which are subdivided into counties and cities. The rural areas of counties are subdivided into townships and, further down, into administrative and natural villages.

Shifting Livelihood Strategies of Small Cotton Farmer in Southern Xinjiang 223

Xinjiang has hardly been adjusted in the years since the introduction of the Household Responsibility System of the early 1980s. Shache county is a case in point here, with initially not more than 2.0 mu/capita average arable land availability.7 During the 1980s and 1990s this area was actually stimulated to ‘open up’ land by reclaiming wastelands, which did expand the cultivated land area slightly. However, in the past decade there have been high-level government decisions on water supplies for downstream areas which caused substantial cuts on the upstream water users in the Tarim river basin.8 Wasteland reclamation is therefore still possible, but only if no extra water resources are used; hence the limitations on expanding cotton or grain production are substantial.9 Based on joint field research in March 2008,10 we furthermore show that there is wide implementation of the “6311”11 policy towards greater farmproduce diversification through the introduction of fruit and nut trees – inter-cropped with cotton and grain – in recognition of the limitations of cotton as the main farm income generator for small farmers and its continued pressure on the limited land and water resources of the region. The massive introduction of fruit trees and vegetable production (amongst other means, through the introduction of green houses), with accompanying technical assistance and other stimulation measures is clearly visible in the counties mentioned. Whether internal (regional and extra-regional, national) markets will be able to absorb the rapidly growing supply of fruit and nut production (including apricots, jujubes, almonds and walnuts) is however still to be seen, but there are clear signs of alternative livelihood strategies being implemented by small peasants. These are largely policyinduced, but freedom of choice for these producers is also somewhat increasing, indicating a substantial change in government policies. The chapter is divided into four sections. The following, second, section will analyze rural income inequality in Xinjiang, connecting it to a high rate of income dependency particularly on crop agriculture (mainly cotton 7 8 9

10

11

1 Hectare = 15 mu. In the Chinese translations the river is called Talimo. In the statistics on land use (see Statistics Bureau XUAR 2006; 2007) it seems that in reality this policy has been partly ignored, as there are some counties that have rapidly expanded the amount of reclaimed land. The joint academic research team, as part of the above mentioned KNAW-financed research project, visited several counties, townships and villages in southwestern Xinjiang, between 16-27 March 2008, consisted of Prof. Pu Chunling, Dr. Zhulifia, Mr. Liu Wensheng (CEM/XAU), Dr. Shi Xiaoping (NJAU), Dr. Murat Arsel, Dr. Kristin Komives and Dr. Max Spoor (ISS). This is short-hand for the plan in Shache county to expand fruit and nut production towards six hundred thousand mu of almond, three hundred thousand mu of apricot, and one hundred thousand each for walnut and other fruits, such as jujube.

224

Spoor, Shi, and Pu

and to a lesser degree grain) and the existence of only limited options for non-farm activities. Given that most small farmers have similar levels of technology, farming income is in turn largely dependent on farm size, which can be quite different between counties, as original allocation was based on available arable land and the number of rural inhabitants residing there. Hence the variance in land size for small farms is substantial, not only between prefectures or counties, but also within them. Inequality in land size at county level is shown for two prefectures, Aksu and Kashgar, and is discussed using recent data from a township in Shache county (Kashgar prefecture), showing that differences between villages are still substantial: though, surprisingly, less than between counties. Other factors are the availability of water and the positioning of a farm in relation to the main irrigation canal or system.12 However, these are beyond the scope of this chapter to analyze. The third section will further analyze the noted income dependency from crop agriculture by focusing on the two primary crops in the already mentioned prefectures, cotton and grain, with cotton being the main cash crop and income earner. Some case-study data, derived from our field visit to several villages in Awati and Shache counties in March 2008, will illustrate this. We also discuss the process of market liberalisation that is taking place, with the rural market environment improving through more competition in factor markets and between processing units, of which new private ones (such as cotton gins) have appeared, though it seems that the cultivated areas (and corresponding production targets) for cotton, and to a lesser extent wheat, are still being operated by ‘advising’ farmers. Nevertheless, there is awareness among the local and regional authorities that cotton is not a sustainable way out of poverty, except for those who have more land and sufficient levels of technology. The fourth section, therefore, analyzes the massive move in the past few years towards the introduction of inter-cropping of cotton with fruit and nut trees, in particular apricot, jujube, almond and walnut, and the promotion of vegetable production in greenhouses. It will be argued that this can potentially improve incomes of small farmers, but that there are also dangers on the horizon, as it is unclear where the markets for jujubes, almonds, walnuts and apricots would be, showing a possible ‘fallacy of composition’ effect. In our conclusion we will combine this observation 12

Those farmers who are at the tail end of a large irrigation canal, such as those we noticed in a visit to Yawalik township (Awati county), have much less access to water, at least when they need it. Distance to desert areas might also be another influence. Interestingly, the disadvantage of low water supply is compensated (in the particular Ulucherle township) by slightly larger farms, possibly caused by initially lower population density in the 1980s (Fieldnotes, 21 March 2008).


with the move away from quasi-dependency on cotton towards a more diversified livelihood strategy for small Uyghur farmers in Xinjiang. Its success will depend on the development of rural markets (labour, finance, input and output) in these relatively isolated regions and will further depend on what impacts these shifts have on land and water resources.

2 Income inequality, poverty and agriculture Xinjiang province, in the northwest of China, is still largely rural and has a relatively high poverty headcount ratio, in particular in prefectures that are in the far southwest of the province, such as in Aksu, Kashgar and Khotan (see Spoor and Shi 2009). In 2006, Aksu and Kashgar, which we will study more closely in this paper, had rural populations which represented, respectively, 57.4 and 67.8 percent of the total inhabitants of their respective prefectures (Statistics Bureau XUAR 2007). According to official statistics, the Uyghur population is still the largest ethnic group in the province, although not anymore a majority because of the influx of Han Chinese, which started in the 1950s (see Becquelin 2004; Moeller 2006) and was recently complemented by migration towards the booming oil industry in, for example, the city of Korla and the economic opportunities provided by the ‘Western Development Strategy’ (Xibu da kaifa), which was initiated in 2000. However, in many of the rural areas of Xinjiang, particularly in the southwest, the population is nearly exclusively Uyghur. For example, in Aksu and Kashgar these shares were, respectively, 72.7 and 90.0 percent in 2006, while in our largely rural field visit areas, Awati and Shache, these were 78.4 and 95.4 percent (Statistics Bureau XUAR 2007). Most of these rural households are small farmers who received a land allocation (“contract” land) in the early 1980s; these were in most cases very small, because of the limited available arable land (with water for irrigation) in densely populated areas, complemented by some “reclaimed” land in the following two decades. Rural net income in Xinjiang is low by Chinese standards, with an average for the whole of XUAR at 2,737.3 Yuan per capita in 2006. In Figure 1, we can see that it is not the lowest, which is found in the province of Guizhou (1,984.6 Yuan), as Xinjiang’s rural net income per capita in 2006 is number 7 of 31 provinces in China, ordered from lowest to highest income (NBSC 2007).13 13

It is surprising to note that in 1990 Xinjiang had a rural net income per capita which was number 19 from the lowest, while in 1995 it had dropped to number 7, a position which (in spite of overall improvements in rural incomes all over China), remained stable until 2006. The between-province differences have grown in the meantime. In

226

Spoor, Shi, and Pu

Intra-regional (provincial) inequality is also substantial, although slightly less than inter-regional, as we can note in Xinjiang province. For 8 out of 11 prefectures there, average rural net income is derived from household survey-based data, published by the Statistical Bureau of XUAR (see Table 1). In 2006, the average level was given as 2,737.4 Yuan, while the lowest was found in Khotan and Kashgar prefectures (1,417.3 Yuan and 1,872.9 Yuan, respectively) and the highest in Changji Autonomous Prefecture (5,003.7 Yuan).

Rural Net Income/Capita

10.000 9.000 8.000 7.000 6.000 5.000 4.000 3.000 2.000 1.000 0

Figure 1: Per capita net income for rural households in China, 2006 Source: NBSC (2007); Authors’ Calculations.

Aksu prefecture, one of the regions we discuss in great detail in this paper, shows an above average regional rural net income, namely 3,360.0 Yuan (see Table 1). We assume that income inequality can be explained by the following factors: • Spatial differences in infrastructure (such as roads and communication); • Closeness to or distance from fast-growing industrial (or natural resource extracting) centres, with positive spin-offs towards rural areas (migration and supply of primary commodities); • Availability of a non-farm rural economy (processing companies, services etc.); • Water availability within the overall scarcity of this resource in the XUAR region, the influence of seasonal fluctuations in water supply, and the positioning of small farms in relationship to the main irrigation canal; and 1990 the ratio of rural net income/capita between the poorest and richest provinces (Gansu-Shanghai: 4.43:1) has slightly increased to 4.60:1 (Guizhou-Shanghai).


• The size and quality of original (arable) land allocation, as no policy of regular re-distribution has been followed. In Xinjiang the dependency of rural incomes on agricultural (crop-related) production is great, as we can also note in the data presented in Table 1. In Aksu and Kashgar prefectures, where ‘cotton is king’, this share is even 81.4 and 65.6 percent, respectively. What can therefore be expected is that there is a situation of “many people, little land”.14

Table 1: Rural income differentiation and dependency on agriculture (2006) (RMB/Capita)

Rural Net Income 5003.7

Agri Share of Rural Gross Income Population % 71.2 734807

Rural Pop Share %

Income Share %

7.3

13.4

Changji Autonomous Prefecture Counties Directly Under Ili Prefecture Tacheng Administrative Offices Altay Administrative Offices Bayangol Mongol Administrative Offices Aksu Administrative Offices Kashgar Administrative Offices Khotan Administrative Offices

3164.7 4003.3 3302.6

57.0 72.4 43.3

1410660 489123 274149

14.1 4.9 2.7

16.3 7.2 3.3

4459.3 3360.0 1872.9 1417.3

72.3 81.4 65.6 44.0

420831 1326607 2576260 1500051

4.2 13.3 25.8 15.0

6.9 16.3 17.6 7.8

Local Regions (Total)

2737.4

68.0

10002681

100.0

100.0

Source: Based on Household Survey Data published in aggregate by Statistical Office XUAR (2007).

This in turn then leads to an unequal distribution of income, in which not only average incomes show a substantial variance: those prefectures with the highest rural populations, such as Kashgar and Khotan (representing 25.8 and 15.0 percent of the total rural population, respectively), only earn 17.6 and 7.8 percent of the overall net rural income of the XUAR. In rural prefectures such as Kashgar and Khotan we can indeed see that average farm size is smaller than elsewhere, something which is to be expected with the low incomes and the large dependency from farming activities. Although there are many problems in the official Chinese data in terms of the precise measurement of the agricultural and rural populations, we have tried to estimate in the most consistent manner the land size per capita, here defined as cultivated/arable land per ‘rural inhabitant’.15 Our 14

15

In our fieldwork of March 2008, this rather telegram-like style of translation in English from the Chinese language became a regular part of our joint terminology, as it explained quite a lot in a few words. The official data which can be found in the Xinjiang Statistical Yearbooks of various years includes data on ‘agricultural population’. However, these data come from the

228

Spoor, Shi, and Pu

initial expectation was that the land size per capita variable would decrease over the years, but using the official data this seems not to be the case. For Aksu and Kasghar (Figure 2) there is substantial inter- and intra-prefecture average land size inequality (Table 2).

Aksu Prefecture

Kashgar Prefecture

Figure 2: Map of Xinjiang, with Aksu and Kashgar prefectures For example in Aksu prefecture, where the original land allocation was higher, there was in 2001 an average cultivated land size/rural inhabitant (µ) of 4.3 mu at county level, with a standard deviation (σ) of 1.4. In 2006 µ was 4.2 mu, with a σ of 1.2. In the case of Awati county, which will be taken as an example from Aksu prefecture in this paper because of its large dependency on cotton, the average cultivated land/rural inhabitant was higher, namely 5.1 mu in 2001 and even 5.3 mu in 2006.16 For Kashgar prefecture, where land availability is much lower, there was an increase

16

Public Security Bureau and are likely to misjudge reality as they have not measured or updated for a long time. Hence, many people might live in urban areas (and actually administrative units called ‘city’ rather than ‘county’), while they are still listed in the books as ‘agricultural’. Although we have used them in Spoor and Shi (2008), we have now tried to use the data on ‘rural population’ (We acknowledge the suggestion made by Andrew Fischer of ISS, who noted that the rural population data after the agricultural census of 2001 is much more reliable than the earlier-mentioned data on agricultural population). However, even with these there are problems, as the Statistical Yearbooks have two different series for this variable, without explaining the difference. In the statistics this is explained by two effects, namely a starting trend of outmigration and some addition of reclaimed land. The latter is rather surprising as, during our fieldwork, farmers stressed that since 2001 there has been a strict prohibition of allocating more water if land is reclaimed, and even an overall reduction of water supplies in Awati county (Fieldnotes, March 2008).


from 2.7 mu in 2001 to 3.2 mu in 2006, although with a substantial increase of variance as well. Some of the most populated counties in Kashgar prefecture, such as Shufe and Shule (close to Kashgar city), and Yengisar (between Shache and Kashgar city), have very limited arable land availability, between 1.6−2.2 mu/rural inhabitant (= 0.10−0.15 ha).

Table 2: Arable land/rural capita in Aksu & Kashgar (2001-2006) mu/cap 2001 4.78 6.34 3.10 4.98 4.51 5.08 2.86 5.06 1.81 µ 4.28 (σ) 1.41 2001 Kashgar City 0.89 Shufu 1.76 Shule 2.27 Yengisar 1.92 Zepu (Poskam) 3.97 Shache (Yarkant) 2.63 Yecheng (Kagilik) 2.72 Makit 4.40 Yopurga 2.98 Jiashi (Payzawat) 2.20 Bachu (Maralbexi) 3.72 µ 2.68 (σ) 1.04 Source: Based on Xinjiang Statistical Yearbook 2002, 2005, 2007. Aksu City Wensu (Onsu) Kuqa Xayar Xinhe (Toksu) Baicheng (Bay) Wushi (Uxturpan) Awati Kalpin

2004 4.58 5.89 2.64 4.48 4.26 4.17 2.32 4.38 3.00 4.02 1.15 2004 0.74 1.71 2.19 1.87 3.83 2.54 2.49 4.40 2.77 2.14 3.28 2.54 1.02

2006 5.25 6.08 3.15 4.83 4.47 4.09 2.19 5.29 3.37 4.30 1.23 2006 0.93 1.89 2.19 1.62 4.67 2.99 2.51 6.86 3.17 2.52 5.37 3.16 1.78

In Shache county, our second field visit area, which is the most populated county in Kashgar prefecture, the land size variable increased slightly from 2.6 mu in 2001 to 3.0 mu in 2006. Nevertheless, these data are not very reliable, as land data as well as population data are different, depending on the source.17 17

In our field visit to Kosherik township in Shache county (from which we also present some village-level data on land), it turned out that the Public Security Agency stated that the population was 15,020, while the One Child Policy Committee gave a figure of 14,039 and the Agricultural Bureau told us 14,549 (Fieldnotes, 24 March 2008).

230

Spoor, Shi, and Pu

At township level, hence within a possibly more homogeneous area in terms of population density, we also found variability in land availability – which often explained much regarding the ‘poverty’ status of a village – but less than at county level. For comparison, the coefficient of variation σ µ is 0.29 for Aksu and 0.56 for Kashgar prefecture in 2006 (with data at county level), while it is 0.21 in Kosherik township (with data at village level; see Table 3).

Table 3: Land size of farms in Kosherik township (Shache County) Village 1 2 3 4 5 6 7 8 9 10 11 12 13

Total Contract Orchard 1419 1169 46 3438 3027 183 2566 2248 240 2620 2200 242 2135 1830 128 2265 2010 145 1454 1132 132 1560 1257 99 2187 2088 54 1973 1754 127 2831 2290 82 1017 832 48 792 764 23

Fruits Reclaimed 119 86 0 228 0 76 162 16 116 60 0 111 0 190 0 205 0 45 0 93 327 132 24 113 0 5

HH 155 278 448 426 276 214 167 184 327 230 366 111 96

Mu/HH Mu/Capita 9.2 2.06 12.4 2.78 5.7 1.29 6.2 1.38 7.7 1.74 10.6 2.38 8.7 1.96 8.5 1.91 6.7 1.50 8.6 1.93 7.7 1.74 9.2 2.06 8.3 1.85 µ 1.89 σ 0.40 Source: Based on data provided by the township administration, Field visit March 2008.

If one looks at the data in Table 3 with somewhat more attention, it can be seen that the “contract” land distribution (based on the original land allocation) is slightly more equal than it is for the total distribution, as there is higher variance in the additional “reclaimed” land, which most likely will again depend on water availability.

3 Income dependency from land and cotton In order to push our argument further, namely that rural income inequality18 can be largely explained by relatively fixed land allocations generating limitations on the availability of cultivated land/rural inhabitant and the water needed to cultivate it, we will focus in this section on the dependency of agricultural income on crop production: in particular from cotton and, to a lesser extent, grain. In Table 4 the sown areas and output are given for

18

There also variable figures on cultivated land. Nevertheless, the differences are not so great that we cannot analyze the overall trends in land availability and land use. Again, we are discussing spatial inequalities here (This issue was clarified in a personal communication by Dr. Andrew Fischer of ISS.)


these two strategic crops. The data shows that the cotton areas have expanded substantially over the past decade in both Awati and Shache counties (in Aksu and Kashgar prefectures, respectively), combined with a substantial rise in output (and much less increase in yield, in particular in Kashgar). The grain area in Awati has been reduced substantially, as farmers have been growing more cotton. This happened also in Aksu prefecture as a whole, as was reported by high officials of the Aksu Reform and Development Commission.19

Table 4: Shache and Awati counties (2001-2007): agricultural production

Shache [Yarkant]

2001 2004 2006 2007

Grain Sown Area (Hectare)

Grain Output (Tons)

Grain Yield (Tons/Hectare)

61370 60690 79680 64668

348263 344384 429273 421594

5.67 5.67 5.39 6.52

Cotton Cotton Sown Output Area (Tons) (Hectare) 43790 38570 50000 ..

54683 52500 78750 81553

Cotton Yield (Tons/Hectare)

1.25 1.36 1.58 ..

Awati

2001 13110 79961 6.10 30000 42000 1.40 2004 12850 82166 6.39 28000 45972 1.64 2006 11290 74938 6.64 35670 53575 1.50 2007 4627 31247 6.75 43333 63213 1.46 Source: Xinjiang Statistical Yearbook 2002; 2005; 2007; Aksu Prefecture (2008); Shache Prefecture (2008).

Actually these officials were quite worried that the grain area was being reduced so fast that it was weakening the food self-sufficiency of the region. The regional government had already decided on wheat subsidies to promote grain production, as it turned out that the sown cotton area for 2008 was even higher than planned. The latter is an interesting issue in itself, as farmers cannot always freely decide what to grow. Although the centrally planned production quotas have been largely relaxed, it seems that there is still a system of planned size of planted cotton and grain at prefecture level, which is then ‘translated downwards’ to the county, township and administrative village levels. Finally, these quotas are distributed within natural villages, with some liberty for farmers to decide on crop mix and ‘what to grow where’. It was often stressed that farmers are not forced, but are ‘advised’ to grow this or that. In Shache county, the administration provided another analysis of the income dependency of farmers on cotton production. In a meeting with the governor, he said that, 19

Interview by the authors, 19 March 2008, Aksu (Fieldnotes).

232

Spoor, Shi, and Pu

according to the county leadership and based on an analysis of costs and possible yields, there were limitations to income improvements with only cotton. Even with improved technology, the maximum gross income from cotton was being estimated at 2,000 Yuan /mu. He added that there were new instructions by the prefecture government to gradually reduce the cotton area. One improvement that has been made in the prefecture’s cotton production was expansion of the area under drip irrigation. Also, previously the county administration only had cotton gins, but with the help of private investors they have now established processing capacity for yarn production in the region.20 To complement farming incomes, widespread introduction of fruit and nut trees has been pursued in the region, mostly inter-cropped with cotton, while intensive vegetable production has been promoted through the building of large green houses, which were indeed visible in the vicinity of towns or villages.

Table 5: Average income structure Awati farmer (2006-2007) Crop Agriculture Cotton Grain Fruit Forestry Animal Husbandry Non-Farm Industry

2006 79.0

2007 80.6 65.6 8.3 4.7

73.1 3.0 4.1

2.8 12.8 5.4

2.5 12.2 4.7

Total 100.0 Source: Awati County Agricultural Bureau (2008).

100.0

At the moment, income dependency on cotton for small Uyghur farmers is still very high. In Awati county (Aksu prefecture), the contribution to income from crop agriculture was 79.0 percent in 2006 and even 80.6 percent in 2007. The income shares from cotton are 65.6 and 73.1 percent, respectively (see Table 5), which seem to be coherent with the expanding cotton areas in Awati during these past years. Although we lack specific figures for cotton income dependency in Shache county, taking into account that there is little in the way of non-

20

It is quite difficult to give an average cost picture, but data provided by the Shache county Agricultural Bureau estimated that in, for example, Taghaqi Township the net benefit per mu for cotton was 756.2 Ұuan, with input costs at 440.0 Yuan, hence 1,206.2 Yuan as gross income. However, in other townships, such as Ygenqi and Kosherik, gross income was substantially less, respectively 928.1 and 800.0 Yuan/mu. (Data provided in a meeting with Agricultural Bureau officials of Shache county, 24 March 2008; see Table 6 for further details on costs).


farm rural economy and relatively low migration21 in that area, the shares could well be similar or even higher. Box 1: Visit to Village Nr. 28, Taghaqi Township, Shache County, 23 March 2008 This village was located in the north-west of the township. The first farmer we visited in this village was a tall, very friendly, and quiet man. His family consisted of the farmer and his wife, respectively 49 and 47 years old, the latter serving us almonds, dried apricots and yoghurt during our visit; 2 daughters; and 1 son. His son had been married but was now divorced. Currently they have a grandchild (from a daughter) in the house, but only three adults lived there: themselves, together with the son who had returned. The farmer had 10 mu of arable land, but also cultivated 10 mu of land, for which he had paid 100 Yuan/mu to a younger brother who had migrated. He sometimes did work for a neighbouring Bin Tuan farm. He grows cotton and fruits on 4.5 mu, apricots on 2.5 mu and almonds on 2 mu. He had no grassland, but would like to have some wasteland assigned to him, so he could reclaim and improve its quality. Since 1992 the couple have planted fruits on 5 mu, with 1 mu of apricots, giving them around 1,000 Yuan/mu in output. The son works off-farm as well, but according to the father was not very eager to learn new skills. His father had tried to encourage him, but to no avail. Only later in the conversation did we understand the reason why. When the boy was only 9 years old, the father had been involved in a fight with some people in the village and had killed a person. He was arrested, convicted, and spent 10 years in jail. This meant that, after he was released, his son did not want to accept his authority. During his stay in prison, his wife had been responsible for the animals. He had lost the right to cultivate his land, which was itself taken from him during his prison stay. Now he is again involved in animal husbandry, presently with 2 cows, 1 donkey and 9 sheep. They were selling sheep for 480 Yuan (market price), while an adult cow could fetch even 3,000 Yuan, with a young cow costing around 600 Yuan. His crop output was estimated at 5,500 Yuan for 7.5 mu of wheat, 9,000 Yuan (gross) for 10.0 mu of cotton (of which 5 mu had been negatively affected by strong winds), fruit 2,500 Yuan, corn 3,200 Yuan (net), indicating around 25,000 Yuan of gross income. He also had some income in the past from picking cotton, including 3,200 Yuan in 2003, 1,700 Yuan in 2005, and recently in 2006, 1,900 Yuan. In 2007, with his two daughters he managed to earn 2,500 Yuan in the cotton harvest (of others). We asked him whether he took a loan to buy farm inputs such as fertilizers. He said that it was possible, but that he had sold 8 sheep and 1 cow in order to buy them, he did not borrow money. He had a colour TV(since 2002), but no fridge. September 2006 his son married, but recently divorced (June 2007). Did he have any savings? His answer was that he did have them, but that in a short period of time his savings ran out. One daughter had married in 2003, and divorced recently. She recently married again. His son had married, and also his other daughter married, all three in a matter of less than 2 years. He paid for his son’s marriage an amount of between 9,000 and 10,000 Yuan, while the marriages of his daughters cost between 2,500 and 3,000 Yuan.

For both counties we have conducted some interviews with small Uyghur farmers which clearly showed the importance of cotton (and to a lesser extent grain) for their farm incomes. An interesting case was the farmer we visited in Taghaqi township, Shache county (Box 1). He had been in jail for a serious crime and had lost his allocated “contract” land of around 10 mu. After he left jail and had reinserted himself into the village, he got his rights back and was even cultivating 10 mu of land for a brother who had migrated. In this case one notes the strong dependency on cotton (and 21

In the three townships of Shache County which we visited in March 2008, namely Igenqi, Taghaqi and Kosheri, the share of labour that was permanently working outside the township in the labour force (compared between brackets with the share of the total population was): 23% (8) for Igenqi, 2% (1) for Taghaqi and 5% (2) for Kosherik [Data provided by Shache county officials, March 2008].

234

Spoor, Shi, and Pu

wheat), with the need to complement this income with animal husbandry and fruit output, which we will discuss below. Income from cotton also came partly from cotton picking earned by him with his two daughters (in other farms in the region). Another farmer we visited, in this case in Kosherik township, lived in a village (nr. 4) which had on average rather low available cultivated land per capita (see Table 2). He was a slightly better off farmer, with 11 mu, of which 6 mu was used for cotton and 5 for wheat (see Box 2). He had already started quite some years ago planting almond, walnut and apricot trees, mostly in the wheat fields, though in the interview he considered the results not to have been very good, especially from the almond trees, which he planted 10 years ago, but had not cared for (Fieldnotes Visit Kosherik township, 24 March 2008). He revealed the extremely limited possibilities for manoeuvre with 11 mu of cultivated land, from which the farmer calculated that he generated 3,500 Yuan gross income from cotton and 2,000 Yuan from wheat, while animal husbandry gave him 13,000 Yuan in 2007. Although this meant that animal husbandry provided a large part of income, this seems to be overstated, as he needed to sell much more than normal, reducing his assets. He estimated that his net income was 18,200 Yuan, from which a loan of 5,000 Yuan for inputs had to be repaid. With 8 members of the extended household, net income was not more than 1,500 Yuan per year, slightly under the average that was reported for Kashgar prefecture in 2006 (see Table 1). Box 2: Visit to Village Nr. 4, Kosherik Township, 24 March 2008 Arriving in the village we simply took the first house. When we invited inside, we noticed it had been rebuilt quite recently, with very nice woodwork. Only the male head of the family was there, everybody else had gone to the local bazaar. The farmer had injured his leg and was now handicapped in many things. His extended household was in total 8 people, with wife, daughter and child, and son with daughter-in-law, and two grandsons. They have 11 mu: 6 mu for cotton and 5 for wheat, with 8.8 being contract land and 2.2 mu reclaimed. He had 2 mu of almond, 3 mu of walnut and 1 mu of apricot, mostly in the wheat fields. In the cotton fields he also has some additional spices. With walnuts he started 6 years ago, with almonds longer (10 years), but since the trees are not very well taken care of the results are not very good (only producing 8-10 Kg or 250-300 Yuan/tree). The conversation was very detailed, but also quite influenced by the village leader, who was making his first visit to farms of the village. He often answered instead of the farmer. In terms of income the following data was given by the farmer. His wheat yield was 300400 Kg/mu at a price of 2.2 Yuan/Kg. He kept quite a lot for his own consumption. Cotton gave him 3,500 Yuan (with around 350 Yuan/mu as costs to be subtracted) and wheat around 2,000 Yuan (with 200-300 Yuan/mu a costs); animal husbandry still generated 13,000 Yuan as net income. His son does not contribute any income to the farmer, but rather spends it all on his own family (as part of the extended family), estimated at 2,000-3,000 Yuan/annum. Estimating his full income, it would be around 13,000 (animal husbandry) + 3,400 (wheat) +1,850 (cotton) = 18,200 Yuan, but minus loans only 13,200 Yuan. With 8 members (disregarding the son’s income), this is only 1,500 Yuan/capita/year, well under the PPP 1.08 USD/capita/day poverty line (although above the Chinese national one). Looking at the household they did not really belong to the poor, but were ‘vulnerable to poverty’. As the farmer stated, his injury had caused him to “sell his sheep to the bank”.


The latter case confirmed the data we later received from the agricultural bureau of the Shache county administration, which gave us detailed insight into the costs and benefits of cotton and, within the context of small farms, the limitations it poses with respect to income. According to these data the net income per mu was 488.1 and 756.2 Yuan for the townships of Ygenqi and Taghaqi, respectively, and only 360.0 Yuan for Kosherik (see Table 2 for the average land size per village in Kosherik township). The net benefits per capita were, respectively, 798.6, 893.9 and 371.2 Yuan.

Table 6: Cotton cost-benefit analysis (Shache county, 2007) Unit mu mu tons Million Yuan

Ygenqi

Taghaqi

Total Cultivated Land Cotton Land Cotton Output

Total Cultivated 65000 Land 36000 Cotton Land 4176 Cotton Output

Cotton Output Sold in Markets Input Costs Seeds Plastic Film Fertilizer Pesticides Water Equipment Hired Labour Small Eq/Maint Insurance

33.41 32.41 15.84 1.26 2.34 3.60 0.54 1.08 3.60 2.88 0.36 0.18

Net benefit Net benefit/mu Net benefit/cap

Kosherik Total Cultivated 78000 Land 37000 Cotton Land 4477 Cotton Output


17.57 Net benefit 488.06 Net benefit/mu 798.64 Net benefit/cap

44.26 42.83 16.28 1.30 2.40 3.70 0.56 1.11 3.70 2.96 0.37 0.19

33000 15000 1500


12 11.64 6.6 0.53 0.98 1.5 0.23 0.45 1.5 1.2 0.15 0.08

27.98 Net benefit 756.22 Net benefit/mu 893.87 Net benefit/cap

5.40 360.00 371.19

Source: Data provided by the Shache county agricultural bureau (Field visit, March 2008).

The average cultivated land availability was 2.95 mu for Ygenqi, 2.49 mu for Taghaqi, and only 2.27 mu for Kosherik, which reveals some negative influence of the low land/rural inhabitant ratio on income, although the difference could also be caused by factors such as water availability or land quality.

4 Diversification and policy-induced change In other regions in China, the process of resource-use intensification by small farmers had already taken off much earlier, induced by the increased freedom to decide about their crop-mix, in combination with very small landholdings, leading to a process of resource intensification with possibly

236

Spoor, Shi, and Pu

negative environmental consequences (see Tilt 2008). However, the rapid development of factor markets and the current process of diversification in the southwestern prefectures of Xinjiang towards the production of more fruit, nuts and vegetables has been much more policy induced. The provincial, prefecture and county governments have pushed plans such as the already mentioned ‘6311’ policy and the ‘Fruit and Forest Planting around Tarim Basin’ (FFPTB) project. For the past few years, the intercropping with cotton or wheat of fruit trees, such as apricots, jujubes, and nut trees (e.g. almonds and walnuts), has been strongly ‘advised’ as a good route for farmers, with massive technical assistance being provided by the county government.

Table 7: Fruit production in Aksu and Kashgar (2001-2006)

Aksu

Apricots hectares 2001 13300 2004 27457 2006 35413

Jujubes tons 68835 160728 224214

2001 19817 166042 2004 56757 355350 2006 65902 554364 Source: Xinjiang Statistical Yearbooks 2002, 2005 and 2007. Kashgar

hectares 480 9574 30827

tons 1290 2194 6464

686 1710 7508

2264 2772 8650

In Table 7 we show some aggregate data on apricot and jujube acreage (although most of this is inter-cropped with cotton and, to a lesser extent, with wheat) and output, with growth over the period 2001-2006 having been nothing less than spectacular. Similar data for these fruits can be provided at county level, as well as for the more traditional almond and walnut production, which also expanded in this period. Apart from the fruit and nut production, in the southwestern prefectures the re-conversion of some farmland (especially near towns and other urban areas) into greenhouses for vegetable production has also been promoted. Finally, in Shache county (just outside of Shache city in the direction of Yangisar) private investors, with help from the county government, have set up a fruit-processing company, which had just begun to function when we were there (Fieldnotes from visit to Shache, March 2008). These developments seem to confirm the hypothesis that, within the context of very small farms such as those of the Uyghur minority farmers in southwestern Xinjiang, cotton – which had been a near mono-crop for several decades – is now seen as not being a sustainable income generator; therefore, diversification of income for such smallholders is called for,


including the increase of temporary or permanent migration, which is still limited in the region. They also indicate that the regional government (at different levels) is well aware of this problem and has decided to massively promote the intercropping of cotton and wheat with fruit and nut trees. One downside, apart from possible impact on the intensity of resource use and demand on water22, can lie in the impact of the use of pesticides during the cotton production cycle on the fruit trees.23 There is also the possible ‘fallacy of composition’ problem, as it remains relatively unclear where the markets for apricots, jujubes, almonds and walnuts are to be found, domestically and externally.24

5 Conclusions In this chapter we have analyzed the relationship between low net rural incomes in several prefectures and counties of southwestern Xinjiang and their dependency on cotton (and to a lesser extent) on grain farming, within the context of the very small amount of arable land available to these farms, a situation normal for large numbers of farming households of mostly Uyghur origin, the largest minority in the province. Having to farm 22

23

24

Cotton particularly needs water in the month of June, while fruits and vegetables need water in the spring, when it is often not supplied because of the absence of regulatory dams and water reservoirs upstream in the Yarkant river (Data provided in a meeting with Shache county officials, in this case from the Water Bureau, 24 March 2008). The question of possible pesticide residues on the fruits or nuts produced in the cotton fields was asked many times in our fieldwork, but never answered satisfactorily. For example, in Ygenqi Township, when visiting an inter-cropped field, the head of the township said that they use a tractor for spraying pesticides on the cotton “as much as possible low to the ground”. He also stated that “until recently the pressure to plant cotton had been quite strong, but this has decreased and they now have many more options; for farmers this is particularly so on reclaimed land, but not so much on ‘contract’ land” (Fieldnotes, Visit to Igenqi and Taghaqi townships, Shache county, 23 March 2008). Asking the county officials of the Agricultural Bureau about the potential markets for apricots, nuts and jujubes, there was very little clarity at that level. The analysis given was that, for almonds there would be no problems in selling more, as they were only produced in this region. For fruits like apricots and jujubes, these should be sold (fresh or processed) to other regions and, as one official concluded, “in Xinjiang they should start eating more fruit”. As other regions (take Sichuan) are implementing similar diversification programs for smallholders (see Tilt 2008), the question remains whether domestic demand will be sufficient. Export markets for several of the targeted fruits and nuts are already dominated by countries such as Turkey (Personal communication with Dr. Murat Arsel during our joint field visits in Shache county).

238

Spoor, Shi, and Pu

on average between 2-5 mu (0.13-0.33 hectare) per rural inhabitant, with often scarce water resources, we have argued that these farmers’ quasiexclusive dependence on cotton as income generator will not lead them out of poverty, which is still widespread in the region, in particular in the southwestern prefectures of Kashgar and Khotan. We have carefully compared secondary data from the Statistics Bureau of the Xinjiang Uyghur Autonomous Region (XUAR) with data gathered in a field visit undertaken by a team of scholars from the Institute of Social Studies, the Xinjiang Agricultural University and the Nanjing Agricultural University to Awati and Shache counties (respectively of Aksu and Kashgar prefectures) in the second half of March 2008. Our conclusion is that there seems to be a rather linear relation between the initial arable land allocation, that in densely populated areas (such as Shache and the neighbouring county of Yengisar) departed from an initially low allocation of ‘contracted’ land. The dependency on agricultural production in net rural income is very high, also because there is little migration and a very weak non-farm rural economy to complement it. Only animal husbandry seems to add significantly to rural incomes. Although according to official data the effect of husbandry is estimated to be relatively marginal (see Table 5), in interviews with farming households it seems to be more important (see Box 1 and 2). Finally, in agricultural production the dependency on cotton remains very high, completing the link between low incomes, little land, and cotton. During our fieldwork we observed that for the past few years, in particular since 2004 and as very clearly illustrated in Shache county (the most populated county of Kashgar prefecture), the regional government has pushed forward a plan to diversify farm produce, with inter-cropping of fruit (apricot and jujube) and nut (almond and walnut) trees, and the conversion of some farmland into greenhouses for vegetable production. This reveals an awareness of the regional and local governments that diversification is needed and more HVA production needs to be undertaken by these smallholders; otherwise their incomes will not substantially improve. This is a policy-induced change, but it also opens avenues for more specific decisions by small farmers to diversify and move out of their quasi-dependency on cotton. Massive production of fruits and nuts, however, also faces some problems. In the form of practiced inter-cropping there are limitations, as the water needs of various plants are different; hence there would be a need for (still expensive) drip irrigation. Furthermore, pesticide spraying on cotton might negatively affect the quality of fruits and nuts that are harvested in the same field (at least from a consumer protection perspective). Finally, there is a danger of a ‘fallacy of composition’ effect, as at the time of writing it remains rather unclear


where the markets for all these (largely unprocessed) commodities are to be found. There has been some investment in fruit processing plants, but hope was laid on increased demand from other provinces. However, all over China there are similar processes of diversification towards fruits, nuts, spices and vegetables being implemented; therefore, it still remains to be seen whether this strategy will be successful.

Acknowledgements The research underlying this paper is part of the project “Changing Livelihood Strategies in Rural Xinjiang: Cotton Production, Environment and Poverty Reduction” (Project No. 07CDP028; 2007-2010), financed by the Royal Dutch Academy of Sciences (KNAW), and approved by the Chinese Ministry of Science and Technology, executed by the Institute of Social Studies/Erasmus University Rotterdam, in cooperation with the Xinjiang Agricultural University, in particular the College of Economics and Management (CEM/XAU), and with support of the Nanjing Agricultural University. The valuable assistance rendered by Dr. Zhulifia, Associate Professor at CEM/XAU, and Mr. Liu Wensheng, PhD Scholar at CEM/XAU, and the work of Dr. Murat Arsel and Dr. Kristin Komives (both ISS) in the preparation and execution of the joint fieldwork in March 2008 is gratefully acknowledged here. The authors are also grateful for the comments provided by Dr. Andrew Fischer of ISS.

References Becquelin, N. (2000). Xinjiang in the Nineties. The China Journal 44: 65–90. Becquelin, N. (2004). Staged Development in Xinjiang. China Quarterly 178: 358–78. Chen, J., Chai, J., Chen, K., Jiang, Z. (2004). Causes of Regional Poverty and Direction of Investment in Xinjiang, China. College of Economics and Management, Xinjiang Agricultural University of China (In mimeo). Heilig, G. K., Zhang, M., Long, H., Li, X., Wu, X. (2005). Poverty Alleviation in China: A Lesson for the Developing World? Paper presented at the International Conference on the West Development and Sustainable Development, Urumqi, China, August 2–4. Ho, P.P.S (2009). Land Markets, Property and Disputes in China. In: Spoor, M. (ed.), The Political Economy of Rural Livelihoods in Transition Economies: Land, Peasants and Rural Poverty. London and New York: Routledge, 200−224. Ho, P.P.S., Spoor, M. (2006). Whose land? The Political Economy of Land Titling in Transition Economies. Land Use Policy 23 (4): 580−87. NBSC (2007), China Statistical Yearbook 2007. Beijing: China Statistics Press, published for the National Bureau of Statistics of China Spoor, M., Shi, X. (2009). Cotton and Rural Income Development in Xinjiang. In: Spoor, M. (ed.), The Political Economy of Rural Livelihoods in Transition Economies: Land, Peasants and Rural Poverty. London and New York: Routledge, 225−243.

240

Spoor, Shi, and Pu

Statistics Bureau XUAR (2001). Xinjiang Statistical Yearbook 2001. Urumqi/Beijing: China Statistics Press. Statistics Bureau XUAR (2002). Xinjiang Statistical Yearbook 2002. Urumqi/Beijing: China Statistics Press. Statistics Bureau XUAR (2005). Xinjiang Statistical Yearbook 2005. Urumqi/Beijing: China Statistics Press. Statistics Bureau XUAR (2007). Xinjiang Statistical Yearbook 2007. Urumqi/Beijing: China Statistics Press. Tilt, B. (2008). Smallholders and the ‘Household Responsibility System’: Adapting to Institutional Change in Chinese Agriculture. Human Ecology 36: 189−99.

Part IV Agricultural Intensification: Pesticide Use and IPM

Chapter 13 Reaping Bitter Fruit? Farmers’ Health and Pesticide Use in the Mekong Delta, Vietnam

Nguyen Huu Dung1, Max Spoor2,3, and Lorenzo Pellegrini2 Abstract: This paper analyses the impacts on farmers’ health of pesticide use in rice production in the Mekong Delta, Vietnam. The data are provided by two household surveys carried out in 1996/97 and 2000/01 which offer insights on changes in pesticide use over time and the health impacts of such use. Furthermore, a health-cost model is used to estimate the costs associated with pesticide-related health impairments. The application of chemical inputs per hectare per crop diminished during the study period, but incidence of double and triple cropping increased, resulting in rising total use of agrochemicals. At the same time, some farmers adopted integrated pest management, and an increase was recorded in the surveyed farmers’ awareness of the possible health impacts of toxic substances. Finally, pesticide use was indeed found to be associated with acute poisoning symptoms and with substantial private health costs. Keywords: Pesticide use, Integrated Pest Management, Farmers’ health, Vietnam

1 Introduction The Mekong River Delta is considered the ‘rice basket’ of Vietnam. In recent years, intensification of rice production in the delta has led to substantial increases in land productivity and overall output. Intensification has been marked by rising use of pesticides, chemical fertilizers, and improved seeds as well as by a movement towards two to three rice crops per year, in some cases combining rice production with one dryland crop. Though the tendency in rice prices over the past decade has been negative, the relatively higher increment in yields has nonetheless caused farmers’

1 2 3

University of Economics, Ho Chi Minh City, and Centre for Environment Economics, UEH Institute of Social Studies, The Hague/Erasmus University Rotterdam Barcelona Institute for International Studies


244

Dung, Spoor, and Pellegrini

incomes to rise.4 Intensification of rice production, however, relies heavily on High Yielding Varieties (HYVs) and agrochemicals. This study, which covers the four-year period between the agricultural seasons 1996/97 and 2000/01, estimates the impact of use of some of these chemicals, in particular pesticides, on human health and computes the associated private health costs. In general, there are two possible routes of exposure of humans and wildlife to pesticides: (i) direct exposure after spraying or granulate treatment via spray drift and gases and (ii) indirect exposure via (residues in) food derived from crops on which pesticides have been applied directly and via pesticide-contaminated drinking water. The current study focuses on the effects of direct exposure on farmers’ health. It uses a health-risk model to estimate risks to farmers’ health by combining information on direct exposure to pesticides with their associated health impacts. The analysis begins with a statistical description of farmers’ exposure to pesticides and health problems reported by the respondents to surveys that we conducted in 1996/97 and 2000/01. The health problems include acute poisoning symptoms that could be related to pesticide exposure. As a second step, a conservative estimate is calculated of the health costs associated with pesticide use to approximate the private health costs related to the use of agrochemicals. Lowering these expenditures could be considered to be a result of farmers’ substantially reducing pesticide use, a form of monetary benefit accruing to them, and might also be interpreted as a lower-boundary estimate of the true value of reducing pesticide-related illness. However, the true value would include a number of values in addition to our estimates, most notably, state expenditures on health and externalities. Part of Vietnam’s health expenses are covered by the state, which subsidizes health care. Furthermore, pesticides leave the system in the form of leakages, surface runoff, spray drift, and gases. Emissions of pesticides applied to fields pose potential risks to human health (both applicators and consumers), beneficial organisms, groundwater, and ecosystem integrity. These effects constitute a significant cost to society and to individual rice farmers – affecting the long-term sustainability of rice production – and can be viewed as externalities: they are unintended consequences of production, for which farmers are neither economically penalised nor rewarded. Although these externalities are likely to have a substantial negative impact on human welfare, they have been omitted from this analysis. 4

This article was written before the price hikes of rice in 2008, which will affect the income of rice producers in the Mekong Delta (and their production decisions, including the use of pesticides).

Farmers’ Health and Pesticide Use in the Mekong Delta, Vietnam

245

Another factor that lowers our estimates of health effects is that we have studied only acute symptoms. Although pesticide use has also been found to be associated with chronic diseases (Antle and Pingali 1994; Pingali et al. 1994; Huang et al. 2001; Mancini et al. 2005), their costs have been excluded from the analysis, though they certainly do comprise part of the cost of pesticide use.5 The study sites for the two surveys are located in the Mekong Delta. The 1996/97 survey covered six villages in five districts of four provinces, and the 2000/01 survey was conducted in six villages of six districts of the same provinces as the 1996/97 survey. The districts are situated along the Mekong River, 120 km west of Ho Chi Minh city, between the border with Cambodia and the city of Can Tho. The survey sites have varied production environments (in terms of access to irrigation, soil fertility, etc.) and agrochemical application per hectare. Villages were selected according to the following criteria: 1. provinces, districts, and villages where rice is the prevalent crop and different levels of intensification are represented;6 2. a relatively wide geographical distribution of sites; and 3. location away from cities and district centres, since agricultural land in urban areas may be allocated to other uses in coming years and may not be representative of the numerous rural villages in the Mekong River Delta. The Mekong River Delta is a large and diverse area, with a geographically dispersed population. Stratified purposive sampling was therefore employed to ensure that smaller groups of farmers matching the sampling criteria were adequately represented in the survey sample. A total of 30 farm households were interviewed in each of the six villages, making a total sample of 180 respondents per survey. The production and healthimpairment data in the surveys concentrated on the winter-spring season, and health-cost data was collected only in the 1996/97 survey. Patterns of agrochemical use were studied by comparing households across the two surveys. A group of 76 households in the four villages were present in both surveys, enabling direct comparisons to be made over time. This panel data set of 76 households provided rather complete information and is denoted as the ‘same households group’ (SHH group), as distinct 5 6

One example of these externalities is contamination of drinking water (e.g. CSE 2003). Data on land use for selection purposes were provided by the provincial and district authorities.

246


from ‘all farmers’ or ‘all farm households’, which form a substantially larger group. Analysis of the survey data on pesticide use shows that the intensification of rice production, as it is being pursued in the Mekong Delta and more particularly at the study sites, has had serious health implications. Farmers’ health costs suggest that pesticide use has also had a significant monetary cost for rice-producing farm households due to related health impairments.

2 Farmers and pesticides The risk of adverse health effects is a function of pesticide toxicity and exposure during application. Toxicity refers to the ability to cause injury or illness. It can be either acute, causing illness that develops soon after exposure, or chronic, causing illness that develops over a long period after exposure. Chronic effects of pesticide poisoning are often irreversible and may include reduced body weight, anaemia, kidney disorders, central nervous system disorders and cardiovascular disorders. A pesticide with a high level of acute toxicity can be very hazardous to health, even when a small amount is absorbed. However, in Vietnam there is no monitoring of the health impacts of pesticide use and no adequate health impact statistics are available at either the local or national levels. In our survey, we collected evidence of perceived acute illness and combined it with other household-level information. Since 1992, the national integrated pest management (IPM) programme, has helped farmers to make better use of pesticides and to improve their knowledge and skills in pest control and safety. The programme has focused on the farm level and included a broad range of training and research activities. IPM methods were being introduced in the study areas as well, and during the four-year study period increasing numbers of farmers began implementing IPM. Our data distinguish between IPM and non-IPM farm households. 2.1 Frequency of pesticide application Exposure to pesticides can threaten health depending on, among other factors, the nature of the pesticide, frequency of exposure and quantities used. We follow the World Health Organization in disaggregating pesticides into categories I, II, III, and IV, with higher numerical order denoting decreasing risk.74 Frequency of pesticide application refers to the 7

The World Health Organization classifies pesticides into four categories based on the toxicity of the chemical compound and on their formulation: extremely hazardous

247


number of sprayings or other forms of treatment with pesticides during a growing season and acts as another indicator of pesticide exposure. Table 1 shows that the average number of pesticide applications increased from 3.71 in 1996/97 to 4.06 in 2000/01, an increase that is statistically significant at the 5% level. Farmers typically applied herbicides once, insecticides once, and fungicides twice (PPD 1999). Farmers often applied more than one pesticide at a time, especially for fungicides. Therefore, the total number of exposures to hazardous categories of pesticide is somewhat different from the number of applications. The frequencies of exposure to pesticides in categories I and II (NA1) and pesticides in categories III and IV (NA2) are defined as the number of times that farmers had contact with that level of pesticide. Because each farmer could be exposed to more than one type of pesticide during a single application, the sum of NA1 and NA2 is equal to or larger than the number of applications per season. This distinction was made to better reflect the link between pesticide use and health impairments. A ‘vertical’ comparison between seasons shows a substantial change in the application of pesticide categories. Table 1 shows that, during the four-year period, the farmers surveyed reduced the frequency of application of pesticides in categories I and II from 2.53 to 1.80 applications per crop, but raised that of pesticides in categories III and IV from 2.65 to 3.53 applications per crop. All of these changes are statistically significant.

Table 1: Mean frequency of pesticide usage (number of applications per crop) Pesticide type All pesticides Categories I and II (NA1) Categories III and IV (NA2)

1996/97

2000/01 Non-IPM farmers (N=166) 3.72 4.15 2.70 1.91 2.60 3.60

t-test -1.83* 2.75*** -3.26***

IPM farmers (N=170) All pesticides Categories I and II (NA1) Categories III and IV (NA2)

3.69 2.16 2.76

All pesticides Categories I and II (NA1) Categories III and IV (NA2) Source: 1996/97 and 2000/01 surveys.

3.71 2.53 2.65

4.02 1.75 3.50 All farmers (N=336) 4.06 1.80 3.53

-1.69* 1.82* -2.85*** -2.47** 4.34*** -4.76***

(Ia) and highly hazardous (Ib); moderately hazardous (II); slightly hazardous (III); and unlikely to present acute hazard in normal use (IV).

248


Although the situation varies from case to case, many factors have contributed to the relative increase in categories III and IV use. Pesticide companies and retailers have conducted marketing activities, daily broadcasting numerous advertisements for fungicides (which belong to categories III and IV) via television and radio. Furthermore, low fungicide prices relative to insecticide and herbicide prices have contributed to an increase in their use. IPM farmers, through their participation in farmer field schools, training courses, and experimental fields, have altered their perception of insecticide spraying, thereby reducing the overall number of applications. Within a season, the average frequencies of pesticide application and use of pesticides in both the NA1 and NA2 categories were lower for IPM farmers than for non-IPM farmers (except for the application of NA2 in 1996/97). However, the difference was statistically significant only for the application frequency of NA1 pesticides in 1996/97 (see Table A-2 in the annex). Though the average number of pesticide applications per crop for IPM farmers was lower than that for non-IPM farmers, data in Table 2 show that both groups increased the number of applications during the four-year period. In 2000/01, 47% (all farm households) to 55% (SHH group) of the non-IPM farmers sprayed five times or more per season, while only 33% (SHH group) to 35% (all farm households) of IPM farmers sprayed that often. A survey by the Plant Protection Department, PPD (1999) similarly found IPM farmers to have a lower number of pesticide applications than non-IPM farmers. Farmers with higher educational levels also tended to have a lower frequency of pesticide application, implying that those farmers had better access to IPM practices and innovation (see Table A-1 in the annex).

Table 2: Frequency of pesticide application in rice production (share of surveyed farm households) Frequency (Applications/crop)

1996/97

Non-IPM IPM All All farm households 0-2 19.3 15.5 18.1 3-4 57.1 67.2 60.5 >=5 23.5 17.2 21.5 SHH-group 0-2 19.1 20.7 19.7 3-4 57.4 72.4 63.2 >=5 23.4 6.9 17.1 Source: 1996/97 and 2000/01 surveys.

2000/01 Non-IPM

IPM

All

19.1 34.0 46.8

12.5 52.7 34.8

14.5 47.2 38.4

13.6 31.8 54.5

13.0 53.7 33.3

13.2 47.4 39.5


249

2.2 Farmers’ perceptions of the effects of pesticide use on personal health The perception that long-term pesticide use can contribute to health ailments was relatively common among the respondents. Most of the farmers interviewed said that their spouses, children, and other family members participated in rice growing. Field activities include planting, weeding, applying fertilizer, spraying pesticides, and harvesting rice crops. However, no farmer reported allowing children or women to directly apply pesticides. Some children were allowed to participate in pesticide application by carrying pesticide bottles or boxes, sprayers, and water for applicators. Nonetheless, because pesticides drift and find their way into the water of the rice fields, they may nonetheless lead to health effects on female and child workers involved in other activities, such as weeding.8

Table 3: Health effects of prolonged pesticide use as perceived by surveyed farmers 1996/97 survey Measurement scale No. of % of farmers farmers No effect 8 4.5 Very little effect 25 14.1 48.0 Little effect 52 29.4 Large effect 40 22.6 Very large effect 33 18.6 52.0 Extremely large effect 19 10.7 177 100.0 Source: 1996/97 and 2000/01 surveys.

2000/01 survey No. of % of farmers farmers 4 2.5 20 12.6 34.0 30 18.9 54 34.0 29 18.2 66.0 22 13.8 159 100.0

Most farmers perceived pesticides as having ill effects on their health (Table 3), with more experienced farmers believing pesticides to have a stronger negative effect on health, compared to farmers who had been using pesticides for a shorter time. Data in Table 3 indicate that recognition of pesticides’ potential ill effects on health has increased over time, a trend that might be associated with training that was part of the IPM programme. Farmers who said that pesticide application had a ‘large effect’ on their health believed that, if they used pesticides for an extended period, their physical condition would become weaker, their lifespan would be reduced, 8

Mancini et al. (2005) found that also women –who typically do not apply pesticides directly– are affected by pesticides and show symptoms of acute poisoning. These impacts would be additional to the ones analysed here –that relate only to direct spraying– but none of the risky behaviour mentioned by these authors are common in our study area (e.g. in the Mekong Delta we found that no women mix pesticides by hand with water, before application).

250


and they would face other ‘unknown’ health problems. Those who foresaw ‘no effect’ on their health were predominantly young farmers who had only recently begun growing rice and had never experienced poisoning symptoms. They believed acute poisoning symptoms, such as fatigue, headaches, and itching skin, were normal and short-lived, saying that these signs normally ‘disappeared after bathing’. 2.3 Evidence of health impairment Farmers reported many visible symptoms of pesticide poisoning. Since signs or symptoms of pesticide poisoning can be confused with other ailments (e.g., the flu or food poisoning), the interviewers asked respondents a question to confirm that the reported symptoms appeared right after or within 24 hours of spraying. Therefore, the cases reported in this study can be viewed as ‘actually poisoned’ observations. Research has often used visible health impairments as evidence of the effects of pesticide use on farmers’ health (Huang et al. 2001).

Table 4: Farmers’ perceptions of whether their symptoms were truly pesticide poisoning Assessment scale

1996/97 survey (N=177) % of No. of farmers* farmers 3 2.3 9 7.0 9 7.0 85 65.9 90.7 23 17.8

2000/01 survey (N=159) No. of % of farmers farmers 4 4.6 5 5.7 14 15.9 56 63.6 89.8 9 10.2

No opinion Maybe Sure Very sure Completely sure No. respondents with symptoms 129 100.0 88 100.00 Note: * based only on respondents who reported signs and symptoms of poisoning. Source: 1996/97 and 2000/01 surveys.

However, to ascertain the true proportion of farmers suffering pesticide poisoning, medical research is needed to verify acute cases and uncover hidden or chronic cases of accumulated poisoning. In this study, no medical tests were conducted; hence, health impairments are likely underestimated. Nevertheless, health impacts were consistently observed among the farmers surveyed. The 1996/97 and 2000/01 surveys found that farmers who applied pesticides experienced a host of complaints after spraying that they had not suffered beforehand. To identify reporting bias, respondents were asked whether they believed pesticides could cause such health ailments.

251


Respectively, 91% and 90% of our sample farmers in 1996/97 and 2000/01 said that they were sure their symptoms were the result of poisoning, as they did not occur after other field activities (e.g. fertilizer application) (Table 4). Post-spraying symptoms such as blurred vision, body tremors, muscle fasciculation (eyelid twitching), skin itching and irritation, and even vomiting were considered to be due to pesticide exposure. Table 4 shows that 73% (129 of 177) and 55% (88 of 159) of the farmers experienced pesticide poisoning in the 1996/97 and 2000/01 seasons, respectively. The lower proportion of farmers reporting poisoning symptoms in 2000/01 may be the result of reduced doses of pesticide applied, better knowledge of safe use and a shift from pesticide categories I and II to the less toxic categories III and IV. However, it should be noted that a farmer can simultaneously suffer from more than one acute poisoning symptom. Symptoms reported during or shortly after applying pesticides included headache, eye irritation, fatigue, shortness of breath, vomiting, skin irritation, coughing, diarrhoea, convulsions, among others (e.g. stomach cramps, body tremors, dry throat, chest pains, and dizziness). Table 5 presents the signs and symptoms of pesticide poisoning reported by all farmers.

Table 5: Signs and symptoms of pesticide poisoning reported by all farmers Sign/symptoms No. of farmers

1996/97 % of farmers

Headache 70 Skin irritation 47 Fatigue 45 Eye irritation 31 Shortage of 20 breath Heart complaints 19 Vomiting 12 Cough 5 Fever 4 Diarrhea 4 Convulsion 4 Other 27 Source: 1996/97 and 2000/01 surveys.

2000/01 % of farmers

No. of farmers 39 27 25 18

62 38 41 34

39 24 26 21

11

10

6

11 9 3 2 2 2 15

9 4 1 2 1 0 21

6 3 1 1 1 0 13

The most mentioned symptom was headache, though this is possibly associated with other conditions as well, such as hard work (Table 5). However, no headache case was reported to be severe; rather, respondents reported headaches of medium severity, requiring only a short rest. Fatigue

252


is another neurological effect of pesticide poisoning; headache, dizziness, and fatigue are among the central nervous system effects associated with mild poisoning by organophosphates, organochlorines, carbamates and high doses of pyrethroids (Sodavy et al. 2000). Skin irritation was also widespread, especially in the 1996/97 survey. Most farmers reported that during application, hands, legs, face, and eyes were often exposed to chemicals during mixing and via spray drifts. Another frequent complaint is eye irritation after spraying; which also shows a substantial increase over time. Many respondents did not wear eyeglasses during spaying, resulting in eye exposure to spray drifts. Shortage of breath, heart problems, vomiting, and convulsion were less common and substantially reduced in the second survey.9 1996/97 WS season

2000/01 WS season

e Ey ue tig Fa

m Vo

50.0 45.0

Percentage of farmers

40.0 35.0 30.0 25.0 20.0 15.0 10.0 5.0 0.0 rt ea io n t ra pi es

s er th O on lti vu on C ea rrh ia D r ve Fe gh ou it C

H

R

in e Sk ch da ea H

Figure 1: Reported signs and symptoms of pesticide poisoning, SHH group Figure 1 displays the proportion of farmers in the SHH group reporting the various signs and symptoms of poisoning. The symptoms, namely headache, skin irritation, fatigue, and eye irritation, were reported at levels similar to those of the all-farmers group, with fatigue and eye irritation showing an unexplained rise in the second survey. World Bank research on the health effects of pesticide use in the Mekong River Delta, especially from organophosphates and carbamates, showed pesticide poisoning symptoms to be severe among rice-growing farmers. Medical tests done in that study found rates of cardiovascular, respiratory, and skin diseases of 11%, 2%, and 38%, respectively. These rates of poisoning are even higher 9

Almost all farmers wear masks (but most often made from simple cloth) to prevent inhalation of pesticides during spray drifts.


253

than our findings. In the World Bank study, blood tests (n=67) found acute and chronic pesticide poisoning rates among rice-growing farmers of 42% and 58%, respectively.10 Such results strengthen the idea that the visible impairments presented in this study underestimate the impacts of agrochemical exposure on those rice farmers who directly apply chemicals in the Mekong River Delta. Dasgupta et al. (2007) furthermore reported that there was indeed a high incidence of cholinesterase enzyme inhibition from exposure to organophosphate and caramate pesticides. In their sample, they found that 35% had poisoning symptoms (with 25% of this group with acute and 21% with even chronic features). However, they also concluded that there was a weak composite correlation between the index of equally weighted (selfreported) symptoms and tested blood poisoning. In our research we did not have the possibility to have clinical tests done, hence our data is limited to self-observed and reported phenomena.

3 Determinants of acute health impairments This study examines the impacts of pesticide exposure on farmers’ health using a health-risk model with the following empirical formulation: HRISK i = α 0 + β1 AGE i + β 2WTHT i + β 3 SMOKE i + β 4 DRINK i + β 5TOCA 1i + β 6TOCA 2 i + β 7 NA1i + β 8 NA 2 i + ei

HRISK is the health impairment indicator; it is equal to 1 if any health impairment occurred or 0 if no health impairment was reported. Three separate health-risk models were estimated, one each for headache, skin irritation, and eye irritation. The independent variables are from standard health-risk models (e.g. Antle and Pingali, 1994). AGE is the farmer’s age (years since birth). WTHT, a ratio of farmer’s weight (kg) to height (m), is a proxy for the health status and the level (and quality) of nutrition. SMOKE is a dummy for smoking, and equals 1 if the farmer was a regular smoker and 0 otherwise. DRINK is a dummy for habitual alcohol consumption, and equals 1 if the farmer drank regularly and 0 otherwise. TOCA1 denotes total dose of pesticides in categories I and II in grams of active ingredient 10

The World Bank research project ‘Poverty and Pesticide Use in Vietnam’ was conducted in 2003 in collaboration with the University of Economics of Ho Chi Minh City and the Centre of Occupational and Environmental Health. The survey covered 10 communes in five provinces of the Mekong Delta, with 626 rice growing farmers that were interviewed (and clinically examined). Skin tests were performed with a group of 214 and blood tests with 190 of these. The data was presented at the Poverty Environment Nexus (PEN) Workshop in Hanoi, April 2004.

254


(a.i.) used per hectare per crop. TOCA2 denotes the total dose of pesticides in categories III and IV in grams a.i. used per hectare per crop. NA1 is the number of applications of categories I and II pesticides per season. NA2 is the number of applications of categories III and IV pesticides per season. The i denotes individual farmers. Finally, ei is an error term. We look in turn at the three symptoms most frequently reported by farmers (from the previous section), but omit fatigue since its cause is more difficult to pinpoint. 3.1 Headache In both surveys, headache incidence is significantly associated with drinking habits and health status (Table 6). Farmers who drink experience this symptom after spraying more often than non-drinking farmers.

Table 6: Estimates of health impairment (headache) of rice-growing farmers 1996/97 WS rice season 2000/01 WS rice season Coefficient Wald test Coefficient Wald test Constant -0.096 (2.252) 0.002 -0.891 (2.523) 0.125 AGE 0.053*** (0.019) 7.440 0.003 (0.018) 0.019 WTHT -0.156** (0.066) 5.435 -0.131* (0.067) 3.802 SMOKE -0.328 (0.382) 0.739 0.797 * (0.434) 3.381 DRINK 1.020 ** (0.408) 6.227 0.789 * (0.453) 3.024 TOCA1 0.025 ** (0.012) 4.340 0.004 ** (0.001) 5.310 TOCA2 0.002 ** (0.000) 6.073 0.001 (0.001) 0.702 NA1 0.105 (0.126) 0.701 0.473 * (0.263) 3.241 NA2 -0.005 (0.117) 0.002 0.396 * (0.238) 2.773 Chi square (8 d.f.) 36.707*** 52.342*** Predicted probability 0.40 0.39 Note: The figures in parentheses are standard errors of estimates; ***, **, and * denote significance at the 1%, 5%, and 10% levels, respectively. Variables

Smoking also significantly increases the probability of suffering headache in the 2000/01 survey, but not in that of 1996/97. Headache is related to the use of pesticides in both categories I and II (TOCA1) and categories III and IV (TOCA2), although the latter relation is not significant in the 2000/01 survey. The results are understandable, since most pesticides are neurotoxicants, especially pesticides in categories I and II. A more frequent application of pesticides increases the probability of suffering a headache after spraying. Farmers in the two surveys have a probability, estimated at the sample mean level of all variables, of 0.40 of suffering headaches.


255

3.2 Skin effects Skin problems are commonly reported among rice-growing farmers who are exposed to pesticides. The logit regression estimates in Table 7 indicate that the incidence of skin problems is indeed positively and significantly related to the doses of pesticides applied from categories I and II (TOCA1) and III and IV (TOCA2) in the two surveys. The influence of the number of applications (NA1) and (NA2) on skin effects is not consistent between the two surveys, though we had expected positive signs. The number of exposures to pesticides in categories I and II is significantly related to skin irritation in the 2000/01 survey. Finally, farmer’s health status (WTHT) with a negative sign, as expected, is significantly related to skin effects. Incidence of skin abnormalities was not significantly related to age or smoking. At the sample mean value of all variables, the estimated probabilities of skin problems for farmers in the 1996/97 and 2000/01 surveys are 0.27 and 0.24, respectively.

Table 7: Estimates of health impairment (skin irritation) of ricegrowing farmers Variables Constant AGE WTHT SMOKE DRINK TOCA1 TOCA2 NA1 NA2

1996/97 WS rice season Coefficient Wald test -0.161 (2.618) 0.004 -0.010 0.215 (0.021) -0.129 * (0.077) 2.812 -0.518 1.293 (0.456) 1.995*** (0.593) 11.296 0.001* 2.939 (0.0007) 0.002** (0.0009) 3.567 0.111 0.589 (0.145) 0.245 * (0.129) 3.577

2000/01 WS rice season Coefficient Wald test -0.0878 (3.044) 0.0008 0.0197 (0.021) 0.817 -0.202** (0.086) 5.547 0.122 (0.497) 0.060 0.939* (0.565) 2.769 0.003* (0.002) 3.262 0.003** (0.002) 4.486 0.457* (0.280) 2.686 0.224 (0.276) 0.662

Chi square (8 d.f.) 39.313*** 43.776*** Predicted probability 0.27 0.24 Note: The figures in parentheses are standard errors of estimates; ***, **, and * denote significance at the 1%, 5%, and 10% levels, respectively.

3.3 Eye effects Table 8 presents estimates of eye irritation due to pesticide exposure. Chronic eye irritation can lead to the formation of a vascular membrane over the cornea, which diminishes visual capacity and eventually reduces the sufferer’s productivity (Antle and Pingali, 1994). Few farmers use eye protection during spraying, indicating a general lack of attention to the ill

256


effects of pesticides on their eyes in the long run. The incidence of eye irritation increases significantly with drinking and exposure to herbicides and fungicides (TOCA2).

Table 8: Estimates of health impairment (eye irritation) of rice growing farmers 1996/97 WS rice season 2000/01 WS rice season Coefficient Wald test Coefficient Wald test Constant -1.356 (2.917) 0.216 0.374 (2.946) 0.016 AGE 0.116 0.001 (0.022) (0.057) 0.411 0.003 WTHT -0.147 * (0.083) 3.108 -0.166** (0.082) 4.243 SMOKE 0.496 (0.565) 0.771 0.374 (0.508) 0.541 DRINK 0.969 * (0.573) 2.859 0.402 (0.541) 0.552 0.003* (0.002) 2.689 TOCA1 0.005 (0.008) 0.391 TOCA2 0.025*** (0.008) 8.985 0.003* (0.001) 3.532 NA1 0.249 * (0.147) 2.882 0.531* (0.279) 3.616 (0.278) 0.044 NA2 0.118 (0.144) 0.672 0.058 Chi square (8 d.f.) 27.473*** 36.321*** Predicted probability 0.18 0.21 Note: The figures in parentheses are standard errors of estimates; ***, **, and * denote significance at the 1%, 5%, and 10% levels, respectively. Variables

As expected, the ratio of weight to height (WTHT) carries a negative sign in terms of eye abnormalities among farmers in both surveys. In addition, the number of exposures to pesticides in categories I and II (NA1) contributes significantly to eye irritation, whereas the number of exposures to pesticides in categories III and IV (NA2) has no significant effect. The probability of experiencing eye irritation among the sample farmers in both the 1996/97 and 2000/01 surveys was 0.18 and 0.21, respectively, calculated from parameters estimated from the logit function at the mean level of all variables.

4 Private health costs associated with pesticide use 4.1 Model specification Health costs related to pesticide use are commonly computed using estimates of expenses related to treatments for restoring a farmer’s health. This study looks at the private costs associated with acute health impairments caused by exposure to agrochemicals. Included in our healthcost calculations are the following items (in Vietnamese dong or ‘VND’): • the opportunity cost of work days lost due to illness (assumed to be equal to the average daily wage in 1996/97 multiplied by the number of days


257

lost) and days of restricted activity (assumed to be equal to one-third of the average daily wage); • recuperation expenses (medicines, doctors, and hospital costs, including meals) incurred by farmers who directly spray pesticides; • costs for protective equipment (avoidance costs). Due to the difficulty of estimating the costs related to treatments for restoring the health of farmers to their normal state, the health costs discussed here are for a single rice season only, limited to treatments for visible health impairments. The explanatory factors for health costs are linked to four broad classes of variables: (i) farmer’s general health, (ii) pesticide exposure, (iii) farmer characteristics, and (iv) rice-farming practices. The health variables include proxies for a farmer’s physical health (as measured by the weight to height ratio) and the two known voluntary health hazards, namely smoking and alcohol consumption. Pesticide exposure variables include total pesticide dose per hectare per crop and the number of applications (proxy for the number of times a farmer is exposed). The age of the farmer is a farmer characteristic. Finally, rice-farming practices are represented by the variable ‘IPM’. Following the health economics literature (Antle and Pingali 1994; Pingali et al. 1997: 110-116), the health cost function was modelled as a logarithmic form of the hypothesized determinant factors. The log-log cost function can be interpreted as a first-order approximation of the true cost function and is globally well behaved (Pingali et al. 1997: 110-116). LnHC i = α + β 1 LnAGE + β 2WTHT + β 3 SMOKE + β 4 DRINK +

β 5 LnDOSE + β 6 NA + β 7 IPM + ei where (LnHC) is the health cost shouldered by farmers (the natural logarithm of VND). LnAGE is the natural log form of farmer’s age. WTHT is a proxy for health and nutrition, measured as the ratio of farmer’s weight (kg) to height (m). SMOKE and DRINK are dummies for smoking and drinking (0 for non-smokers, non-drinkers and 1 for smokers, drinkers). LnDOSE is the natural log of total dosage of pesticides used (gram a.i./crop/ha). NA is the number of pesticide applications per season. IPM is a dummy variable taking a value of 1 if the farmer in question practises IPM and a value of 0 otherwise. Descriptive statistics of variables with continuous values, such as health cost, age, pesticide doses, and physical fitness (WTHT), are presented in Table A-3 in the annex. Box-plot diagrams accompanying this table show

258


that these variables are normally distributed and not skewed. An ordinary least squares method was then employed to estimate the model parameters. 4.2 Estimated pesticide-related health costs Table 9 presents the estimated parameters for farmers’ health costs. Results of the overall test for significance of the model (F-test) and adjusted R2 show the model is significant, and independent variables on the right-hand side explain 55% of the change in pesticide-related health costs (the dependent variable). While all other explanatory variables are significant and consistent with expectations, the ‘smoking’ dummy variable has a negative sign, contrary to our expectation, as this variable had been found to significantly affect farmers’ health in a similar study undertaken in the Philippines (Antle and Pingali 1994). Though bearing an unexpected sign, the estimated coefficient is not statistically significant, indicating that smoking does not affect farmers’ health costs. This result is not surprising once we consider that most smoking-related health problems are chronic in nature, while here we considered only acute symptoms. Habitual alcohol consumption contributes significantly to a rise in farmers’ health costs.11 Since drinking is common among farmers in the study sites (reported by 61% of the respondents in the 1996/97 survey), this is a matter of concern. Meanwhile, the health status variable is statistically significant, with a negative sign as expected. This finding indicates that the nutritional/physical status of farmers is an important factor in reducing their health costs. It also reveals high health costs for those farmers with ‘weak’ physical condition in the data set. The estimated coefficient of the age variable shows it to be associated with increased health costs. In other words, the older the farmer the higher the health costs, a relationship that was to be expected. Considering the impact on farmers’ health costs of the total quantity of pesticides used, estimates show that health costs increase by 0.93% for every 1% increase in total dose. This significant impact of pesticide exposure on farmer’s health costs is consistent with previous studies in other countries (Antle and Pingali 1994; Dung and Dung 1999; Huang et al. 2001). Figure 2 shows that the health costs reported by farmers increases as the dose of pesticides applied increases. Health costs due to pesticide exposure are higher for non-IPM farmers.

11

This result is contrary to that in the study of Pingali et al. (1994), who found an insignificant but still negative sign, which seems to be counter-intuitive.


259

120,000

Health costs (VND)

100,000

80,000

60,000 40,000

20,000

0

1

2

3 4 Pesticide use quintiles

All farmers

5

Non-IPM farmers IPM farmers

Figure 2: Pesticide-related health costs reported by farmers, 1996/97 The coefficient of frequency of application is also statistically significant, indicating that the more frequent the pesticide exposure, the higher the health costs incurred. This finding is consistent with regression results in the health-risk models presented in the previous section. We assumed that farmers applied a specific dose of a given pesticide (measured in grams a.i.) over the season as a whole. We then expected more frequent use of pesticides – and, therefore, a smaller amount of pesticide applied each time – to be associated with lower health costs (Huang et al., 2001). However, this assumption may not be valid, since toxicity or hazardousness is not the same between pesticides or even within one hazard category. Farmers at the study sites used a combination of herbicides, insecticides, and fungicides. When farmers used a larger dose of pesticides in categories I and II, and applied them more frequently, more health problems were likely to occur, and hence higher health costs. Finally, application of IPM techniques has a beneficial impact on farmers’ health. The negative and significant coefficient of the IPM dummy variable indicates that farmers practising IPM techniques incur significantly lower costs due to health impairment. In addition to lower pesticide use (see Dung and Spoor 2007), IPM farmers reported a better understanding of the health hazards involved in pesticide use, and hence followed more precautions during applications. Our results indicate that IPM farmers have lower health costs, even when we control for pesticide use. The pesticide-related health cost was estimated at 54,720 VND per farmer per season, using sample mean values (i.e. average values of the sample in the model and the mode for dummies). This is a relatively low

260


cost value, but must be multiplied by a factor of two or three, as this is the number of rice crops planted annually on a plot.

Table 9: Estimated heath costs to farmers due to pesticide exposure, Mekong Delta 1996/97 Variables Coefficient Standard error t-ratio Constant 1.806 1.017 1.775* LnDOSE 0.934 0.092 10.178*** NA 0.179 0.037 4.882*** LnAGE 0.774 0.164 4.728*** WTHT -0.028 0.016 - 1.778* SMOKE -0.048 0.076 - 0.641 DRINK 0.109 0.064 1.695* IPM -0.201 0.070 - 2.865*** Predicted 54,720 VND; Min: 18,958 VND; Max: 154,517 VND health cost Source: Estimated from 1996/97 survey. Adjusted R square = 0.55. Regression F value (7,136) = 37.464; ***. ***,**, and * denote statistically significant at the 1%, 5%, and 10% levels, respectively.

The parameters estimated using the health-cost regression results were also employed to calculate the average and marginal farmer health cost per applied pesticide dose. Two scenarios with differing assumption values are presented here for illustration purposes. In scenario 1, the health cost is based on estimates for a non-smoking, but drinking, farmer population. It assumes an average farmer, aged 41 years, with a weight of 51 kg, and a height of 1.60 m (WTHT ratio of 31.87). Pesticide doses applied are simulated at levels of 750, 850 and 950 grams a.i. per hectare per crop. The number of pesticide applications is four per season. Finally, health costs are predicted for farmers who use IPM techniques and those who do not. In scenario 2, all assumption values are the same as in scenario 1, except the number of pesticide applications is five per season. The results are summarized in Tables A-4 and A-5 in the annex. A number of observations can be made from the estimated results. For IPM farmers, when pesticide doses are applied four times, and increased from 750 (average dose in 2000/01 ) to 850 and 950 grams a.i. (the average dose in 1996/97 ), health costs increased from 40,135 to 45,115 and 50,032 VND per farmer per season, respectively. The marginal health costs indicate that each additional gram of pesticide a.i. causes a health-cost increase of 49.98, 49.56 and 49.19 VND, respectively.12 12

To understand the limited magnitude of these (marginal) costs, the exchange rate of the USD versus the Vietnamese đong was 11,270 đong during the 1996-97 season,


261

For non-IPM farmers with the same doses of pesticide use and other characteristics, the pesticide-related health costs per season (per gram a.i. of pesticide) and marginal health costs are all higher than those of IPM farmers (Table A-4 in annex). When frequency of application increased from four to five times per crop per season, the estimated health cost increased by about 16.4% (Table A-5 in annex).

5 Conclusion This study has demonstrated the link between pesticide application – in terms of both quantity and frequency – and acute poisoning symptoms. Farmers’ perceptions are in line with our findings, and a high percentage of the farmers did in fact associate their symptoms with pesticide applications. Furthermore, private health costs are related to the total dose of pesticides applied and the frequency of application. Meanwhile, adoption of IPM techniques and farmers’ nutritional/physical status are important factors in reducing farmers’ health costs. Most notably, introduction of IPM techniques has a significant positive influence on health costs, partly due to the use of less poisonous substances and partly because of a more cautious attitude towards agrochemical applications in general on the part of IPM farmers.

References Antle, J.M., Capalbo, S.M. (1994). Pesticide, Productivity and Farmer Health: Implications for Regulatory Policy and Agricultural Research. American Journal of Agricultural Economics 76 (3): 598-602. Antle, J.M., Pingali, P.L. (1994). Pesticides, Productivity, and Farmer Health: A Philippine Case Study. American Journal of Agricultural Economics 76 (3): 401-430. Carpenter, S.R., Carado, N.F., Correll, D.L., Howarth, R.W., Sharpley, A.N., Smith, V.H. (1998). Nonpoint Pollution of Surface Waters with Phosphorus and Nitrogen. Ecological Application 8(3): 559-568. Centre for Occupational and Environmental Health (2004). Poverty and Pesticide Use in Vietnam’, Report presented at the Poverty Environment Nexus (PEN) Workshop in Hanoi, organised in cooperation with the World Bank, April. CSE (2003). Analysis of Pesticide Residues in Bottled Water. Centre for Science and Environment, CSE Report, India. Dasgupta, S., Meisner, G., Wheeler, D., Xuyen, K., Lam, N.T. (2007). Pesticide Poisoning of Farm Workers: Implications of Blood Test Results from Vietnam. International Journal of Hygiene and Environmental Health 210: 121-132. and 14,460 đong during the 2000-01 season [see< http://www.oanda.com/convert/ fxhistory>, accessed 15 May 2008]

262


Dung, N.H., Spoor, M. (2007). Intensification of Rice Production and Negative Health Effects for Farmers in the Mekong Delta during Vietnam’s Transition. In: Spoor, M., Heerink, N., Qu, F. (eds.), Dragons with Clay Feet? Transition, Sustainable Land Use, and Rural Environment in China and Vietnam, Lanham/New York: Rowman & Littlefield, Lexington Books, 229-250. Dung, N.H. (1994). Profitability and Price Response of Rice Production system in the Mekong Delta-Vietnam’, Unpublished Master Thesis, Chiangmai University, Thailand. Dung, N.H., Dung, T.T.T. (1999). Economic and Health Consequences of Pesticides Use in Paddy Production in the Mekong Delta-Vietnam. Research Report, Economy and Environment Program for Southeast Asia (EEPSEA). Dung, N.H. (1995). Agricultural Inputs and Rice Marketing in the South-East Area and Mekong Delta Vietnam. Research Report under the State Project KX03-21C. Dung, N.H. (1997). Role of Rural Women in Sustainable Agricultural Development and Hunger Alleviation and Poverty Reduction in the Mekong Delta Vietnam. Research Report, Vietnam- Netherlands Research Program (VNRP). Dung, N.H., Thien, T.C., Hong, N.V., Loc, N.T., Minh, D.V., Thau, T.D., Nguyen, H.T.L., Phong, N.T., Son, T.T. (1999). Agro-chemicals, Productivity and Health in Vietnam. Research Report, Economy and Environment Program for Southeast Asia (EEPSEA). Fang, B., Wang, G. , van den Berg, M., Roetter, R. (2005). Identification of Technology Options for Reducing Nitrogen Pollution in Cropping Systems of Pujang. Journal of Zhejiang University SCIENCE 2005 6B (10): 981-990. FFTC (2002). Effects of Intensive Fertilizer Use on Groundwater Quality. Heong, K.L., Cuc, N.T.T., Binh, N., Fujisaka, S., Bottrell, D.J. (1995). Reducing EarlySeason Intersection Applications through Farmers’ Experiments in Vietnam. In: Denning, G. L., Xuan, V.T. (eds), Vietnam and IRRI : A Partnership in Rice Research, IRRI, The Philippines, and Ministry of Agriculture and Food Forestry, Hanoi, Vietnam, 217-222. Huan, N.H., Mai, V., Escalada, M.M., Heong, K.L. (1999). Changes in Rice Farmers’ Pest Management in the Mekong Delta, Vietnam. Crop Production 18: 557563. Huang, J., Qiao, F., Zhang, L., Rozelle, S. (2001). Farm Pesticide, Rice Production, and Human Health. Research Report, Economy and Environment Program for Southeast Asia (EEPSEA). Hung, N.N., Singh, U., Xuan, V.T., Buresh, R.J., Padilla, J.L., Lap, T.T, Nga, T.T. (1995). Improving Nitrogen Use Efficiency of Direct-Seeded Rice on Alluvial Soils of the Mekong River Delta. In: Denning, G.L., Xuan, V.T. (eds), Vietnam and IRRI: A Partnership in Rice Research. IRRI, The Philippines, and Ministry of Agriculture and Food Forestry, Hanoi, Vietnam, 137-148. Mancini, F., van Bruggen, A.H.C., Jiggens, J.L.S., Ambatipudi, A.C., Murphy, H. (2005). Acute Pesticide Poisoning Among Female and Male Cotton Growers in India. International Journal of Occupational and Environmental Health 11 (3), 221-232.


263

Pathak, B.K., Kazama, F., Iida, T. (2004). Monitoring of Nitrogen Leaching from a Tropical Paddy Field in Thailand. Agricultural Engineering International: The CIGR Journal of Scientific Research and Development. Manuscript LW 04 015, Vol. VI, December. Pingali, P.L., Marquez; C.B., Palis, F.G. (1994). Pesticides and Philipinne Rice Farmer Health: A Medical and Economic Analysis. American Journal of Agricultural Economics 76 (3): 587-92. Pingali, P.L., Roger, P.A. (eds.) (1995). Impact of Pesticides on Farmer Health and The Rice Environment. Massachusetts: Kluwer Academic Publishers and the International Rice Research Institute, Losbanos, Laguna, Philippines. Pingali, P.L., Hossain, M., Gerpacio, R.V. (1997). Asian Rice Bowls: The Returning Crisis? Wallingford: CAB International. PPD (1999). Farmers’ Perceptions, Beliefs and Practices in Rice Pest Management in High Production System. Survey PPD-MARD Vietnam, November. Roy, R.N., Misra, P.V. (2002). Economic and Environmental Impact of Improved Nitrogen Management in Asian Rice-Farming Systems, Sodavy, P., Sitha, M., Nugent, R. (2000). Farmers’ Awareness and Perceptions of the Effect of Pesticides on Their Health. FAO Community IPM Program, Field Document, Cambodia . Son, T.T (1998). The Economics of Fertilizer Use: The Case of Rice Production in the Mekong Delta, Vietnam. Master’s Thesis. Ho Chi Minh City University of Economics. Xing, G.X., Zhu, Z.L. (2000). An Assessment of N Loss from Agricultural Fields to the Environment in China. Nutrient Cycling in Agroecosystem 57 (1): 67-73. Zhu, J.G., Han, Y., Lin, G., Zhang, Y.L., Shao, X.H. (2000). Nitrogen in Percolation Water in Paddy Fields with a Rice/Wheat Rotation. Nutrient Cycling in Agroecosystems 57(1): 75-82.

264


Appendix Table A-1: Number of pesticide applications in rice production, rainy season, Mekong Delta 1999, classified by farmers’ level of education and use of IPM No. of Applications

IPM training

Farmer education

Trained Not trained Elementary Secondary High school Insecticide applications (percentage of farmers) 0 10.6 1.8 4.4 5.1 10.3 1 50.8 30.2 30.7 46.4 44.3 2 24.6 34.9 33.3 26.1 30.9 3 10.6 14.8 14.0 13.0 10.3 4 3.4 13.0 13.2 7.2 3.1 >=5 0.0 5.4 4.4 2.1 0.0 Fungicide applications (percentage of farmers) 1 6.3 3.4 2.4 7.5 4.3 2 38.0 28.1 30.9 36.0 32.8 3 41.6 46.1 44.7 43.5 43.1 4 10.9 17.4 18.7 9.3 14.7 5 3.2 3.9 3.3 3.1 4.3 6 0.0 .6 0.0 0.0 0.9 7 0.0 .6 0.0 .6 0.0 Herbicide applications (percentage of farmers) 0 4.1 1.2 4.1 2.6 1.8 1 93.6 90.1 87.7 91.6 97.4 2 2.3 8.8 8.2 5.8 .9 Total pesticide applications (percentage of farmers) 2 0.5 0.6 0.0 1.2 0.0 3 18.1 2.8 4.1 16.1 12.1 4 28.1 15.7 19.5 21.1 27.6 5 26.7 23.0 24.4 24.8 25.9 6 13.1 20.8 14.6 17.4 18.1 7 9.0 18.5 21.1 9.9 9.5 8 3.2 7.3 5.7 5.0 4.3 9 0.5 6.2 8.1 1.2 0.0 10 0.9 2.2 2.4 1.2 0.9 >10 0.0 2.9 0.0 1.8 1.7 Source: Calculated from 1999 survey by the Plant Protection Department (PPD), Southern Division.

All farmers

6.3 40.7 29.8 12.6 8.0 2.6 5.0 33.5 43.8 13.8 3.5 0.3 0.3 2.8 92.1 5.1 0.5 11.3 22.5 25.0 16.8 13.3 5.0 3.0 1.5 1.4

265


Table A-2: Mean frequency of pesticide use (nr. of applications per crop), IPM and non-IPM farmers Pesticide types

non-IPM (N=166)

IPM (N=170)

t-test

1996/97

All pesticides

3.72 (1-9)

3.69 (1- 8)

Categories I and II

2.70 (0-8)

2.16 (1-7)

Categories III and IV

2.60 (1-7)

2.76 (1-6)

0.17NS 2.08** -0.57NS

2000/01 All pesticides Categories I and II Categories III and IV

4.15 (1-7) 1.91 (0-5) 3.60 (1-7)

Source: 1996/97 and 2000/01 surveys.

4.02 (1-7) 1.75 (0-5) 3.50 (0- 6)

0.56NS 0.50NS 0.35NS

266


Table A-3: Variables used in the health-cost model 12.5

12.0

11.5

11.0

10.5

10.0

9.5

9.0 N=

LnHC Mean Median Std. Deviation Minimum Maximum Skewness

Statistic Std. Error 10.931 0.006 10.829 0.668 9.59 12.28 0.233 0.207

LnDOSE Mean Median Std. Deviation Minimum Maximum Skewness

Statistics Std. Error 6.9287 0.003 6.9832 0.3626 6.14 7.64 -0.496 0.207

LnAGE Mean Median Std. Deviation Minimum Maximum Skewness

Statistics Std. Error 3.778 0.002 3.810 0.1950 3.26 4.11 -0.612 0.207

WTHT Mean Median Std. Deviation Minimum Maximum Skewness

Statistics Std. Error 31.483 0.172 31.250 2.009 28.00 36.00 0.188 0.206

137

LNHEALTH

8.0

7.5

7.0

6.5

6.0 N=

137

LNTOPEST

4.2

4.0

3.8

3.6

3.4

3.2 N=

137

LNAGE05

38

36

34

32

30

28

26 N=

137

VARWT


267

Table A-4: Predicted health costs of pesticide use in rice production (scenario 1) Parameters CONSTANT LnDOSE NA LnAGE WTHT SMOKE DRINK IPM Ln of health cost Estimated health cost 1 Average heath cost 2 Marginal health cost 3

IPM farmers 2 1.806 6.299 0.716 2.879 -0.892 0 0.109 -0.201

3 1.806 6.403 0.716 2.879 -0.892 0 0.109 -0.201

Non-IPM farmers 1 2 1.806 1.806 6.183 6.299 0.716 0.716 2.879 2.879 -0.892 -0.892 0 0 0.109 0.109 0 0

3 1.806 6.403 0.716 2.879 -0.892 0 0.109 0

10.6

10.716

10.820

10.801

10.917

11.021

40,135

45,105

50,032

49,070

55,147

61,171

53.51

53.06

52.67

65.43

65.18

64.63

49.98

49.56

49.19

61.11

60.60

60.14

1 1.806 6.183 0.716 2.879 -0.892 0 0.109 -0.201

Note: Scenario 1 uses coefficients estimated in the health cost model and the following assumption values: a 41-year-old farmer, with a weight of 51 kg and height of 1.60 ms (WTHT ratio of 31.87); a non-smoking, but drinking, farmer. Pesticide doses applied at 750, 850, and 950 grams a.i./ha/crop presented are in columns 1, 2, and 3, respectively. Health costs are predicted for farmers who use IPM and those who do not. The number of pesticide applications is four per crop per season. 1 Estimated heath cost of farmers (VND per season); 2 Average health cost of farmers (VND/ gram a.i.); 3 Marginal health cost (MC) calculated as MC = estimated pesticide coefficient X (health cost/pesticide dose).

268


Table A-5: Predicted health costs of pesticide use in rice production (scenario 2) Parameters CONSTANT LnDOSE NA LnAGE WTHT SMOKE DRINK IPM Ln of health cost Estimated health cost 1

1 1.806 6.183 0.895 2.879 -0.892 0 0.109 -0.201 10.779

IPM farmers 2 3 1.806 1.806 6.299 6.403 0.895 0.895 2.879 2.879 -0.892 -0.892 0 0 0.109 0.109 -0.201 -0.201 10.896 10.999

Non-IPM farmers 1 2 1.806 1.806 6.183 6.299 0.895 0.895 2.879 2.879 -0.892 -0.892 0 0 0.109 0.109 0 0 10.980 11.097

3 1.806 6.403 0.895 2.879 -0.892 0 0.109 0 11.200

48,002

53,947

58,689

73,161

59,840

65,956

Average heath cost 2 64.00 63.46 62.99 78.25 77.96 3 Marginal health cost 59.78 59.28 58.83 73.09 72.47 Note: Scenario 2 uses coefficients estimated in the health cost model. Assumption values are similar to those of scenario 1, except the number of pesticide applications is five per crop per season. 1 Estimated heath cost of farmers (VND per season); 2 Average health cost of farmers (VND/gram a.i.); 3 Marginal health cost (MC) calculated as MC = estimated pesticide coefficient X (health cost/pesticide dose).

77.30 71.93

Chapter 14 Training and Visit (T&V) Extension vs. Farmer Field School: The Indonesian Experience Budy P. Resosudarmo1 and Satoshi Yamazaki2 Abstract: For several decades the effective and efficient dissemination of new agricultural knowledge among farmers in developing countries has been problematic. Two major programs were implemented in Indonesia, namely The Training and Visit (T&V) Extension Program or The Massive Guidance (BIMAS) Program, from the mid 1960s until the end of the1980s, and the Farmer Field School (FFS) Program, during the 1990s. The main difference between these two programs is that, where the T&V was concerned, farmers were instructed what to do, while the FFS program encouraged and stimulated farmers to make their own decisions. This paper aims to discuss and compare the effectiveness of these two programs with reference to rice production in Indonesia. This paper would like to argue that, for regions where the level of development is still very low, implementing a T&V program instructing farmers what to do is probably more appropriate than an FFS. As for regions where agriculture is relatively developed, an effective FFS program is more appropriate. Keywords: Food policy, Agricultural economics and policy, Development policy, Public policy

1 Introduction One of today’s major challenges remains how to feed the world population, particularly in developing countries (FAO 2004; Sachs 2005; Thirtle and Lin 2003; World Bank 2007). Being able to boost food production in countries where famine and undernourishment exist is an important strategy towards meeting this challenge. Numerous genetically altered plants have continuously been invented, even for difficult regions such as Sub-Saharan Africa, with these new varieties being expected to increase 1

2

Associate Professor, The Arndt-Corden Division of Economics, Crawford School of Economics and Government, The Australian National University, Canberra, Australia. Lecturer, School of Economics and Finance, University of Tasmania, Hobart, Australia.


270

Resosudarmo and Yamazaki

food production significantly (Hoisington et al. 1999; Khush 1999; Pinstrup-Andersen et al. 1999; Karshenas 2001; Thirtle and Lin 2003). The most important issue, then, is how these new plants and technologies should be disseminated in regions where undernourishment and poverty are the main problems. Several scientists have discussed this dissemination issue and consider that East Asia’s successful Green Revolution during the 1970s, 1980s and 1990s should be the model to follow (Otsuka and Kalirajan 2006; Otsuka 2006). Of particular interest is the case of rice production in Indonesia, where dissemination of agricultural intensification methods led to annual average growth rates of more than 4 percent during the 1960s, 1970s and the beginning of the 1980s. Two major programs in particular were implemented in Indonesia to disseminate new rice production knowledge among farmers, namely The Training and Visit (T&V) Extension Program—in Indonesia this is called the Massive Guidance (Bimbingan Masal or BIMAS) Program—and the Farmer Field School (FFS) Program. The main difference between these two programs is that, where the T&V program was concerned, farmers were instructed on what to do and were given incentives through the provision of cheap credit to follow these instructions (Mosher 1976; Feder and Slade 1986; Birkhaeuser et al. 1991), while the FFS program encouraged and stimulated farmers to make their own decisions (Oka 1991; Pincus 1991; Kenmore 1992). The performance of these programs has been well discussed in a number of previous studies for various regions.3 However, there has been less discussion of the extent to which these programs are different in terms of their implementation processes and outcomes. Hence, this paper aims to discuss and compare these two programs and their performance along those lines. There are at least five difficulties in achieving the goal of this paper. First, Indonesia no longer actively conducts any of the T&V or FFS programs. Second, Indonesia never actively conducted these two programs simultaneously: the T&V or BIMAS program was conducted from the mid 1960s until the end of the 1980s, the FFS program during the 1990s. Third, the Indonesian T&V program mostly, though not only, dealt with introducing farmers to new high-yielding varieties, namely IR8, IR5 and their derivatives, while the FFS program was mostly concerned with appropriate ways to control pests, namely the integrated pest management 3

For example, the T&V extension program worldwide has been reviewed by Birkhaeuser et al. (1991) and Anderson et al. (2007). Van den Berg and Jiggins (2007) provided a comprehensive review of the FFS program. Röling and Van den Fliert (1994) discussed in more detail the institutional settings of the FFS program. Both the T&V and FFS programs were discussed by Van den Fliert et al. (1996) and Eicher (2007), but not in great detail.

Training and Visit (T&V) Extension vs. Farmer Field School

271

approach. Fourth, consistent and comparable data reporting the performance of these two programs is relatively limited. Hence, it is difficult to compare the two programs quantitatively. Lastly, how to precisely measure the performance of these programs is itself a controversial and complicated issue, still the focus of on-going debate (Feder et al. 2004a; Van den Berg and Jiggins 2007). This paper will conduct a narrative-descriptive analysis, describing chronologically the implementation of these programs and their outcomes, including the unintentional ones, and explaining why it was possible to implement each program and why the outputs were as such. To compare the performance of these programs, this paper will particularly focus on seven indicators: 1) impact on yield, 2) impact on chemical use, 3) risk of crop failure, 4) ability to sustain the technology disseminated, 5) cost, 6) up-scaling issues, and 7) corruption issues. Our conclusions will be drawn based on these indicators. The outline of this paper is as follows. In the next section, we present the overall performance of rice production in Indonesia. Then the implementation of T&V or BIMAS and its performance is described, followed by a section on the FFS program. Finally, we compare these two programs and draw conclusions.

2 Overview of rice production in Indonesia4 Rice has been the most important food crop in Indonesia for decades. Nevertheless, maize, cassava, and sweet potatoes have also been important. We compare the performance of rice production with that of cassava, maize and sweet potatoes in Indonesia since the mid 1960s. Figure 1 shows that, in the mid 1960s, the level of rice production was more or less the same as that of cassava, and that of maize and sweet potatoes was around one fifth that of rice. Until 2006, rice and maize production grew relatively steadily and reached around 3–4 times their 1966 levels. Cassava production grew moderately, while that of sweet potatoes declined. Figure 1 also shows the average annual growth in production for cassava, maize, rice, and sweet potatoes for 4 different periods. The first period is from 1966 until 1983. This is the period when BIMAS (including its credit program) was actively being implemented. The second period is from 1984 until 1989. BIMAS’s credit program was abolished around 1984, and the whole BIMAS program had been put aside by 1989. The third period comprises when implementation of the FFS began (1990) until the economic crisis (1997). The fourth period is from 1998 until 2006, a 4

From now on, we will use the term BIMAS program for the Indonesian T&V program.

272


post-crisis period. This periodisation is expected to give us some insight into how the production of food crops has changed over time in Indonesia under different regimes. 8

Average Annual Grwoth Rate (in %)

Production (in 106 ton)

60

40

20

0 1966

1971

1976 Cassava Rice

1981

1986

1991

1996

Maize Sweet Potatoes

2001

2006

6

4

2

0

-2 1966-1983 Cassava

1984-1989 Maize

1990-1997 Rice

1998-2006

Sweet Potatoes

Figure 1: Production (in 106 metric tons) and average annual growth Rates (in %) of several Indonesian food crops Source: FAOSTAT [http://faostat.fao.org/default.aspx].

During the first period, the rice production growth rate was well above 5 percent annually — the highest growth rate of the periods covered. The growth rate of maize production was close to 2 percent annually, while cassava and sweet potatoes experienced almost no growth. To understand the reasons behind the dynamics of production of these crops, it is necessary to observe their yield per ha performance (Figure 2). Although it is true that the area under rice production has continued to expand along with the production level since the early 1960s, as can be seen in Figure 2, the yield per ha performance of rice also significantly improved during this period. This situation has typically been associated with the success of the BIMAS program. Around the end of this period, Indonesia became selfsufficient in rice (Simatupang and Timmer 2008). The slow growth of cassava and maize production were mostly due to the slow expansion of plantation areas for these crops. The negative growth of sweet potatoes was due to a reduction of areas under sweet potato production.

273

Training and Visit (T&V) Extension vs. Farmer Field School 20

Average Annual Growth Rate (in %)

6

12

3

Yield (in 10 kg/ha)

16

8

4

0 1966

1971

1976 Cassava Rice

1981

1986

1991

1996

Maize Sweet Potatoes

2001

2006

4

2

0 1966-1983 Cassava

1984-1989 Maize

1990-1997 Rice

1998-2006

Sweet Potatoes

Figure 2: Yield (in 103 hg/ha) and average annual growth rate (in %) of several Indonesian food crops Source: FAOSTAT [http://faostat.fao.org/default.aspx].

During the second period, the average annual growth rate of rice production declined to slightly above 3 percent. Cassava’s performance significantly improved; together with maize, they grew at annual rates of close to 4 percent and slightly above 3 percent, respectively (Figure 1). From the perspective of yield per ha, average annual growth of yield per ha for cassava, maize and sweet potatoes had improved compared to the first period, but not for rice, which dropped by more than half its average annual growth rate for the first period. This indicated that Indonesia was no longer self-sufficient in rice. The BIMAS program was typically criticised for this situation, and during this period alternative programs were being seriously debated. During the third period, the growth rate of rice production continued to decline to well below 2 percent, while the production of cassava and sweet potatoes grew negatively. Only maize performed rather well, relatively speaking (Figure 1). Average annual growth rates of cassava, rice and sweet potatoes were very low (Figure 2). Although the FFS introduced during this period was considered ineffectual in improving the annual growth rate of rice production, it was credited with preventing an even worse situation from occurring. These issues are still being debated (Yamazaki and Resosudarmo 2008). Whatever verdict may be reached on that score, it is generally agreed that the first severn years of the 1990s were a gloomy period for Indonesia’s food crop production (Simatupang and Timmer 2008). During the fourth period, after the economic crisis, rice production did not improve. Its growth was relatively the same as during the previous

274


period. Cassava production performed well, but not maize and sweet potatoes (Figure 1). The numbers for average annual growth rate of yield per ha do, nonetheless, show improvements for cassava, maize, rice and sweet potatoes. For rice, however, the average annual growth rate of yield per ha was still far below that of the first period. Let us now compare Indonesia’s performance in food crop production, namely rice, with some other developing countries in Asia. Figure 3 shows that Indonesia’s rice yield per ha in 1966 was similar to that of other developing countries in Asia, except for China. But from the mid 1960s to the mid 1980s, Indonesia’s productivity grew faster than that of most other countries in Asia. Hence Indonesia’s rice yield per ha was relatively higher than most other developing countries in Asia, but still not as high as that of China. As mentioned above, many associated this achievement with the implementation of the BIMAS program. Since then, however, Vietnam's rice yield per ha has increased faster than Indonesia's, so by the mid 2000s the former’s rice yield per ha outperformed that of the latter.

Yield (in 103 hg/ha)

8

6

4

2

0 1966

1971

1976

1981

1986

1991

1996

2001

2006

China

India

Indonesia

Philippines

Thailand

Viet Nam

Figure 3: Yield (in 103 hg/ha) of rice in several developing Asian countries Source: FAOSTAT [http://faostat.fao.org/default.aspx].

The conclusions that can be drawn from the discussion in this section are the following. First, rice production in Indonesia performed well during the 1960s, 1970s and early 1980s. It has been suggested that the


275

improvement in yield per ha during this period was due to the implementation of the BIMAS program. Second, since the mid 1980s, Indonesia has not been able to maintain the performance of high growth in rice production, due to its inability to improve the yield per ha significantly. Critics typically claimed that the BIMAS program had become ineffective. It might be true that the alternative program, the FFS, was able to stop the declining growth of yield per ha, but it did not seem able to improve it significantly. The next two sections will discuss in more detail and give some explanations for the differences in performance of the two rice technology dissemination programs implemented since the mid 1960s, namely the BIMAS and FFS programs.

3 BIMAS: The Indonesian T&V program 3.1

Historical perspective

During the Japanese occupation, throughout its early independent years, and until the beginning of the 1960s, Indonesia faced serious food shortages. The first serious Indonesian effort to improve rice production was the establishment of the Padi Sentra program in 1959, which would run for 5 years. Learning from international experience, various padi centers were established to supply farmers with inputs they needed on credit. From the beginning of the program, logistics were the main problem; when it ended, there had been almost no improvement in rice production, and farmer rate of repayment was low (Roekasah and Penny 1967; Birowo 1975). The main reason for the failure of the Padi Sentra program was not because local agricultural scientists were unable to develop techniques to improve rice production. Education and research systems in the field of agriculture in Indonesia were relatively developed; many Dutch researchers were still in the country, and links existed with international communities. The most likely reason for its failure was lack of experience in engaging in a large-scale program, particularly the lack of adequately trained and experienced personnel to handle various management activities (Birowo 1975; Mears and Moeljono 1981). A glimmer of hope came from the two consecutive projects Action Research and Mass Demonstration (Demonstrasi Masal or DEMAS) — the embryos of BIMAS — conducted by the Bogor Institute of Agriculture in West Java in 1963–64 and 1964–65, respectively. Around 440 students were sent to about 220 villages, covering approximately 10 thousand hectares of rice fields, to guide (read “instruct”) farmers in implementing the Five Farming Efforts system (Panca Usaha Tani) and acquire credits for the farmers’ cooperative (Koperta). The Five Farming Efforts system

276


was concerned with (1) intensive use of high yielding varieties (HYVs) which had recently been developed, such as Arimbi, Dara and Shinta (Fox, 1993); (2) appropriate and timely use of fertilisers; (3) pest and disease management; (4) improvement in cultivation methods; and (5) improvement in irrigation and drainage systems. Farmers received loans, mostly in kind, in the form of slips or release orders which were presented to the PN Pertani kiosk for the delivery of farm supplies such as seeds, fertilisers and insecticides. After the harvest, farmers were also to return their loans in kind (Roekasah and Penny 1967; Rieffel 1969; Birowo 1975; FAO 2001). This was also the early period of T&V development worldwide (Birkhaeuser et al. 1991). Indonesian researchers were certainly in communication with international experts in developing this DEMAS program. Despite several cases where fertilisers did not arrive in time or repayments were problematic, overall the programs were considered successful. Increased rice yields among farmers involved in the program were recorded at around 50 percent more than previous harvests. In 1965 these activities were adopted as a nationwide program, called BIMAS, and were organised by the Department of Agriculture (DOA). That year around 1,200 students were sent to an area covering around 140 thousand hectares of rice fields. In the following year the BIMAS area coverage increased to around 480 thousand hectares and continued to increase after that. Gradually the students were replaced by agricultural extension staff of the Department of Agriculture. The Indonesian People’s Bank (BRI or Bank Rakyat Indonesia, then called Bank Koperasi Tani dan Nelayan) was the main source of loans, which were mostly extended through farmer cooperatives or village heads (lurah). KOLOGNAS (Komando Logistik National), formed in 1966 to maintain domestic distribution of food crops, was also asked to provide credit to BIMAS farmers (Roekasah and Penny 1967; Rieffel 1969). As the program expanded significantly, it became more apparent that the logistics of timely and appropriate provision of fertilisers and pesticides were problematic, and intensive supervision was often not available. Hence, two modifications took place in 1967. First, the loans BIMAS farmers received also included cost of living and transportation, and they were required to pay back their loans in cash. The second was the establishment of another massive supervision program called INMAS (Intensifikasi Masal or mass intensification). With INMAS, farmers still received supervision, though less than compared with BIMAS, and no credit facility was provided. Farmers were expected to find their own source of financial support. INMAS planners would merely arrange for fertilisers, insecticides and sprays to be available for cash purchase close to


277

the farm. Initially, INMAS was meant to be a follow-up program for successful BIMAS farmers who no longer needed financial support or were able to find their own sources. Later on, it turned out that being ex-BIMAS was not in fact a criterion for joining INMAS (Mears and Afiff 1968; Birowo 1975). It is also important to mention that during this period BIMAS started to introduce the IR8 and IR5 varieties of rice to farmers, the HYVs that had just been developed by the International Rice Research Institute (IRRI) and which started the Green Revolution in South and Southeast Asia (Fox 1991 and 1993). While there is no clear report on the performance of farmers who joined INMAS, there were optimistic reports relating to the performance of BIMAS farmers. Data in 1966–67 showed that the average yield increase among BIMAS farmers was around 1–2 tons of rice per hectare, and the average yield of BIMAS farmers was around 50 percent higher than that of non-BIMAS farmers (Mears and Afiff 1968). This was most likely due to the introduction of new HYVs. When Soeharto resumed the presidency in 1967/68, he and his economic team, recognising the large amount of foreign exchange needed to import rice, made BIMAS one of the top national priorities. In 1968, Soeharto and his cabinet modified BIMAS into BIMAS Gotong Royong (Cooperative BIMAS). In this program, the government contracted seven foreign companies to supply fertilisers, pesticides and some equipment at subsidised prices on a one-year deferred payment basis to BIMAS farmers (Pearson et al. 1991). These companies were paid a fixed price for every hectare they supplied with production inputs. BULOG, formed in 1967 to replace KOLOGNAS, organised the payments to these companies and repayments from farmers. The main reason for contracting foreign companies is that the government was running out of foreign exchange for the importation of necessary production inputs. In distributing these inputs to farmers, foreign contractors were assisted by agricultural extension people. By the wet season of 1970, the program covered approximately 780 thousand hectares or around 10–20 percent of the total rice field area in Indonesia (Birowo 1975). BIMAS Gotong Royong was considered a failure. Reported yields during its implementation were below expectations and the average rate of repayment was as low as around 20 percent. One main argument commonly mentioned for this lack of success was that the guidelines for farmers were too strict. Instead of suggesting that farmers adopt the BIMAS formula with some flexibility, they were instructed to follow it strictly, whereas, in reality, technical changes cannot be made mandatory for farmers. The entire system was also open to abuse: from mark-up pricing of material inputs and cheating over the quantity/quality distributed

278


to black-market sale of supplies obtained from the program (Birowo 1975; Piggott et al. 1993). Despite the failure of BIMAS Gotong Royong, abandoning the program was impossible. Although rice importing began in the early 1970s, so that Indonesia became the world’s largest rice importer — around 20 percent of the world rice trade — food shortages were still a problem. Hence a new rice intensification program was established, called Improved BIMAS (BIMAS yang Disempurnakan), in which BRI played a much more significant role. The state bank set up (1) village and mobile units to overcome problems of lending to small farmers, (2) village retailers of fertilisers and insecticides to reduce problems of late delivery, and (3) village warehouses to store rice awaiting sale, so as to use the stored rice as a warranty for further credits. The program increased the number of extension workers to replace the BIMAS students fully and distributed HYVs widely (IR8, IR5 and C4 of IRRI and Indonesian Pelita I-1 and Pelita I-2) and, later on, pest-resistant varieties (IR26 and IR30 in 1975, IR24, IR28, IR32 and IR34 in 1976) to farmers. Any private sector company, in general, was allowed to sell fertilisers and pesticides to the BIMAS market. However, the government heavily subsidised the prices of these inputs. BULOG, later on supported by BUUD (Badan Usaha Unit Desa or rural semi-cooperatives) and KUD (Koperasi Unit Desa or rural cooperatives), was asked to actively purchase and sell rice to establish floor and ceiling prices (Mears and Moeljono 1981; Manning 1987; Pearson et al. 1991; Fox 1993; Timmer 1996). It was thanks to bonanza oil revenues received during the 1970s that necessary funding was available to allow for expansion of the program. During this period, significant rehabilitation and expansion of irrigation systems was also conducted throughout Indonesia, vital to the implementation of HYVs through the BIMAS program. International agencies such as USAID and World Bank were also involved in supporting the implementation of BIMAS and rehabilitation of irrigation systems (Pearson et al. 1991; Booth 1977a and 1977b). Throughout the 1970s, the BIMAS program was considered to be successful. The area coverage of BIMAS by the mid 1970s was around 4 million hectares or around 70–80 percent of the rice growing area (Mears and Moeljono 1981). The program continued throughout the 1980s, though the achievements were not as impressive as in the 1960s and 1970s. Another program, similar to BIMAS, was also developed in 1979, namely INSUS (Intensifikasi Khusus or Special Intensification). This program was geared more toward developing extension activities, including farmer (discussion) groups and cooperatives rather than incorporating the BIMAS credit component. INSUS was modified into OPSUS (Operasi

279


Khusus or Special Effort Program) in the early 1980s and finally into SUPRA INSUS in 1987, typically to form better farmer discussion groups and cooperatives. There was some argument to the effect that the performance of INSUS farmers was better than that of BIMAS farmers, but not significantly so (Sawit and Manwan 1991; FAO 2004). 3.2

Accomplishments of BIMAS

Firstly, rice production increased substantially during BIMAS, and it contributed to ensuring secure food supplies. BIMAS farmers were also reported to have received higher income than non-intensified farmers. Hence, the intensification program was argued to have contributed to poverty reduction in rural areas (Tabor 1992). Secondly, when the financial situation of farmers was very poor, BIMAS provided a relatively easy channel for necessary capital. Thirdly, BIMAS was able to disseminate new cultivation knowledge widely, especially by making farmers adopt important new inputs, such as HYVs, fertilisers and pesticides. Fourthly, up-scaling the program to encompass rice fields nationwide was not too difficult and could be achieved in a relatively short period.

kg/ha

%

250

100

200

80

150

60

100

40

50

20

0 1969

1972

1975

1978

1981

1984

% of area under intensification

1987

1990

Fertilizer use kg/ha

Figure 4: Rice intensification area and fertiliser use Source: Hill (2000).

280


Ten year after the BIMAS program was begun, around 45 percent of rice areas in the country were under some sort of intensification program; after 20 years, around 75 percent; and after 25 years, more than 80 percent. The various intensification programs employed were BIMAS, INMAS, INSUS or SUPRA INSUS (Figure 4), which resulted in a significant, relatively steadily increasing yield of rice production in the 1960s, 1970s and 1980s (Figure 2). In 1979, Indonesia was still the largest rice-importing country in the world, with annual imports of 2.9 million metric tons, costing about USD 600 million. However, by 1983, Indonesia became a rice selfsufficiency country for the first time in its history (Sawit and Manwan 1991; Pearson et al. 1991; Tabor 1992; Piggott et al. 1993; Hill 2000). 3.3

Issues with BIMAS

The first concern regarding BIMAS was its high cost, as argued by Mears and Moeljono (1981), Barbier (1989), Tabor (1992) and Panayotou (1993). There are several reasons to support this argument. First of all, while the market nominal monthly interest rates in the 1970s were around 5 percent, those of BIMAS were around 1 percent.5 Second, as has been mentioned before, BIMAS operations were prone to abuse, such as selling BIMAS supplies on the black market, corrupting the quantity/quality of BIMAS supplies and stealing farmer repayments. Third, in several cases, rates of credit repayment were low. For example, in the 1973/74 budget year alone, the amount of unreturned credits was about USD 4 billion (DOF, 1977). Fourth, BIMAS encouraged the increased use of pesticides and fertilisers up to an unnecessary level. From the outset, the government subsidised these inputs so as to maintain the farmer profit margin and to keep the price of rice relatively low. In the mid 1980s, the rates of subsidy for fertilisers and pesticides reached around 50 and 80 percent of their market prices, respectively. Hence, the total fertiliser and pesticide subsidy in 1986/87 was around USD 725 billion, that is, around 66 percent of the total agricultural sector development budget for that fiscal year (Barbier 1989; Tabor 1992). The second concern regarding BIMAS, still related to the first issue, was collusion between agricultural chemical companies and high-ranking officers in the Department of Agriculture. BIMAS encouraged the use of fertilisers—typically non-organic fertilisers—and pesticides which benefited suppliers of these chemical products, in this case chemical companies. In a way, BIMAS provided a guaranteed amount of sale each year for these companies (Tabor 1992). To protect this lucrative income and to smooth the processes for obtaining necessary permits, most of these 5

In real term, interest rates of BIMAS’ credits were probably negative.


281

chemical companies put high-ranking officers from the Department of Agriculture and retired generals on their payrolls. Furthermore, through these high-ranking officers, chemical companies influenced the BIMAS program (including INSUS) to instruct farmers to use increasing amounts of fertilisers and pesticides. The third concern regarding BIMAS was pest resistance as a result of this overuse of pesticides. The first major pest outbreak was when brown planthoppers damaged more than 450 thousand hectares of rice fields in 1976/1977. The estimated yield loss was 364,500 tons of milled rice, which could have fed three million people for an entire year. In 1980, the outbreak of green leafhoppers damaged at least 12 thousand hectares of rice fields in Bali alone. The reaction to these pest outbreaks, instead of reducing the use of pesticides, was to instruct farmers to use even more. Hence, in 1986 there was another brown planthopper outbreak, destroying approximately 200 thousand hectares of rice (Useem et al. 1992; Oka 1997). The fourth concern regarding BIMAS was the human health impact caused by the use of pesticides. In 1988, it was recorded that there were around 1,300 cases of acute pesticide poisoning in 182 general hospitals throughout the islands of Java and Bali. An observation on the health of farmers in that same year indicated that approximately 20 to 50 percent of the farmers who utilised pesticides contracted chronic pesticide-related illnesses. These illnesses included headaches, weakness, insomnia, and difficulty with concentrating (Achmadi 1991). The above-mentioned concerns regarding the BIMAS approach became apparent by the mid 1980s. In 1984, consistent with the on-going economic reforms due to declining oil revenues, the BIMAS credit package was abolished and the KUPEDES market-oriented program was introduced. In 1989, the pesticide subsidy was eliminated, though the fertiliser subsidy continued. By the early 1990s, the glory of BIMAS was over. The Department of Agriculture (DOA) decided it would no longer implement BIMAS and its derivatives actively.

4 Farmer Field School: The IPM program 4.1

Historical perspective

In response to alarm concerning the negative side effects of the overuse of pesticides, by the end of the 1970s Indonesian scientists had learned from their research and various worldwide reports of many more problems relating to the use of pesticides in agriculture (Oka 1978, 1979; Soekarna 1979; Pimentel et al. 1992; Antle and Pingali 1994). Based on these findings and information from international agricultural communities,

282


Indonesian scientists concluded that Indonesia had to stop relying solely on pesticides and needed to employ several control tactics, including synchronised planting, crop rotation, natural predators, as well as pesticides, but only as a last resort, that is, to adopt the strategy commonly known as integrated pest management (IPM). But the tactic need not be applied uniformly, as farmers themselves, either individually or sometimes collectively, needed to make their own decisions as to the best strategy. It was also concluded that conducting a Farmer Field School (FFS) in which farmers could ‘learn by doing’ was the appropriate method of learning IPM skills. Resistance to moving from a BIMAS to an FFS approach was very strong in the DOA. First, many officials in the DOA still believed chemical pesticides to be the easiest, most reliable and effective method of pest control. Second, several high officials in the DOA were closely associated with pesticide companies that still wanted to promote the intensive use of their products (Oka 1997). Third, the supporters of intensive use of pesticides were politically strong: particularly chemical companies that received the most benefit from their intensive use and subsidisation. Several retired generals with strong political influence had vested interests in these companies. The second national brown planthopper outbreak in 1986 aroused the concern of the Indonesian National Planning Agency (BAPPENAS), at that time the most powerful government agency, which quickly sought advice from scientists in the DOA, leading universities and international organisations, who recommended implementation of the IPM program at a grassroots level. This led to the establishment of the “IPM by Farmers” program (Oka 1997). BAPPENAS consulted intensively with the president Soeharto concerning the need to implement the IPM program, and this resulted in the launching of Presidential Decree (Keppres) No. 3/1986, supporting the implementation of the IPM and providing national political support to establish the IPM program as a national policy requiring the support of all government agencies, including the military. It was also a signal from the president to all retired generals to retract their political support of pesticide companies. The main activity of the IPM by Farmers program was conducting field schools to train farmers in IPM practice. To achieve this goal nationwide, three steps were taken: training for trainers, training for farmers by these trainers, and training for farmers by farmers. The last two types of training were undertaken at the FFS, whose graduated farmers were expected to change their beliefs and practices from exclusive use of pesticides more towards management of the ecosystem, growing healthy crops, and


283

preserving beneficial natural enemies, as well as being capable of making their own decisions as to the best way to grow their plants and to control pests in their fields, rather than rote compliance with instructions to spray pesticides regularly. Farmers were also expected to develop a habit of regularly conducting observations in their fields and skills to identify pests and their natural predators (Dilts 1985; Kenmore 1992; Useem et al. 1992). In the FFS, participants were asked to observe and discover, by themselves, pests and their natural enemies. Participants discussed their findings with one another. Then they were encouraged to derive practical conclusions and implement them. In this training there was no clear-cut distinction between trainers and trainees. Trainers only acted as facilitators. Most of these activities were conducted in the field, where half of the field was planted using techniques that farmers had normally practiced and the other half was planted following the IPM practices being learned. Realising that it was still very hard to expect the DOA actively to implement IPM training to extension workers and farmers, BAPPENAS undertook this role from mid 1989 to mid 1994. This was unusual, since BAPPENAS is supposed to be concerned only with planning, not with implementation. By the end of 1991, 2,000 extension workers and 1,000 field pest observers had trained approximately 100,000 farmers. By 1992, approximately 200,000 farmers, most of them rice farmers, were trained in IPM practice. Approximately ten percent of these 200,000 farmers were chosen to receive further training to become trainers. Funding for the first two years of this activity, 1989–1991, was mainly from USAID, which extended funding until mid 1992. From mid 1992 till mid 1994, the program also received some support through a World Bank loan for other existing agricultural training projects not particularly designated for IPM training (SEARCA 1999). Reports on the success of this IPM by Farmers program were available, and typically they mentioned that the program had been able to empower farmers in making their own decisions and that there was evidence of yield increases among farmer graduates of the program. In mid 1994, the program was transferred to the DOA, reflecting a declining interest in this program by BAPPENAS and a growing interest among officers in the DOA, as well as a desire to make the program truly a national one. For the next five years, until mid 1999, the program was mostly (approximately 75 percent) funded through a loan from the World Bank that was specifically targeted to support the IPM Program. The Indonesian government provided the other 25 percent, as the matching fund for the World Bank loan. The total cost of the second stage of the program was approximately Rp. 112 billion (around USD 35 thousand), reflecting the intention to up-scale the previous program (DOA, 1999). By 1999,

284


around 1 million farmers had been trained in the FFS. The number was still relatively small, however, since there were around 20 million rice farmers in the country. Despite a growing number of farmers attending the program and its larger coverage, a negative view of the second stage (1994–1999) of the IPM emerged. It never received the strong political support accorded to the first stage of the program (1989–1994) and had to face problems, such as funding delays and other bureaucratic obstacles, that would have been overcome had top leadership been strongly supportive of the policy. It was suspected that training quality declined during the 1994–1999 period of the program. Hence, there were some doubts that up-scaling and sustaining the efforts of the IPM program would ever be successful (Pincus 2002). In 1997, economic crisis hit Indonesia, resulting in a huge drop in the country’s GDP in 1998. During this period, the number one priority of the government, including foreign donors, was to restructure the financial sector to prevent it from bringing down the national economy even further and to soften the impact of this crisis on poor people. Suddenly, the IPM Program was no longer a national priority, losing all of its by then only moderate political support, and was terminated at the end of 1999. 4.2

Accomplishments of FFS

First, various case studies find evidence of an increase in rice yield, at least in the short-run (SEARCA 1999; Van den Berg 2004; and Van den Berg and Jiggins 2007). Second, the FFS was able to disseminate new and better cultivation knowledge to farmers, in particular changing their attitudes toward insects, pesticides and pest control. The FFS was able to encourage farmers to reduce pesticide use and convince them to use it properly when needed (Oka 1991; Pincus 1991; Useem et al. 1992; Darmawan et al. 1993; and Deybe et al. 1998). Also, the FFS increased farmers’ confidence in making their own decisions as to how to best cultivate their plants without instructions from agricultural extension workers. Furthermore, farmers understood the need to activate farmer groups, since collective actions in controlling pests are much more effective than individual ones. As a result, there was an increase in the quantity and quality of discussions in farmer groups concerning pest control and growing healthy crops (Oka and Dilts 1993; and Pontius et al. 2002). Before enrolling in the FFS, almost all farmers thought of most insects as pests that therefore should be killed. After the FFS, farmers realised that there are harmless insects and, most importantly, there are natural predators for most pests in their fields. Furthermore, farmers now understood that there is an economic threshold of pest population, below which the pests will not have any significant impact on the amount to be harvested.


285

Regarding pesticide use, farmers also understood that inappropriate and excessive use of pesticides is dangerous and harmful, because they not only kill pests, but also their natural enemies and other living creatures in the fields; overuse of pesticides leads to pest resistance to pesticides and then increased risk of pest outbreaks; and pesticides are poisons that are also very harmful to humans. Hence, to control pests below their economic threshold, farmers preferred to implement synchronised planting and crop protection, field sanitation, and the use of resistant plants as their first measure. Further action involved conducting physical control measures and preserving natural enemies, before, if circumstances necessitated, as a last resort using the least dangerous pesticides. Lastly, appropriate control of pesticides also meant there were fewer opportunities for officials and chemical companies to abuse the program. 4.3

Issues with FFS

The major concern regarding the FFS program is related to up-scaling and maintaining the quality of the program. The rate of expansion of the program had been slow. In the 10 years of implementation, only 5 percent of rice farmers had the opportunity to join the FFS and learn the IPM method. It is also suspected that the training quality declined during the 1994–1999 period of the program. Availability of funding seems not to have been the main reason, however, as it has been reported that the DOA was not able to spend all funding available for the FFS program during that period (DOA 1999). The second concern relates to keeping farmers consistent in implementing the IPM method. There is evidence that FFS graduate farmers returned after a while to the old method of routinely spraying pesticides and conducted field observations less often than they should have. There are reasons for this. First, routine pesticide spraying seems much easier than conducting observations and making a decision to develop a strategy to control pests without using pesticides. Second, pesticide companies kept finding ways to influence farmers to use more pesticides. One of their strategies was to develop a program named “IPM Plus”, which involved routinely spraying pesticides. Third, many field extension workers had not mastered the IPM method and, when a ‘crisis’ came about, they quickly resorted farmers to spraying pesticides. Lastly, due to changes in labour organisation over time, the opportunity cost of learning and implementing the IPM strategy could have become too high, inducing farmers to return to calendar-based spraying (Beckmann and Wesseler 2003). The third concern is related to an expectation that FFS graduate farmers would share their knowledge of the IPM with their neighbours. Given the

286


complexity of the information and farmers’ limited ability to convey complex decision-making skills effectively to other farmers through informal communication, the diffusion process was possibly limited and curtailed (Feder et al. 2004b). The fourth concern is that the cost of implementing an FFS program is high (Quizon et al. 2001; Anderson and Feder 2004; Eicher 2007; Van den Berg and Jiggins 2007). The estimated start-up and recurrent costs per graduate amounted to between USD 21 and USD 62. The average total cost per school was around USD 532 for a trainer-to-farmer school and around USD 586 for a farmer-to-farmer school, with the latter being slightly more expensive since two experienced farmers were needed to lead the school, whereas the trainer-to-farmer school only needed one official trainer (Braun et al. 2000). It is important to note that there are about 20 million rice farmers in Indonesia. This high cost slowed down the program’s expansion, and made it hard to maintain program quality. The fifth concern involves farmers’ willingness to enroll in a FFS, which required farmers attending the school to spend some time in joint discussions. Poor farmers tend to use their off-farm time for other jobs. Hence, while there were strong incentives for many farmers to enroll in and graduate from the program (Oka 1997), attending an FFS was not attractive to some farmers; if they enrolled, they could not finish the program (SEARCA 1999). Finally, there is no evidence yet whether FFS graduate farmers have experienced higher yields than those who did not attend the FFS, especially in the medium- to long-run. Various case studies in Sumatra, Java, Bali and Lombok reported that IPM farmers were able to increase yields by approximately 10 percent and to reduce the use of pesticides by approximately 50 percent, resulting in a reduction of costs by approximately 11 percent (MET 1993; Oka 1997; Kuswara 1998a and 1998b; Paiman 1998a and 1998b, Susianto et al. 1998; SEARCA 1999; and Van den Berg 2004). However, a study by Feder et al. (2004a), using a panel data system, argued that there is no evidence that the FFS induced an increase in yields and a reduction in the use of pesticides. It is important to note that most case studies observed farmers who had relatively recently graduated from the FFS, typically observing a small group of FFS graduates and comparing their performance with a small adjacent group of non-FFS farmers. Conversely, Feder et al. (2004a) observed a larger sample of farmers (around 320 observations) and their performance throughout a medium period of time by comparing their performances in 1991 and in 1999. Hence, one possibility is that the case studies actually report the short-run impact of the FFS, while the panel data studies observe its medium-run impact. It appears that immediately after


287

graduating from the FFS or being exposed to its influence, farmers did significantly improve their performance, but they were not able to maintain this standard over time, with their performance declining in the mediumterm (Yamazaki and Resosudarmo 2008).

5 Discussion Let us compare the performance of BIMAS and FFS programs based on the seven indicators mentioned above: 1) impact on yield, 2) impact on chemical use, 3) risk of crop failure, 4) ability to sustain the technology disseminated, 5) cost, 6) up-scaling issues, and 7) corruption issues. In terms of the impact on yield, many seem to agree that the BIMAS program implemented from the mid 1960s till around 1983 was able to significantly improve the yield per ha of BIMAS farmers, so that the average annual growth of yield per ha was more than 4 percent during these years (Mears and Moeljono 1981). This annual growth rate was high by world standards. After 1984, many started to doubt that the BIMAS program was still effective in improving farmers’ yields. Where the FFS program is concerned, the debate over whether or not the program actually improved farmers’ yields per ha is still ongoing. In the immediate period after farmers graduated from the school, it seems there was evidence that their yields improved. However, there was less or even no evidence of this in the medium- and long-term after farmers had graduated from the school. In terms of chemical use, there was a tendency for BIMAS farmers to overuse both fertilisers and pesticides, leading to instances of acute and chronic human poisoning. The FFS program educated farmers to minimise the use of chemical inputs, particularly pesticides, while maintaining/improving yields. Nevertheless, whether or not farmers who graduated from the FFS were actually able to reduce their use of chemical inputs significantly is still debatable. The overuse of pesticides under the BIMAS regime was also seen to cause pest resistance and so induce pest outbreaks in various BIMAS areas in the mid 1970s and 1980s. Hence, the BIMAS program was blamed for imposing a higher risk of crop failure. The FFS program, on the other hand, introduced an integrated pest management (IPM) technique aiming, among other things, to avoid the problems of pest resistance. Experience in the early 1990s shows that this technique effectively overcame this issue and thus reduced the risk of crop failure due to pest outbreaks (Oka 1997). In the end, both the BIMAS and FFS programs, for different reasons, seemed to fail to sustain the technology they introduced. Implementation of HYVs with regular use of chemical inputs introduced by the BIMAS program was not sustained, since the overuse of chemicals induced a higher risk of crop failure and human health problems. Neither was the IPM

288


technique introduced through the FFS program carried out fully, since there is evidence that FFS graduate farmers returned after a while to the old method of routinely spraying pesticides and conducting field observations less often than they should have. Both programs are costly, at least by a developing country’s standard. Typically, implementation of these programs is only feasible when there is a windfall revenue or significant external donors. As has been mentioned before, the cost of the chemical input subsidy under the BIMAS program reached about USD 725 billion in the 1986/87 budget year. BIMAS was heavily financed by Indonesia’s bonanza oil revenues in the 1970s, which turned out to be unsustainable, and some support from external agencies such as USAID and World Bank. As the revenues of oil declined in 1980s, so did the BIMAS activities. The FFS program cost per farmer was also high. The Indonesian FFS program was supported mainly by USAID during the first phase (1989– 1991) and by the World Bank in the second phase (1992–1999) of its implementation. When the economic crisis hit Indonesia in 1997/98, priorities of external agencies changed mainly to the issues of banking and industrial restructuring. Without available external funding, the FFS program had to be terminated. The Indonesian experience showed that, when funding was available, it was rapidly possible to scale-up the BIMAS program to reach about 80 percent of targeted areas or farmers. This was not the case for the FFS program. Extension workers’ limited knowledge and ability to transfer relatively complex knowledge and decision-making processes to farmers, as well as farmers’ limited ability to transfer this complex knowledge through informal communication to other farmers were the main challenges to scaling-up the FFS program, which Indonesia failed to do, and the program never reached the majority of targeted farmers. Lastly, and importantly, BIMAS was prone to abuse, such as where officials marked-up prices of inputs, reduced the quantity or quality of supplies distributed to farmers, or sold supplies on the black market. There were fewer opportunities for abuse in the FFS program. BIMAS also created an opportunity for collusion between chemical companies and officials. Chemical companies were willing to provide kickbacks to officials as long as more of their chemicals were used in the program. This kind of opportunity did not exist in the FFS program.

6 Conclusion This paper has reviewed Indonesia’s experience in conducting the Training and Visit (T&V) Extension Program—in Indonesia this is called the Massive Guidance (Bimbingan Masal or BIMAS) Program—and the


289

Farmer Field School (FFS) Program to accelerate dissemination of new knowledge to and among farmers. Although these two national programs were different, there are key similarities underlying their degrees of success. First, there was strong local research, supported by strong links with international research communities, before the implementation of the programs. This contributed significantly to their design and made the initial implementation of the programs possible. Several modifications of the programs reflected this strong local knowledge. Second, there was strong national political will to implement the programs. Both Soekarno, the president before Soeharto, and Soeharto himself placed rice intensification as a top priority. Explicit support from these leaders signaled for every agency in the country to support the programs, which made their smooth implementation possible. Without this strong political will, the budget would most likely have been more restricted, and the programs could hardly have succeeded. Third, there were administrative breakthroughs to start the programs. In the case of BIMAS, the program was initiated by the two projects organised by the Bogor Institute of Agriculture. At that time, the Indonesian economy was in a very bad state and the political situation was not stable, so government departments were not functioning properly. Conditions at the university were much better, and urging students to contribute to the program was easier, as they were mostly idealistic young people who wanted to so some good for their country and they were also eager to obtain field experience. In the case of the FFS, the administrative breakthrough was the BAPPENAS decision to start the program, as there was much resistance to it in the Department of Agriculture, including the from Minister of Agriculture, at that time. Fourth, farmers could quickly observe the positive impact of the program in terms of a significant increase in yields in the case of BIMAS and an improvement of their cultivation knowledge, a feeling of confidence in making decisions and better health in the case of the FFS. In the previous section, we have discussed the benefits and costs of these two programs in terms of their impacts on yield, chemical use and risk of crop failure, their ability to sustain the technology disseminated, their costs, as well as issues related to scaling-up the programs and corruption. Based on this discussion, the main overall conclusion is the following. For regions where the level of development is still very low, such as in off-Java islands and some of the Sub-Saharan regions, implementing a T&V program instructing farmers what to do is probably more appropriate than an FFS. Within a relatively short period of time employing the former, large numbers of farmers will do as they are told. But because of this it is

290


also important to protect the program from officials who might abuse it and from collusive behaviours between chemical companies and officials. After the program is significantly developed, it is necessary to switch to a program such as the FFS, allowing farmers to make their own decisions as to what is best to do in their fields. As for regions where agriculture is relatively developed, such as in Java and other regions in Asia, there is no need to go back to implementing programs such as BIMAS. A more effective FFS program followed by a new program to maintain farmers’ knowledge is more appropriate. Whatever the choice, strong local research with links to international communities, a national political will and administrative breakthroughs are most likely needed.

Acknowledgements The authors would like to thank participants in the International Seminar on Economic Transition and Sustainable Agricultural Development in East Asia, Ho Chi Minh, Vietnam, 21–22 June 2007, and the seven unknown referees for their valuable comments and suggestions. All errors, however, are the authors’ responsibility.

References Anderson, J.R., Feder, G. (2004). Agricultural Extension: Good Intensions and Hard Realities. The World Bank Research Observer 19 (1): 41-60. Anderson, J.R., Feder, G., Ganguly, S. (2006). The Rise and Fall of Training and Visit Extension: An Asia Mini-Drama with an African Epilogue? In Van den Ban, A.W., Samanta, R.K. (eds.). Changinf Roles of Agricultural Extension in Asian Nations, New Delhi: B.R. Publishing Company, 149-172. Antle, J.M., Pingali, P.L. (1994). Pesticides, Productivity, and Farmer Health: A Philippine Case Study. American Journal of Agricultural Economics 76: 418430. Ardiwinata, A.N., Umar, N., Handayani, N. (1997). Residu Insektisida Organoklorin, Organofosfat dan Karbamat dalam Beras dan Kedelai di Beberapa Pasar di DKI Jakarta. Prosiding Seminar National PHT, Jakarta: Program National PHT, 346-354. Barbier, E.B. (1989). Cash Crops, Foods Crops, and Sustainability: The Case of Indonesia. World Development 7(6): 879-895. Beckmann, V., Wesseler, J. (2003). How Labour Organization May Affect Technology Adoption: An Analytical Framework Analysing the Case of Integrated Pest Management. Environmental and Development Economics 8 (3): 437-450. Birkhaeuser, D., Evenson, R.E., Feder, G. (1991). The Economic Impact of Agricultural Extension: A Review. Economic Development and Cultural Change 39: 607650.


291

Birowo, A.T. (1975). BIMAS: A Package of Program for Intensification of Food Crop Production in Indonesia. SEADAG Papers, The Asia Society, New York. Braun, A.R., Thiele, G., Fernández, M. (2000). Farmer Field School and Local Agricultural Research Committees: Complementary Platforms for Integrated Decision-Making in Sustainable Agriculture. Agricultural Research and Extension Network Paper No 105, Overseas Development Institute, London. Booth, A. (1977a). Irrigation in Indonesia: Part I. Bulletin of Indonesian Economic Studies 13(1): 33-74. Booth, A. (1977b). Irrigation in Indonesia: Part II. Bulletin of Indonesian Economic Studies 13(2): 45-77. Central Statistical Agency (Badan Pusat Statistik or BPS). various years. Statistical Yearbook of Indonesia. Jakarta: BPS Publication. Darmawan, D.A., Yusdja, Y., Kusmayadi, A., Saleh, C., Purwoto, A., Siagian, V. (1993). Kajian Aspek Sosial Ekonomi Pengendalian Hama Terpadu. Bogor: Pusat Penelitian Sosial Ekonomi Pertanian. Department of Agriculture (DOA) (1999). Laporan Akhir Proyek Pengendalian Hama Terpadu Pusat. Ministry of Agriculture, Jakarta, Indonesia. Department of Finance (DOF) (1977). Nota Keuangan dan Anggaran Pendapatan dan Belanja Negara Tahun 1975/1976. Ministry of Finance, Jakarta, Indonesia. Deybe, D., Vaissayre, M., Gouyon, A., Chaume, J., Goze, E. (1998). Study on the Indicators to Evaluate Integrated Pest Management Training Program in Indonesia. CIRAD, Bogor, Indonesia. Dilts, R. (1985). Training: Reschooling Society. Prisma 38: 79-90. Eicher, C.K. (2007). Agricultural Extension in Africa and Asia. Literature review prepared for the World AgInfo Project, Cornell University, Itchaca, New York. Feder, G., Murgai, R., Quizon, J.B. (2004a). Sending Farmers Back to School: The Impact of Farmer Field Schools in Indonesia. Review of Agricultural Economics 26(1): 45-62. Feder, G., Murgai, R., Quizon, J.B. (2004b). The Acquisition and Diffusion of Knowledge: The Case of Pest Management Training in Farmer Field Schools, Indonesia. Journal of Agricultural Economics 55(2), 221-243. Feder, G., Slade, R. (1986). The Impact of Agricultural Extension: The Training and Visit System in India. Research Observer 1(2): 139-161. Food and Agriculture Organization (FAO) (2001). Rice Self-sufficiency: Indonesia. In FAO (ed). Sharing Innovative Experience Vol 5: Example of Successful Initiatives in Agriculture and Rural Development in the South. Rome: FAO. [available at http://tcdc.undp.org/sie/experiences/vol5/content5new.asp] Food and Agriculture Organization (FAO) (2004). The State of Food and Agriculture 2003-2004. Rome: FAO. Fox, J. (1991). Managing the Ecology of Rice Production in Indonesia. In Hardjono, J. (ed.). Indonesia: Resources, Ecology, and Environment. Singapore: Oxford University Press, 61-84. Fox, J. (1993). The Rice Baskets of East Java: The Ecology and Social Context of Sawah Production. In Dick, H., Fox, J., Mackie, J. (eds), Balanced Development: East java in the New Order, Singapore: Oxford University Press, 120-157.

292


Hill, H. (2000), The Indonesian Economy. Cambridge: Cambridge University Press. Hoisington, D., Khairallah, M., Reeves, T., Ribaut, J-M., Skovmand, B., Taba, S., Warburton, M. (1999). Plant Genetic Resources: What can They Contribute Toward Increased Crop roductivity? Proceedings of the National Academy of Sciences 96: 5937-5943. Karshenas, M. (2001). Agriculture and Economic Development in Sub-Saharan Africa and Asia. Cambridge Journal of Economics 25:315-342. Kenmore, P.E. (1992). Indonesia’s IPM – A Model for Asia. FAO Intercountry Programme for the Development of Integrated Pest Control in Rice in South and Southeast Asia, Manila, the Philippines. Khush, G. (1999). Green Revolution: Preparing for the 21st Century. Genome 42: 646655. Kuswara, E. (1998a). IPM in Marga Sub-District, Tabanan District, Bali. Community IPM: Six Cases from Indonesia. Semi-Annual FAO IPM Technical Assistance Progress Report, Jakarta, Indonesia, 100-137. Kuswara, E. (1998b). IPM in Tulang Bawang Udik Sub-District, Tulang Bawang District, Lampung. Community IPM: Six Cases from Indonesia. Semi-Annual FAO IPM Technical Assistance Progress Report, Jakarta, Indonesia, 221-248. Manning, C. (1987). Public Policy, Rice Production and Income Distribution: A Review of Indonesia’s Rice Self-Sufficiency Program. Southeast Asian Journal of Social Science 15(1): 66-82. Mears, L.A., Afiff, S. (1968). A New Look at the BIMAS Program and Rice Production. Bulletin of Indonesia Economic Studies 4(10): 29-47. Mears, L.A., Moeljono, S. 1981. Food Policy. In Booth, A., McCawley, P. (eds.). The Indonesian Economy during the Soeharto Era, pp. 23-61. Oxford: Oxford University Press. Monitoring and Evaluation Team (MET) (1993). The Impact of IPM Training on Farmers’ Behaviour: A Summary of Results from the Second Field School Cycle, IPM National Program, Indonesia. Mosher, A.T. (1978). An Introduction to Agricultural Extension. NY: Agricultural Development Council. Oka, I.N. (1978). Quick Method for Identifying Brown Planthopper Biotypes in the Field. Interntional Rice Research Newsletter, 3(6): 11-12. Oka, I.N. (1979). Cultural Control of the Brown Planthopper. In: International Rice Research Institute (IRRI). Brown Planthopper: Threat to Rice Production in Asia. Los Banos: IRRI, 359-369. Oka, I.N. (1991). Success and Challenges of the Indonesian National Integrated Pest Management Program in the Rice-based Cropping System. Crop Protection 10 (3): 163-65. Oka, I.N., Dilts, R. (1993). Program Nasional PHT dalam Usaha Pengembangan Sumber Daya Manusia, Mempertahankan Kelestarian Lingkungan dan Efisiensi Produksi. Paper presented in Lokakarya Pembangunan Berkelanjutan di Tingkat Lokal dalam Menanggulangi Kemiskinan, Bogor Institute of Agriculture, Bogor, Indonesia.


293

Oka, I.N. (1997). Integrated Crop Pest Management With Farmer Participation in Indonesia. In: A. Krishna, N. Uphoff, M.J. Esman (eds), Reasons for Hope. Connecticut: Kumarian Press. Otsuka, K. (2006). Why Can’t We Transfrom Traditional Agriculture in Sub-Saharan Africa? Review of Agricultural Economics 28 (3): 332-337. Otsuka, K., Kalirajan, K. (2006). Rice Green Revolution in Asia and Its Transferability to Africa: An Introduction. Developing Economies 44 (2), 107-122. Paiman (1998a). An IPM Case Study in Gerung Sub-District, West Lombok District, West Nusa Tenggara. In: Community IPM: Six Cases from Indonesia, SemiAnnual FAO IPM Technical Assistance Progress Report, Jakarta, Indonesia, 138-166. Paiman (1998b). An IPM Case Study in Ngantang Sub-District, Malang District, East Java. In: Community IPM: Six Cases from Indonesia, Semi-Annual FAO IPM Technical Assistance Progress Report, Jakarta, Indonesia, 168-189. Panayotou, T. (1993). Green Markets: The Economics of Sustainable Development. San Fransisco: Institute for Contemporary Studies Press: Pearson, S., Naylor, R., Falcon, W. (1991). Recent Policy Influences on Rice Production. In: S. Pearson, W. Falcon, P. Heytens, E. Monke, R. Naylor. Rice Policy in Indonesia. Ithaca: Cornell University Press, 8-21. Piggott, R.R., Parton, K.A., Treadgold, E.M., Hutabarat, B. (1993). Food Price Policy in Indonesia. ACIAR Monograph No. 22, ACIAR, Canberra. Pimentel, D., Acquay, H., Biltonen, M., Rice, P., Silva, M., Nelson, J., Lipner, V., Giordano, S., Horowitz, A., D’Amore, M. (1992). Environmental and Economic Costs of Pesticide Use. BioScience 42: 750-60. Pincus, J. (1991). Farmer Field School Survey: Impact of IPM Training on Farmers’ Pest Control Behavior. Integrated Pest Management National Program, BAPPENAS, Jakarta, Indonesia. Pincus, J.R. (2002), “State Simplification and Institution Building in World BankFinanced Development Project.” In J.R. Pincus and J.A. Winters (eds.), Reinventing the World Bank. Ithaca: Cornell University Press, pp. 76-100. Pinstrup-Andersen, P., Pandya-Lorch, R., Rosegrant, M.W. (1999). World Food Prospects: Critical Issues for the Early Twenty-First Century. Food Policy Report, International Food Policy Research Institute, Washington, D.C. Pontius, J.C., Dilts, R., Bartlett, A. (2002). Ten Years of IPM Training in Asia – From Farmer Field School to Community IPM. FAO, Bangkok. Quizon, J., Feder, G., Murgai, R. (2001) Fiscal Sustainability of Agricultural Extension: The Case of the Farmer Field School Approach. Journal of International Agricultural and Extension Education 8 (1): 13-24 Resosudarmo, B.P., Kuncoro, A. (2006). The Political Economy of Indonesian Economic Reform: 1983-2000. Oxford Development Studies 34 (3): 341-355. Rieffel, A. (1969). The BIMAS Program for Self-Sufficiency in Rice Production. Indonesia 8: 103-133. Roekasah, E.A., Penny, D.H. (1967). BIMAS: A New Approach to Agricultural Extension in Indonesia. Bulletin of Indonesia Economic Studies 3 (7): 60-69.

294


Röling, N., Van den Fliert, E. (1994). Transfroming Extension for Sustainable Agriculture: The Case of Integrated Pest Management in Rice in Indonesia. Agriculture and Human Values 11 (2): 96-108. Sachs, J. (2005). The End of Poverty: How We can Make It Happen in Our Lifetime. London: Penguin Books. Sawit, M. H., Manwan. I. (1991). The Beginning of the New SUPRA INSUS Rice Intensification Program: The Case of the North Coast of West Java and South Sulawesi. Bulletin of Indonesia Economic Studies 27 (1): 81-103. SEARCA (1999). Integrated Pest Management Program Training Project: Impact Evaluation Study. Internal Report for Indonesian Ministry of Agriculture and the World Bank, Jakarta, Indonesia. Simatupang, P., Timmer, C.P. (2008). Indonesian Rice Production: Policies and Realities. Bulletin of Indonesian Economic Studies 44(1), 65-80. Soekarna, D. (1979). Pengaruh pestisida bentuk EC dan WP terhadap beberapa predator wereng coklat, Nilaparvata lugens. Paper presented in Kongres Entomologi I, January 9-11, Jakarta, Indonesia. Susianto, A., Purwadi, D., Pincus, J. (1998). Kaligondang Sub-District: A Case History of and IPM Sub-District. In: Community IPM: Six Cases from Indonesia. Semi-Annual FAO IPM Technical Assistance Progress Report, Jakarta, Indonesia, 15-99. Tabor, S.R. (1992). Agriculture in Tansition. In: Booth, A. (ed), The Oil Boom and After: Indonesian Economic Policy and Performance in the Soeharto Era. Oxford: Oxford University Press, 161-203. Thirtle, C., Lin, L. (2003). The Impact of Research-led Agricultural Productivity Growth on Poverty Reduction in Africa, Asia and Latin America. World Development 31 (12): 1959-75. Timmer, C.P. (1996). Does Bulog Stabilise Rice Prices in Indonesia? Should It Try? Bulletin of Indonesian Economic Studies 32(2):45-74. Van den Berg, H. (2004). IPM Farmer Field Schools: A Synthesis of 25 Impact Evaluations. Report prepared for the Global Facility, Wageningen University, the Netherlands. Van den Berg, H., Jiggins, J. (2007). Investing in Farmers: The Impacts of Farmer Field School in Relation to Integrated Pest Management. World Development 35 (4): 663-686. Van den Fliert, E., Pontius, J., Röling, N. (1996). Searching for Strategies to Replicate a Successful Extension Approach: Training of IPM Trainers in Indonesia. European Journal of Agricultural Education and Extension 1(4): 41-63. Useem, M., Setti, L., Pincus, J. (1992). The Science of Javanese Management: Organizational Alignment in an Indonesian Development Programme. Public Administration and Development 12: 447-471. World Bank (2002). The World Bank’s Health, Nutrition and Population data platform”, World Bank, Washington D.C. [http://devdata.worldbank.org/ hnpstats/]. World Bank (2007). World Development Report 2008: Agriculture for Development, World Bank, Washington D.C.


295

Yamazaki, S., Resosudarmo, B.P. (2008). Does Sending Farmers Back to School Have an Impact? Revisiting the Issue. Developing Economies 46 (2): 135-50.

Chapter 15 Determining Factors of IPM Adoption: Empirical Evidence from Longan Growers in Northern Thailand Chapika Sangkapitux,1 Pornsiri Suebpongsang,1 Sakdamneon Nonkiti,1 and Andreas Neef 2 Abstract. This paper aims at identifying the determining factors of IPM adoption, drawing on a cross-sectional, primary data set obtained from a farm household survey of 154 longan growers under the Good Agricultural Practice (GAP) program in Lamphun, a northern province of Thailand. The most important reasons given by longan growers for joining the IPM training programs were ‘suggestion from an extension worker’ and ‘following a neighbor’s example’. Astonishingly, the number of training courses attended did not have a significant influence on IPM adoption. Our results show that a higher level of knowledge of negative health effects from pesticide application significantly increases the degree of IPM adoption of these longan growers, but a significantly lower adoption level is found among farm households with a high ratio of off-farm income. As regards labor organization, hired labor plays a crucial role in longan production; yet, when it comes to pest management activities, family labor is much more important than hired labor. Evidence from the survey further suggests that family-operated farms have a stronger tendency to adopt IPM strategies than those farms that rely to a great extent on hired labor. Keywords. Integrated Pest Management, Labor organization, Technology adoption, Good Agricultural Practice, Longan, Thailand

1 Introduction Rising concerns about public health risks and the environmental burden caused by pesticide application has led to a re-evaluation of chemical-based pest management practices (Rola and Pingali 1993). Integrated Pest Management (IPM) emerged as an alternative approach to pest control methods relying exclusively on chemicals. There are a large number of conceptual definitions of IPM, most of which include using natural or ecologically sound principles or techniques, preventing pests from reaching 1 2

Department of Agricultural Economics, Faculty of Agriculture, Chiang Mai University, Thailand University of Hohenheim, Stuttgart, Germany


298

Sangkapitux, Suebpongsang, Nonkiti, and Neef

economically damaging levels, and using multiple pest control strategies, including cultural, biological and chemical methods (Fernandez-Cornejo and Jans 1999). The various objectives of IPM, such as enhanced sustainability and reduced reliance on hazardous pesticides, are achieved through 1) enhancing the efficiency of current measures, for example by the use of adequate monitoring techniques to ensure that pesticides are only applied when absolutely necessary, 2) substituting hazardous substances, for instance by replacing pesticides with mechanical control and biopesticides or by developing transgenic varieties, and 3) redesigning crop production systems, for example by adopting alternative crop rotations, implementing area-wide management schemes or alternating direct seeding with transplanting (IRRI 2007). One important aspect of IPM is to apply the “economic threshold” concept, which is most relevant for decisions that are made when the level of pest attack can be assessed through adequate monitoring. The economic threshold is the level of attack where the estimated benefits of treatment cover the costs of that treatment. If the severity of the pest occurrence is below the threshold, the costs of treatment would exceed the benefits and the farmer would sustain a loss when applying the treatment. Being a major producer and exporter of agricultural goods, Thailand has relied heavily in the past on the use of pesticides, many of which are classified as hazardous. Residues have been identified in soil, water, and agricultural products throughout the country (Thapinta and Hudak 2000; Kunstadter 2007). Since the Thai government has recognized the growing problem, it has enacted various environmental laws and programs aimed at minimizing the negative effects of pesticides on the environment and on human health. In 1998, the Good Agricultural Practice (GAP) program was instigated. The Thai government also declared 2003 the ‘year of food safety’ and launched a campaign to introduce the country as the ‘kitchen of the world’. The GAP program aims to ensure that food crops produced in Thailand are safe, healthy and meet the standards and requirements of the country’s food regulations, focusing on alternative ways to control pests in crop production. An integral part of the GAP program is Integrated Pest Management (IPM), an important means to achieve the standards required. IPM in longan growing was introduced in Thailand more than a decade ago through collaboration between IPM DANIDA (a Danish research and development initiative) and the Thai Departments of Agricultural Extension and Agriculture. Several training programs have been established to raise awareness among farmers about harmful pesticide practices and to encourage them to adopt less hazardous pesticides, with applications based on economic thresholds and complemented by biological and mechanical control methods. However, the degree of IPM

Determining Factors of IPM Adoption

299

adoption in longan production has remained disappointingly low. This paper attempts to identify factors supporting and hindering IPM adoption by longan growers in northern Thailand. Knowledge of the key factors driving the adoption of IPM will facilitate policy formulation, program planning and targeting, and help in diagnosing constraints in existing methods of IPM dissemination. Therefore, the study addresses a challenge faced by researchers, extension workers and policy makers involved in the development and implementation of an appropriate IPM strategy for fruit growers. Our results also provide insights into the prospects for widespread implementation of IPM in Thailand. We hypothesize that adoption is mainly limited by farmers’ lack of knowledge of the adverse health effects of pesticide application. A second hypothesis is that the availability of offfarm employment opportunities may present serious constraints on IPM adoption. Third, it is hypothesized that labor organization may also influence the degree of adoption of alternative pest management practices. The remainder of this paper is organized as follows. First, theoretical and empirical studies on IPM adoption are reviewed. Second, the evolution and adoption of IPM in Thailand are highlighted. Third, the longan farms surveyed in this study are described, followed by the presentation of an economic behavioral model for IPM adoption and specification of the empirical model. Fourth, results of the econometric estimation are presented and discussed. The final section summarizes the the paper's main findings and discusses key policy implications.

2 Supports for and barriers to Integrated Pest Management adoption: a review The effectiveness of IPM in reducing overuse of pesticides, thus improving agricultural productivity, human health and the environment has been demonstrated in a number of studies: conducted mostly in Western countries, but also in Asia and South America (Maumbe and Swinston 2000). According to Feder et al. (1985), factors influencing adoption of agricultural technology can be divided roughly into four categories, namely (1) characteristics of the technology, (2) farmers’ socio-economic conditions, (3) physical environment, and (4) institutional environment of the farm. Empirical studies focusing on examining off-farm income, labor organization and knowledge on health effects as major factors determining IPM adoption are reviewed hereafter in more detail. 2.1 Off-farm income Adoption of agricultural technologies such as IPM has been rather slow, despite explicit economic and environmental advantages (Fernandez-

300


Cornejo and McBride 2002). It is hypothesized that IPM adoption is driven by ‘non-quantifiable’ advantages, such as simplicity and flexibility, which lead to a reduction in time requirements for managerial activities. An obvious use of farm managers’ time is off-farm employment. The study of Blasé (1960, cited in Fernandez-Cornejo et al. 2007) found that off-farm income influenced the adoption of ‘conservation’ practices by providing ‘supplemental income’ to finance conservation expenditures. On the other hand, off-farm income can be a factor inhibiting the implementation of conservation practices, e.g. when the additional income is spent on family expenses and for essential farm production expenses other than conservation (Ervin and Ervin 1982, cited in Fernandez-Cornejo et al. 2007). McNamara et al. (1991) used empirical evidence from peanut producers to conclude that IPM required substantial time for management and that off-farm employment may present a serious constraint on IPM participation. 2.2 Labor organization Beckmann and Wesseler (2003) theoretically analyzed the interaction between IPM adoption and farm labor organization using a benefit-cost model and argued that the adoption of IPM also depended on farm labor organization. In this case, farm labor organization was classified into (1) owner-operated, (2) owner-operated in combination with family or permanently hired labor and (3) owner-operated in combination with shortterm hired labor. This was further examined through an empirical study of durian farmers in Chanthaburi province, Thailand. Their empirical results confirm that labor organization significantly determines the adoption of IPM as hypothesized in the theoretical model. Small farms under owner and family operation adopted IPM at a higher level than others. This stems from the fact that the transaction costs of small farms under family operation are usually lower than that of larger farms, which make them more adaptable to IPM practices, with their intensive monitoring and managing tasks (Beckman et al. 2006). 2.3 Knowledge of pesticide effects on human health The influence on IPM adoption of farmers’ knowledge of pesticide effects on human health has been explored with regard to a variety of aspects. The study of Maumbe and Swinston (2000) depicts health problems in forms of experience of pesticide-related acute symptoms, application of preventive measures and knowledge of the level of toxicity of pesticides. The authors found that previous experience with and knowledge of health-related effects of pesticides did not have a significant influence on IPM adoption


301

and concluded that more awareness about the health risks of pesticide application needs to be created among farmers targeted by IPM programs to promote adoption.

3 Pest management in longan farming systems in northern Thailand The principal pests causing economic damage to either young leaves, trunks, branches, flowers or fruits of longan trees in the Lamphun province of northern Thailand are the Longan Stink Bug (Tessaratoma papillosa Drury), Leaf Eating Weevil (Hypomeces squamosus Fabricius), Leaf Roller (Archips micaceana Walker), Leaf Eating Looper (Oxyodes scrobiculata Fabricius), Bark Eating Caterpillar (Cossus sp.), Longan Fruit Borer (Conopomorpha sinensis), Clerck (Othreis Fullonica), Nipaecoccus sp., Drepanococcus chiton, Cornegenapsylla sinica, and Scirtothrips dorsali (Duangsaard 2000). These pests can cause damage in longan orchards in various stages, ranging from leaf shooting to flowering and fruiting. The occurrence of each pest can also reach an epidemic level in different stages, but only some pests like the Longan Stink Bug can cause damage during both flowering and fruiting stages (Leelapiromkun 2006). Under conventional pest management it is mainly recommended to spray chemical pesticides for controlling pests. Although the application of chemical pesticides has alleviated pest problems in the short term, it has entailed a number of negative externalities, such as secondary pest outbreaks, development of pesticide resistance and the destruction of natural enemies. Therefore, IPM has been proposed as an alternative approach to conventional pest management. The GAP program, introduced in Thailand in the late 1990s, aims to ensure that food crops produced in Thailand are safe to eat and meet the standards of the country’s food regulations. This program is based on various aspects of crop management in order to produce safe agricultural products, with pest control being an important component. The program focuses on alternative ways to control pests in crop production, e.g., by using pest-resistant varieties, biological control methods, or improving cultural practices. Here, the use of pesticides would be the method of last resort to control pests, and farmers are encouraged to use pesticides only when absolutely necessary. Hence, IPM, which is based on similar ideas, is an integral part of the GAP program. A basic principle is regular monitoring of pests and applying ‘preventive measures’ in order to reduce pest infestation and, consequently, the use of chemicals. Currently, two types of substances, petroleum oil spray and wood vinegar, have been introduced to longan farmers as a preventive measure. The list of IPM practices considered within this study is based on the recommendations of the research report on longan by the Office of

302


Agricultural Research and Development (Region 1) and supplemented by activities identified during the survey. There are 41 IPM activities which have been categorized into six important groups, namely (a) monitoring of infestation of pests and diseases (b) application of threshold levels (c) multiple strategies, such as record keeping, using traps and so-called ‘stinky balls’, sterilizing tools being used and wrapping fruits (d) cutting, burning, pruning, and weeding (e) applying biological methods, and (f) regular inspection of the sprayer.

4

Sample selection

The empirical model for this study is based on data from a farming system survey of 154 longan growers under the GAP program in Lamphun, a northern province in Thailand where longan trees are extensively grown in commercial orchards. As of 2005, longan farmers in Lamphun contributed 37.8% to national production (Office of Agricultural Economics 2005). Field research was concentrated on five sub-districts in Muang Lamphun district, where the highest number of GAP farmers can be found. As of September 2006, 3,007 longan farmers in Muang Lamphun district had applied for a GAP certificate, and most of them had already obtained it (Office of Agricultural Research and Development (Region 1), 2006). The study sample size is 5.1% of total longan farmers who applied for the GAP certificate in Muang Lamphun district. In this study we surveyed a sample of 154 households, which exceeds the determining sample size of 97 households3 needed at the 90% confidence level. GAP farmers’ names from a list provided by the Office of Agricultural Research and Development Region 1 were subdivided according to subdistrict location. The sampling procedure followed the stratified random sampling method. The sample size in each sub-district in Muang Lamphun district was calculated based on a weighted average of the number of GAP farmers in each sub-district compared to the total number of GAP farmers in Muang Lamphun district.

5

Model specification

5.1 Defining IPM adoption The successful evaluation of any IPM strategy has to begin with a clear definition of what is being assessed. Typically, IPM involves a number of pest management practices that are both location- and crop-specific. There 3

The determination of the sample size equation is based on the Taro Yamane equation (Yamane, 1973). The equation is n = N / [1+ N(e)2] where n = sample size; N = population; e = confidence level.

303


is no consensus in the literature as to which specific pest management practices are part and parcel of IPM, though IPM definitions have been classified as either “input-oriented” or “output-oriented” (Swinton and Williams 1998). The latter focus on desired outcomes such as profitability, human health and environmental quality, while the former relate to specific IPM practices. For the purpose of this study we use the “input-oriented approach”: the degree of IPM adoption is captured in the form of the number of IPM activities applied by longan growers, according to the set of recommended IPM practices mentioned above. 5.2 Factors hypothesized to determine IPM adoption: The factors included in the model to estimate the adoption levels of longan growers are 1) farm characteristics, such as farm size, number of plots, cropping patterns, and intensity; 2) socio-economic conditions of the farmers, such as age, gender, education, household members, off-farm income, and farm labor organization; 3) knowledge of health effects; 4) knowledge of IPM; and 5) IPM training attendance. Description of these independent or explanatory variables is summarized in Table 1.

Table 1: Descriptive statistics of longan farm households Variable

Definition/Unit

Mean

Standard Deviation

Minimum

Maximum

Farmers’ characteristics Gender

= 1 if female

Age

years

Education of farm owner Family members in farming

0.88

0

1

56.25

10.86

29

88

years of schooling

6.22

3.90

0

18

number

1.75

0.80

0

1

ratio of off-farm income per total family income

0.55

0.33

0

1

Plot

number

2.19

1.50

1

9

Longan orchard

size in rai

4.05

3.74

0.25

31

Multiple cropping

1= if multiple

0.19

0

1

Cropping Intensity

Revenue per productive tree (Thai Baht)

7

2,333

Farm and off-farm income Off-farm income

Farm characteristics

201.12

282.25

304


Table 1: Continuation Variable

Definition/Unit

Mean

Standard Deviation

Minimum

Maximum

IPM training and knowledge IPM training

number of attended courses

0.69

IPM knowledge

1=if farmer has high IPM knowledge

0.55

number of IPM activities

19.94

2.31

0

24

0

1

6

35

IPM practices IPM adoption

5.62

Awareness of pesticide effects on human health Knowledge of pesticide effects on human health

1= if farmer has knowledge

0.80

0

1

Illness from chemicals

1 = sick

0.09

0

1

1= if very high pest pressure

0.55

0

1

Farm labor organization

share of hired labor day per total labor day

0.44

0

0.99

Farm labor market conditions

1= if easy to hire laborers (to spray pesticides)

0.56

0

1

Pest pressure Pest pressure Labor organization 0.31

5.3 Economic decision model of adoption: Count model For this study, we have applied a Poisson maximum-likelihood regression model to predict the discrete, but non-categorical, pest management strategies used by longan growers in northern Thailand. Following Greene (2003), the predicted values Y1,Y2,…,Yn are assumed to have independent Poisson distribution with parameters λ1 , λ2 ,…, λn respectively. The basic equation for the Poisson regression is represented as follows: e −λ λ i Pr ob (Υ i = y i x i ) = y i!

where yi = 0, 1, 2, 3,……

yi

(1)

305


The parameter λi is assumed to be log-linearly related to regressors xi. Hence, Ln (λ i ) = x i ' β (2) The log-likelihood function is given by n

[

]

ln L = ∑ − λi + yi xi' β − ln yi !

(3)

i =1

The expected number of IPM activities practiced by the longan growers is given by E [Y = y i x i ] = Var [Y = y i x i ] = λ i = e x iβ '

(4)

Apart from the Count model employed to examine the factors determining the level of IPM adoption, a Tobit model was used for solving the problem of endogeneity of farm labor organization captured in the form of the ratio of hired labor per total labor as adopted from Beckmann et al. (2006).

6

Empirical results

The specific information gathered on the characteristics of the farm households under study and longan farming systems in general is presented below, providing an overview of the data used in the econometric model. These descriptive results are followed by a presentation of the outcome of the regression analysis. 6.1 Descriptive results Farmers’ characteristics The households of the surveyed farmers can be characterized as nuclear families composed on average of 3.61 people with 2.47 adult members who were able to work, of which only 1.75 people were working on the family farm. The average education attainment of farmers in the research location was 6.22 years of schooling (Table 1). Farm and off-farm income According to the information provided by farmers for the crop year 2005/2006, the average family income was 214,506 Baht4 per year, being 4

1 Thai Baht was equivalent to around 0.03 US$ in March, 2008 (Bank of Thailand, 2008)

306


composed of farm and off-farm income. To assess how much the farm income and longan production in particular contributes to family income, the farmers were asked to attribute their income in percentage to three categories: from longan, from other crops and from off-farm work. The results showed that on average the off-farm income contributed the largest share of family income with 55% (Table 1), while farm income from longan was ranked second with 40%. Other crops played a minor role with only 5%. Farm characteristics The average total farm size per household was 7.47 rai5 with an average of 2.19 plots per household. The average area under longan cultivation was 4 rai per family, ranging from a minimum of 0.25 rai to a maximum of 31 rai (Table 1). In cases where farmers had many plots in their farms, they were asked to choose only one of them for in-depth interview about gross margin and labor organization. The selected plot size of most households varied from 2 to 6 rai. Half of the longan orchards had 40 to 100 trees, generating about 201 Baht per productive tree, varying from a minimum of 7 Baht per year to a maximum of 2,333 Baht per year with a standard deviation of 282.25. The Edor6 variety was grown in all selected plots. The average age of productive longan trees was 16 years. 81% of the longan selected plots were single cropping systems, while only 19% were multiple cropping systems, where longan trees were mixed with other fruit trees or/and vegetables. IPM training and knowledge The average number of IPM training courses attended by the total sample of farmers was only 0.69 courses (Table 1). From 24% of the total sample who had attended IPM training programs, the household head was the main participant in the training. These farmers attended 2.95 courses on average, ranging from 0 to 24. The most important reasons farmers gave for joining the training were ‘suggestion from an extension worker’ and ‘following the example of a neighbor’. Most of the farmers who joined the training rated the training as being of ‘high’ or ‘very high’ value. The farmers’ knowledge of IPM was assessed by asking the farmers to mention the IPM strategies they were aware of. Farmers were given a high IPM knowledge score if they could mention at least three IPM strategies. Under this coding system, 55% of the farmers received a high IPM knowledge score (Table 5 6

1 rai = 0.16 hectare Edor is the most popular cultivar in Thailand and about 73% of the total longan area in the country is planted with this cultivar. It is an early-maturing cultivar as indicated by the name 'Dor' meaning early.


307

1). The small percentage of farmers who attended IPM training programs could be the major reason why farmers in the sample had limited knowledge of IPM. To confirm this hypothesis, the correlation between attendance of IPM trainings and IPM knowledge scores was analyzed. The value of the Pearson correlation coefficient was significant at the 0.01 level, which can imply that these two variables were positively related. The lower the number of attended IPM trainings, the lower was the IPM knowledge score. Even though low attendance at trainings and their IPM knowledge scores showed that farmers were not very familiar with IPM, this lack of knowledge could also be attributed to the fact that the technical term ‘IPM’ was not well understood by the respondents. IPM practices As regards the 41 recommended IPM activities, the mean number of activities practiced by the longan growers is 19.94, and the maximum and minimum are 35 and 6, respectively (Table 2). The IPM activities were categorized into six important groups, namely (a) monitoring of infestation of pests and diseases, composed of 14 activities in total, of which 11.51 activities were applied by farmers on average; (b) application of pest control threshold level, comprising nine activities in total, of which 3.15 activities were applied by farmers on average; (c) multiple strategies, such as using traps, record keeping, using so-called ‘stinky balls’, sterilizing tools being used and wrapping fruits, comprising seven activities in total, with only one activity being applied by farmers on average; (d) cutting, burning, pruning, and weeding, comprising a total of five activities, of which 3.23 activities were applied by farmers on average; (e) applying biological methods, such as using neem (Azadirachta indica) extract as a biological pest control, spraying wood vinegar, release of predators and maintaining a rearing station for predators, comprising five activities in total, with only 0.18 activities being applied by farmers on average; and (f) regular inspection of the sprayer, comprising only one activity, but with a comparatively high rate of 0.86 activities being applied by farmers on average. Table 2 shows which categories of IPM activities farmers applied more frequently, seen in relation to the other categories. The ratio of the sum of IPM activities applied within each IPM category to the total recommended IPM activities in that category is used as an indicator. A high ratio means that farmers applied many activities in that IPM category. The ratio of the category “regular inspection of the sprayer” was the highest, with 0.86, followed by the category “monitoring infestation of pests and disease”, and the category “cutting, burning, pruning, and weeding”, at ratios of 0.82 and 0.65, respectively. The other categories, namely

308


“threshold level” and “using biological control methods” had ratios of only 0.35 and 0.04, respectively.

Table 2: IPM activities in longan farms according to different categories Category of activities

Average ratio of activity applied per total IPM activities (41 activities)

Monitoring infestation of pest and disease

0.82

Application of pest control threshold level

0.35

Multiple strategies, such as cutting and burning, pruning, weeding and sterilizing tools being used

0.65

Using biological methods, such as neem extract, spraying wood vinegar, release of predators and maintaining a rearing station for predators

0.04

Regular inspection of the sprayer

0.86

Source: own calculation

Awareness of pesticide effects on human health Farmers’ opinions about five possible risks from using pesticides, namely (a) health damage, (b) harmful effects to fish and other aquatic species, (c) harmful effects to birds, (d) harmful effects to mammals and farm animals, and (e) toxicity to beneficial insects were assessed by asking the farmers to give a score from 1 for ‘not risky’ or ‘not important’ to 5 representing ‘very risky’ or ‘very important’ for each possible risk category. Under this scoring system, the farmers were separated into two groups: farmers with awareness of health risks (above the mean score of 2.5) and without awareness (below the mean score of 2.5). According to this classification, 80% of the farmers were categorized as risk-aware (Table 1). In all, the results indicate that farmers (1) are aware of the human health risks of pesticide application and (2) care more about their own health than about the effects on animals or the environment in general. Pest knowledge and pest pressure The knowledge of pests can be a determinant of pest management practice and an indicator of the seriousness of the pest problem. Therefore, the study also assessed farmers’ knowledge of pests by asking for the names of the pests depicted in four pictures that an enumerator showed to each farmer. Farmers were given a very high pest knowledge score if they could name all four pests. According to this scheme, most of the farmers had a


309

moderate to high pest knowledge score, though only 11 percent of the farmers received the highest pest knowledge score. When considering the most problematic insect pests and diseases in the study area, we found that 21 percent of farmers indicated the leaf-eating looper as the most serious problem, followed by the bark-eating caterpillar and witches’ broom. Farmers’ opinions about the severity of pest problems were also assessed by asking the farmers to give a score from 1 to 5 indicating the level of each pest problem. Through this scoring system, the farms were separated into two groups—the farms with high pest pressure (above the mean score) and the farms with low pest pressure (below the mean score)—with 55% of the farms considering the severity of their pest problem to be at a high level (Table 1). This factor was then fed into the regression model as dummy variable. Labor organization The average number of annual labor days per rai for longan orchards was about 44 (Table 1). In general, family labor was slightly more important than hired labor, with an annual average of around 28 days per rai per year as compared to 16 days, respectively. Family labor particularly plays an important role in pest management activities, while hired laborers, especially casual laborers, spend most of their time in harvesting and selling activities. On average family members spent about 18 labor days per rai per year on pest management activities, whereas hired labor accounts for only 2 labor days per rai per year for such activities. Family laborers are mainly responsible for pest monitoring. On average they spent 7 labor days per rai and per year for pest monitoring, while hired workers spent 0.02 labor days per rai per year for the same activity. Likewise, IPM activities are mainly conducted by the owner and/or family members. On average hired laborers spent 1.5 labor days per rai for IPM activities, whereas owners and family members spent in total around 10 labor days per rai annually. The amount of labor spent on pesticide activities is more evenly distributed between owners and hired laborers: owners spent 0.9 labor days and hired laborers spent 0.7 labor days per rai for pesticide activities (Table 3).

310


Table 3: Labor organization

Activities Pest monitoring (a) Pesticide activities (b) IPM activities ( c ) Fertilizing (d) Irrigation (e) Tree management* (f) Harvesting and Selling (g) Total Pest Management (a+b+c)

Family labor (days per rai) 7.07 0.90 9.84 1.45 3.83 1.49 3.38 27.96 17.81

Hired labor (days per rai) SubPermanent Casual Total 0.01 0.01 0.02 0.02 0.67 0.69 0.19 1.30 1.49 0.01 0.35 0.36 0.02 1.28 1.30 0.01 0.24 0.25 0.00 11.99 11.99 0.26 15.84 16.10 0.22 1.98 2.20

Total (days per rai) 7.09 1.59 11.33 1.81 5.13 1.74 15.37 44.06 20.01

* flower induction, bamboo sticks and tying branches

For this study, farm labor organization is represented by the ratio of hired labor days per total labor days. The average ratio of the hired labor days per total labor days was 0.44 ranging from 0 to 0.99 with a standard deviation of 0.31. Since hired labor plays an important role in pest management activities, the possibility of finding available hired labor, especially for spraying pesticide, was assessed by asking the farmers whether it was relatively easy or difficult to find such labor, with a score of 1 indicating “very easy” to 5 meaning “very difficult”. The percentages of farmers who gave lower and higher than average score were about equal at 56%. This implies that longan farmers’ experiences with finding hired labor for spraying activities varied significantly within the sample. 6.2 Regression results For this study, two empirical models are estimated. The first is a Tobit model testing the exogeneity of labor organization, and the second is a Count model to determine factors influencing IPM adoption. The regression result of the first model, as shown in Table 4, provides evidence that farm labor organization is an endogenous variable. Farm size and number of plots positively determined farm labor organization, with significance at the 1% level, while multiple cropping pattern and number of family members involved in farming negatively affect farm labor organization at the 5% and 10% significance levels, respectively. These three factors were then used as instrumental variables for estimating farm labor organization, which were then integrated into the second model.

311


Table 4: Regression results Tobit Model Labor organization

Count Model IPM Adoption

Coefficient (Standard Error)

Coefficient (Standard Error)

Education of farm owner

0.00 (0.00)

0.00 (0.00)

0.06

IPM training

-0.00 (0.00)

-0.00 (0.00)

-0.11

IPM knowledge

0.03 (0.05)

0.01 (0.03)

0.24

Knowledge of pesticide effects on human health

0.04 (0.06)

0.14*** (0.04)

2.80***

Size of longan orchard

0.02*** (0.00)

-

-

Multiple cropping

-0.16** (0.06)

-

-

Crop Intensity

0.00 (0.00)

-0.00 (0.00)

-0.00

Pest pressure

-0.02 (0.05)

0.06 (0.03)

-0.00

Farm labor market condition (spray pesticide)

0.06 (0.05)

0.03 (0.04)

0.71

Off-farm income

-0.03 (0.07)

-0.15*** (0.05)

-3.16***

Number of plots

0.04*** (0.01)

-

-

Family members in farming

-0.09*** (0.03)

-

-

Illnes from chemicals

-0.03 (0.08)

0.04 (0.06)

0.81

Gender

-0.08 (0.07)

-0.01 (0.00)

-0.04

Age

-0.00 (0.00)

-0.00 (0.00)

-0-04

Constant

0.18** (0.46)

3.04*** (0.15)

60.71***

Independent Variables

Ho: exogeneity Reject ***,** Note: *, **, ***: 10% , 5%, 1% significance. Source: own calculation.

Marginal Effect

312


Results of the Count model on the determinants of IPM adoption practices are summarized in Table 4. The major results of the Count model suggest that knowledge of pesticide effects on human health and off-farm income are significant at the 1% level. The knowledge of pesticide effects on human health positively determines the level of IPM adoption of longan farmers. Surprisingly, the number of training courses attended did not seem to have a significant influence on IPM adoption. On the other hand, the level of IPM adoption significantly decreased as the contribution of offfarm income to family income increased. Farm organization was found to be insignificant in determining the level of IPM adoption of the longan growers.

7

Discussion and conclusion

Drawing on a cross-sectional, primary data set obtained from a farm household survey of 154 longan growers under the GAP program in Lamphun, Thailand, the purpose of this study was to test several hypotheses related to adoption of IPM practices in longan production. The finding that a higher level of knowledge of negative health effects from pesticide application significantly increases the degree of IPM adoption of the longan growers underscores the findings of various studies pointing to the importance of improving farmers’ knowledge as regards negative health impacts of pesticides. Astonishingly, the number of training courses attended did not appear to have a significant influence on IPM adoption. This may reflect a poor quality of the agricultural extension service or inappropriate modes of disseminating information in the field of IPM. The policy implication is that the quality of training courses provided by the government needs to be improved. A starting point would be to assess the training needs of farmers and to understand their particular social networks and modes of communication and knowledge acquisition. As regards labor organization, results from the econometric model suggest that hired labor plays a crucial role in longan production; however, when it comes to pest management activities, family labor is much more important than hired labor, pointing to a tendency of assigning such important tasks to relatives who can be trusted. In contrast to a study of IPM adoption in durian growers by Beckman et al. (2006), the labor organization variable is insignificant in determining IPM adoption among the longan growers studied. This can be explained by the fact that the individual value of longan fruits is much lower than that of durian fruits, which are sold by the piece and are the most highly priced fruits in Thailand. Hence, durian farmers are likely to provide much more individual care to their trees than longan farmers. In addition, hired labor


313

for pesticide spraying is available at relatively low cost in the longangrowing region. The finding of a significantly lower adoption level among farm households with a high ratio of off-farm income supports those studies that have identified off-farm income as a major barrier to IPM adoption. The managerial demands and time requirements of IPM are particularly high in the case of longan production, making it hardly possible for farmers engaged in off-farm activities to adopt such labor-intensive practices. This does not imply, however, that policy-makers should refrain from supporting off-farm alternatives to the farming sector in northern Thailand. A sustainable rural development strategy not only has to create a strong agricultural sector with environmentally friendly forms of crop production, but also a viable system of small and medium enterprises that can add value to agricultural raw material.

Acknowledgement Financial support from the Deutsche Forschungsgemeinschaft (German Research Foundation) for carrying out this survey is gratefully acknowledged.

References Beckmann, V., Wesseler, J. (2003). How labor organization may affect technological adoption: an analytical framework analyzing the case of integrated pest management. Environment and Development Economics 8(3): 1-14. Beckmann, V., Iravan, E., Wesseler, J. (2006). The effect of farm labor organization on IPM adoption: empirical evidence from Thailand. Contributed paper presented at the International Association of Agricultural Economists Conference, Gold Coast, Australia, August 12-18, 2006. Blasé, M.G. (1960). Soil erosion control in Western Iowa: progress and problems. Unpublished Ph.D. Dissertation. Iowa State University, Ames, IA. Channarong Duangsaard (2000). Pests in longan. In Noppadon Charansumrit, Pawin Manochai, Nopmanee Tophunnanon, Teeranut Chantrachit, Winai Viriyaalongkorn and Pichai Somboonwong (eds.) Longan production. Centre of Longan and Litchi Research and Development, Maejo University, Thailand (in Thai). Ervin, C.A., Ervin, D.E. (1982). Factors affecting the use of soil conservation practices: hypothesis, evidence, and policy implications. Land Economics 58(3): 277292.

314


Feder, G., Just, R.J., Zilberman, D. (1985). Adoption of agricultural innovations in developing countries: A survey. Economic Development and Cultural Change 33(2): 255-298. Fernandez-Cornejo, J., Jans, S. (1999). Pest management in U.S. agriculture. United States Department of Agriculture, Economics Research Service. Agricultural Handbook No. AH717. available at http://www.ers.usda.gov/publications /Ah717.pdf. [accessed March 2007] Fernandez-Cornejo, J., McBride, W.D. (2002). Adoption of bioengineered crops (Agricultural Economic Report No. 810). United States Department of Agriculture Economic Research Service, Washington, DC, USA. Fernandez-Cornejo, J., Ashok Mishra, A., Nehring, R., Hendricks, C., Southern, M., Gregory, A. (2007). Off-farm income, technology adoption, and farm economic performance. Economic research report number 36, United States Department of Agriculture, Washington, DC, USA. Greene, W.H. (2003). Econometric analysis, 5th Edition. Prentice Hall, New Jersey, USA. Kunstadter, P. (ed.) (2007). Pesticides in Southeast Asia: Environmental, biomedical, and economic uses and effects. Silkworm Books, Chiang Mai, Thailand. Maddala, G.S. (1992). Limited-dependent and qualitative variables in econometrics. Cambridge University Press, Cambridge, UK. Maumbe, B.M., Swinton, S.M. (2000). Why do cotton growers in Zimbabwe adopt IPM. The role of pesticide related health risks and technology awareness. Paper presented at the Annual Meeting of the American Agricultural Economics Association, Tampa, FL, July 30 - August 2, 2000. McNamara, K.T., Wetzstein, M.E., Douce, G.K. (1991). Factors affecting peanut producer adoption of integrated pest management. Review of Agricultural Economics 13(1): 129-139. Office of Agricultural Economics (2005). Agricultural statistics of Thailand, crop year 2004/2005. Ministry of Agriculture & Cooperatives, Bangkok, Thailand. Office of Agricultural Research and Development Region 1 (2006). List of farmers who applied for GAP. Office of Agricultural Research and Development Region 1, Chiang Mai, Thailand. Patcharaporn Leelapiromkun (2006). Research report on longan. Office of Agricultural Research and Development Region 1, Department of Agriculture, Thailand (in Thai), Chiang Mai, Thailand. International Rice Research Institute (IRRI) (2007). Key concepts or tools for thinking about IPM. IRRI: Rice Knowledge Bank. available at http.//www. knowledgebank.irri.org/IPM/keyconcepts/PrintDoc/KeyConcepts.doc [accessed March 2007]. Rola, A.C., Pingali, P.L. (1993). Rice productivity and farmers’ health: an economic assessment. World Resources and International Rice Research Institute, Washington DC, USA. Swinton, S.M., Williams, M.B. (1998). Assessing the economic impacts of integrated pest management: lessons from the past, directions for the future. Staff Paper 98-12. Department of Agricultural Economics. Michigan State University, East Lansing, USA.


315

Thapinta, A., Hudak, P.F. (2000). Pesticide use and residual occurrence in Thailand. Environmental Monitoring and Assessment 60(1): 103-114. Yamane, T. (1973). Statistics: an introductory analysis. Harper International Edition, Tokyo, Japan.

Chapter 16 Transaction Cost Analysis of Hired Labor Use in Pest Management: An Empirical Study of Fruit Tree Farming in Thailand Evi Irawan1, Volker Beckmann1,2, and Justus Wesseler3 Abstract: This paper reports an empirical investigation of the effect of pest management tasks on the likelihood of hired farm labor employment, using a transaction costs economic framework. It is argued that hired farm laborers are not randomly assigned to perform pest management tasks, but rather to the tasks that are easier to monitor, are applied repetitively, and do not require specific skills. Accordingly, the main hypothesis derived is that pest management tasks labeled as part of an Integrated Pest Management (IPM) strategy are most likely to be performed under farm labor organization comprised largely of family laborers. We test this hypothesis based on farm-level data on pesticide management and labor use in Durian and Tangerie production systems in Thailand. Keywords: Fruit trees farming, Hired labor, Pest management, Transaction cost economics

1 Introduction Recent theoretical and empirical studies on the use of hired farm labor in farm production activities have been considering the presence of transaction costs (e.g. Schmitt 1991; Evenson et al. 2000) and have identified it as an important determinant of adoption of labor-saving agricultural innovations (e.g. de Janvry et al. 1989). Nonetheless, empirical studies on the use of hired labor in pest management activities are very limited. Most studies are concentrated on farm labor organization at farm level rather than on detailed farm labor organization for farm production 1

2

3

Department of Agricultural Economics, Faculty of Agriculture and Horticulture, Humboldt University of Berlin, Germany Department of Environmental Management, Chair of Environmental Economics, Brandenburg, University of Technology Environmental Economics and Natural Resources Group, Wageningen Univerty, The Netherlands


318

Irawan, Beckmann, and Wesseler

activities, such as pest management. Thus, a detailed picture of farm labor– task interaction within farms is often overlooked. Pest management is an activity in fruit tree farming that may be performed during almost all growth stages. Having a look at the details shows that pest management activity involves tasks with distinct attributes. Depending on the strategy being employed, a set of tasks may vary from one farm to another. A calendar-based spraying strategy, for instance, may involve simple tasks such as pesticide spraying, while others, such as Integrated Pest Management (IPM) strategies, may consist of a wide ranging set of tasks, from simple to more complex ones. From a farm-household economic viewpoint, demand for hired labor to perform farm production activities is determined, among other things, by the opportunity costs of using family labor, the marginal product of labor on-farm work and the number of working family members (Schmitt, 1991; Ellis, 1993). The model predicts that, if the value of the marginal product of the first unit of hired labor is less than the prevailing wage rate for farm labor, a farmer is more likely not to hire farm labor. Similarly, if the value of the marginal product for the first unit of family labor applied on the farm is less than the opportunity cost of using a family member, then it is more likely that there will be no family labor used for on-farm production. When family labor has two different uses, working on the family farm and offfarm, the returns to labor in both activities are equalized at the margin. Family labor is applied on-farm until the point where the value of the marginal product is equal to the market wage rate for off-farm work. This view, however, is insufficient to account for which tasks hired farm laborers are being employed. Empirical farm level studies of small-scale rice farming in Southeast Asia show that farming tasks are not allocated randomly between family and hired laborers (e.g. Roumasset and Smith 1981; Kikuchi and Hayami 1999). Hired laborer is mainly assigned to perform arduous tasks such as transplanting, weeding, and harvesting. These tasks demand large quantities of farm labor and their performance can be monitored by ex post inspection on the field or visible effort outcomes (e.g, transplanted areas, weeded areas, and harvested quantities). In contrast, crop care tasks tend to be performed by family labor without immediate visible outcomes (Kikuchi and Hayami 1999). This, along with the principle of comparative advantage, explains the tendency of family labor to concentrate on higher level tasks such as supervision, which is less arduous, leaving the more arduous tasks such as harvesting to hired labor (Bautista 1993). This paper is concerned with looking at the interaction between decisions to use external labor and pest management tasks in fruit tree farming systems and aims to identify the determinant factors of task

Transaction Cost Analysis of Hired Labor Use in Pest Management

319

assignment to hired labor in pest management. We use a transaction cost economics framework to predict the extent to which farm households use hired labor to perform pest management tasks in fruit tree farming.

2 Theoretical framework Transaction cost economics theory might offer at least partial explanation of farm household labor–task arrangements (Roumasset and Smith 1981; Pollak 1985). The basic idea of this approach is that a governance structure is a result of transaction costs economizing (Coase 1952; Williamson 1981, 1985). The main focus is on the characteristics of a transaction itself. With respect to farming task accomplishment, the transaction is the farmer’s decision concerning whether the task will be performed by hired labor or by family labor. Thus, the characteristics of the task may play a role in the farmer’s decision-making process. The transaction cost perspective suggests that, in the presence of bounded rationality and the potential for opportunistic behavior on the part of one or more parties to a transaction, the choice between family and hired labor depends on three critical dimensions of transactions: asset specificity, uncertainty/complexity, and frequency. In examining each characteristic of tasks in pest management activity, we will highlight hypotheses generated from the model. 2.1 Asset specificity Asset specificity refers to the specialization of assets. Generally, the value of highly specialized assets diminishes when they are shifted to other alternative uses. Therefore, large investments in specialized assets increase risk when confronted with hold-up problems or with opportunistic behavior by market partners. In the specific transaction of hiring farm laborer to perform pest management tasks, asset specificity can be mainly related to crop-pest specific tasks. Hiring farm laborers possessing crop-pest specific skills may be burdened with high transaction costs, if not to say impossible. Lepak and Snell (1999) stated that, by their nature, the development of workers with firm-specific skills relies on training acquired within a given firm rather than on a search of the open labor market. Thus, what farmers can do is to train farm laborers and provide long-term employment contracts. However, these may not come without costs. If the farm laborer contracts with another farm, the first farmer will lose his training investment. Beside that, long-term employment will increase hoarding costs during the slack periods of farm production. These may negate an incentive for the farmer to train hired farm laborer rather than family laborer.

320


2.2 Frequency Frequency is the second important dimension in the transaction cost approach. The idea is that the more frequent the occurrence of a transaction, the more easily the related fixed (sunk) costs can be recuperated and the more economical it is to perform the task oneself (Vernimmen et al. 2000). For pest management tasks, this idea should be interpreted carefully. High frequency tasks in pest management activities such as pruning, weeding, and spraying pesticides are both time and (manual) labor intensive. Take spraying pesticides as an example. In durian farming, if the farmer employs a calendar pesticides spraying strategy as a mean of pest control, he has to apply pesticides eight times per season (Ruay-Aree et al. 2001, c.f. Vonsaroj 2007) and, since the average labor used for each pesticide application is 1.10 labor days/rai in average, then the farmer is likely to spend 8.80 labor days/hectare for one production season4. In tangerine farming, the amount of time required for calendar pesticide spraying is much higher than in durian farming as, for one production season, a farmer has to apply pesticides at least 17.7 times (Ruay-Aree et al. 2001, c.f. Vonsaroj 2007), and the average for each pesticide application is 1.92 labor days/rai5. Hence, the major trade-off for a farmer is between alternative uses of his time and family labor available. This may lead a farm household with higher opportunity costs to hire external labor to perform arduous tasks and allocate family labor to managerial tasks or work off-farm. The likelihood of using external labor is probably much higher when the farmer is aware of the side effects of pesticides on human health (Beckmann and Wesseler 2003). The frequency of other pest management tasks, especially tasks associated with IPM, is relatively low or incidental contingent upon the severity of pest problems on the farm. Thus, from the hired laborer viewpoint this means that the labor market for IPM tasks is relatively small and very uncertain compared to the labor market for manual workers that do not require specific skills such as weeding or spraying pesticides. This may negate the advantage of being a specialist on specific pests. Likewise, it may be very difficult for a farmer to find suitable hired laborer to perform IPM tasks in the labor market. Combining human asset specificity and the uncertain nature of IPM tasks, the transaction costs become very high or may even block a transaction. 4

5

This information is derived from an empirical survey of durian farming in Chanthaburi, Thailand in March – April 2005, conducted by the authors. This information is derived from empirical survey of tangerine farming in Fang District – Chiang Mai, Thailand in October – November 2006, conducted by the authors.


321

2.3 Uncertainty A third dimension of the transactions costs approach is uncertainty. With respect to pest management tasks, uncertainty is associated with the difficulties in monitoring performance and obtaining hired farm labor in a timely manner. These difficulties can increase the adverse effects of bounded rationality and opportunism. The problems of monitoring performance of some tasks like weeding, picking dropped fruits or pesticide applications are relatively minimal. The other tasks associated with IPM, such as fruit thinning or cutting infested branches, are however very difficult to monitor. For example, it is difficult to monitor whether farm laborers only cut infested branches and take into account the future productivity of the tree or whether leaving infested branches and thereby possibly decreasing future productivity. If the farm laborers are not residual claimants of production yield, it is more likely that they may not perform their tasks with as much care, precaution, or wise judgment as it they were. Thus, the likelihood of behaving opportunistically increases with the difficulties in monitoring performance. The problems of obtaining hired farm labor in a timely manner are associated with the nature of those IPM tasks which are contingent on pest problems and sensitive to hold-up problems of task implementation. Finding hired farm laborer with expertise in a specific pest is most probably costly, if available at all. Thus, uncertainty may increase for a farm household relying on hired labor to perform such tasks. Using the line of reasoning provided by the transaction cost approach as mentioned above, we derive hypothetical combinations between the characteristics of tasks and the likelihood of external labor employment in pest management tasks (Table 1). We suppose, to simplify, that each task characteristic can take two values: general or crop-pest specific skills (asset specificity), occasional or repetitive (frequency), and monitorable or nonmonitorable (uncertainty). However, since our analysis is static, any factor such as opportunity costs of family labor, labor capacity constraints, accessibility of labor market, or budget constraints allegedly affected farmers’ decisions to hire external labor is assumed to be fixed or given.

322


Table 1: Tasks characteristics and the likelihood of external labor use

Low uncertainty

High uncertainty

General skills Occasional Repetitive Hired casual or Hired casual permanent labor labor (1) (2) Family labor Family labor (5) (6)

Crop-pest specific skills Occasional Repetitive Family labor (3)

Hired permanent or family labor (4)

Family labor (7)

Family labor (8)

Table 1 shows that if a task does not require specific skills and is monitorable, regardless of the frequency, it is hypothesized that the farmer will hire external labor to perform it (category 1 and 2). This changes, however, if a task requires specific skills (category 4). The use of external labor in these cases is still possible. But, the farmer has to train the laborer with skills specific to the crop and pests. To secure his training investment, the farmer then has to provide a long-term employment contract to the laborer. Furthermore, external labor may not be a good option for the farmer when the task is non-monitorable (category 5, 6, 7 and 8) or monitorable but occasional (category 3).

3 Methods 3.1 Data source Data is derived from surveys of durian and tangerine farming in Thailand. The durian farming survey was conducted in Chanthaburi Province in 2005, while the tangerine farming survey was conducted in Fang District, Chiang Mai, in 2006. The two research locations and two crops were selected purposively by considering the existence of different forms of farm labor organization within the two groups of farm households. The data consists of 157 durian and 160 tangerine farm households, selected from registered IPM-trained durian farmers in the Office of Agriculture Extension in Chanthaburi Province and from a list of tangerine farmers engaged in the Good Agricultural Practices (GAP) project of the Department of Agriculture (DOA) in 2005. The samples were derived using stratified random sampling. The survey itself forms part of an IPM adoption and farm labor organization project. 3.2

Data of analysis

To analyze the data, we use descriptive statistics and an econometric model. The descriptive statistics include a t-test to compare means of variables in question. The econometric model is applied to estimate the


323

effect of the attributes of pest management tasks as well as control explanatory variables on the likelihood of hired labor use in pest management. Variables used in the empirical model are described in Table 2. Farm households’ decisions about employing hired labor in pest management are typically a dichotomous choice problem, for which binary choice models are most suitable (Verbeek 2000; Greene 2000). A binary choice model explains a dichotomous dependent variable in terms of a unobserved latent variable yi* that represents the latent preference of farm household i for the choice yi = 1 . Thus the model is generally expressed as yi* = xi' β + ε i

(1)

where yi* is the value of the response variable for farm household i , xi' is the vector of independent variables of farm household i that determine yi* , β is a parameter that indicates the effect of vector xi' on yi* , and ε i is a stochastic error term. This model takes the possibility into account that farm households with the same observed characteristics x may make different choices because of unobserved individual effects. Our assumption is that a farm household chooses to employ hired labor in pest management if the utility difference exceeds a certain threshold level, which can be set to zero without loss of generality. Therefore, we observe yi = 1 (employing hired labor) if and only if yi* > 0 and yi = 0 (not employing hired labor) if otherwise. Thus, we have P ( yi = 1) = P ( yi* > 0 ) = P ( xi' β + ε i > 0 ) = P ( −ε i ≤ xi' β ) = F ( xi' β )

(2)

where F is the cumulative distribution function of ε i . The exact distribution of F depends on the distribution of error term, ε i . The probit model arises from assuming a normal distribution, and a logit model arises from assuming a logistic distribution. Under the standard assumptions about the error term, there is no a priori reason to prefer probit to logit estimation (Greene, 2003). Accordingly, probit and logit models typically yield very similar results in empirical work (Verbeek 2000; Wooldridge 2001; Greene 2003). Considering all these aspects, a probit model was developed to study the factors affecting the likelihood of employing hired labor in pest management. According to the probit model, the probability of a farm household employing hired labor in pest management given the independent variables,

324


such as the operator’s characteristics, household’s characteristics, farm’s characteristics and labor market, P ( yi = 1 xi' ) can be defined as P ( yi = 1) = Φ ( xi' β )

(3)

where Φ is the standard cumulative normal probability distribution or cumulative distribution function (CDF) and xi' β is called the probit score or index (Greene 2003). The probability of not employing hired labor P ( yi = 0 xi' ) is therefore P ( yi = 0 ) = 1 − Φ ( xi' β )

(4)

Following Wooldridge (2001) and Greene (2003), the probit model then can be estimated using the following log-likelihood function:

(

)

n

{

}

ln L β xi' = ∑ ln Φ ( xi' β )  + ln 1 − Φ ( xi' β )  i =1

(5)

Unlike ordinary least squared (OLS) method, the coefficients in a probit model cannot be interpreted directly. One way to interpret the parameters is to look at the derivative of the probability that yi equals one with respect to the j -th element in vector xi . Thus, the marginal affect of a particular variable on the likelihood of employing hired labor in the probit model can be calculated using the following equation: ∂Φ ( xi' β ) ∂xij

= φ ( xi' β ) β j

(6)

where φ (.) denotes the standard normal density function. In addition, probit model estimation can be is misleading when the residuals are not normally distributed and heteroscedastic. According to Greene (2003), those problems can be detected by applying a likelihoodratio test (LR-test) for normality and homoscedasticity. If the LR-test rejects the null hypothesis, this indicates that the residuals are either not normally distributed or heteroscedastic and, hence, need to be corrected. Furthermore, the goodness of fit of a probit model is identified from McFadden’s R2 and predictive quality (Verbeek 2001; Greene 2003).


325

Table 2: Description of variables used in empirical model Variable Dependent variable - HLAB Independent variable Operator characteristics - AGE - EDUC - OFF - SEX - EXP Household characteristics - NUMOFF - ELDER - INFANT Farm characteristics - ORCH - AGETREE - PESTPRESS -MIXED - TASK-i Farm labor market - MART - WAGE

Description =1 if the farm household employed hired labor in pest management Operator’s age (years) Operator’s years of schooling =1 if operator has off-farm employment =1 if operator is male Operator experience in durian or tangerine farming (years) Number of household members having off-farm employment Number of elderly people (more than 60 years old) in household Number of children less than 6 years old Size of fruit orchard in rai Average age of trees on the orchard (years) Operator’s perception of current pest problems, =1 if pest pressure is high =1, if mixed farming; only for durian Number of performed tasks in category i, where i=1,2, and 4 Operator’s perception of hiring casual labor for spraying pesticides, =1 if hiring is easy or very easy Natural logarithm of hired labor wage rate per labor day (in Thai Baht)

4 Results 4.1 Hired farm labor characteristics and wage rate Fruit tree farming in the study locations is primarily a family endeavor, but hired farm laborers make a significant contribution. The survey reveals that 70.70 percent of surveyed durian farm households employed hired farm labor, while the number of tangerine farm households that employed hired farm labor even amounted to more than 90 percent. In durian farming, hired farm laborer is comprised largely of local people living nearby the farm households or from neighboring villages, while in tangerine farming the hired farm laborer is comprised largely of migrant laborer from Burma. These hired farm laborers are employed either as permanent or casual labor. In addition to wages, the farm households usually provide settlements inside their orchards for permanent laborers and their families.

326


Table 3: Characteristics of hired farm laborers in Durian and Tangerine Farming Hired labor characteristics Durian - Age (years) - Formal education (years) - Farming experience (years) - Wage rate (baht/labor day) Tangerine - Age (years) - Formal education (years) - Farming experience (years) - Wage rate (baht/labor day)

Permanent

Casual

35.80 4.92 9.19 167.74

29.97 5.63 7.57 161.60

29.05 n.a 4.06 111.03

n.a n.a 5.52 111.95

t-test for equal means *** ** **

***

Note: n.a. means data is not available. ***,**, mean statistically significant at 1% and 5% levels, respectively.

The characteristics of permanent and casual hired labor employed in durian farming, in term of age, education and farming experience, are statistically significantly different. Permanent hired laborers are older and less educated than casual laborers, but they have greater experience with durian farming than casual laborers. However, a means comparison between wage rate of permanent and casual laborers is not statistically significantly different. It seems likely that the wage rate of farm laborers in durian farming is not determined by the characteristics of farm laborers themselves. The experience of permanent and casual laborers employed in tangerine farming is statistically significantly different, where permanent laborers have less experience than casual laborers. However, both kinds of laborer are paid at a similar wage rate. A means comparison of wage rate between permanent and casual laborers cannot reject the null hypothesis that these laborers are paid at an equal wage rate. The reason is the wage rate of hired farm laborers is set through the bargaining process between farmer and hired laborers. Unless the hired laborers can disclose their skills, the wage rate seems unlikely to be correlated with the characteristics of hired laborers. 4.2 Task attributes and farm labor organization Pest management in durian and tangerine farming involves 10 and 12 tasks, respectively. In accordance with their transaction qualities, these tasks can be grouped into 4 distinct attribute categories6. The attributes of tasks in category 1 and 2 are similar in terms of skills required and monitorability 6

The assignment of the attributes of each task is derived from observation in the field and in discussion with farmers during survey implementation.

327


of performance. The only difference is the frequency of tasks. Similar patterns are also found in the attributes of tasks in category 4 and 5. Detailed information concerning the attributes of pest management tasks in durian and tangerine farming is provided in tables 4 and 5.

Table 4: Task attributes and farm labor organization for pest management in durian farming No

Characteristics

1

Repetitive General skill Monitorable

2

Occasional General skills

3

4

Monitorable Repetitive Specific skills Nonmonitorable Occasional Specific skill Nonmonitorable

Task name

No. farm

Pruning Spraying chemical pesticides Spraying biopesticides Manual weeding Installing insect traps Spraying trees using water jet

137

Pest Monitoring*

Cutting and burning infested branches Separating adjacent fruits Fruit thinning

Conditional probability of farm labor organization: Family labor Hired labor Family and hired only only labor 0.52 0.28 0.20

141

0.54

0.21

0.25

26 118 11

0.65 0.49 1.00

0.19 0.40 0.00

0.15 0.11 0.00

20

0.80

0.05

0.15

149

0.99

0.01

0.00

66

0.83

0.06

0.11

0.97

0.03

0.00

0.93

0.03

0.03

30 29

Note: * means that there is only one farm household that employed permanent labor to perform pest monitoring. Source: own calculation.

Columns 5, 6, and 7 of tables 4 and 5 show the conditional probability of a particular farm labor organization being employed for a particular pest management task. The conditional probability of employing farm labor organization O j , where j = 1, 2,3 , respectively, is estimated in the following cases: (1) farms employ only family labor  Pr ( O1 Ti ) ; (2) farms employ only hired labor  Pr ( O2 Ti )  ; (3) farms employ both family and hired labor  Pr ( O3 Ti )  . Ti is defined as the event where a farm household performs pest management task i . According to Pal (1999), the conditional probability of O j given Ti can be approximated by the relative frequency. Then the conditional probability can generally be defined as: Pr ( O j Ti ) =

n ( O jTi ) n (Ti )

(7)

328


where n ( O jTi ) is the number of joint occurrences of O j and Ti , and n (Ti ) is the number of occurrences of Ti .

Table 5: Task attributes and farm labor organization for pest management in tangerine farming No

Characteristics

1

Repetitive General skill Monitorable

2

Occasional General skills Monitorable

3

Repetitive Specific skills Nonmonitorable

4

Occasional Specific skill

Task name Collecting dropped fruits Collecting snails Pruning Spraying chemical pesticides Spraying biopesticides Manual weeding Banding trunk bottom Installing insect traps Spraying[[??]] trees using water jet Pest monitoring

Cutting and burning infested branches Eradicating severely infested trees

No. farm

Conditional probability of farm labor organization: Family labor Family and Hired labor only only hired labor

26

0.54

0.04

0.42

12 158

0.58 0.33

0.17 0.12

0.25 0.55

158

0.35

0.14

0.51

25 151 2 10

0.60 0.38 0.00 0.90

0.24 0.35 0.00 0.00

0.16 0.27 1.00 0.10

1

1.00

0.00

0.00

152

1.00

0.00

0.00

127

0.72

0.06

0.22

103

0.66

0.12

0.22

Nonmonitorable

Source: own calculation.

In durian farming the probability that farms employ farm labor organization containing only family laborer in pest management tasks is higher than that of employing only hired laborer or that of employing family and hired laborer. A similar finding is also found in tangerine farming, except for pruning and spraying chemical pesticides, where the probability of farms employing both family and hired laborer is higher that that of farms employing only family laborer or that of farms employing only hired laborer. Generally speaking, the probability of farms employing family laborer increases with the skill specificity and monitorability of performance and decreases with the repetitiveness of task performance. Hired farm laborers are not likely to be randomly assigned to perform pest management tasks, but rather in a focused way to the tasks that are attributed to be repetitive, require general skills, and are monitorable. However, their assignments are likely to be under supervision of farm household laborers. This is implicitly supported by the fact that the probability of farms employing family and hired labor is generally higher than that of employing only hired labor, with the exception of manual


329

weeding, where the probability of farms employing only hired labor is just slightly lower than that of employing only family labor. Pest monitoring tasks seems likely to be a farm household’s business, with no hired laborers being assigned to perform them confirming the prediction of the theroretical model (case 7 and 8 in table 1). Farmers select a set of pest management tasks based upon the information they have gathered from monitoring the pest in their orchards. With regard to asymmetric information associated with hired laborers’ skills and performance levels, employing hired farm labor to perform pest monitoring may involve risks, such as yield reduction due to inappropriate pest monitoring, that are most probably beyond the farmers’ acceptance level. Since hired laborers are not the residual claimants, it is more likely that they will not perform as carefully as when the family members do these tasks by themselves. 4.3 Estimated probit model We further estimate the effect of tasks and other variables on the likelihood of employing hired labor in pest management both in durian and tangerine farming. The estimates only include the farms that perform at least one pest management task, excluding pest monitoring, since this task is only performed by family members. We have also ruled out outliers and missing data. In the end, the number of observations includes 150 durian and 159 tangerine farm households. Before estimating the data, we also checked for the presence of a multicollinearity problem by calculating the variance inflation factors (VIF), with the largest value for VIF being 2.29 and 2.43 for durian and tangerine farming models, respectively, indicating that there is no multicollinearity problem. Likelihood Ratio (LR) tests to detect the presence of heteroscedasticity and non-normality distribution of errors were also conducted after estimation. The LR tests for heteroscedasticity of errors in the durian and tangerine models cannot reject the null hypothesis that the errors are homoscedastic at the one percent level of significance. Similarly, the LR tests for non-normality distribution of errors cannot reject the null hypothesis that the errors are normally distributed at the one percent level of significance. A chi-squared statistic of the log-likelihood of durian and tangerine farming indicates that the composite effect of the independent variables differs from zero at the one percent and five percent levels of significance, respectively. The pseudo R2 measure of goodness-of-fit of durian and tangerine farming models is, respectively, 0.187 and 0.197. Furthermore, the predictive quality of durian farming is 74 percent, while the tangerine

330


farming model correctly predicts 77.99 percent of cases. For cross-section data, these values are considered reasonable (Greene 2003). 4.4 Durian farming model The coefficients of most variables are not statistically significant, except for variables TASK 1, AGE and TASK 2 which are statistically significant, respectively, at the one percent, five percent, and 10 percent levels. AGE and TASK 1 have expected positive signs, meaning that an older farm operator is more likely to employ hired farm labor in pest management. Similarly, the probability of employing hired labor is more likely to increase when the farm household is deciding to increase the number of pest management tasks which are “repetitive, do not require specific skills and are monitorable”. TASK 2 unfortunately turns up with a negative sign. The reason for this result may be due to the attribute of the tasks which are occasionally performed and the data we have only captures a small number of farm households applying those tasks. Employing hired farm labor on tasks attributed with low frequency (occasionally performed) may increase transaction costs and hence become a disincentive for the farm household to employ hired farm labor.

Table 6: Probit regression estimates for durian farming (N=150) Variable Coefficient Standard error Marginal effect AGE 0.032 (-0.016) ** 0.013 EDUC 0.062 (0.040) 0.024 SEX 0.475 (0.480) 0.188 OFF -0.297 (0.275) -0.118 EXP -0.004 (0.013) -0.002 NUMOFF -0.034 (0.206) -0.013 INFANT 0.354 (0.328) 0.140 ELDER -0.177 (0.215) -0.070 ORCH 0.032 (0.032) 0.012 ORCH2 -0.000 (0.001) -0.000 AGETREE 0.024 (0.019) 0.010 PESTPRESS 0.102 (0.246) 0.040 MIXED -0.093 (0.355) -0.036 TASK 1 0.493 (0.155) *** 0.194 TASK 2 -0.613 (0.326) * -0.241 TASK 4 0.028 (0.132) 0.011 MART -0.097 (0.336) -0.038 WAGE 2.696 (2.167) 1.062 CONSTANT -18.007 (11.281) Log-likelihood value -84.008 Chi-squared value (df=18) 38.62 *** Pseudo R-squared 0.187 Percent correctly predicted 74.00 Note: standard errors are reported in parentheses; * , **, *** indicate significant at 10%, 5%, 1% levels, respectively.


331

A closer look at the marginal effect of the significant variables shows that the marginal effects of TASK 1 and TASK 2 are considerably higher than that of AGE. This result at least supports what we have hypothesized: that transaction costs matter in determining farm households’ decisions on employing hired farm labor in pest management.

Table 7: Probit regression estimates for tangerine farming (N=159) Variable Coefficient Standard errors Marginal effect AGE 0.011 (0.018) 0.002 EDUC 0.081 (0.052) 0.017 SEX -0.626 (0.767) -0.096 OFF -0.165 (0.295) -0.037 EXP 0.041 (0.119) 0.009 NUMOFF -0.093 (0.147) -0.020 INFANT 0.122 (0.345) 0.026 ELDER 0.240 (0.292) 0.052 ORCH 0.133 (0.048) *** 0.029 ORCH2 -0.001 (0.001) * -0.000 AGETREE 0.002 (0.139) 0.000 PESTPRESS 0.119 (0.365) 0.027 TASK 1 -0.153 (0.197) -0.033 TASK 2 0.152 (0.172) 0.033 TASK 4 0.723 (0.663) 0.156 MART -0.160 (0.281) -0.035 WAGE -7.438 (2.973) ** -1.607 CONSTANT 34.964 (14.153) ** Log-likelihood value -66.282 Chi-squared value 32.48 ** (df=17) Pseudo R-squared 0.197 Percent correctly 77.99 predicted Note: standard errors are reported in parentheses; * , **, *** indicate significant at 10%, 5%, 1% levels, respectively.

4.5 Tangerine farming model Similar to the results of our probit estimation of the durian farming model, the coefficients of most variables in the tangerine model are not statistically significant, except for variable orchard size (ORCH and ORCH2) and wage rate of hired labor (WAGE). The coefficients of variable ORCH and ORCH2 are statistically significant at the one percent and 10 percent levels and have the expected positive and negative signs, respectively. This means that the likelihood of employing hired farm labor increases with the size of a tangerine orchard and then, due to the presence of transaction costs, decreases when the size of a tangerine orchard is larger than the optimal size of employing hired labor. This result confirms what has been found in

332


other studies (e.g. Frisvold 1994; Evenson et al. 2000). The marginal effect of ORCH on the likelihood of employing hired farm labor is 2.9 percent. The variable WAGE is statistically significant at the five percent level and has the expected negative sign, indicating that the increase in wage rate of hired labor is likely to reduce the likelihood of employing hired farm labor in pest management tasks. If the wage increases by one percent the likelihood of employing hired farm labor decreases by 1.6 Percent. Probit estimation of the tangerine model cannot provide statistical evidence supporting the hypothesis that the attributes of tasks matter with regard to farm households’ decisions about employing hired labor. The reason may be that the pest management tasks applied by farm households that employed hired farm laborer are not statistically significantly different from those that did not employ any.

5 Conclusion Using a transaction cost economic framework, this paper has empirically tested the effect of pest management tasks on the likelihood of hired farm labor employment in durian and tangerine farming in Thailand. We have argued that hired farm laborers are not randomly assigned to perform pest management tasks, but rather to the tasks that are easier to monitor, are applied repetitively, and do not require specific skills. Accordingly, the main hypothesis we derived is that pest management tasks labeled as part of an IPM strategy are most likely to be performed under farm labor organization comprised largely of family, rather than hired, labor. The study has generated the following results: (1) The differences between permanent and casual hired laborers are more obvious in durian farming than in tangerine farming. In general, permanent laborers in durian farming are older, more experienced, and less formally educated. But, the wage rate is not statistically different. In tangerine farming, the difference between permanent and casual laborers centers on tangerine farming experience, which casual laborers generally have more of than permanent ones do. Similar to the result found in durian farming, the wage rate between permanent and casual labor in tangerine farming is not statistically different. (2) The conditional probability of the form of farm labor organization shows that (i) pest monitoring seems likely to be a family labor activity, (ii) pest management tasks labeled as part of an IPM strategy are more likely to be performed under farm labor organization comprised largely of family labor, and (iii) the use of hired labor seems likely to be under supervision of family labor. These results confirm the hypotheses derived from transaction costs economics and the theoretical framework developed by Beckmann and Wesseler (2003). (3) Estimates of a probit model of durian farming furthermore confirm that the attributes of pest


333

management constitute the most important factor determining the likelihood of employing hired farm labor in pest management. Unfortunately, due to lower variation of pest management tasks applied by tangerine farm households, the estimated probit model of tangerine farming cannot provide statistical evidence similar to the durian farming model.

References Bautista, E.D. (1993). The Impact of Technical Change on Rural Labor Markets in the Philippines. Journal of Philippine Development 20(1): 31-53 Beckmann, V., Wesseler, J. (2003). How Labour Organization May Affect Technological Adoption: An Analytical Framework Analysing the Case of Integrated Pest Management. Environment and Development Economics 8:114. Ellis, F. (1993). Peasant Economics: Farm Households in Agrarian Development 2nd Edition. Cambridge, UK: Cambridge University Press Evenson, R.E., Kimhi, A., DeSilva, S. (2000). Supervision and Transaction Costs: Evidence from Rice Farms in Bicol, The Philippines. Discussion Paper No.814. Economic Growth Center, Yale University, New Haven, Connecticut, USA. Firsvold, G.B. (1994) Does Supervision Matter? Some Hypothesis Test Using Indian Farm Level Data. Journal of Development Economics 43: 217-238. Greene, W. (2003). Econometric analysis 5th Edition. Prentice Hall. de Janvry, A., Sadoulet, E., Fafchamps, M., (1989), Agrarian Structure, Technological Innovations and the State. In: Bardhan, P.K. (ed.), The Economic Theory of Agrarian Institutions. Oxford: Oxford University Press Kikuchi, M., Hayami, Y. (1999). Technology, Market, and Community in Contract Choice: Rice Harvesting in the Philippines. Economic Development and Cultural Change 47 (2): 371-386. Lepak, D.P., Snell, S.A.. (1999). The Human Resource Architecture: Toward a Theory of Human Capital Allocation and Development. Academy of Management Journal, 24: 31-48. Lichtenberg, E., Berlind, A.V. (2001). Does it Matter Who Scouts? Working Paper No. 01-11, Department of Agricultural and Resources Economics. The University of Maryland, College Park. Pollak, R. (1985). A Transaction Cost Approach to Families and Households. Journal of Economic Literature 23: 581-608. Roumasset, J.R., Smith, J. (1981). Population, Technological Change, and the Evolution of Labor Markets. Population and Development Review 7(3): 401-419. Schmitt, G. (1991). Why is Agriculture of Advanced Western Economies still Organized by Family Farms and Will this be also in the Future? European Review of Agricultural Economics 18: 443-458. Verbeek, M. (2000). A Guide to Modern Econometrics. John Wiley & Sons Ltd. Vernimmen, T., Verbeke, W., Van Huylenbroeck, G. (2000). Transaction Cost Analysis of Outsourcing Farm Administration by Belgian Farmers, European Review of Agricultural Economics 27: 325–345.

334


Vongsaroj, P. (2007). Economics of IPM. In: Kunstadter (ed.). (2007). Pesticides in Southeast Asia: Environmental, Biomedical, and Economic Uses and Effects. Chiang Mai: Silkworm Books.

Appendix Table A1: Means of variables for durian farming model Variable

All farm With hired Without hired t-test for equal households labor labor means N=150 N=82 N=68 0.55 HLAB (0.50) 51.95 52.33 51.5 -0.58 AGE (11.14) (11.96) (10.18) 7.03 7.59 6.37 -1.59 EDUC (3.71) (3.99) (3.23) 0.93 0.95 0.91 -1.11 SEX (0.25) (0.22) (0.29) 0.24 0.22 0.26 0.40 OFF (0.43) (0.42) (0.44) 24.40 24.59 24.18 -0.34 EXP (10.77) (10.99) (10.56) 0.32 0.28 0.37 0.87 NUMOFF (0.62) (0.50) (0.73) 0.14 0.16 0.12 -0.56 INFANT (0.40) (0.46) (0.32) 0.49 0.48 0.51 0.51 ELDER (0.70) (0.69) (0.72) 17.52 21.15 13.15 -3.35 *** ORCH (14.74) (16.47) (10.95) 15.55 16.35 14.59 -0.54 AGETREE (6.87) (7.08) (6.53) 0.60 0.60 0.60 -0.10 PESTPRESS (0.49) (0.49) (0.49) 0.86 0.87 0.85 -0.29 MIXED (0.35) (0.34) (0.36) 2.79 2.95 2.60 -3.68 *** TASK 1 (0.82) (0.72) (0.90) 0.21 0.13 0.29 1.84 * TASK 2 (0.47) (0.34) (0.57) 0.83 0.79 0.87 0.01 TASK 4 (1.01) (0.90) (1.14) 0.14 0.13 0.15 0.29 MART (0.35) (0.34) (0.36) 5.09 5.09 5.08 -1.78 * WAGE (0.06) (0.06) (0.07) Note: standard errors are reported in parentheses; * , **, *** indicate significance at 10%, 5%, 1% levels, respectively.


335

Table A2: Means of variables for tangerine farming model Variable

All farm households

With hired labor

Without hired labor

t-test for equal means

N=160 N=125 N=35 3.49 HLAB (2.17) 49.87 49.71 50.43 0.38 AGE (10.39) (10.67) (9.47) 5.26 5.46 4.53 -1.78 * EDUC (3.27) (3.43) (2.48) 0.96 0.96 0.97 0.34 SEX (0.19) (0.20) (0.17) 0.31 0.30 0.34 0.43 OFF (0.46) (0.46) (0.48) 4.21 4.35 3.73 -2.11 ** EXP (1.69) (1.73) (1.48) 0.68 0.64 0.80 0.91 NUMOFF (0.87) (0.86) (0.93) 0.17 0.17 0.20 0.36 INFANT (0.43) (0.42) (0.47) 0.34 0.36 0.26 -0.92 ELDER (0.64) (0.66) (0.56) 8.64 9.87 4.24 -4.81 *** ORCH (10.40) (11.34) (3.46) 4.29 4.40 3.89 -1.97 * AGETREE (1.40) (1.40) (1.34) PESTPRES 0.87 0.88 0.83 -0.73 S (0.34) (0.33) (0.38) 3.31 3.30 3.37 0.56 TASK 1 (0.65) (0.62) (0.73) 1.44 1.46 1.34 -0.83 TASK 2 (0.75) (0.75) (0.76) 0.08 0.10 0.03 -1.66 TASK 4 (0.30) (0.32) (0.17) 0.46 0.44 0.54 1.07 MART (0.50) (0.50) (0.51) 4.72 4.71 4.73 2.54 ** WAGE (0.05) (0.05) (0.04) Note: standard errors are reported in parentheses; * , **, *** indicate significance at 10%, 5%, 1% levels, respectively

Part VI Natural Resource Endowment and Trade

Chapter 17 The Feasibility of Founding OREC and China’s Countermeasures Zhonghui Wang1 and Yujie Chan2 Abstract: Along with the price of foodstuff hiking in 2008, the Organization of the Rice Exporting Countries (OREC) was initiated by Thailand’s Government. In this paper, the authors have first analyzed the causes, feasibility and barriers of founding OREC. Then the authors have examined the possible impacts of the establishment of OREC on China. Finally, the authors have provided some countermeasures for China to deal with this issue. The authors have argued that the really important consideration for China with regard to the establishment of OREC should be food Security although the establishment of OREC might be good for developing China’s agriculture and increasing the income of China’s peasants. Keywords: OREC, Feasibility, Barriers, Impacts, Countermeasures

1 Introduction At the beginning of May 2008，Thailand had approved in principle the establishment of the Organization of the Rice Exporting Countries (OREC) together with Myanmar，Laos，Vietnam and Cambodia. They promised to negotiate rice prices in the international market, so as to acquire pricing rights. In this way, they could regulate the supply of rice, deal with food crises, and, finally, protect their own interests. However, only five days later, former Thailand foreign minister Nuobadun announced that Thailand’s government had abandoned the founding of OREC. With skyrocketing rice prices and intense food crisis in 2008, more and more people have been discussing and attaching great importance to whether it is necessary to establish OREC and whether it can solve the world food crisis after its establishment. In this vein, here we first analyze the causes of the establishment of OREC，including a look at the history and current situation of the Organization of the Petroleum Exporting Countries 1

2

WTO Research Centre, Nanjing University of Finance and Economics, Nanjing, China Shanghai Subsidiary Company, China Grain Reserves Corporation, Shanghai, China

Beckmann, V., D.H. Nguyen, X. Shi, M. Spoor, J. Wesseler (eds.) (2010). Economic Transition and Natural Resource Management in East- and Southeast Asia. Aachen: Shaker Publisher, pp. 339-350.

340

Wang and Chan

(OPEC), also presenting the features of the world rice market and the causes of the food crisis in 2008. Then, we examine the feasibility and barriers of founding OREC. Finally, we analyze the possible impacts of the establishment of OREC on China and China’s countermeasures to deal with this issue.

2 The reasons for creating OREC In the first half year of 2008, the whole world was facing a serious crisis over the price of grain, though people had a little hope that an organization would be established to solve the crisis. However, it is still unknown whether such an organization would actually release the pressure of rising grain prices and the demand for grain, while giving help to the declining world economy. The purpose of establishing OREC is to adjust the price of grain by controlling the supply-demand relations. 2.1 The experience of OPEC The first international organization of broad scale that has been trying to control commodity prices through collective behavior is the Organization of the Petroleum Exporting Countries (OPEC). It is an organization established by some developing petrol-producing countries to coordinate petrol policy among member countries in order to protect the economic right and interests of their nations. It was founded in September 1960 in Baghdad by Iraq, Saudi Arabia, Iran, Kuwait and Venezuela. Later, Qatar, Indonesia, Libya, The United Arab Emirates, Algeria, Nigeria, Gabon, Ecuador (which withdrew in 1992) among others have joined gradually. Its general headquarter is in Vienna, Austria. The purposes of OPEC are to coordinate policy among member countries, to take collective action to negotiate with foreign petrol companies. OPEC aims to make sure of the stability of prices in the international petrol market, attempting to generate conditions for stable income under any conditions for member countries by eliminating the dangerous and unnecessary fluctuation of petrol prices. At the same time, OPEC aims to provide enough perennial petrol supply. This target was mainly realized in the first ten years after the establishment of OPEC. However, with the expansion of scale and the increasing proportion of controlled petrol resources, OPEC’s obligation to adjust the price of world petrol has become heavier as it plays the role of lever of the world economy. When the rising price of petrol hampers the development of the world economy, OPEC has the obligation to increase the output of petrol to balance the price, though declining petrol prices shrink the profits of petrol-producing countries. When there is too much petrol for world demand, OPEC has the obligation to reduce output to keep

The Feasibility of Founding OREC and China’s Countermeasures

341

the stability of petrol prices. As long as petrol prices are out the range regulated by OPEC, the member countries will bear countless losses without getting any compensation. Thus, although OPEC has had a certain influence on keeping the stability of the world petrol market and guaranteeing profits for the member countries over the past 40 years, it still has many problems. 2.2 The current situation of the international rice market 2.2.1 The main structures and features of the international rice market. As one of the chief crops throughout the world, rice feeds about half of the world’s population. In recent years, however, the size of the rice trade has decreased because of a high degree of monopolization, government intervention and intense price fluctuation in the international rice market. Meanwhile, international rice production is concentrated in certain areas: China, South Asia and Southeast Asia are the top three regions, where almost 90% of the world’s rice is grown.3 Many residents of these regions are extremely dependent on rice because of the soil and the climate conditions where they live. For most developing countries, the governments interfere with the rice market mainly to keep abundant rice supply and low consumption prices, while in some developed Asian countries or regions one of the main goals for the governments is to protect the interests of domestic rice producers. The international rice exporting countries are also concentrated, while the rice importing countries are much more dispersed. Thailand, Vietnam, Pakistan, the US, Australia and China are the top six rice exporting countries: altogether, their rice production accounts for 60% of the world’s export. The main importing countries are Bengal, Indonesia，Ivory Coast, Nigeria，Iraq，Iran，Saudi Arabia，Brazil，the Philippines, Japan, Malaysia, Cuba, North Korea, and so on. However, the amount of these countries’ rice importation added together only accounts for 40% of the world’s total rice import in recent years.4 2.3 The world food crisis and its causes For a long period of time, world food prices have remained relatively stable. However, beginning January 2008, food prices in Southeast Asia rose sharply. In two months, the price of rice rose from 600 US dollars per ton to 900 US dollars per ton. Moreover, the rising price wildly spread 3

4

Wang, Mingli.2006.The Import and Export Situation of China’s Rice and the Outlook of Future’s Trend. Agricultural Outlook 4(7): p.3. ibid.

342

Wang and Chan

from North America to the huge area of Africa. The rising price of rice deteriorated the situation around the world, affecting the world economy. At the same time, skyrocketing rice prices brought about new problems concerning political and social stability in the rice importing countries. Some supermarkets in developed countries like USA and the UK have even restricted the amount of rice that can be purchased. The Philippines, with a population of about 90 million, was the first country affected by the food crisis and had to meet its domestic rice demand by importing rice from other countries. Although its weather is suitable for growing crops, the Philippines neglected agricultural production for a long period of time, so the quantity of rice imported increased from 0.72 million tons in 1997 to 1.87 million tons in 2007. In 2007, the GDP growth rate of the Philippines reached a record high of 7.3% for the first time over the past 30 years. In order to develop the economy by reinforcing the export of industrial products, the supports for agriculture from the Philippines government have been declining. Yet rice consumption in the country has increased drastically along with the high growth of population: 12.05 million tons in 2007, compared with 10.59 million tons in 2005.5 The increase of local rice demand combined with insufficient local rice supply has forced the Philippines to increase its rice import. China is the biggest rice growing country in the world, but it can only basically be self-sufficient. Although India is the second largest rice growing country in the world, its rice exports only rank fifth or sixth in the world. The Philippines can only mainly import rice from Thailand. However, Thailand began to raise their rice price, and other countries followed suit. At the same time, some countries curtailed their rice export volume. As a result, the food crisis of Southeast Asia took place and, since then, the international food price has increased rapidly. The reasons why world rice prices have increased dramatically can be analyzed as follows: First, most governments of the world have turned their attention to industry in recent years. As a result, less capital was devoted to agriculture, which led to the decline of grain production. Second, from a global point of view, with the rise of world oil prices, many countries have been trying to replace petrol with organic oil extracted from crops, which increased the demand for rice. Thirdly, the floating capital in international capital markets hedged rice. This also led to a dramatic rise of rice prices. Fourthly, besides the man-made causes, there were some natural disasters which can not be neglected. For instance, a hurricane hit Myanmar in April 2008 that half of the country’s population suffered from. Myanmar is one of the large rice growing and exporting countries of the world. As a result 5

The Reality and Future of China’s Food Security, http://www.foodqs.com/ news/gnspzs01/2008423101011544-5.htm.


343

of the hurricane, Myanmar was not only unable to export rice to the international market, but also had to seek international aid for rice. Moreover, in the same year a plague of insects took place in Vietnam, the world’s second biggest rice exporter, and a huge snowstorm occurred in China.

3 The feasibility of the establishment of OREC 3.1 The five countries planning on joining OREC are the prime producers and exporters of rice The establishment of OREC is somehow feasible. Asian production and consumption of rice accounts for 90% of the world’s rice production and consumption, and nine Asian countries, including China, India, Indonesia, Bangladesh, Vietnam, Thailand, Myanmar, the Philippines, Japan, are in the list of the world’s top ten rice producers.6 Of the five countries that put forward the proposal for the establishment of OREC, Thailand, Vietnam and Myanmar take the first, second and sixth places for rice exporting countries, respectively, while Laos and Cambodia are also important rice exporting countries. In 2006, the total output of rice production in Thailand was 29.65 million tons. Its rice export is between 7 and 10 million tons, which accounts for 25% to 35% of the total rice trade in the world, making it the number one rice exporting country.7 So, the combined rice production of the five countries in the union could control 90% of the world’s rice export market. From this point of view, OREC is thought to be much more powerful than OPEC, because even at its strongest OPEC has only controlled 53% of the world’s petrol exporting.8 3.2 The five countries are for the establishment of OREC The leaders of the five countries had expressed that they supported the establishment of OREC. For example, the Cambodian government believed that such an organization is necessary and said that it would help Cambodia to avoid a price war and promote the exchange of rice information by founding OREC. Meanwhile it can be used to fight against OPEC. Cambodia even thought that the world food crisis would bring a chance for its farmers and appealed to them to grow more foodstuffs, including rice, as 6

7

8

Wu, Jinpu. 2004. General Introduction of World’s Rice, Cereal & Feed Industry, No.10, p.12. 2006’s Situation and 2007’s Forcast of Rice Production and Export in Thailand http://www.21food.cn/html/news/35/153055.htm Bai, Yongxia. 2008. OREC Confronting the Pressure of Failure. www. www.thebeijingnews.com, May 6. .

344

Wang and Chan

soon as possible. The government thought that 80% of the local farmers would benefit from the world food crisis.9 The spokesman for Laos’s diplomacy department announced that the government would consider seriously the proposal of establishing OREC, as put forward by Thailand, and thought such an organization would endow the five countries with negotiation power. Myanmar officials said that they were researching the proposal and it was possible that they would join in. Although Vietnam withheld its decision at first, claiming there was no response to establishment of OREC and denying that establishment of OREC could be expected soon, its government also approved joining OREC a few days later.10

4 The barriers to establishing OREC 4.1 Because of the world food crisis, many countries could not agree on the establishment of OREC. In the first half of 2008, the global rice price doubled. This rapid rise of price had led Vietnam, India, Egypt and Cambodia to restrict rice exportation to insure domestic supply. The other main export countries also started to curtail rice export. As a result, the food shortage became a serious problem for many governments in 2008, because of which social unrest occurred in Haiti and some African counties. The food shortage has become a worldwide crisis, and high food prices may threaten the achievement of eliminating poverty. Soaring food prices could lead 0.1 billion people around the world into hunger and threaten the stability of the international food market. Because of this situation, the OREC proposal which was put forward by Thailand’s Government was doomed to be rejected by other countries. 4.2 Fierce opposition to the establishment of OREC have been from rice importing countries. OREC would, first and foremost, hurt the interests of neighboring countries in Asia, because rice consumption there accounts for 90% of the world consumption. Although the members of OPEC are countries which also import rice, the amount of rice import is no more than one million tons per country. On the contrary, some countries in Southeast Asia have expanded their rice importation in recent years. As a result, some Southeast Asian 9

10

Fu, Yunwei. 2008. Rice’s OPEC :to Protect or to Retaliate? www. Xinhuanet.com. May 5. Wang, Lu. 2008. Thailand Transferring from OREC to CRTC for A New Rice Union. http://www.sina.com.cn. March 8.


345

countries would take every means to object to the establishment of OREC in order to achieve the objective of protecting their own interests. The Philippines, as the first opposed the proposal, thought that OREC would bring more serious problems instead of solving the world food crisis, because the organization is on the opposite side as the poor and it would only worsen the current situation instead of solving famine and poverty. If food is manipulated by a few rice growing countries, the consumers would not be able to afford to buy rice anymore. At the same time, some rice importing countries hold the opinion that OREC will be unable to control the rising of rice prices and, instead, may facilitate the existence of a rice monopoly which would become the dominant factor for rising prices. 4.3 The power of OREC would be weak after its establishment Another barrier to establishing OREC is that its actual power to control the world rice market would be weak after its establishment, for the following reasons: First, OREC would hardly affect the world like OPEC as rice production in the countries of OREC is decentralized. The reason is that rice production is part of a small scale peasant economy, and the governments can hardly intervene into the rice market efficiently, because all decisions such as when, how many and how much to sell are made by millions of dispersed farmers. In addition, the yield of crops is easily affected by natural conditions. Secondly, the attitudes of member countries to establish the union are inconsistent. OREC, as advanced by Thailand, would not include the second major rice-exporting country, India, so it is out of question that America as the fourth or other large rice yielding countries such as China and Pakistan could be included. So OREC is not as representative as OPEC. Furthermore, the five countries which are ready to establish OREC have varied in political standpoint, political structure and economic policy, which might increase the difficulty of establishing OREC. In addition, the five countries in OREC have not cooperated so well together. For example, Vietnam which exports rice of low quality showed itself to be half-hearted about Thailand’s proposal and has not accepted it completely, because of its inferior rice quality Vietnam’s competitive ability would be weakened, as people would be willing to buy rice of higher quality if prices are the same. Thirdly, interference towards the establishment of OREC has also been coming from other organizations. Many countries in the world could not bear that the international rice price is dominated by a few countries because of the extraordinary strategic importance of rice. The controlling

346

Wang and Chan

power of OPEC is manifested not only in its production and market dominance, but also in its close connections with many syndicates of the main big economies (the US and Europe). As the power to control rice prices that would be assumed by OREC conflicts with the profits of the most wealthy and strong countries, the creation of OREC has been strongly criticized. Although Japan’s rice production is not very big, 85% of its 2.3 million farmers are engaged in rice planting. In order to protect these farmers’ interests, the Japanese government has gotten into disputes with the US government quite often. It is thus apparent that Japan is unlikely to agree to the establishment of OREC.

5 Impacts of the establishment of OREC on China and China’s countermeasures 5.1 Impacts of the establishment of OREC on China China is the largest rice growing country and consumes the most rice in the world. It has a long history of growing rice and is one of important rice exporting countries. Although the share of China’s rice export is not more than 20% of the world export market, it accounts for 23% of world rice planting area and 37% of total world rice production output.11 Nevertheless, the share of China’s total rice export is no more than 1 percent of its total output. And China has sufficient rice reserves. As a result, when the world price skyrockets, China can still maintain the stability of the rice price, and it would not be difficult for China to find necessary countermeasures if it were to think that the world rice price is being controlled in a way that would endanger its own interest. China’s rice exports are higher than imports in most years, but the quantity of rice export is not so great. As can be seen from Table 1, rice exporting was at its zenith in 1998 and then declined. In 1998 China exported the largest quantity of rice in history, about 3.74 million tons, which accounted for 19 percent of the world’s total rice export value (924.03 million US dollars), generating 803.99 million US dollars of trade surplus for that year. During the last decade, the trade value of rice exports has been more than for rice imports, except for a 15.18 million US dollar trade deficit in 2004. In 2005, the quantity and value of China’s rice exports declined, dropping to 0.69 million tons with a value of 232.34 million US dollars. Since 2005, however, China’s rice exports have increased again.

11

Bai, Yongxia.ibid.

347


According to research12, the total amount of the world’s rice production is about 415 million tons and the total consumption is 417 million tons. Since 2000, the quantity of world rice trading has risen from 24 million tons to about 29.4 million tons in 2006, which amounted to only 23 percent of China’s own rice consumption. So the total amount of world rice trading was less than a fourth of China’s rice consumption in a year. Moreover, the world’s rice storage has declined consecutively for many years, standing in 2006 at only 42.38 million tons (excluding China), which was about one third of China’s rice consumption. Therefore, it is difficult for China to rely on international rice markets for food security. Fortunately for China, it has had good rice harvests in recent years, so the establishment of OREC has limited impact on it.

Table 1: The quantity and value of China’s rice imports and exports since 1997 Money Unit: Millions of U. S. Dollars Year Import of rice Quantity

Quantity Unit: Millions Tons Export of rice

Money amount

Quantity

1997 0.33 139.77 0.94 1998 0.24 120.04 3.74 1999 0.17 78.15 2.71 2000 0.24 112.71 2.95 2001 0.27 98.85 1.86 2002 0.24 80.31 1.99 2003 0.26 97.18 2.62 2004 0.77 254.64 0.91 2005 0.52 199.45 0.69 2006 0.73 293.68 1.24 2007 0.49 228.04 1.34 Source: China’s Statistical Yearbook 1998 to 2008.

Money amount 264.57 924.03 652.02 561.05 329.01 385.01 501.75 239.46 232.34 409.68 487.87

The quantity of China’s imported rice is not very much, and it only imports rice from a few countries, mainly from Thailand (90%). If OREC is established, a monopoly over rice prices might be formed and the price of imported rice would rise. Consequently, the costs of China’s rice imports might go up. However, since China’s rice exports were more than its imports over the last decade (except in 2004), the establishment of OREC might actually turn out to be good for China’s economy. Moreover, after OREC is established, future rice prices in the international market could be 12

“Rice Plays a Very Important Role in http://www.bdhnyck.com/shownews.asp?id=348

National

Food

Security”,

348

Wang and Chan

more stable. Thus, the establishment of OREC could be good for developing China’s agriculture and increasing the income of China’s peasants. Nevertheless, compared to China’s total goods trade surplus, the trade surplus from rice export is very small. In 2007, China’s total trade surplus was 261,830 million US dollars, but China’s trade surplus from rice was only 259.83 million US dollars, less than 0.1 percent of the total trade surplus. Therefore, from China’s perspective, the really important consideration with regard to the establishment of OREC should be food security, rather than the trade surplus from China’s rice exports. 5.2 China’s countermeasures against the founding of OREC Whether OREC should be established is still an important issue deserving profound research. There are many factors which should be considered here, such as which countries can join OREC, how to control the world rice prices after its establishment and so on. At the present time, the establishment of OREC does not seem to be a good idea for solving the world food crisis. The urgent issues for China today are to increase food output, promote rice productivity, strengthen capacities for dealing with natural disasters and increasing investment in agriculture. Rice is certainly the most sensitive grain in China. In recent years, its rice supply has basically been kept balanced and rice prices have been comparatively stable, but the gap between China’s rice output and consumption is increasing continuously. Thus, China has had to use its rice reserve to meet demand. At present, China’s reserve is about 40 to 50 million tons.13 Because of this, China avoided the food crisis of 2008. But if China continues only to rely on its rice to remedy the gap, a food crisis may emerge one day in China as well. China’s food security has to rely on the whole country’s grain productivity and farmers’ incentives for farming. Moreover, rice growing requires good conditions in terms of water, soil and climate as well as the infrastructure of irrigation works. China has to take some measures to deal with these problems and needs better management of its natural resources. First, China has to increase investment in agriculture. Although China’s investment in agriculture is increasing, it still falls relatively behind the other two industrial sectors. Thus it is necessary to increase the investment in agriculture. Meanwhile, the development of high quality rice is an important way to improve its productive competitiveness. Secondly, China has to develop diversified and standardized rice for export. The international market needs diversified grades of rice. But China 13

The Reality and Future of China’s Food Security, www.foodqs.com. April 23, 2008.


349

still lacks rice of high quality and diversified into different grades. The classification of China’s rice is not closely connected with the demand of the international rice market. Thus, China’s rice can only compete with other countries in the international rice market at the middle or poor quality grade. It is still hard for China’s rice to compete with other countries in the high quality rice market. Therefore, in order to develop rice for export, China must diversify production. No matter whether for the domestic market or foreign markets, China has to improve rice quality and the level of rice processing so that China’s rice will meet the demands for rice diversification in the international market with high-quality, more types and a greater spectrum of grades. Thirdly, China should strengthen its export of rice to the existing market, while also developing export of rice to new markets. At present, countries in Africa and Southeast Asia are the main markets for China’s rice export. However, the quantity of China’s rice export to the Middle East and Europe has been increasing in recent years. Thus, China has to work on continuing to win over new rice export countries which heavily depend on the international rice market, such as the Middle East and Europe. Fourthly, China should maintain a sufficient area for rice planting and strengthen capacities to prevent natural disasters. According to water and climate conditions, China’s Northern Region is not suitable for expanding rice-planting area blindly, especially in the Northeast Region where water is not sufficient. Thus, China’s Northern Region could even have some of its rice-planting area reduced. Meanwhile, China’s Southern Region could increase acreage for growing rice to insure sufficient area for rice planting in the whole country. Natural disasters have frequently occurred in recent years in China. Thus, China should develop capacities for preventing natural disasters, so that it can secure a sufficient supply of rice for itself.

References: Chen, Y. (2003). The Trade Characters of International Rice and China’s Rice Export.China Rice 15(1): 7-10. Cao, Y. (2007). The Import and Export of China’s Food and the Strategic Choice of Grain’s Security. Journal of the Party School of the Taiyuan s Committee of the C.P.C. 10(4): 46-48. Li, T., Wang, Z. (2007). An Analysis of OPEC’s Influence on International Oil Price. Northeast Asia Forum 16(3): 32-35. Lou, C. (2001). OPEC in 21st Century and Its Role in the World s Oil Market. Modern Petroleum & Chemical Industry 9(12): 17-19. Niu, Q., Jia, S. (2006). Analysis of OPEC’s Monopoly in Crude Oil Market. Petroleum Science 3(4): 26-30.

350

Wang and Chan

Wang, M. (2006). The Current Situation of China’s Rice Import and Export and the Outlook of the Future’s Trend. Agricultural Outlook 2(7): 3-7. Wang, Z., Xu, J., Wang, H. (2006). A Game Theory Analysis of OPEC’s Influence on World Oil Price. Petroleum Science 4(3):21-25. Wu, J. (2004). A General View of World Rice. Cereal & Feed Industry 30(10): 12-15. Wu, Q., Zhou, K. (2007). The Analysis of the Competition Situation for Chinese Paddy Output. Chinese Agricultural Science Bulletin 23(12): 447-450. Yu, J. (2006). The Evolution of OPEC and Its Challenges in the New Century. Arab World Studies 107(6): 20-26. Zhu, M., Chen, W. (2002). The Trade Situation of World Rice and China’s Countermeasure. World Agriculture, 29(5): 32-35.

Chapter 18 Competitiveness Analysis for Major Agricultural Product in the Mekong Delta: The Case of Tien Giang Province Nguyen Trong Hoai1 Abstract: This paper considers the international competitiveness of agricultural production in the Mekong River Delta using the case of Tien Giang Province. Competitiveness was measured in terms of domestic resource cost (DRC) ratios for three commodities (rice, citrus and shrimp) at the farm level. The results indicate that, for the period of 2004, rice and shrimp have been potentially competitive, while citrus has been somewhat uncompetitive. To strengthen the competitiveness analysis, focus-group interviews of farmers and an expert survey were carried out. The results suggest that, to sustain competitiveness in the context of regional integration with a world trade practice, information on market prices and trends and on technology should be disseminated with greater effort. Keywords: Mekong river delta, Domestic resource cost, Competitiveness analysis, Focus-group interview

1 Introduction In Vietnam, during the process of transition from central planning to market-orientation, the contribution of agriculture to GDP declined from about 40% in the late 1980s to 20% in 2004, and the share of agricultural exports also fell from 60% to 30% (World Bank 2005). However, with about 75% of the population living in rural area and about two-thirds of the labour force participating in agricultural production, agriculture should receive greater concern from policy-makers, since the development of agriculture also means improving living standards for people in the nation. During 1998 – 2002, the growth rate of agriculture stayed at about 4% per year which is considered to be high by international standards (World Bank 2005). Until recently, rice, fruit, shrimp and fish have played a very important role in exports, accounting for about 40% of the exported value of agriculture (World Bank 2005). The process of trade liberalization and the country’s accession to AFTA and WTO are offering the agricultural sector new opportunities as well as imposing challenges. As a low-income 1

University of Economics, HoChiMinh City, Vietnam


352

Hoai

country, together with the instability of the world market for agricultural products, the competitiveness of agricultural products in both the domestic and international markets has been a critical problem for the agricultural sector. In rice, fruit, shrimp and fish production, the Mekong River Delta (MRD) is always considered to be the biggest supplier for both the domestic and international markets, but poverty is still a pressing problem in this region; although the number of people in poverty has declined significantly, this trend has been unstable. MRD is the region where the highest percentage of people easily fall back into poverty when there are abrupt changes in economic conditions (AusAID 2004). In early August 2006, a forum was held in Can Tho city addressing the potential for development of agriculture in MRD, and experts all agreed that MRD possesses potential for agricultural development, but the competitiveness of its major agricultural products is still a serious concern (Nguyen Kiem 2006). Hence, it is necessary to study the competitiveness of major agricultural products in the MRD to more deeply understand the strengths and weaknesses of these products in the world market, especially now that MRD will be facing a stronger level of competition since Vietnam officially agreed to become a WTO member at the end of 2006.

2 Literature review 2.1 Competitiveness theory: from comparative advantage to competitive advantage Competitiveness has received significant attention from researchers since the 1980s, but due to its complexity, until now there has not yet been a concept of competitiveness that is universally accepted. This section attempts to briefly introduce the basics of competitiveness theory, from classical to modern approaches. 2.1.1 Traditional trade theories Traditional trade theories evolved from absolute and comparative advantage developed by Adam Smith and David Ricardo to the HeckscherOhlin model. Although no single theory can adequately explain the competitiveness of countries in international trade nowadays due to the complexities involved, the traditional theories have set up a foundation for modern theories which attempt to better explain competitiveness in a complicated world.

Competitiveness Analysis in the Mekong Deltat

353

2.1.2 Competitiveness theories Since the end of the 1980s, researchers have been trying to develop a concept of competitiveness, but they have not yet reached an agreement on either a concept or its measurement. But there is a mainstream view, led by Michael Porter, that dominates some centers of decision-making in the United States and Europe. In his book The Competitive Advantage of Nations, Porter sets the foundations of a theory on competitiveness (Coy 2006). Porter believes that a country’s prosperity does not grow from natural endowments, as traditional trade theories thought, but rather that prosperity is created, not inherited, and a country’s competitiveness depends on its industries’ capacity to innovate and upgrade. Through acts of innovation, companies achieve competitive advantage (Esterhuizen 2006). However, like other theories, Porter’s model has been criticized by other authors which have been reported by Esterhuizen(2006). An extension of Porter’s diamond model is the nine-factor model of Cho, who criticizes Porter’s model for its inadequacy for developing countries like Korea. Cho divides the sources of international competitiveness into two broad categories named “physical” and “human” factors. The physical factors include endowed resources, the business environment, related and supporting industries and domestic demand. The human factors include workers, politicians and bureaucrats, entrepreneurs and professional managers and engineers (Esterhuizen 2006). 2.2 Measurements of competitiveness in agriculture To analyze potential rather than revealed competitiveness, the most common measure is Domestic Resource Cost (DRC). The DRC for a certain commodity compares the opportunity cost of domestic resources used in production for that commodity to the value-added it generates at international prices: n

∑ aijV j DRCi =

j = k +1 k

(1)

Pi − ∑ aij Pj j =1

where aij (j = k +1 to n) are the technical coefficients for domestic resources and non-tradable inputs;

354

Hoai

Vj is the shadow prices of domestic resources and non-tradable inputs; Pj is the border price of traded output; and aij (j=1 to k) are the technical coefficients for traded inputs and the border price of traded inputs. If the DRC is smaller than 1, domestic production is considered internationally competitive, because the opportunity cost of domestic resources is smaller than the net foreign exchange it gains in export or saves by substituting for imports. 2.3 Related research The adoption of DRC in analyzing competitiveness in agriculture can be found in studies by Gorton et al. (2001), Gorton and Davidova (2001), Gorton et al. (2006) and Bajramovic et al. (2006). These authors mentioned here applied DRC to their micro surveys for some agriculture products. Gorton et al. (2001), in their study on the international competitiveness of Polish agriculture, apply DRC ratios for three farm sizes and eight commodities. The results show that for the period 1996 to 1998 Polish crop production is more internationally competitive than livestock farming. Rapeseed and potatoes are considered as the most internationally competitive crops of those analyzed. However, during the period, the international competitiveness gets worse as international commodity prices fell. There is an inverse relationship between DRCs and farm size. They believe that this is an important result as Polish production is relatively fragmented and the degree of structural change has been slow. Gorton and Davidova (2001) study the price competitiveness of agricultural production in Central and East European Countries (CEECs). It combines empirical works conducted by the authors and other studies that have estimated DRC ratios for agriculture in various CEECs. The paper finds that in general CEEC crop production is more internationally competitive than livestock farming. Through the mid-1990s, wheat production in Bulgaria, the Czech Republic, Hungary, Romania and Slovakia gained international competitiveness. Meanwhile, also during this period, milk production was not internationally competitive. In this paper, the authors also study variations in DRCs by farm type and conclude that larger private farms in Hungary and the Czech Republic are more internationally competitive than smaller private farms in crop production. However, conclusions should be treated with caution due to sensitivity of DRC ratios to changes in international prices and the selection of the shadow prices for non-tradable inputs.


355

Gorton et al. (2006) study the past and future international competitiveness of Hungarian agriculture using DRC ratios with data for 2000–2002 as a base. Future international competitiveness is estimated for 2007 and 2013 using three scenarios: baseline (no accession), accession with historical rates of productivity growth and accession with dynamic improvements in productivity. The study indicates that accession will have a negative impact on the international competitiveness of Hungarian agriculture by increasing land and labour prices. Therefore, to maintain competitiveness in the arable sector, Hungary will need to achieve dynamic improvements in productivity to offset the effect of higher factor costs. Bajramovic et al. (2006) research on the competitiveness in the agricultural sector of Bosnia and Herzegovina. The authors apply both quantitative and qualitative approaches to study the competitiveness of the country’s agricultural sector. Three products are examined including milk, pepper and raspberries. In this research, DRC is used as the main quantitative indicator for competitiveness analysis. Besides, the authors also approach the determinants of competitiveness from macroeconomic conditions, trade policy to agricultural policy. The conclusion on competitiveness is favourable for pepper and milk (but considered as potential).

3 How to measure competitiveness This section develops a framework for assessing the competitiveness of agricultural products in Tien Giang. DRC has been selected because from the literature review it was seen that many researchers adopt this indicator when studying competitiveness in the agricultural sector. Because of its versatility and easiness to be interpreted, DRC has become the dominant indicator in World Bank-funded structural adjustment programs and in the analysis of transition economies (Bajramovic et al. 2006). In addition, DRC is also considered appropriate for product-based analysis (Coy 2006). DRC formula and its explanation in details has been mentioned on part 2.2 in this paper, and now we would like to consider that how it will be measured in our empirical study. n

∑ aijV j DRCi =

j = k +1 k

Pi − ∑ aij Pj j =1

According to Bajramovic et al. (2006: 17), “shadow prices for domestic resources and international prices for tradable outputs and inputs are referred to as social prices because they represent opportunity costs and

356

Hoai

opportunity benefits for the nation engaging the scarce resources in alternative production activities. They differ from the private prices (financial prices) faced by the producers due to the effects of policy distortions and market failures (for instance monopoly pricing and high transaction costs).” So, in our analysis we will use social price instead of shadow price even thought these two prices are sometime interchangeable. To reduce complexity and make up for the unavailability of data, some assumptions are set up for DRC with reference to Gorton and Davidova (2001: 4-5) and Bajramovic et al. (2006: 15-16): • Social prices for outputs and tradable inputs are measured as border prices (export / import parity prices) and adjusted to the farm level. For products for which the country in question was a net exporter during the analyzed period, an average FOB export parity price is usually taken as the unadjusted reference price. For products for which the country was a net importer, average CIF import parity prices are applied. The adjustment of prices from border to farm should account for, where appropriate, port and handling charges, transport, storage and maintenance costs. An alternative approach, in the absence of reliable border prices, is to adjust farm prices for the cost of transportation to the border. • Social cost of labour should be measured in terms of opportunity cost and the average wage paid in manufacturing as a proxy for this. But Bajramovic et al. (2006) argue that persistent and high unemployment clearly indicate that a labour market is distorted and not in equilibrium. So, the observed wage rate may represent an overestimation of opportunity costs. • Social price of land is typically measured by its rental value in the most profitable alternative agricultural use. • Technical coefficients necessary for calculation of the DRCs may systematically differ among farms of different sizes. Where data are available, attempts to consider variations between farm types should be made. For the case of Tien Giang, data for calculating DRC was derived from the following assumptions: • Prices are from 2004 in thousands of VND. • Tradable and non-tradable input use, private prices and yields are taken from the Vietnam Household Living Standard Survey (VHLSS) 2004 (Section 4B. Agricultural, Forestry and Aquacultural Production Activities).


357

• For the case of shrimp, due to only one observation on shrimp production in VHLSS 2004 for Tien Giang Province, observations for the whole MRD were selected instead. This problem has occurred because VHLSS 2004 is not a survey especially for aquaculture, so the number of observations for specific activities is sometimes very low.

4 Research design for focus group interview With the purpose of determining the key factors contributing to success as well as imposing constraints on the competitiveness of agricultural products in Tien Giang, factors coming from Cuevas (2004) were used as the foundation. The focus group discussions were carried out in 2007 in districts of Cai Be, Cai Lay, Tan Phuoc and Go Cong Dong in Tien Giang province. Farmers participated in groups of five or seven people. The numbers of participants are fifty for group of paddy, forty for group of citrus and twenty-five for group of shrimp farming. Results from the discussions were summarized by enumerators on the spot. In addition, ten experts in agribusiness were also interviewed to get their opinions on farmers’ ability to produce competitive products.

5 Product analysis: rice 5.1 Measurement of DRC The result of the DRC for rice is 0.20, revealing that Tien Giang’s rice production possesses a high potential to be competitive. This also seems true from a commonsense perspective, as MRD supplies the largest quantity of rice for export. 5.2 Results from focus-group interviews Market and consumer-oriented plan: More than three fourths of farmers responded that they do not know anything about AFTA, while the remaining farmers had heard about it from mass media like TV or radio but failed to identify AFTA correctly. One hundred percent of farmers sell products to local traders who influence their production plans greatly. Traders often offer low prices and have great power in bargaining. Quality control in production: Farmers believe that Vietnamese rice is lower in quality compared to that of competitors due to seeds, techniques and land. Hence, they began investing in better seeds and adopting new techniques to increase the competitiveness of their product. Most farmers admit that overuse of chemicals may result in reducing the quality of paddy. However, when a disease appears to be very serious, they would

358

Hoai

rather use a lot of chemicals to protect their plants. Beside that, since local traders are farmers’ biggest customers, there is little motivation to establish standards, thus discouraging farmers from upgrading product quality. Production process: One hundred percent of farmers said that their production process has remained virtually unchanged for at least 5 years. Technology and technological transfer: One hundred percent of farmers believe their technology is up-to-date. Beside their neighbours, assistance from extension officers plays an important role in disseminating knowledge to farmers. Mechanization is taking place widely, and many farmers are buying machines to use on their fields and others as service activities; thus, using buffaloes or oxen for ploughing no longer exists. Financial management: Farmers do not develop a written plan for production or have a reserve in case of crop failure. Due to limited funds, about fifty percent of farmers take loans from a bank, and most of their produce is sold to pay debts from inputs like fertilizers and pesticides. General management: Due to limited managing capacity, farmers wish to establish a link between themselves, businesses and government (extension officers, scientists, policy-makers) so that they can receive better prices and have a more stable market. 5.3 Results from expert interviews Market- and consumer-oriented plan: About twenty percent of farmers know rice is exported, while most of them are not concerned about price fluctuation in the world market. As a matter of fact, AFTA and WTO are not the farmers’ concern either. Farmers often sell their products to local traders who do not care about the quality of paddy. Prices are set according to negotiations between farmers and local traders in which the bargaining power belongs to the traders. About thirty percent of farmers rely on market signals in formulating their production plans. Being aware that consumers require rice of good quality, they are more careful in selecting seed, in using fertilizers and pesticides under the instructions from extension officers. Meanwhile, the remaining seventy percent of farmers do not base their decisions on consumer requirements in making production plans. In the end, experience and traditional practice still play a dominant role in the production process.


359

Quality control in production: About eighty percent of farmers do not take quality control into account in production. Though aware that chemicals are being used unscientifically, farmers feel reluctant to reduce them since they want to make sure their plants keep growing rather than being destroyed by insects. However, the recent rapid increase in the price of pesticides has made farmers reconsider how to use them more wisely. Technology and technological transfer: Mechanization is taking place rapidly, and most farmers have learned how to put new technology into their production, as transferred from extension officers, but only one fifth of them actually put the knowledge into practice or they just apply the new technology on a small part of their land. Financial management: Most of the farmers do not keep records of their costs of production on a regular basis. At best, a financial summary is often made after every season. In general, the financial management capacity of farmers is rather limited. General management: Farmers manage to select seed, take care of plants and sell their products. There is, however, virtually no organization to support farmers in terms of market information, inputs and outputs.

6 Product analysis: Citrus 6.1

Measurement of competitiveness

The DRC value of 1.15 for citrus indicates that it is generally not competitive. It should be noticed that the reference price has been calculated for the group of citrus including grapefruit, oranges and tangerines. If oranges are excluded from this group, the DRC result may be reversed, indicating the potential competitiveness of grapefruit and tangerines. 6.2

Results from focus-group interviews

Market- and consumer-oriented plan: Most farmers do not pay attention to international market information about where their products are sold and know very little about AFTA or related issues. To sell their products, three options are available for farmers: cooperatives, traders or local markets. Cooperatives require products of high quality, whereas traders do not set up quality requirements, but have stronger bargaining power instead. Local markets are the usual solution for farmers when their products do not meet

360

Hoai

quality requirements from cooperatives and traders are offering rather low prices. Quality control in production: Though appreciating Integrated Pest Management (IPM) for its effectiveness, farmers do not apply it to their whole production process. There are many brands of chemicals sold, but farmers still do not know clearly about the instructions of each chemical they are using. Production process: All farmers agreed that their production process has remained unchanged for five years. Technology and technological transfer: IPM was introduced to farmers five years ago, yet farmers cannot apply everything they have learned, because it is time-consuming and costly. One other source of information farmers often make use of is a regional television channel from Can Tho city, showing programs on production techniques for fruits. Financial management: Farmers do not have a written plan for production. For grapefruit, seventy to eighty percent of farmers often get loans from banks. General management: As smallholders, the quantities farmers can supply are also small. In addition, collaboration with each other in production and marketing is relatively weak; thus farmers cannot improve their weak bargaining position. 6.3

Results from expert interviews

Market- and consumer-oriented plan: Farmers do not have a good plan for production; rather they often follow temporary signals from the market. This production without a clear vision has resulted in: (1) increases in input costs and (2) prices often fall sharply during harvest periods due to excess supply. Quality control in production: Although awareness of quality control has been raised with reference to consumers’ demand, farmers still prefer habits and experience, so their products are somehow not in a same standard format. Good Agricultural Practice (GAP) was introduced to farmers, but, due to small and scattered landholding together with high costs for implementation, the application of GAP is rather limited.


361

Technology and technological transfer: Production techniques have been improved toward lowering costs and increasing productivity. Over eighty percent of farmers want to use appropriate production technology. However, technological transfer does not take place as expected in reality. Many farmers still follow experience and habits, so the production process remains virtually unchanged. Also, small and scattered landholdings are a significant reason for the limited application of scientific innovations into production. Financial management: Keeping record of expenses for production is not done regularly; farmers estimate costs and profits by recalling figures in their memory. General management: Cooperation among farmers in marketing for their products is still rather weak. There exist cooperatives to buy farmers’ high quality products, but farmers still need support in terms of product orientation, technological transfer and market information.

7 Product analysis: shrimp 7.1

Measurement of competitiveness

For shrimp, DRC was calculated using data from VHLSS 2004 for the whole MRD, because in VHLSS 2004 there is only one observation for shrimp farming. The value of 0.22 for DRC suggests that shrimp is also a potentially competitive product. 7.2 Results from focus-group interviews Market- and consumer-oriented plan: More than three fourths of farmers responded that they do not know anything about a world trade practice , others hears from mass media like TV, radio but failed to identify the world trade practice correctly due to their lackings of information and knowledge. Contract farming with businesses does not exist, while local traders are the only buyers for their products, so farmers can hardly improve their bargaining position. Quality control in production: Finding reliable suppliers of breeds is a difficult task for farmers; they do not even trust national hatchery centers for the quality of breeds supplied. Local traders, the monopolistic buyers, do not establish a clear set of standards for shrimp to be sold. However, all farmers say they do not abuse antibiotics when their shrimp face diseases.

362

Hoai

Production process: The production process varies among farmers, each farm following a different approach. Technology and technological transfer: Farmers raising shrimp actively participate in workshops on shrimp farming, and over two thirds of them put the knowledge gained there into practice. Farmers appreciate the closed production system, but they do not follow this approach strictly due to high costs. Financial management: Shrimp farming requires a lot of money compared to other farming activities; thus farmers carefully keep record of expenses, so that they can calculate costs and profits for every season. General management: Contract farming with businesses is what farmers hope to achieve in reality, so that the market for their output will be more stable and with better prices. Farmers say they would commit to meeting all conditions in such contracts. 7.3 Results from expert interviews Market- and consumer-oriented plan: About twenty percent or less of farmers formulate market-oriented plans in production, actively selecting breeds and studying techniques to supply products that meet market demand. Others make plans the same as the previous year or imitating successful farmers of the past year. Over three fours of farmers do not know about any world trade practice and have no information about the market. They also do not care much about the charge about Vietnamese shrimp being blamed for dumping prices. Their most important customers are local traders who buy shrimp from over ninety percent of the farmers. Quality control in production: Quality control is now farmers’ great concern to prevent disease and to lower costs. However, every farmer follows their own way. This raises a threat of water contamination, a channel for spreading disease from farm to farm. Technology and technological transfer: The technology now is more modern and more suitable for mass production than previously, and technological transfer to farmers has been implemented for several years. Most of the farmers are enthusiastic about participating in workshops on new technology.


363

Financial management: Book-keeping is carefully done by farmers for every season, so that they can analyze their effectiveness and learn from experience for the next season. General management: In general, farmers manage production according to their own methods and cooperation among them is still weak.

8 Conclusion In the case of Tien Giang province, rice and shrimp seem to have a high potential to become competitive. However, there are doubts about the sustainability of such competitiveness. Farmers still do not pay attention to or are lacking of information about upcoming competition from imported agricultural products, and the competition level is going to increase over time as Vietnam gradually begins following free trade principles, based on international trade commitments, in the near future. Besides, cooperation among them and with companies is not yet set up in a sustained manner. Therefore, indirect supports to farmers based on WTO regulations are needed, for instance, in terms of technological transfer and information dissemination. In addition, legal measures to enforce contracts between farmers and companies should be considered and enacted by policy makers and local government as well.

Acknowledgement The research for this paper was conducted under the sponsorship of the Seed Fund for Strategic Research and Training (SFRT) Program. The author would like to acknowledge SFRT for the valuable support.

References AusAID (2004) MeKong Delta Poverty Analysis – Final Report, http://www. ausaid.gov.au/research/pdf/mekong_poverty_report_04.pdf, accessed on 26 March 2006. Bajramovic S., Davidova S., Gorton M., Ognjenovic D., Pettersson M., Rabinowicz, E. (2006) Competitiveness in the Agricultural Sector of Bosnia and Herzegovina, Livsmedelsekonomiska institutet. http://www.sli.lu.se/pdf/SLI_Rapport_20065.pdf, accessed on 11 February 2007. Coy J.E.L. (2006). Competitiveness and Trade Policy Problems in Agricultural Exports: A Perspective of Producing/Exporting Countries in the case of Banana Trade to The European Union, PhD Dissertation, University of Göttingen.

364

Hoai

http://deposit.ddb.de/cgibin/dokserv?idn=982418833&dok_var=d1&dok_ext= pdf &filename=982418833.pdf, accessed on 14 February 2007. Cuevas R. (2004) Food Engineering, quality and competitiveness in small food industry systems – with emphasis on Latin America and the Caribbean, FAO Agricultural Services Bulletin, FAO. Esterhuizen D. (2006) An Evaluation of the Competitiveness of the South African Agribusiness Sector, PhD Dissertation, University of Pretoria. http://upetd.up.ac.za/thesis/available/etd-12082006144349/unrestricted/00front.pdf, accessed on 02 January 2007. Gorton M., Danilowska A., Jarka S., Straszewski S., Zawojska A. and E. Majewski (2001) The International Competitiveness of Polish Agriculture, http://www.staff.ncl.ac.uk/matthew.gorton/polfull.pdf, accessed on 03 August 2006 Gorton M., Davidova S., Banse M., Bailey, A. (2006). The International Competitiveness of Hungarian Agriculture: Past Performance and Future Projections. Post-Communist Economies 18(1), 69-84. Gorton M., Davidova, S. (2001) The International Competitiveness of CEEC Agriculture. Paper presented to the British Association of Slavonic and East European Studies (BASESS) Conference, Cambridge, 7th – 9th April 2001. Nguyen, K. (2006) Mekong Delta on Accession to WTO: Developing Advantage of Agricultural and Fishery Exports, People Army, 08 August. http://www.quandoinhandan.org.vn/right.php?id_new=65450, accessed on 12 August 2006. World Bank (2005) Accelerating Vietnam's rural development: Growth, Equity and Diversification – Volume I: Overview, East Asia and Pacific Region Rural Development and Natural Resources Sector Unit, World Bank. http://siteresources.worldbank.org/INTVIETNAM/Resources/vn_rsu_vol_1_en .pdf, accessed on 24 March 2006.

Part VII Natural Resource Risks and Coping Strategies

Chapter 19 Analysis of Risk and Coping Strategies of Fishing Communities in the Philippines when facing Natural Calamities Maria Rebecca Campos1 Abstract: More than half a million small fishers in the Philippines have been availing of loans from QUEDANCOR, the credit arm of the Philippine Department of Agriculture. The financing scheme has been quite successful, with a repayment rate at 95%. However, the occurrence of natural calamities, such as typhoons as well as pests and diseases, has affected the productivity of fisheries, thus hindering fishers from paying and renewing their loans. Failure to access credit could greatly inhibit them from continuing to venture on fishing activities and could eventually jeopardize the welfare of their entire households. The inability of creditors to pay their loans and meet their obligations also impairs, to a large extent, the financial operation and viability of the lending institutions. This study analyzes the natural risks and risk management practices of these fishers and recommends mitigation mechanisms to minimize the impact of natural calamities. Moreover, it suggests a bridge financing scheme that can be an effective and efficient instrument for enabling fishers to carry on their livelihood activities and support their families’ basic needs, while slowly recovering from their losses. Keywords: Risk, Coping strategies, Philippines, Fishing, Natural calamities

1 Introduction The Quedan and Rural Credit Guarantee Corporation (QUEDANCOR) of the Philippine government, operated under the Department of Agriculture, is tasked to provide and distribute credit resources and initiate programs to help accelerate rural growth, generate employment and establish enterprises that provide better livelihood and income opportunities to small fishers. To support the government’s overall goal of achieving food security and economic stability, QUEDANCOR developed a variety of innovative financing schemes to help intensify and sustain production of fisheries and aquaculture products. 1

Faculty of Development and Management Studies, University of the Philippines Open University, College, Laguna, Philippines


368

Campos

While QUEDANCOR has the critical responsibility of providing and improving credit assistance to fishers, it also has the task of helping its beneficiaries meet the repayment obligations of their loans. One reason for defaults can be attributed to the devastating impact of natural calamities. Schemes currently in place at time of writing are still insufficient to help safeguard lending programs and operations from non-repayment of loans due to production losses and damages to personal property. Natural calamities include the uncertainties and vagaries of weather and climate that bring about typhoons, floods, and droughts; earthquakes; volcanic eruptions as well as pests and diseases that affect the productivity of fisheries. Weather and climate disturbances are still considered major sources of risk, especially when they affect production and other related activities, taking heavy tolls among small fishers, laborers and traders. While the frequencies of disease and pest attacks have been decreasing due to continued development of improved technologies, they still inflict significant losses when they do occur. When natural calamities hit, small fishers are unable to pay their loans from QUEDANCOR; moreover, they have difficulty renewing their loan applications from QUEDANCOR or accessing credit from other sources. Failure to access credit could prevent them from being able to continue to venture on fishing activities and could eventually jeopardize the welfare of their entire household. The inability of creditors to pay their loans and meet their obligations also impairs, to a large extent, the financial operation and viability of the lending institutions. Risk management schemes currently employed include price stabilization measures, targeted relief to typhoon and drought victims, and crop insurance systems, to name a few. Some of these schemes, however, are becoming very expensive to implement. Moreover, they fail to enable fishers to regain sufficient resources so that they may continue production. One potential avenue that has not been explored is the continued extension of credit to fisher-borrowers who have suffered from natural calamities so that they can carry on their livelihood activities and slowly recover. Fishers usually require two to three cropping seasons to overcome losses due to inadvertent failure in production. The idea of a bridge fund is to provide assistance to fisher-borrowers who suffer major crop losses due to natural calamities. Such assistance can be provided in the form of loan restructuring and/or provision of low-interest (perhaps interest-free loans) to make it more affordable for borrowers to meet their loan obligations within a prescribed grace period and, at the same time, provide support to their families.

Risk and Coping Strategies of Fishing Communities in the Philippines

369

2 Objectives The purpose of this study is to analyze the natural risks and risk management practices of QUEDANCOR fisher-borrowers and to come up with appropriate measures to mitigate their adverse impacts on income and welfare. Specifically, the objectives are as follows: a. To estimate fishers’ production losses due to natural calamities; b. To document the coping strategies and mitigating measures taken by fishers during natural calamities; and c. To recommend mitigation mechanisms to minimize the impact of natural calamities.

3 Methodology and analytical approach 3.1 Risk and management assessment framework

Prediction / Forecasting

Development and Management Interventions

Natural Natural Calamities Calamities Natural Typhoons, Typhoons, Floods, Droughts, Droughts, Pests Pests & & Diseases, Diseases, Volcanic Volcanic Eruption, Eruptions, Earthquake, Heavy Earthquakes, Rains (LaLandslides, Niña), Tornados, Tsunam etc… is.

Agricultural AgriculturalProduction Systems Systems AQUACULTURE CROPS LIVESTOCK AQUA and CULTURE FISHERIES

Impacts on Lending Institutions (QUEDANCOR, Banks)

Impacts on Small Borrowers

Mitigation and Adaptation Measures

(Farmers, Fishermen, Traders)

Risk and Vulnerability Assessment Environmental Impact Assessment

Figure 1: Conceptual framework for assessment and management of risk due to natural calamities in agricultural and aquacultural production systems Figure 1 shows the conceptual framework for the assessment and management of risks due to natural calamities such as typhoons, floods,

370

Campos

droughts, extreme climatic events (e.g., El Niño and La Niña), volcanic eruptions, earthquakes, and occurrence of pests and diseases. The impacts of these natural calamities on agricultural production systems such as fisheries differ with respect to intensity, frequency, time of occurrence, and condition of each production system (e.g., age of fishery). Risks associated with these natural events as well as with pests and diseases may also differ across different locations and times of occurrence within the year. Thus, there is a need for risk assessment analysis to have both the spatial and temporal dimensions to be able to determine the appropriate intervention measures and management strategies to minimize production losses and other damages. Also necessary is the socio-economic characterization of the different areas of operation, particularly income and credit sources, including those coming from non-farm activities, etc. Employing the above framework, the study was undertaken in two stages. The first stage was the quantification of risks or expected losses to help determine appropriate interventions for strengthening QUEDANCOR programs. This involved analysis of the bio-physical and economic aspects of the agricultural production systems (e.g. fisheries and aquaculture), particularly those relevant to QUEDANCOR, to estimate agricultural commodity losses and to determine the impacts of natural calamities on the national, regional and, at best, provincial levels. Secondary data and statistics on historical crop production (yields and costs of production) and on historical occurrence of natural calamities were also collected and used. The second stage was the identification of potentially effective mitigating measures and coping strategies to manage risks. The analysis here was more on a micro-level, using survey data to evaluate the vulnerability of different agricultural production systems, assess risk characteristics and their impact on farmer-borrowers and their households as well as their impact on credit repayment and needs, identify specific coping mechanisms and guidelines for managing agricultural risks by commodity (e.g., fish/aquaculture), and develop common guidelines in determining the most suitable crop or agricultural commodity mixes that could be adopted in different areas covered by the QUEDANCOR programs. The results of these analyses were used as inputs in rationalizing the decision options for QUEDANCOR, especially in relation to its lending operations. The resulting characterization provided the information needed to determine the features and coverage of the bridging fund scheme. With the availability of a new credit window, QUEDANCOR clients and stakeholders could have better chances of continuing and sustaining agricultural production in the event such livelihood activities are affected by typhoons and other calamities. The identified mitigation measures and


371

coping strategies will help fishers minimize the risks in production brought about by the occurrence of natural calamities. At the same time, these measures would help secure QUEDANCOR’s lending operations. The study involved field surveys and field data verification, validation, and consultation with local QUEDANCOR program field implementers, QUEDANCOR member-borrowers who have experienced natural calamity disasters, and other stakeholders including other fishers, local government units, NGOs, etc. 3.2 Quantification and evaluation of risk Risk is defined as the average or expected value of economic loss from a commodity due to the occurrence of a natural calamity. It comes in various types and from various sources and is usually expressed or measured in weight or value of product (e.g. kilograms or tons, or pesos per hectare). Analysis of risks involves evaluation of the hazards associated with a particular calamity as well as the vulnerability of a specific agricultural crop or commodity to the hazard. Risk is location-specific, time-dependent, and may vary from one commodity to another. This is primarily due to the fact that the probability of occurrence and the relative strength of natural calamities usually vary from location to location and usually occur only during certain periods of the year. For example, the occurrence of typhoons during the rainy season is more frequent in areas along the typhoon belt than in areas near the equator. Total risk for a particular commodity is the sum of all risks associated with each source. In this study, only risks associated with the occurrence of natural calamities are considered. Risk due to occurrence of natural calamities is determined as the product of four factors, namely: (1) probability of occurrence of the natural calamity; (2) probability of hazard associated with the occurrence of the calamity; (3) degree of vulnerability of the commodity or agricultural enterprise; and (4) the level of exposure of the commodity in the area. That is, Risk = Probability of occurrence of calamity x Hazard x Vulnerability x Exposure Hazard refers to potential damage due to the occurrence of a natural calamity that may cause the loss of life or injury, damage to property, social and economic disruption and dislocation, or environmental degradation (International Strategy for Disaster Reduction, 2002). Vulnerability refers to the propensity of an area or a commodity to a particular hazard whenever a calamity occurs. It may be reduced or even

372

Campos

eliminated if there are some coping or adaptation strategies and/or mitigating measures implemented in the agricultural production systems. Exposure for a particular commodity may be represented by the land area planted with the crop, the number of people or population involved in the commodity, gross value-added, etc. Understanding of the concept and measurement of risk requires appreciation of the different components that define it. 3.3 Sources of risk Risk can be differentiated in terms of its sources, which include natural phenomena (e.g. weather and climate variability; natural calamities like typhoons, floods and droughts, pests and diseases, etc.) or socio-economicpolitical factors (e.g. price fluctuations, changes in demand, trade, etc.). Fisheries and aquaculture production are intrinsically risky, and natural calamities are the major sources of risk. Depending on the strength and duration of these calamities, they could lead to total loss of production, which could have devastating effects on the welfare of small fishers. For some calamities, the speed of occurrence is so fast that fishers are left unprepared to face the aftermath. Risks from other sources could also lead to losses in agricultural production, but their speed of occurrences is not usually as fast. In such cases, fishers could possibly devise some measures to reduce any expected negative impact. Measurement, Estimation and Evaluation of Risks due to Natural Calamities Quantification of risks due to natural calamities involved the measurement of the four factors mentioned above. The probability of occurrence for each parameter is estimated based on the relative frequency of occurrence of the event, records of which are usually available over a period of time. For instance, the relative frequency of occurrence of typhoons in a year in an area estimates the probability of occurrence of a typhoon in the locality. The probability of occurrence or prevalence of pests and diseases can similarly be determined from historically recorded data. The hazard parameter associated with the occurrence of each natural calamity was also measured from available past records or accounts in the area, or from results published in the literature. For example, typhoons are usually recorded with corresponding intensity, which could provide an indication of the level of hazard or danger to the commodity in question (e.g. fisheries and aquaculture). In some cases, records that show the actual damages from natural calamities to fisheries are also available. These data are often collected and published by government agencies, like the Department of Agriculture (DA) Bureau of Agricultural Statistics (BAS).


373

The incidence of typhoon occurrences in the different locations (provinces) in the Philippines was based on a previous study by the Manila Observatory in 2004. As can be seen, the frequency of occurrence differs across locations throughout the country (Manila Observatory, 2004; PAGASA, 2003). The northern part of the Philippines is most frequently visited by typhoons, with the frequency of occurrence being estimated at about 32% of the time. The provinces in Central Luzon, Southern Luzon and the Bicol Region are visited by typhoons around 25% of the time, the central Philippines between 7% to 18% of the time, and the Mindanao provinces at a very low frequency of 1% of the time. The same study (Manila Observatory, 2004; PAGASA, 2003) also analyzed the risk due to typhoons. Locations in the eastern parts of the central Philippines and the provinces in Luzon face the highest risk from the occurrence of typhoons. Risks faced in relation to the occurrence of typhoons are not, however, directly correlated with the frequency of their occurrence. Hazards due to natural calamities—such as climate variability as exemplified by El Niño events, La Niña phenomena, etc.—were estimated based on their relative frequency of occurrence. The severity of extreme dry episodes experienced in the Philippines during the last 50 years was observed. The worst El Niño event that brought significant damage to agricultural production systems was experienced during the dry season of 1997-1998. Another factor that determines risk relates to the vulnerability of a commodity to natural calamity. Vulnerability is sometimes considered synonymous to risk or expected value of loss in relation to the hazards brought about by a specific natural calamity (e.g. typhoon, El Niño episode, or La Niña event, etc.). Vulnerability may be defined as the susceptibility to stresses or hazards and the capacity (or lack thereof) to adapt, cope with or mitigate, and recover from such hazard. Figure 2.5 shows the spatial differences in vulnerability to typhoons of different locations throughout the country, with the provinces of Mindanao and the western portion of the central Philippines being generally less vulnerable. There are three defining features of vulnerability, namely: (a) locationspecific; (b) scale-dependent; and (c) dynamic or time-dependent. The vulnerability of a commodity may differ with respect to each kind of natural calamity. It should be noted that areas most likely to be affected by natural calamities such as typhoons or climate variability may not necessarily be the most vulnerable, if effective and efficient adaptation and mitigation strategies and measures are well in place. Vulnerability is also time-dependent, since a commodity may be more susceptible to a particular hazard at a certain development stage of the fish. It is also important to note that there is a level of uncertainty in vulnerability.

374

Campos

Analysis of the different elements of vulnerability is a pre-requisite for the development of policies that can be expected to promote sustainable development and equity. It also requires consideration of adaptation and mitigation measures as well as the capacity to cope with or withstand natural stresses during the agricultural production process. The vulnerability of a commodity or agricultural enterprise may be close to zero if adaptation or coping measures are available when the natural calamity occurs, but it may be near 100% if no coping or mitigating measures are ready to be implemented. There are several parameters that are used to represent vulnerability or the impact of exposure of the commodity to different hazards, namely: yield levels, area planted and net profits from agricultural production activities. However, one should be careful in using these parameters as surrogates, because these variables can vary greatly according to sources of data as well as the methods used to collect them. A list of exposure levels by commodity in the different provinces in the Philippines was used, based on data from various sources. Estimation of these parameters was based on secondary data published in the literature (e.g. BAS records), historical records of fishers, and from NGOs and other institutions, especially those that work in the localities studied. They may also be estimated from primary data gathered in socio-economic surveys, agricultural censuses, etc. 3.4 Data limitations for assessing risk Insufficiency of data often hinders achieving satisfactory estimation of the parameters that determine risk. Secondary data are not always available for all locations and/or commodities. Many secondary data, if and when these are available, are usually not in the form that can be directly plugged into a risk analysis framework. They are often used as surrogates or proxies for the required data. Collecting primary data is very expensive and time consuming. In addition, extra care has to be exercised in order to ensure the quality and consistency of responses. The output in this study vis-à-vis its objectives has been primarily influenced by the availability of the secondary data in the areas covered and the completeness and accuracy of the primary data collected. Agricultural commodities covered are those that currently have significant shares in QUEDANCOR’s loan portfolio. A few additional commodities were included because of their great potential to grow and expand in the near future. Area coverage by commodity similarly considered representative locations only, despite the fact that risk coefficient estimates are very area-specific. Some adjustments were made, when necessary, to


375

show the implications of the risk coefficient estimates on wider or more global (e.g. regional or national) scales. 3.5 Secondary data analysis Identification of risks due to natural calamities was conducted using secondary data that were collected form different national and local government and non-government offices across the country. Secondary data on damages and losses associated with natural calamities, pests and disease infestations were gathered from the National Disaster Coordinating Council (NDCC), Bureau of Agricultural Statistics (BAS), and from different Local Government Units (LGUs) in both regional and provincial offices. Both data sets collected from this activity were used to support the following analyses. 3.6 Quantification of risk due to natural calamities This particular activity included determining the vulnerabilities of the selected areas or provinces to natural calamities such as typhoons, floods and droughts including extreme climatic phenomena, like El Niño and El Niña. In addition to the secondary data, information from published literature as well as the primary data that were gathered from the field survey has been used. The review of literature and other historical records of concerned government agencies focused on the frequency and intensity of typhoons and the frequency, duration and time of occurrence of floods and droughts, including reported amounts of damages to fisheries and other related information. The reliability of reported statistics, especially on the extent of damages, were verified and crosschecked from other sources, primarily from key informants. The expected output of this analysis was risk profiles for the different production systems (i.e. fisheries) in the selected province. 3.7 Estimation of aquaculture yield losses due to natural calamities. One approach to quantifying yield loss due to specific causes or factors is to determine the yield reduction coefficient (YRC) associated with a particular factor (e.g., typhoon, flood, drought, pest, disease, etc.). YRC can be estimated from historical data or from simulation results (Lansigan et al., 1995). The actual yield can then be obtained by adjusting potential or attainable yield by the corresponding yield reduction coefficient(s) associated with the cause(s) of yield losses, as defined below for the cases specifically investigated by this study: Actual crop yield = Potential Yield x Yield Reduction Coefficient

376

Campos

where Potential Yield = Attainable yield in the location Yield Reduction Coefficient = (1 – RTτ) · (1 – RFτ) · (1- RDτ) · (1- RPDτ) where RTτ = reduction coefficient due to typhoon during period τ RFτ = reduction coefficient due to flood during period τ RDτ = reduction coefficient due to drought during period τ RPDτ = reduction coefficient due to pests or diseases during period τ 3.8 Identification of risk management and coping strategies Advances in science and technology, including geophysical and meteorological observation networks, have facilitated the reasonable and reliable prediction of seasonal climate trends three to six months ahead of time. Such adequate lead-time provides opportunities for establishing or developing an early warning system based on reliable seasonal climate forecasts. However, this requires careful interpretation and evaluation of risks due to climate variability and prediction. 3.9 Field verification and validation activities for primary data collection Ideally, analysis of the risks and assessment of the impacts of typhoons, floods and droughts (including El Niño and La Niña) should have been done in all areas covered by QUEDANCOR. But, considering time and budgetary constraints, area selection was done in two steps. First was to focus on areas (regions and provinces) that are most vulnerable to natural calamities. Second was the inclusion only of areas where QUEDANCOR programs are intensive in terms of loan exposure. Field verification site selection was based on the following criteria: (1) Number of loan beneficiaries and amount of loan exposure; (2) frequency of occurrence of natural calamities in the area, with the three major disasters considered being typhoons, ENSO (El Niño Southern Oscillation), and pests and diseases; (3) vulnerability to and extent of losses from damages (presence of “hot spots”); (4) situation regarding peace and order in the area; and (5) proximity and accessibility for research purposes, as represented by total travel costs. A survey of QUEDANCOR fisher borrowers was conducted for the following commodities: tilapia, milkfish, grouper, and seaweed. Validation of primary data taken from the survey was done by comparing these with benchmark data coming from the Bureau of Agricultural Statistics as well as regional offices of the Philippine government.


377

4 Results and discussion The commodities covered by this study are classified under aquaculture, which is the leading fish-producing sector in the Philippines, contributing 44 percent (1,717,026.7 mt) to the total fish production of 3,926,173.3 mt in 2004. Aquaculture registered the highest growth (18%), followed by municipal (2.4%) and commercial fisheries (1.6%) in 2003. In terms of employment, in 2003 aquaculture and fisheries provided employment to around one million or 5 percent of the country’s total labor force; in aquaculture, 26 percent (258,480) aqua farmers were engaged in using different cultivation methods. To date, QUEDANCOR has awarded about half a million small fishers the benefit of its credit facilities. The abovementioned figures make aquaculture a very promising industry when viewed with regard to decreasing catches from natural sources, due mainly to rapid human population growth. Constraints on orderly development of aquaculture tend to be political and administrative rather than scientific and technological. However, other constraints on its economic viability and sustainability have been identified, including the occurrence of natural calamities (typhoons, floods and droughts) and their consequences (diseases and water-quality degradation). Coupled with a need to understand other complex and diverse barriers to aquaculture, these constraints continue to present obstacles to its rapid and orderly expansion. These natural calamities and their consequences have caused gargantuan problems for all of us in the Philippines, but particularly for aquaculture fishers. The summary of technical information presented below on aquaculture production for each species surveyed does not include information from respondents giving unreliable data/information for some of the items sought during the survey. 4.1 Production and profitability The average production and net income per cropping season of the commodities produced by the QUEDANCOR fisher-borrowers sampled was computed. Table 1 indicates that the net income obtained by the respondents is above the poverty threshold of PhP 12,265 (NSO, 2005). It further shows that more than half of the borrowers are not really impoverished; hence, the QUEDANCOR borrowers are quite better off than most of the fishers in the country, although this could not be solely attributed to availment of QUEDANCOR loans. Borrowers also engaged in non-farm activities such as off-farm employment, retailing and vending are the most well off, with virtually all of them being above the poverty line.

378

Campos

Table 1: Productivity, net income and selling price by commodity Region

Province

Commodity

Productivity (Average yield)

Net income (Pesos/ha)

Selling price (Pesos)

725kg/ha 3.5 t/ha

15,900 91,290

37/kg 35/kg

4,900kg/ha

91,600

27/kg

3.25 t/ha

277,000

800/kg

I IV

Pangasinan Bangus Batangas Tilapia Zamboanga IX Seaweeds del Norte Surigao del CARAGA Grouper Norte Source: QUEDANCOR Risk Study, 2005.

Problems in Production The majority (83%) of respondents claimed that they encountered production problems, with the most cited being the occurrence of pests and diseases (64%) followed by the occurrence of natural calamities (38%). Other reasons that affect production are lack of capital (11%) and relatively high prices of inputs (10%). 4.2 Average loss due to natural risks The average loss of QUEDANCOR borrowers due to natural risks is estimated in this section and was obtained by multiplying expected revenues as reported through the survey (by commodity and by location) with percentage losses arising from natural risks. All commodities are affected by natural risks, though in varying degrees and depending on farm location. Commodities are also subject to risks ranging from pests and diseases to ENSO and typhoons. Seaweeds grown in Zamboanga del Norte have incurred losses to QUEDANCOR borrowers of up to about 75%. Although a very attractive business, commanding a net income of PhP 277,000 per hectare, Table 2 shows that it is also a very risky venture. If left unabated, these losses in income on the part of QUEDANCOR borrowers in turn deter them from paying their loans to QUEDANCOR.

379


Table 2: Average loss (in pesos) due to natural risks by commodity Region

Province

Commodity

Natural risk

Loss (%)

Average loss (PhP/ha)

Region I Region IV

Pangasinan Batangas

Milkfish Tilapia

Typhoons Typhoons

87.8 84.39

13960.2 77039.63

Region IX

Zamboanga del Norte

Seaweeds

Pests & diseases

75

Surigao del Norte

Grouper

Pests & diseases

64.95

Typhoons ENSO

72.68 6.96

CARAGA

68,700 179911.5 201,323.60 19,279.20

*per head

Source: QUEDANCOR Risk Study, 2005.

4.3 Coping and adaptive practices Coping strategies as defined by Davis (1993) are individual or community responses to changes in environmental conditions, or responses to its consequences, such as declining food availability. They are short-term responses for securing a livelihood system when subject to periodic stress. Mitigating or adaptive strategies, on the other hand, refer to the way in which, in order to meet their livelihood needs, individuals, households and communities have changed their mix of productive activities and modified their community rules and institutions over the long term in response to economic or environmental stresses or shocks. Coping strategies Seaweed growers in Zamboanga del Norte harvest their crop as soon as they become aware of signs of pests or disease. They also clean the growing area by removing lumot (algae) and lapa-lapa (epiphytes). Surigao del Norte’s grouper fishers usually transfer cages to deeper water during periods of continuous rain, preventing abrupt changes in temperature and salinity. The tilapia growers in Batangas harvest the stock before an announced strong typhoon arrives. They also select sites where the terrain of the surrounding shore areas can weaken or deflect strong winds and waves. The milkfish fishers in Pangasinan harvest their stock when typhoons are announced that may directly affect their fishponds. Table 3 presents the different mechanisms adopted in aquaculture production to mitigate and cope up with the effects of natural calamities.

380

Campos

Table 3: Practices Adopted by QUEDANCOR Borrowers to Safeguard and Minimize Impact Before and After Natural Calamity Occurrence Type of Aquaculture Production Seaweed (Kappaphycs and Eucheuma) farming

Grouper cage culture

Tilapia cage culture Brackish water milkfish pond culture

Mitigating Measures • manual removal of algae (“lumot”), epiphytes (“lapa-lapa”), and mud • financially better-off growers transfer their farmed seaweed to a less crowded area where current flows freely • lower plants further from the water surface to prevent too much exposure to sunlight, especially during low tide • for enlarged thallus tips, loosen or untangle string of filamentous plants • harvest plants as soon as disease occurs • locate cage so as to make it accessible, especially in times of natural calamity, yet secure from vandals and poachers • transfer cages to deeper water during periods of continuous rain, preventing abrupt changes in temperature and salinity • use strong, weather- and pest-resistant, non-corrosive, and non-abrasive surfaces for cages • select sites where the terrain of the surrounding shore areas weakens or deflects strong winds and waves • harvest stock before an announced strong typhoon arrives • secure fish stock by putting a net-fence on top of perimeter dike • harvest stock before an announced, strong typhoon arrives

4.4 Proactive practices of QUEDANCOR borrowers to mitigate impacts of natural calamities In terms of proactive practices, the most common responses of borrowers are to be alert and to prepare their farms for calamities (see Table 4). Preparing farms means that farm structures are repaired (e.g., seaweed growers in Zamboanga del Norte).


381

Table 4: Proactive safeguards to mitigate impacts of natural calamities by QUEDANCOR borrowers, commodity and province. Commodity

Province

Proactive Safeguards of Borrower

Milkfish Tilapia Grouper Seaweeds

Pangasinan Batangas Surigao del Norte Zamboanga del Norte

create drainage or canal / use of nets be alert / prepare farm for calamity be alert / prepare farm for calamity fix / replace / repair / clean if possible

Urgent assistance needed immediately after a calamity The QUEDANCOR borrowers were asked about the most important types of assistance they need immediately after a major natural calamity. The most common responses are food assistance and financial help to meet other most basic needs—clothing and shelter (see Table 5).

Table 5: Most urgent assistance needed after a calamity by QUEDANCOR borrowers, commodity and province. Commodity

Province

Most Urgent Assistance Needed by Borrowers

Milkfish Tilapia Grouper

Pangasinan Batangas Surigao del Norte Zamboanga del Norte

financial assistance food and financial assistance financial assistance

Seaweeds

financial assistance

The types of assistance requested after the occurrence of calamities have almost always been provided. Responses from the sample surveyed indicated that QUEDANCOR delayed loan interest payments to help its borrowers recover financially. Additionally, awareness/orientation programs have always been conducted to impart to the borrowers proper measures and practices to help reduce their being vulnerable to subsequent occurrence of such calamities (see Table 6). Additionally, technical trainings and seminars have been given on a regular basis to further enhance proper production management. Local government units, particularly the Barangay council, have been primarily responsible for the provision and distribution of food and farm inputs, particularly seeds. National government agencies, on the other hand, provide other inputs like fertilizer. The milkfish growers in Pangasinan were given fingerlings from suppliers. The need to enroll in crop insurance also continues to be emphasized.

382

Campos

Table 6: Assistance provided and sources of assistance after a natural calamity for QUEDANCOR borrowers, by commodity and province. Commodity

Province

Assistance Provided

Source of Assistance

Milkfish Grouper Seaweeds

Pangasinan Surigao del Norte Zamboanga del Norte

fingerlings financial assistance financial assistance

supplier QUEDANCOR QUEDANCOR

4.5 Assistance provided by institutions Physical and social infrastructure have been developed by the government to cater to the needs of the impoverished sector of society. Ease of access to these facilities by the rural poor (farming and fishing households) and relative distance from their residences is an indication of efforts of the government to reach out to the poor. The respondents mentioned their need for physical facilities and social infrastructure, such as farm-to-market roads, bridges, irrigation/canal systems, health centers, and school buildings. The nearest farm-to-market road ranged from about a meter to 5 km from the houses of the respondents. The farthest bridge was 10 km away, while the nearest was about a meter from one respondent’s residence. Irrigation/canal systems were also accessible, the farthest being 7 km from one house. Health centers ranged from 2 to 7 km away, while the farthest school was 3 km. All infrastructure/support services are readily accessible to the respondents, just a walking distance away from their residences. However, in terms of support services, awareness was high only for those that were offered by QUEDANCOR, local government units (LGUs), the Department of Agriculture (DA) and non government organizations (NGOs) present in their communities. Support services needed were in the form of trainings/seminars, technology transfer, livelihood projects, animal and seed dispersal, as well as marketing assistance. Trainings were offered by the government—QUEDANCOR, LGU, DA, Department of Science and Technology (DOST), Department of Trade and Industry (DTI), Department of Health (DOH)—as well as NGOs whose offices were less than a kilometer to as far as 38 km away from their houses. Technology transfer came from QUEDANCOR, LGUs, and the Bureau of Fisheries and Aquatic Resources (BFAR), which were 1 to 28 km within their reach. The same thing holds true for livelihood projects and animal and seed dispersal programs. The respondents acknowledged marketing assistance from QUEDANCOR, LGUs, BFAR and the private sector, where they usually have credit ties with institutions whose offices range from 3 to 67 km away


383

from their residences. This means that the borrowers have access to basic social services. It is worth noting that some respondents identified their need for an insurance company to cover their commodity, primarily due to the heavy losses that they incur due to natural risks. They mentioned that the Philippine Crop Insurance Corporation (PCIC), about 128 km away from their residences on average, was just too far away and inaccessible to them. They were banking on the idea that QUEDANCOR would help them along in this area, since QUEDANCOR had already assisted them as a source of credit, provider of inputs and technology transfer, up to marketing of their produce. 4.6 QUEDANCOR calamity bridge funds For QUEDANCOR’s loan operations, a bridge fund can serve as a potential buffer to recoup the loans of defaulting fisher-borrowers that suffer crop losses due to the effect of natural calamities. Such a bridge fund scheme in support of the QUEDANCOR credit program needs to be carefully studied and crafted. The total amount of the bridge fund should be based on the frequency of occurrence of natural calamities in a given area and their potential impact as expressed in terms of the value of crop losses. There are critical issues and concerns that need to be clearly understood and addressed for such a task to succeed. One issue would be the proper identification of borrowers who could and will avail themselves of such support services. Other issues include the circumstances under which extension of credit can be granted and determining appropriate amounts of credit for enabling the continuation of production and other livelihood activities. A final issue pertains to the setting of a reasonable grace period that would be given to the affected borrowers, one that allows easy repayment of total loans taken (regular loan plus the extension loan granted through bridge financing). Different types of fisher-borrowers require different types of credit and risk alleviation assistance in order to recover. Identifying the specific form and amount of assistance for each type requires critical information that needs to be gathered and analyzed in order not to put the operation of the lending institutions at stake.

5 Conclusions When a natural disaster strikes, the QUEDANCOR borrower copes by reducing his vulnerability through the following means: reducing his consumption and social obligations; selling his livestock; in some cases migrating some household members, so they can seek employment elsewhere to augment the family income; withdrawing from his grain

384

Campos

inventory for household consumption and offering it for sale, but at a price lower than usual; and in some cases devising forms of collective action together with the community. Households with larger amounts of assets and other sources of income from non-farm activities are less vulnerable to natural risks and have a better ability to obtain credit from other sources. The worst affected and most vulnerable during calamities are the respondents who are poor and marginalized. Not only are they the worst hit, but also their capacity to recover from a disaster is very low. Any extreme situation traps the poor in a situation of selling off productive assets that become difficult to retrieve and, thereby, reinforce poverty almost permanently. The majority of typhoon-stricken respondents have sustained damages to personal dwellings, loss of personal effects as well as their sources of livelihood: their farms. Those with off-farm income were able to retrieve some of those lost assets. However, those relying only on agriculture and fisheries as their source of livelihood were never able to buy back their assets, even during normal years. In order to repay their loans, the borrowers look for other sources of credit, such as their close relatives or friends, or return to informal creditors offering higher interest rates so that they will have capital to start anew. In the process, they still cannot pay their loans from QUEDANCOR and become even more tied to their creditors. This vicious cycle continues as long as these fisher-borrowers are not provided with a credit scheme which has contingency measures that they can avail themselves of during times of calamity. On the other hand, institutions on disaster management at the LGU level are in place. These offer temporary assistance to the QUEDANCOR fishers in the form of relief goods, disaster shelter, medical assistance, among other things. However, there is no institutionalized credit repayment or restructuring mechanism that assists them with changing their socioeconomic condition, particularly in helping them recover from the losses incurred on their livelihoods and personal effects.

6 Recommendations This study has taken on the challenge of developing the kind of credit support mechanism outlined above. Such credit support should include access to timely and reliable information and forecasts on the possible occurrence of natural calamities and prediction of impending or potential risks and hazards that they may cause to agriculture/aquaculture and the rural community at large. Such forecasts need to be anchored in recent scientific and technological advances. In addition, location-specific coping and mitigating strategies need to be formulated to minimize the risks in agricultural and aquacultural production associated with natural calamities.


385

Such strategies require protective measures that are appropriate and effective for reducing or preventing dangers from approaching calamities, particularly for these modes of production. It should be stressed that successful development of the mitigating strategies needs to be complemented by significant efforts of concerned agencies to generate and make available the statistics required in the prediction of possible occurrences of typhoons and other weather disturbances, incidences of pests and diseases as well as the estimation of damages that these may inflict on fishers’ production activities. The following are recommended coping measures: • Seaweed. The erstwhile ideal seaweed farm has now been transformed into a poor and unsuitable setting through the addition of too many farms in the total environment. To cope with decreased production due to the occurrence of pests, diseases and the intrusion of other plant species, some growers should sacrifice for the good of the majority by relocating their farms. This action is necessary for a sustainable seaweed farming industry in the survey area. • Grouper cage culture. To cope with the damages brought about by strong typhoons, the principles and practices of using low-volume, highdensity (LVHD) fish cages can be followed. Among other advantages, LVHD cages give operators ease of movement to safer areas in times of natural calamities (i.e., typhoons, ENSO, or pests and diseases). • Tilapia cage culture. As with grouper cage culture, the principles and practices of LVHD cages should be promoted to cope with the damages of natural calamities. Another advantage that cannot be overlooked is that, when a small cage gets damaged, fish loss is minimized. Promotion of the submerged fish cage—dubbed the “typhoon-proof fish cage”—must also be pursued. This is the only type of netcage culture structure in Laguna de Bay that withstood Typhoon Rosing in 1995 without damage (Tamayo-Zafaralla et al. 2002). It should be noted that more than 65,000 ha of the Laguna Bay was occupied by fishpens before the typhoon hit and only about 30,000 ha withstood the calamity. Two methods of submerged netcage culture were in operation namely, the completely submerged netcage and the adjustable submerged netcage. The former is completely submerged at about a meter from the sea bottom, each cage provided with a buoy that serves as a marker on the water’s surface. Meanwhile, the latter is adjustable to a desired level: either similar to the former or floating like a traditional netcage. Each cage is tied to a main line that is raised above the water by bamboo poles which are evenly distributed and long enough to support one module.

386

Campos

• Brackish water milkfish pond culture. The foregoing methods of submerged netcage culture could be modified and adopted for operation in pond conditions to inhibit damages from typhoons and floods. Structural Measures • Raisers of grouper in Surigao del Norte usually transfer the cages to deeper water during periods of continuous rain in order to prevent abrupt changes in water temperature and salinity. • Tilapia growers in Batangas select sites where the terrain of the surrounding shore areas weaken or deflect strong winds and waves. Institutional Measures • Quarantines must be implemented as a matter of policy, the net effect of which would be to prevent the entry and invasion of selected areas by pests and diseases have hitherto been left unchecked. This is particularly important as invading or ‘invasive’ organisms are known to be more destructive when entering new habitats. Quarantine policies are crucial for determining the extent and magnitude of QUEDANCOR’s lending portfolio for the area. Seeds of new varieties entering the region must be tested as possible carriers of disease-causing organisms. When indeed these new varieties are carriers, a forecast for a disease scenario can be made, which QUEDANCOR can then use as a basis for planning subsequent lending operations. Further, the movement of these diseasesusceptible varieties must be regulated so as to prevent disease spread that would further increase risk and diminish borrowers’ ability to repay loan. This can only be done through collaboration among government agencies concerned. • Improved seasonal climate forecasting of the occurrence of natural calamities as well as effective dissemination of forecasts for preparedness. Advances in science and technology have led to the development of process-based models, which provide the opportunity to integrate institutional capability and interdisciplinary information and knowledge for more systematic and reliable agricultural planning and development. • Improvement of the early warning system by concerned government agencies. Early warning or forecasting systems can be the primary means of determining risks and reducing losses due to pests and diseases. Currently, for corn leaf spot and leaf blight diseases, corn borer insect pest, and black banana Sigatoka, these systems can either be in the form of simple empirical forecast models or of more complex simulation


387

models. The utilization of these forecasting systems certainly will become more meaningful when considered in the context of the growing season. Further, these systems are usually in the form of computer models that can easily be used by personnel with a minimum training in modeling and simulation. In some cases, it is possible to link them to models of crop growth to obtain estimates of yield losses. • A National Pest and Disease Surveillance and Early Warning System concerned with detecting all major pests and diseases for all economic fisheries in the country has been organized through collaboration among government agencies and several state colleges and universities (SUCs). A link to this network should bolster QUEDANCOR’s lending operations in reducing risks due to pests and diseases and provide borrowers an enhanced capacity for loan repayment. • Effective and efficient information, education and communication strategies are crucial components of successful risk assessment and management. Part of such strategies may involve training of Self Reliant Teams (SRTs) in issues related to risks due to natural calamities. • The Information Education Communication (IEC) strategy and/or the SRT training by QUEDANCOR should recognize different geographical peculiarities as in regional dialects and production management practices. The educational attainment levels of stakeholders must also be considered in developing appropriate training modules. The educational strategy—which QUEDANCOR must develop with its partners in agricultural development, such as the Department of Agriculture and agricultural extension workers of local government units (LGUs)— should effectively utilize local communication networks, such as radio and television. • Through its regional and provincial offices, QUEDANCOR should provide technical advice in terms of advisories and recommendations concerning impending natural calamities. These could be disseminated through formal and informal networks in the area. • It must be reiterated that these challenges require effective and efficient networking and partnerships among QUEDANCOR, rural development agencies in the area, LGUS, and even NGOs. • Provision of a standby Calamity Support Bridge Fund to assist fishers in cases of natural calamities.

388

Campos

Acknowledgements This paper was drawn from the results of research projects conducted by the author for SEARCA and the University of the Philippines at Los Banos Foundation, Inc., funded by QUEDANCOR and the Fisheries Resource Management Project. The author gratefully acknowledges the grant awarded by the Economy and Environment Program for Southeast Asia (EEPSEA) for travel to the International Conference on Greening Growth: Economic Transition and Sustainable Agricultural Development in East and Southeast Asia, from 28-29 October 2008 in Nanjing, China.

References Davis . S. (1993). Are Coping Stratgies a Cop Out? Institute of Development Studies (IDS) Bulletin 24(4): 60-72. International Strategy for Disaster Reduction (2002). Disaster Reduction for Sustainable Mountain Development. Information Kit. Lansigan, F.P., Pandey, S., Bouman, B.A.M. (1995). Combining Crop Modelling with Economic Risk Analysis for the Evaluation of Crop Management Strategies. Field Crops Research 51(1,2): 133-145. Manila Observatory (2004). Simulation of Monthly Rainfall Climatology with the MM5 Model. National Statistics Office (2005). Philippine Yearbook. PAGASA - Philippine Atmospheric, Geographical & Astronomical Service Administration (2003). A Study Of Tropical Cyclone Activity Over Northwest Pacific Before, During And After The 1997-1998 El Niño Episode. Tamayo-Zafaralla, M., Santos; R.A.V., Orozco, R.P.; Elegado, G. C. P. (2002). The Ecological Status of Lake Laguna de Bay, Philippines. Aquatic Ecosystem Health and Management 5(2): 127-138.

Contributors Ancev, Tihomir - Agricultural and Resource Economics Group, The University of Sydney, Sydney, NSW, Australia Azad, Md Abdus Samad - Agricultural and Resource Economics Group, The University of Sydney, Sydney, NSW, Australia Bao, Xiaobin - Rural Development Institute, Chinese Academy of Social Sciences, China Beckmann, Volker - Department of Agricultural Economics, Faculty of Agriculture and Horticulture, Humboldt Universität zu Berlin, Germany/ Department of Environmental Management, Chair of Environmental Economics, Brandenburg University of Technology (BTU) Cottbus, Germany Böber, Christian - Institute for Agricultural Economics and Social Sciences in the Tropics and Subtropics (490), University of Hohenheim, Stuttgart, Germany Campos, Maria Rebecca - Faculty of Development and Management Studies, University of the Philippines Open University, College, Laguna, Philippines Chan, Yujie - Shanghai Subsidiary Company, China Grain Reserves Corporation, Shanghai, China Den, Do Thi - Agricultural and Resource Economics Group, The University of Sydney, Sydney, NSW, Australia Dung, Nguyen Huu - University of Economics, Ho Chi Minh City, and Centre for Environment Economics, UEH, Vietman Feng, Shuyi - China Center for Land Policy Research, Nanjing Agricultural University, China Glauben, Thomas - Institute of Agricultural Development in Central and Eastern Europe (IAMO), Germany Grafton, Quentin - Crawford School of Economics and Government, The Australian National University, Canberra, Australia Guo, Guancheng - China Center for Land Policy, Nanjing Agricultural University, China Harris, Michael - Agricultural and Resource Economics Group, The University of Sydney, Sydney, NSW, Australia.

390

Contributors

Heerink, Nico - Development Economics Group, Wageningen University, Wageningen, Netherlands; and China Center for Land Policy Research, Nanjing Agricultural University, Nanjing, China Hoai, Nguyen Trong - University of Economics, HoChiMinh City, Vietnam Irawan, Evi - Department of Agricultural Economics, Faculty of Agriculture and Horticulture, Humboldt University of Berlin, Germany Kompas, Tom - Crawford School of Economics and Government, The Australian National University, Australia Li, Rui - Institute of Soil and Water Conservation, Chinese Academy of Sciences, China Lu, Kaiyu - Institute of Agricultural Economics and Development, Chinese Academy of Agricultural Sciences, China. Mariyono, Joko - Research Associate for Socio-economics, AVRDC – The World Vegetable Centre, Taiwan. Neef, Andreas - University of Hohenheim, Stuttgart, Germany Nonkiti, Sakdamneon - Department of Agricultural Economics, Faculty of Agriculture, Chiang Mai University, Thailand Opschoor, Hans - Department Economics of Sustainable Development, Institute of Social Studies (ISS), The Hague, Netherlands; Department of Spatial and Environmental Economics, Vrije Universiteit, Amsterdam, Netherlands. Pellegrini, Lorenzo - Institute of Social Studies, The Hague/Erasmus University Rotterdam Piotrowski , Stephan - Institute for Agricultural Economics and Social Sciences in the Tropics and Subtropics (490), University of Hohenheim, Stuttgart, Germany. Pu, Chunling - College of Economics and Management, Xinjiang Agricultural University, Urumqi, China Qu, Futian - Land Policy Center, Nanjing Agricultural University, Nanjing, China Resosudarmo, Budy P. - The Arndt-Corden Division of Economics, Crawford School of Economics and Government, The Australian National University, Canberra, Australia. Sangkapitux, Chapika - Department of Agricultural Economics, Faculty of Agriculture, Chiang Mai University, Thailand

Contributors

391

Shi, Lina - China Centre for Land Policy Research, Nanjing Agricultural University, Nanjing, China Shi, Xiaoping - Nanjing Agricultural University, Nanjing, China Spoor, Max - Institute of Social Studies, The Hague/Erasmus University Rotterdam, The Netherlands, and Barcelona Institute of International Studies Suebpongsang, Pornsiri- Department of Agricultural Economics, Faculty of Agriculture, Chiang Mai University, Thailand Tan, Rong College of Public Administration, Zhejiang University, Hangzhou, China Truc, Chi Mai Thi - Research Institute for Aquaculture No. 2, Ho Chi Minh City, Vietnam Wang, Xiaobing - Institute of Agricultural Development in Central and Eastern Europe (IAMO), Germany Wang, Zhonghui - WTO Research Centre, Nanjing University of Finance and Economics, Nanjing, China Werthmann, Christine - Philipps-University Marburg, Marburg, Germany; CGIAR System-wide Initiative on Collective Action and Property Rights (Capri), WorldFish Center Wesseler, Justus - Environmental Economics and Natural Resources Group, Wageningen Univerty, Wageningen, The Netherlands Yamazaki, Satoshi - School of Economics and Finance, University of Tasmania, Hobart, Australia. Zhang, Yanjie- Institute of Agricultural Development in Central and Eastern Europe (IAMO), Germany, and Department of Agricultural Economics and Rural Development, Georg-August-University Göttingen, Germany