Cristian Vasco was born in Quito, Ecuador in 1978. He obtained his degree of agricultural engineer at the Army Polytechnic School in Ecuador and in 2005 joined the Masters‘ Program in International Ecological Agriculture at the University of Kassel, Germany. His Master thesis on “Ecuadorian out-migration to Spain: causes and economic consequences” was awarded with the annual “German Developing Countries Prize” for out-standing research accomplishments concerning Developing Countries at the University of Giessen, handed over by the German Federal Ministry of International Cooperation. It also received the prize for the best thesis written on non-European topics at the Faculty of Organic Agriculture, University of Kassel in the summer semester 2007. In 2008 he started his PhD studies at the Department of Development Economics, Migration and Agricultural Policy (DEMAP) at the University of Kassel. Over the following years he presented his research at several high-ranking international conferences. He obtained his PhD degree in early 2011.
The Impact of International Migration and Remittances on Agricultural Production Patterns, Labor Relationships and Entrepreneurship The Case of Rural Ecuador
ISBN 978-3-86219-086-7 kassel university press
/// Series edited by Béatrice Knerr ///
Cristian Vasco
Cristian Vasco
The research presented in this book quantitatively analyzes the effects of international migration and remittances on agricultural production patterns, labor relationships and entrepreneurship in rural Ecuador. The results show that migrants‘ households spent more on fertilizers and are more likely to accumulate cattle than their non-migrants‘ counterparts. They also demonstrate that neither international migration nor remittances influence the likelihood to participate in reciprocal work agreements in contrast to the conclusions of most sociological studies. Instead, migrants‘ households are more likely to hire wage laborers than their equivalents without migrants. Regarding entrepreneurship, neither migration nor remittances have any effect on households‘ likelihood to own a small-scale business and on the number of non-household members working in a business. Instead, the number of household members working in a business is positively influenced by international migration. These findings contribute to the debate about the effects of international migration on rural regions and provide policy makers as well as rural development and migration practitioners with information and facts to be taken into account for the design of projects linking international migration with rural development.
The Impact of International Migration and Remittances on Agricultural Production Patterns, Labor Relationships and Entrepreneurship
9
International Labor Migration 9
U_IntLabMig9_druck 22.02.11 09:40 Seite 1
International Labor Migration Vol. 9 Editor: Prof. Dr. Béatrice Knerr
Cristian Vasco
The Impact of International Migration and Remittances on Agricultural Production Patterns, Labor Relationships and Entrepreneurship The Case of Rural Ecuador
kassel university
press
This work has been accepted by the faculty of Organic Agricultural Sciences of the University of Kassel as a thesis for acquiring the academic degree Doktor der Agrarwissenschaften (Dr. agr.). 1 Supervisor: 2 Supervisor: st
nd
Prof. Dr. Béatrice Knerr Prof. Dr. Werner Troßbach
Day of disputation
This publication was funded by the German Academic Exchange Service (DAAD). Gedruckt mit Unterstüztung des Deutschen Akademischen Austausch Dienstes (DAAD).
Bibliographic information published by Deutsche Nationalbibliothek The Deutsche Nationalbibliothek lists this publication in the Deutsche Nationalbibliografie; detailed bibliographic data is available in the Internet at http://dnb.d-nb.de. Zugl.: Kassel, Univ., Diss. 2011 ISBN print: 978-3-86219-086-7 ISBN online: 978-3-86219-087-4 URN: urn:nbn:de:0002-30875 2011, kassel university press GmbH, Kassel www.upress.uni-kassel.de
Cover illustration: Tomasz Henning Cover layout: Bettina Brand Grafikdesign, München Printed in Germany by Unidruckerei, University of Kassel
8 February 2011 th
PREFACE For more and more families in rural regions of poor countries international migration cum remittance strategies are an essential part of income procurement to secure basic needs and invest into household items or productive assets. In view of the large sums which are transferred around the globe, from richer (“developed”) to poorer (“developing”) countries, policy makers as well as development theorists are concerned with the question how these remittances could be turned into a sustainable income base for the migrants’ families and a tool for enhancing the economic development of their home countries. These questions are particularly pertinent since the early 2000s when major iaanternational institutions turned to it. Although meanwhile a large number of investigations have studied the impact of migratory movements and remittances on the regions of origin and the left behind families, the results don’t allow for generalizations, in particular not with regard to rural regions from where a major share of the migrants comes and to where they may return. Ecuador belongs to the countries where the economic implications of these migratory movements on the rural families left behind and as a correlate on the regions of origin have hardly been researched, in spite of its high extraordinarily high rates of international outmigration, and in spite of some characteristics which might allow drawing conclusions which are relevant also to other countries in the Andean region and beyond. In view of the importance which the outflow of labour force and the inflow of transfers from abroad have this knowledge deficit constitutes a major problem for the country’s economic policy management. Against that background Cristian Vasco’s original and intensive research in this subject constitutes a most valuable and timely input for the national policy as well as for international development policy and follow-up studies. With regards to the indicators for household and rural region development the author concentrates on aspects which are essential in the socio-economic context of rural Ecuador. These are transition from agriculture to cattle keeping; expenditures for farm inputs; changes in the labour organization; and the establishment of small and medium enterprises in rural areas,. The empirical research which is the core of the investigation is theoretically rooted in the New Economics of Labour Migration (NELM) and on a high methodological level proceeds along the lines of a thorough analysis of secondary data. The work is well-structured and the research design is appropriate for approaching the research questions at hand. The methodological approach is clearly explained and well
documented. Cristian Vasco demonstrates that he has intensively dealt with the issues to be investigated; all relevant aspects are covered in their essential dimensions. In addition to the coverage of the specific topic, the book also includes a comprehensive description of the economic and social context of Ecuador with special reference to the agricultural sector and the migration context, including a historical review. Causes, development, and importance of international migration and remittances are highlighted against the background of the economic and political events. The specific results are achieved by clear and straightforward econometric calculations. In spite of their abundance in detail and their complexities Cristian Vasco succeeds in presenting them in a clear and transparent way. The author is careful in giving policy recommendations, but those which he provides are a logic conclusion from his empirical results The research shows a high scientific standard and at the same times an evident policy relevance. By this Cristian Vasco has succeeded in providing a substantial contribution to the global migration research which at the same time supplies important findings for development policy in Ecuador and beyond. Altogether this book also demonstrates the author’s strong competence as a promising young researcher.
Prof. Dr. Beatrice Knerr University of Kassel
ABSTRACT This research work analyzes the effects of international migration and remittances on agricultural production patterns, labor relationships, and entrepreneurship in rural Ecuador. Using a cross-sectional data base from the Living Standards Measurement Survey 2005-2006, this study first approaches the potential endogeneity of migration and remittances by testing the explanatory power and exclusion restriction condition of a set of instruments that were selected according to the suspected endogenous variable and the outcome variable under study in each case. Valid instruments for migration and remittances were used to run the Smith-Blundell test which rejected the hypothesis of exogeneity only for migration in the model of fertilizer expenditure. The results of a tobit and an instrumental variable tobit analyses show that everything else being equal, households having one or more migrants abroad spent more on fertilizers than non-migrants’ households. Instead, remittance inflows appear not to influence households’ fertilizer expenditure. These findings suggest that migrants’ households invest more in fertilizers in order to compensate the labor losses that international migration entails. By means of a probit analysis, it was possible to determine that ceteris paribus migrants’ households exhibit a higher propensity to acquire livestock than their counterparts without migrants. However, the monthly amount of remittances received by a household does not have any influence on the likelihood of cattle acquisition. These results are consistent with those of several other studies analyzing the impact of migration on farm activity choice and indicate that migrants’ households tend to switch from crop production to the less labor demanding cattle production in order to cope with household labor losses resulting from migration. Regarding labor relationships, neither migration nor remittances appear to influence reciprocal work in rural Ecuador. In contrast other factors like ethnicity of the household head and road infrastructure appear to play a determinant role in the odds for a household to participate in both community work and labor exchange agreements. Migrants’ households show a higher propensity to hire paid labor than their counterparts without migrants. This finding indicates that migrants’ households demand more wage labor to substitute household labor losses inherent to out-migration. 7
This study also finds that neither migration nor remittances have any effect on the likelihood for a household to own a rural enterprise. Factors such as education, credit and service infrastructure appear to be determinant in the likelihood of business ownership. Instead, international migration is associated with higher numbers of household members employed in a business. This suggests that more household labor is needed to compensate labor losses caused by international migration, or that businesses owned by migrants’ households depend on household labor for their functioning. Finally, the results of a tobit analysis indicate that neither migration nor remittances influence the number of non-household members employed in a rural business.
8
ACKNOWLEDGEMENTS I would like to express my gratitude to my supervisors Prof. Dr. Beatrice Knerr and Prof. Dr. Werner Trossbach for their patience, kindness and support during the completion of this dissertation. I would also like to thank my friends and colleagues from the Department of Development Economics, Migration and Agricultural Policy at the University of Kassel. My special thanks go to Ranjita Nepal, Alber El Tabiep, Wildan Syafitri, Jörg Helmke, Santiago Martínez and David Eche for their advice, support and invaluable friendship. I am also indebted to the German Academic Exchange Service (DAAD) for having funded my studies in Germany. Last but never least, I would like to express my deepest gratitude to my father, my mother, my brother Santiago, my sister Tania, little Sebastian, and little Axel. My modest efforts are dedicated to them.
9
Contents ABSTRACT ............................................................................................................................................ 7 ACKNOWLEDGEMENTS .................................................................................................................... 9 List of tables .......................................................................................................................................... 14 List of figures ........................................................................................................................................ 16 List of acronyms .................................................................................................................................... 18 List of definitions .................................................................................................................................. 19 1
2
3
Introduction ................................................................................................................................... 21 1.1
Statement of the problem and relevance of the research ....................................................... 21
1.2
Research questions ................................................................................................................ 23
1.3
Structure of the study ............................................................................................................ 24
Theoretical framework and research hypotheses........................................................................... 25 2.1
The New Economics of Labor Migration (NELM) ............................................................... 25
2.2
Research hypotheses.............................................................................................................. 26
State of research ............................................................................................................................ 29 3.1
Impact of international migration on agriculture................................................................... 29
3.2
Impact of international migration on labor relationships ...................................................... 30
3.2.1
Labor reciprocity ........................................................................................................... 30
3.2.2
Migration and labor reciprocity ..................................................................................... 32
3.3 4
5
Impact of international migration on rural entrepreneurship................................................. 33
Social and economic framework of migration from Ecuador ....................................................... 36 4.1
Basic information .................................................................................................................. 36
4.2
Development indicators ......................................................................................................... 39
4.3
The late 1990s Ecuadorian crisis ........................................................................................... 43
International migration and remittances in Ecuador...................................................................... 47 10
6
7
5.1
First stage of Ecuadorian international migration: the “American dream” .............................. (1961-1997) ........................................................................................................................... 47
5.2
Second stage of Ecuadorian international migration: the “Exodus to Spain” .......................... (1998-2003) ........................................................................................................................... 49
5.3
Remittances ........................................................................................................................... 53
Data and descriptive statistics ....................................................................................................... 62 6.1
Data base ............................................................................................................................... 62
6.2
Variables and descriptive statistics........................................................................................ 64
6.2.1
Migration and remittances ............................................................................................. 64
6.2.2
Farm characteristics ....................................................................................................... 67
6.2.3
Labor force sources ....................................................................................................... 70
6.2.4
Entrepreneurship............................................................................................................ 71
6.2.5
Household and household head characteristics ............................................................. 72
6.2.6
Access to services .......................................................................................................... 73
6.2.7
Home ownership and credit ........................................................................................... 74
Methodology ................................................................................................................................. 76 7.1
The probit model ................................................................................................................... 76
7.2
The tobit model ..................................................................................................................... 77
7.3
Endogeneity ........................................................................................................................... 80
7.3.1
The potential endogeneity of migration and remittances .............................................. 80
7.3.2
Causes of endogeneity and instrumental variables ........................................................ 81
7.3.3
Instrumental variables in the context of probit and tobit models .................................. 83
7.4 8
Testing for endogeneity and the validity of instruments ....................................................... 83
Effects of migration and remittances on fertilizer expenditure and cattle acquisition .................. 86 8.1
Literature background ........................................................................................................... 86
8.2
Specification and descriptive statistics .................................................................................. 87
8.3
Selection of instruments ........................................................................................................ 91 11
8.4 8.4.1
Endogeneity tests ........................................................................................................... 93
8.4.2
Regression analysis ....................................................................................................... 94
8.5 9
Results for fertilizer expenditure and cattle acquisition ........................................................ 93
Discussion ............................................................................................................................. 96
Effects of migration and remittances on labor relationships ....................................................... 100 9.1
Literature background ......................................................................................................... 100
9.2
Specification and descriptive statistics ................................................................................ 101
9.3
Selection of instruments ...................................................................................................... 103
9.4
Results for labor relationships ............................................................................................. 105
9.4.1
Endogeneity tests ......................................................................................................... 105
9.4.2
Regression analysis ..................................................................................................... 107
9.5
Discussion ........................................................................................................................... 110
10 Effects of migration and remittances on entrepreneurship .......................................................... 115 10.1
Literature background ......................................................................................................... 115
10.2
Specification and descriptive statistics ................................................................................ 116
10.3
Selection of instruments ...................................................................................................... 119
10.4
Results for entrepreneurship ................................................................................................ 120
10.4.1
Endogeneity tests ......................................................................................................... 120
10.4.2
Regression analysis ..................................................................................................... 123
10.5
Discussion ........................................................................................................................... 124
11 Summary and Conclusions .......................................................................................................... 129 11.1
Summary ............................................................................................................................. 129
11.2
Recommendations for policy makers and researchers ........................................................ 133
References ........................................................................................................................................... 135 Annex A .............................................................................................................................................. 147 Annex B............................................................................................................................................... 156 12
Annex C............................................................................................................................................... 170
13
List of tables Table 4.1
Urban illiteracy rates by sex and age group for Ecuador (%), 2008.................................. 40
Table 4.2
Net enrollment rates by region for Ecuador (%), 2006. .................................................... 40
Table 4.3
Health indicators for Ecuador and other Latin-American countries. ................................ 42
Table 4.4
Share of Ecuadorian children affected by malnutrition, and diarrheic and respiratory diseases (%), 2006.. ......................................................................................................... .42
Table 5.1
Education level of the Ecuadorian population (25 years and more) and of Ecuadorian immigrants in Spain (%). .................................................................................................. 50
Table 5.2
Emigration by sex in the cities of Quito, Guayaquil and Cuenca (%), 2004. ................... 52
Table 5.3
Distinctive features of the first and second Ecuadorian migratory stages. ....................... 53
Table 5.4
Remittances by geographical region, 2006-2009 (million US $). .................................... 56
Table 5.5
Use of remittances in 2003 and 2008 (%). ........................................................................ 59
Table 6.1
Description of the data bases used for this research. ........................................................ 63
Table 6.2
Shares of migrant and/or remittance recipient households (%). ....................................... 64
Table 6.3
Migrants’ gender by destination country (%). .................................................................. 67
Table 6.4
Remittances by country of origin...................................................................................... 67
Table 6.5
Descriptive statistics of landholdings for migrants’ and non-migrants’ households. ....... 68
Table 6.6
Average migrants’ and non-migrants’ landholdings for the provinces of Azuay, Cañar, Loja and Tungurahua. ....................................................................................................... 68
Table 6.7
Share of households that crop, have livestock, have acquired livestock, use fertilizers and use pesticides by household’s migratory status (%). ........................................................ 69
Table 6.8
Average number of cattle by household’s migratory status. ............................................. 70
Table 6.9
Shares of households hiring temporary workers, permanent workers, participating in community work and exchanging labor by household’s migratory status (%). ................ 70
Table 6.10 Households owning businesses by household’s migratory status (%). ............................. 71 Table 6.11 Average number of business per household, average number of business workers, average number of household members working in a business and average number of nonhousehold members working in a business by household’s migratory status. .................. 71 Table 6.12 Household head’s sex and ethnicity by household’s migratory status (%). ...................... 72 14
Table 6.13 Average household head’s age and average household head’s education by household’s migratory status (years). ................................................................................................... 72 Table 6.14 Household composition for migrants’ and non-migrants’ households. ............................ 73 Table 6.15 Average years of education of household members for migrants’ and non-migrants’ households. ....................................................................................................................... 73 Table 6.16 Access to electricity, piped water and indoor water system for migrants’ and nonmigrants’ households (%). ................................................................................................ 74 Table 6.17 Home ownership and access to credit for migrants’ and non-migrants’ households (%). 74 Table 7.1
Objectives, indicators, methodologies, and sample sizes. ................................................. 79
Table 8.1
Definitions and descriptive statistics of the variables. ...................................................... 89
Table 8.2
Instruments for migration and remittances having fertilizer expenditure and cattle acquisition as outcome variables. ..................................................................................... 92
Table 8.3
Determinants of fertilizer expenditures. ........................................................................... 95
Table 8.4
Determinants of cattle acquisition. ................................................................................... 96
Table 9.1
Definitions and descriptive statistics of the variables. .................................................... 104
Table 9.2
Instruments for migration and remittances having community work, labor exchange and use of wage labor as outcome variables. ......................................................................... 105
Table 9.3
Determinants of community work participation. ............................................................ 108
Table 9.4
Determinants of labor exchange participation. ............................................................... 109
Table 9.5
Determinants of wage labor use...................................................................................... 111
Table 10.1 Definitions and descriptive statistics of the variables. .................................................... 118 Table 10.2 Instruments for migration, remittances and average remittances at town level having business ownership, number of household and non-household members working in a business as output variables. ........................................................................................... 121 Table 10.3 Determinants of business ownership. ............................................................................. 124 Table 10.4 Determinants of the number of household members employed in a business. ............... 126 Table 10.5 Determinants of the number of non-household members employed in a business. ........ 127
15
List of figures Figure 4.1
Population pyramid for the total population, 2006. .......................................................... 37
Figure 4.2
Population pyramid for the urban population, 2006. ........................................................ 37
Figure 4.3
Population pyramid for the rural population, 2006. .......................................................... 38
Figure 4.4
Ecuadorian GDP 2007 (million US $). ............................................................................. 38
Figure 4.5
Share of exports value 2007 (in US $). ............................................................................. 39
Figure 4.6
Net primary and secondary enrollment rates for Ecuador and other Latin-American countries in 2007 (%). ....................................................................................................... 41
Figure 4.7
Public expenditure on education for Ecuador and other Latin-American countries, 20002008. ................................................................................................................................. 43
Figure 4.8
GDP and GDP variation for the period 1995-2003........................................................... 44
Figure 4.9
Poverty and extreme poverty in Quito, Guayaquil and Cuenca, 1995-2000. ................... 45
Figure 4.10 Income distribution, 1990-1999. ....................................................................................... 45 Figure 4.11 Evolution of the unemployment rate, 1995-2003. ............................................................ 46 Figure 5.1
Number of Ecuadorian migrants, 1976-1995.................................................................... 48
Figure 5.2
Permanent Ecuadorian migrants by country of destination, 1995-2000. .......................... 51
Figure 5.3
Evolution in the number of Ecuadorian migrants, 1996-2004. ......................................... 51
Figure 5.4
Evolution of remittances sent to Ecuador, 1993-2008 (million US $).............................. 54
Figure 5.5
Remittances as share of GDP, 1993-2008 (%). ................................................................ 55
Figure 5.6
Remittances by country of origin, 2005-2008. ................................................................. 55
Figure 5.7
Use of remittances arriving in Ecuador, 2003................................................................... 57
Figure 5.8
Use of remittances in the Ecuadorian province of Loja, 2004 (%). .................................. 58
Figure 5.9
Use of remittances in the southern part of Quito, 2004 (%). ............................................ 58
Figure 5.10 Use of remittances in Quito, Guayaquil, Cuenca, Loja and Azogues, 2008 (%). ............. 59 Figure 6.1
Destination country of migrants........................................................................................ 65
Figure 6.2
Migration to the United States by province. ..................................................................... 66 16
Figure 6.3
Migration to Spain by province. ....................................................................................... 66
17
List of acronyms BCE
Banco Central del Ecuador
CEDATOS
Centro de Estudio de Datos
CEPAL
Comisión Económica para América Latina y el Caribe
DNS
Dirección Nacional de Migración
FLACSO
Facultad Latinoamericana de Ciencias Sociales
GDP
Gross Domestic Product
GNI
Gross National Income
IADB
Inter-American Development Bank
IMF
International Monetary Fund
INEC
Instituto Nacional de Estadísticas y Censos
LSMS
Living Standards Measurement Survey
NELM
New Economics of Labor Migration
UNDP
United Nations Development Programme
18
List of definitions Household: According to the U.S. Census Bureau (1990) a household “includes all the persons who occupy a housing unit. The occupants may be a single family, one person living alone, two or more families living together, or any other group of related or unrelated persons who share living arrangements”. Nevertheless, INEC (2007b) defines as a household the person or group of persons who sleep under the same roof and cook their food separately from others. Low income countries: According to the World Bank (2010) this group includes countries which have a GNI per capita that is below US $ 995. Middle income countries: This category includes lower middle income countries which have a GNI per capita between 996 and 3,945 US $ and upper middle income countries which have a GNI per capita between 3,946 and 12,195 US $ (World Bank, 2010). High income countries: Countries that have a GNI per capita higher than 12,196 US $ (World Bank, 2010). Poverty and Extreme poverty: The World Bank (1990) defines poverty as “the inability to attain a minimal standard of living which is measured not only in terms of income and expenditures per capita but also in terms of nutritional status, life expectancy, under 5 year’s mortality and school enrollment. The World Bank (2001) broadens the concept and includes vulnerability, exposure to risk, voicelessness and powerlessness as forms of poverty. INEC (2006) defines as the extreme poor the households which have an income that is not sufficient to meet the basic nutritional requirements of their members, and as poor the households which have a per capita consumption that is lower than the national poverty line. ational extreme poverty line: Defined by INEC (2006) as the monetary value of a basic food basket to meet 2,141 kcal per person per day. ational poverty line: It results from dividing the national extreme poverty line by the Engel’s coefficient which is determined by the relationship between food and total consumptions. 19
Poverty gap: It is the amount of income required to bring every poor person exactly up to the poverty line. It is useful to show the depth of poverty. Household business: According to INEC (2007b), this is any economic activity including production, transformation, mines exploitation, construction, transportation, retailing and services, carried out by household members. This definition excludes purely agricultural exploitations.
20
1 Introduction International migration from rural areas is a trend that has deserved considerable attention from researchers since the late 1990s. Given the importance of farming activities in rural regions, several studies have focused on analyzing the impacts of migration and/or remittances on rural income (Miluka et al., 2007; Pfeiffer and Taylor, 2007), farm activity choice (Pfeiffer and Taylor, 2007; Wouterse and Taylor, 2008) asset accumulation (Adams, 1998; Lucas, 1987; Taylor, 1992), farm technical efficiency (Mochebelele and WinterNelson, 2000; Wouterse, 2008) adoption of technology (Mendola, 2008; Quinn, 2009), agricultural input expenditure (Gray, 2009; Miluka et al., 2007) and labor force demand (Gray, 2009; Salas Alfaro and Pérez Morales, 2007). In the case of rural Ecuador, international migration is not a recent phenomenon. Several researchers (Gratton, 2006; Jokisch and Kyle, 2006) point out that migratory movements from Ecuador to the United States were started by peasants from the southern Ecuadorian Highlands1 who traveled to New York in the 1960s. Such migratory flows were intensified due to the effects of a severe economic crisis that affected Ecuador in the late 1990s2. This time, international migration was not only directed to the United States but also to Spain and Italy (Gratton, 2006; Jokisch and Pribilsky, 2002). The effects of these migratory movements on sending communities remain to be quantitatively assessed.
1.1
Statement of the problem and relevance of the research
Although the largest share (73%) of Ecuadorian migrants come from urban areas (Herrera et al., 2006), Ramírez Gallegos and Ramírez (2005) report that the counties with the largest shares of international migration are located in rural regions. These facts have motivated a number of social researchers to study the effects of international migration on rural areas of Ecuador. Such works conclude that migration and remittances have led to social differentiation between migrants’ and non-migrants’ households (Borrero, 2002; Caguana,
1
According to (Jokisch, 2001) the southern Ecuadorian Highlands are one of the largest international migrantsending regions in South America. 2
At the end of the 1990s, Ecuador experienced a very serious economic crisis featured by a dramatic fall in incomes, a substantial growth of unemployment rates and a very fast process of impoverishment among the population (Acosta et al., 2006; Ramírez Gallegos and Ramírez, 2005). 21
2008a; Jokisch and Kyle, 2006; Martínez, 2006a), increase of land prices (Borrero, 2002; Caguana, 2008a; Carpio, 1992; Jokisch, 2002; Pribilsky, 2007), loss of community leaders (Caguana, 2008b; Camacho and Hernández, 2009; Martínez, 2002; Martínez, 2006a) and weakening of reciprocity traditions (Caguana, 2008b; Martínez, 2002; Martínez, 2003; Martínez, 2006a). As gains from migration and remittances, some authors address the improvement of the living standards for migrants’ households (Carpio, 1992; Kyle, 2000; Pribilsky, 2007), the empowerment gained by women due to men’s absence (Borrero, 2002; Ortega, 2008) and the enhancement of the social and economic positioning of formerly discriminated indigenous peoples who by means of migration have been able to improve their social status (Camacho and Hernández, 2009; Carpio, 1992). Regarding migration and its effects on agriculture, numerous studies (Caguana, 2008a; Jokisch and Lair, 2002; Kyle, 2000; Martínez, 2006b; Pribilsky, 2007) conclude that labor losses caused by international migration have driven migrants’ households to switch from cropping to livestock production. Jokisch (2002) argues that international migration and remittances have allowed migrants’ households to accumulate more land than their counterparts without migrants. Despite the fact that remittances are not invested in farming activities, their inflow seems not to be associated with abandonment of subsistence farming (Jokisch, 2002; Jokisch and Pribilsky, 2002; Pribilsky, 2007). Special attention has been paid to the implicit loss of labor that comes together with international migration. To illustrate, Martínez (2006a) and Pribilsky (2007) observe an increment of the agricultural wages in migrant-sending areas of rural Ecuador. In addition, ancient traditions involving labor force exchange among community members and communal work would have been undermined owing to intense out-migration (Caguana, 2008b; Camacho and Hernández, 2009; Martínez, 2002; Martínez, 2003; Martínez, 2004a; Martínez, 2006a). Remittances allow migrants’ households to cope with inflated wages (Jokisch, 2002; Kyle, 2000; Pribilsky, 2007). However, that would not be the case for non-migrants’ households for whom abandonment of reciprocal work may seriously compromise their possibilities of farming and hence their livelihood strategies. In the case of rural entrepreneurship, the literature concerning the topic is restricted to some anecdotal evidence or local cases of study. To illustrate, Caguana et al. (2008) as well as Camacho and Hernández (2009) report that remittance inflows and the construction boom 22
associated with it, have triggered the number of hardware stores and vehicles offering transport services in southern rural Ecuador. Similarly, Pribilsky (2007) reports that most grocery stores in his research area were owned by returnees. Despite of the arguments presented above, scarce empirical research has been carried out to analyze the impacts of migration and remittances on rural regions of Ecuador. This research work quantitatively approaches the effects of international migration and the subsequent remittance inflows on: a) agricultural production patterns, b) labor relationships, and c) entrepreneurship in rural Ecuador. In this way, it contributes to the debate about the effects of international migration on rural regions and provides policy makers as well as rural development and migration practitioners with information and facts that should be taken into account for the design of projects linking international migration with rural development.
1.2
Research questions
As mentioned above, the general objective of this thesis is analyzing the effects of international migration on rural areas of Ecuador, more specifically; it provides answers to the following research questions: • Have migration and/or remittances driven migrants’ households to switch from cropping to cattle holding? • Have migration and/or remittances impelled the expenditure on agricultural inputs? • Have migration and/or remittances undermined labor exchange and community work traditions in rural Ecuador? • Everything else held constant, do migrants’ households rely more on wage labor than their counterparts without migrants? • Have migration and/or remittances stimulated the formation of rural businesses in rural Ecuador? • What is the effect of migration and/or remittances on the number of household and non-household members employed in rural businesses?
23
In order to address these research questions, this study will use data from the “Living Standards Measurement Survey” 2005-2006 (LSMS), a survey with national representation conducted by the National Institute of Statistics and Censuses (INEC). The estimation of the impact of migration and remittances on cattle acquisition, labor relationships and business ownership will be carried out by means of probit models. In the case of fertilizer expenditure and the number of household and non-household members employed in a business the analysis will rely on tobit models.
1.3
Structure of the study
As for the rest of this research work, it is organized as follows: Chapter 2 elaborates upon the theoretical framework used for this study. Chapter 3 offers a literature review about the effects of migration and remittances on agricultural production, labor relationships and entrepreneurship. Chapter 4 introduces Ecuador, the research country. The analysis starts by detailing some facts about geography, demography and economy. Moreover, this chapter presents some of the country’s development indicators and analyzes the effects of the 1990s economic crisis. In Chapter 5, the history of Ecuadorian international migration as well as the evolution, importance and uses of remittances are presented. Chapter 6 provides information about the LSMS 2005-2006 and shows the descriptive statistics of the data base. In Chapter 7, the general empirical strategy is approached. The basics of probit and tobit models are reviewed and the use of instrumental variables in a context of limited dependent variable models is analyzed. Chapter 8 individually analyzes the effects of migration and remittances on cattle acquisition and fertilizer expenditure. Chapter 9 deals with migration, remittances and their impact on labor relationships in rural Ecuador. Chapter 10 approaches the effects of international migration, remittances and aggregate remittances on rural entrepreneurship. Finally, Chapter 11 presents the summary and the conclusions.
24
2 Theoretical framework and research hypotheses According to several researchers (de Haas, 2007; Taylor, 1999; Taylor et al., 1996) the New Economics of Labor Migration (NELM) has substantially changed the way in which migration is studied. It interlinks the causes and consequences of migration in a way that positive and negative effects on sending economies are possible (de Haas, 2007). This chapter focuses on reviewing the NELM, explaining why it was chosen as the theoretical background for this study, and establishing the research hypotheses.
2.1 The New Economics of Labor Migration (NELM) The New Economics of Labor Migration (NELM) (Stark, 1980; Stark and Bloom, 1985; Stark and Levhari, 1982) holds that migration is part of a household strategy the aim of which is to cope with credit and insurance market failures often prevalent in developing countries3. By means of remittances, migrants’ households are able to reduce risks by diversifying their income sources, and to access to new technologies and inputs that can enhance agricultural production (Taylor, 1999; Taylor et al., 1996). Therefore, in absence of efficient credit and insurance schemes migrants’ households are able to invest in new production techniques and are protected against income risk by sending one or more migrants abroad. Thus, while in earlier research migration is treated as an individual decision (Harris and Todaro, 1970; Todaro, 1969), in the NELM the decision making unit is the household, which will share the costs of migration and also the returns. To illustrate, if a farm experiences crop failures, remittances become an alternative income source that allows migrants’ households to secure their livelihoods. Similarly, credit constrained households will hardly be able to adopt new technologies through investments. In the case of migrants’ households, such constraints could be overcome if remittances are invested in productive activities. Migration entails household labor losses that may reduce agricultural production in the short run. Nevertheless, this might be compensated or even over-compensated as remittances are invested in inputs or new technologies
3
There is no general international agreement on the definition of “developing countries”. The World Bank (2010) classifies “middle” and “low income” countries as “developing countries”. This research work adheres to this definition. 25
The NELM holds that migrants remit not only with the intention of supporting relatives left behind (altruistic motivation) but also due to self-interest. In this sense, Taylor (1999) implies that households and migrants are attached by informal agreements for mutual benefit and income insurance. To illustrate, Lucas and Stark (1985) find that migrants from Botswana tended to remit more when the expectancies of being favored in future bequests are higher. On the contrary, the authors report that migrants from lower income households do not remit more which is inconsistent with the hypothesis of pure altruism. Similar results were shown by Schrieder and Knerr (2000) in their research on Cameroon. These positive effects on investments in new technologies and inputs may be constrained by the loss of household labor resulting from migration. A number of studies (Lucas, 1987; Taylor, 1992; Taylor et al., 1996; Taylor and Wyatt, 1996) point out that international migration is associated with a decline of production in the short run due to labor losses but with an enhancement of it in the long run when remittances are invested in new technologies and inputs which counteract the negative effects of labor losses. A number of researchers (de Brauw et al., 1999; de Haas, 2006; Kilic et al., 2007; McCarthy et al., 2006; Miluka et al., 2007; Pfeiffer and Taylor, 2007; Quinn, 2009; Taylor, 1992; Taylor, 1999; Taylor et al., 1996; Taylor and Wyatt, 1996; Wouterse and Taylor, 2008) adhere to the “New Economics of Labor Migration” (NELM) as the best theoretical foundation to explain the effects of migration and remittances at micro-economic level. This research follows this stream of literature and will test the research hypotheses using the NELM as the theoretical framework.
2.2
Research hypotheses
Fertilizer expenditure is expected to be increased with migration due to the fact that migrants’ households should invest more in fertilizers to balance labor losses. In the same way, remittances might allow recipient households to buy inputs to be able to increase their incomes as found by Gray (2009). In this sense, the first null hypothesis for this study is: • H1: Migration and remittances do not have any effect on households’ fertilizer expenditure.
26
Having Albania as a case of study and using the NELM as their theoretical foundation, Miluka et al. (2007) and McCarthy et al. (2006) state that migrants’ households tend to switch from crop to livestock production due to the fact that the latter is less labor demanding and yields almost the same income. If such a conclusion is applicable to Ecuador, migration and remittances should increase the likelihood for migrants’ households to buy cattle. In order to address the second research question of this study, the following null hypothesis will be tested: • H2: Migration and remittances have no influence on the probability for a household to acquire livestock. Concerning labor relationships, migrants’ households are expected to be less participative in community work because their labor pool has been reduced as a consequence of outmigration. In addition, it is reported that remittance inflows have weakened community ties in sending regions (Caguana, 2008b; Camacho and Hernández, 2009; Martínez, 2002; Martínez, 2003; Martínez, 2004a; Martínez, 2006a). In order to test if these statements hold for Ecuador, the following null hypothesis is proposed: • H3: Migration and remittances have no effect on the probability for a household to participate in community work. Labor exchange traditions might also be affected by international migration and its resulting household’s labor reduction. Furthermore, remittances could be used to hire wage labor making participation in labor exchange agreements unnecessary for migrants’ households (Pribilsky, 2007). In a different scenario, migration could reinforce labor exchange agreements as reported by Carpio (1992). In order to assess the effect of international migration and remittances on labor exchange, the following null hypothesis will be tested: • H4: Migration and remittances have no effect on the likelihood for a household to take part in labor exchange force arrangements. Remittances are expected to provide migrants’ households with the resources to hire wage labor for agricultural tasks as proposed by Jokisch (2002). However, Carpio (1992) reports that wage labor is avoided in Ecuadorian indigenous communities affected by international migration. In this sense, the null hypothesis to be tested is as follows: 27
• H5: Migration and remittances have no effect on the demand for wage labor. International migration of one or more household members might encourage rural households to start household businesses as a way to diversify their income sources and hence reduce their income risk (Kilic et al., 2007). Remittances can be used to finance such entrepreneurial activities. On the other hand, remittances may prevent migrants’ households from investing in entrepreneurial activities because they may raise the reservation wages of household members, inducing greater spending on housing, education and leisure as proposed by Amuedo-Dorantes and Pozo (2006). Taking these two positions into consideration, the next null hypothesis is proposed: • H6: Migration and remittances have no effect on the propensity for a household to own a rural enterprise. Several scholars (Massey and Parrado, 1998; Taylor, 1999; Taylor et al., 1996) have proposed that remittance flowing into rural regions might not only be beneficial to migrants’ households but also to other members of the community due to the spending and transfer effects of remittances at community level. If that is the case, households in towns receiving more remittances should exhibit a higher propensity to own rural enterprises. To assess this statement the following null hypothesis is proposed: • H7: The aggregate effect of remittances at town level does not influence the likelihood for rural households to start businesses. Migration of one or more members may drive migrants’ households to use more household labor to fill the labor gap left by international migrants as suggested by Canales and Montiel Armas (2004), or may impel migrant enterprises to hire wage labor as implied by Kilic et al. (2007). In the case of this research work, the following null hypotheses will be tested: • H8: Migration and remittances do not influence the number of household members employed in a rural business. • H9: Migration and remittances do not influence the number of non-household members employed in a rural business.
28
3 State of research There are several ways in which migration can affect farmers’ lives. First migration involves household labor losses which can bring negative outcomes regarding crop and livestock production. At the same time, the inflow of migrants’ remittances may enable receiving households to overcome capital and risk constraints which otherwise would prevent them from, for example, investing in new technologies or diversifying their income sources. This chapter focuses on reviewing the academic literature concerning migration and its impact on rural regions as well as the relationship between remittances and farming decisions.
3.1
Impact of international migration on agriculture
Academic literature concerning migration, remittances and their effects on rural regions has been dominated by the debate between two groups of researchers who have analyzed the effects of international migration from the perspective of different disciplines. In the late 1970s and early 1980s several researchers (Reichert, 1981; Stuart and Kearney, 1981; Wiest, 1979) studied the effects of international migration in Mexican sending communities and concluded that rather than promoting development, remittances may hamper it by giving origin to a series of negative effects such as dependency, engagement in conspicuous consumption, social differentiation between migrants’ and non-migrants’ households, inflation of land prices and stagnation of productive activities. During the 1990s a second group of researchers (Durand et al., 1996a; Durand et al., 1996b; Massey et al., 1987) presented a more promising view of remittances and their contribution to development. According to this group, remittances have served to improve agricultural production by allowing recipient households to buy inputs, to grow market demanded cash crops, to expand irrigation, and to overcome credit constraints. Along this line Durand et al. (1996b) state that migration detractors fail to consider the multiplier effects of remittances as they are spent on goods and services locally produced. Furthermore, remittances are indeed spent on productive activities when the proper conditions for investment are given (Durand and Massey, 1992; Taylor, 1999; Taylor et al., 1996). Adams (1991) concludes that migrants’ households in rural Egypt show a higher tendency to invest than their non-migrant equivalents as the acquisition of agricultural and building land are the main choices for investment allocation. In the case of Pakistan, Adams (1998) reveals 29
that international remittances play a significant role in the acquisition of land, while they exhibit no effect on livestock accumulation. In contrast, Lucas (1987) suggests that remittances earned by migrants working in South African mines are associated with livestock accumulation and crop production improvement in the long run. In this line, Mochebelele and Winter-Nelson (2000) emphasize that remittances received by migrants’ households in Lesotho allow them to overcome financial constraints that otherwise would prevent them from carrying out timely and regular farm management activities. Migration is also reported to be positively associated with the adoption of high yield varieties in Bangladesh (Mendola, 2008) and Mexico (Quinn, 2009). For Albania, McCarthy et al. (2006) report a raise of rural income which is linked to the change from staples to livestock production. These results are consistent with Miluka et al. (2007) who find that Albanian migrants’ households invest less in crop production inputs than their counterparts without migrants and that they would rather turn their investments to livestock production. A similar shifting pattern from staple to livestock production is registered by Wouterse and Taylor (2008) for migrants’ households in Burkina Faso. In the case of Ecuador, Jokisch (2002) reports that international migration and remittances have allowed migrant households to accumulate more land than their counterparts without migrants. However, the author finds no relationship between international migration and the amounts of fertilizers applied to crops. Conversely, Gray (2009), who tries to disentangle the effects of migration from those of remittances, finds that remittances from abroad positively affect the expenditure on agricultural inputs but have no effect on the area cultivated with subsistence crops; and that male migration is associated with a reduction in maize production.
3.2
Impact of international migration on labor relationships
3.2.1 Labor reciprocity Reciprocal labor has been reported to be a widespread practice in rural areas of developing countries. Literature about the topic includes descriptions of these activities in countries as diverse as Peru (Erasmus, 1956; Guillet, 1980; Mitchell, 1991), Ecuador (Erasmus, 1956; Ferraro, 2004), Mexico (Cohen, 1999), Cameroon (Geschiere, 1995), Nepal (Adams, 1992), Indonesia (Gilligan, 2004) and Yemen (Aw-Hassan et al., 2000). According to Guillet (1980) 30
almost no socioeconomic study performed in the Andes fails to refer to reciprocal labor in any of its forms4. In this sense, the seminal work presented by Erasmus (1956) identifies two kinds of reciprocal work: exchange and festive labor. In the first case, a household in need of labor will ask neighbors, relatives or friends for help to carry out a task. There is an implicit obligation for hosts to reciprocate the labor they benefited from in the same quality and amount when required. In the context of festive labor, a host invites his peers to a working day in which extraordinary food and drink will be served. The host benefits from the labor needed, e.g. to build a house or to harvest his crops as the attendees are to some extent rewarded with the special attention received while working. In the Ecuadorian context, Ferraro (2004) distinguishes two types of reciprocal labor relationships: labor exchange, mostly known as “prestamanos”; and communal work widely referred to as “minga” in the Ecuadorian highlands. In the “prestamanos” system an individual asks others for labor support in order to carry out agricultural tasks with the compromise of reciprocating the service in the future. Instead, the “minga” is used to accomplish works that benefit the whole community, such as maintenance of irrigation systems or expansion of the electrical network. As the works are of collective benefit, the participation of all community members is compulsory. Regarding the motivations for exchanging labor, some studies (Mayer, 1974; Mayer and Zamalloa, 1974; Sánchez Parga, 1984) point out that by means of this practice, rural Andean households have been able to access the human resources required to carry out agricultural tasks which could not be accomplished without extra household labor. In the case of communal work, the incentives for participation are mainly linked to the collective benefits to be gained from community works or projects (Cohen, 1999; Ferraro, 2004), the possibility of accessing communal natural resources (Mayer, 1974), and even the reinforcement of the community identity (Cohen, 1999; Sánchez Parga, 1984). However, reciprocal labor is reported to be weakened due to the dissemination of cash crops, the difficulty of arranging working days at everybody’s convenience, the demand for off-farm labor and the high costs and poor quality of festive labor (Erasmus, 1956; Mitchell, 1991). Under such conditions, the use of reciprocal labor would be restricted to liquidity constrained households in regions 4
For a detailed analysis of all variants of reciprocal labor in the Andes see: Erasmus (1956), Mayer (1974), Mayer and Zamalloa (1974) and Guillet (1980). 31
suffering from labor scarcity. Challenging this position, Guillet (1980) argued that reciprocal labor survives even in monetized areas with no wage labor scarcity and where cash crops have been already introduced. Guillet’s argument is that peasants are rational actors who first analyze the costs and returns before choosing between reciprocal, wage or even both forms of labor. Using quantitative methods Gilligan (2004) is able to conclude that labor exchange coexists with wage labor in several regions of the world and that it is an alternative to paid work in an environment of high labor transaction costs.
3.2.2 Migration and labor reciprocity Loss of household labor due to migration is reported to deteriorate reciprocal labor activities in Mexico (Cohen, 1999; Mutersbaugh, 2002), Peru (Mitchell, 1991) and Yemen (Aw-Hassan et al., 2000). In the case of Ecuador, most studies analyzing rural out-migration refer to its effects on reciprocal work. A number of sociological studies (Caguana, 2008b; Camacho and Hernández, 2009; Martínez, 2002; Martínez, 2003; Martínez, 2004a; Martínez, 2006a) report a decline of reciprocity practices in migrant-sending regions. The authors suggest that labor losses resulting from migration do not allow migrants’ households to exchange labor due to the fact that they cannot compromise to reciprocate labor that they themselves lack. It is also stated that by means of remittances, migrants’ households are able to overcome liquidity constraints that hitherto prevented them from hiring wage laborers for agricultural tasks. Therefore, engaging in labor exchange agreements when wage labor is affordable would become unsuitable and unnecessary. Furthermore, the inflow of remittances into rural areas is reported to have undermined cohesion and solidarity traditions within communities. In this sense, migrants’ households would exhibit a more individualistic behavior and would not care so much about the labor requirements of their community fellows. In his study of a migrant-sending region in Southern Ecuador, Pribilsky (2007) reports that households are still able to fulfill their labor requirements by turning to extended family members’ cooperation. Nevertheless, when the time for labor-demanding agricultural chores such as plowing and weeding comes, extra labor is needed. Such additional work, formerly acquired via labor exchange, is now met by hiring wage laborers. Similar conclusions are drawn by Jokisch (2002) who also observes the difficulties faced by households which do not receive remittances in raising the money necessary to hire labor. In contrast, Carpio (1992) notes that the loss of labor resulting from international migration encourages reciprocal work 32
as a strategy to meet agricultural labor needs in indigenous communities of the Ecuadorian highlands, and that the use of hired labor remains marginal. Similarly, Gray (2009), in a quantitative analysis, finds that female migration positively influences reciprocal labor participation and is negatively associated with the demand for hired labor. Based on these results, the author associates female migration with a shift from the use of wage labor towards reciprocal labor. In short, the existing literature regarding migration and labor relationships presents mixed results. One group of authors suggests that migration and remittances have undermined reciprocal work practices, a second group observes that reciprocal and wage labor coexist in migrant-sending regions, while a third implies that labor scarcity due to migration stimulates labor exchange participation.
3.3
Impact of international migration on rural entrepreneurship
Inspired by the significant increase that the volume of remittances flowing into developing countries has experienced since the 1990s, a number of researchers and organizations (IADB, 2006; IMF, 2010; Ratha, 2003; Solimano, 2003; World Bank, 2006) have repeatedly highlighted the potential of such international money transfers to drive development in migrant-sending countries. Within such research spheres, it is broadly accepted that remittances can significantly contribute to the economic development of sending countries by maintaining macroeconomic stability, reducing poverty and inequality; smoothing consumption for remittance recipient households, increasing investment in education and health, and promoting small business formation, among other positive effects. Nevertheless, it is also implied that the impact of remittances on development could be stronger if a larger share of them would be invested in productive activities. In this way they would not only benefit remittance receivers but also others through dynamization of local economies and employment generation (IADB, 2006; Solimano, 2003). Several researchers and organizations (OECD, 2006; Petrin, 1990; Petrin, 1994; Stark and Markley, 2008) have upheld the concept of promoting entrepreneurship as an intervention able to prompt development in rural regions. Increasing the number of entrepreneurs may boost development in rural areas by providing local people with off-farm jobs, diversifying local economies, increasing tax revenue for the public sector, promoting the use of local 33
resources, and hence raising living standards within communities (OECD, 2003; Stark and Markley, 2008). In the particular case of Ecuador, Lanjouw (1998) argues that non– agricultural activities offer a way to leave poverty and that an expansion of off-farm jobs would be associated with a decline of income inequality. However, rural entrepreneurs in developing countries generally face several constraints such as: poor infrastructure and services, deficient education schemes, and lack of credit for business formation. Such intrinsic problems critically reduce the odds of success for entrepreneurial activities in rural areas. In this regard, Petrin (1994) and Paulson and Townsend (2004) list the lack of credit as one of the biggest drawbacks for rural business formation. For migrants’ households, such financial limitations can be overcome using remittances as a source of capital for productive activities (Lucas and Stark, 1985; Taylor, 1999). Further on, even when migrants’ households themselves do not invest; the demand for certain goods and services that comes together with remittance flows can drive local entrepreneurial activities and employment generation (Durand and Massey, 1992; Taylor, 1999; Taylor et al., 1996). In this sense, Massey and Parrado (1998) state that international migration may enhance business formation in two ways: by providing migrants’ households with capital to invest and by stimulating the demand for goods and services due to the aggregate effect of remittances arriving in sending regions. Massey and Parrado (1998) conclude that “migradollars”, as they label remittances flowing from the United States into Mexico play a significant role in the process of business formation, both at household and community levels. Furthermore, neither migration nor remittances have any effect on the number of household and non-household members working in a business, which drives the authors to conclude that the migration process itself cannot be blamed for the small size and little employment generation often attributed to migrant businesses. In contrast, Canales (2008) and Canales and Montiel Armas (2004) argue that businesses funded with remittances are characterized by low investment, dependency on household labor, and incapability of generating paid jobs, thus their multiplier effects, under such conditions, are rather limited. Another group of researchers has reported a positive relationship between return migration and entrepreneurship. For instance, Arif and Irfan (1997) conclude that savings accumulated by Pakistani migrants during their working time in Middle Eastern countries allowed them to 34
switch from production and service employment to business and agricultural activities as they returned to Pakistan. De Haas (2006) found that more than 35% of the Moroccan returnees in his sample invest in private businesses. After controlling for the endogeneity of temporary migration with respect to the odds of having a business, Wahba and Zenou (2009) are able to determine that the savings and human capital acquired by Egyptian temporary migrants while abroad increase their probability of becoming entrepreneurs in spite of the loss of local networks that international migration may entail. Using instrumental variable methods to account for the endogeneity of the migration decision with regards to business ownership, Kilic et al. (2007) find a positive correlation between past international migration experience and the likelihood of owning a household business in Albania.
35
4 Social and economic framework of migration from Ecuador This chapter begins by presenting some geographic, demographic, and socio-economic facts for Ecuador, the country which is the focus of this study. Next, Ecuador’s development is discussed and contextualized by presenting several development indicators. Finally, the late 1990s economic crisis and its consequences for Ecuador are reviewed.
4.1
Basic information
Ecuador is an Andean country bounded by Colombia in the North, Peru in the South and the East, and the Pacific Ocean in the West. The 283,561 km² of national territory are distributed in four natural regions: “La Sierra” (The Highland) which is shaped by high-altitude mountain chains and crossed by the Andes range; “La Costa” (The Coast) that comprises the lowlands lying in the western part of the country next to the Pacific Ocean; “El Oriente” formed by the Amazon Rainforest in the eastern part of the country which accounts for almost half of the total national area, and finally the insular region of Galapagos which gathers several small islands in the Pacific Ocean. INEC (2010a) estimates Ecuador’s population to be around 14,212,147. The population density is 50.12 inhabitants per km². Figure 4.1, 4.2 and 4.3 display the population pyramids for the total, urban and rural populations respectively. The pyramids show a genderequilibrium. Nevertheless, the pyramid for rural areas suggests depopulation of those older than 20 years. The predominant ethnic group is the “Mestizo” which accounts for the 65% of the total population, followed by the “Indian” group with 25%, whites with 7-10% of the current population and a minority of Afro-Ecuadorians who are mainly settled on the northern coast of the country. INEC reveals that 61% of the population is concentrated in cities. According to the IMF (2010) Ecuador had a GDP of US $ 57,303 million in 2009 while the GDP per capita for the same year reached US $ 4,059. Figure 4.4 shows the distribution of the Ecuadorian GDP. Oil, the most important commodity in the Ecuadorian export basket, alone accounts for 16% of the total GDP. Underlining the extent of the dependency of the Ecuadorian economy on oil exports, Figure 4.5 shows that oil exports represent 72% of the
36
total export revenues. Other important exportable commodities are bananas5, shrimps, canned fish and cocoa. Figure 4.1 Population pyramid for the total population, 2006.
Source: INEC (2007b) Figure 4.2 Population pyramid for the urban population, 2006.
Source: INEC (2007b)
5
Ecuador is the world’s largest banana exporter. 37
Figure 4.3 Population pyramid for the rural population, 2006.
Source: INEC (2007b) Figure 4.4 Ecuadorian GDP 2007 (million US $).
Source: BCE (2010) The Ecuadorian economy is based on exports of a variety of agricultural commodities and oil. Due to the instability in the prices of primary products in international markets, the Ecuadorian economy is influenced by sudden changes in the supply and demand of these commodities. A clear example of this occurred in the late 1990s when a sharp reduction in the crude oil prices badly hit the Ecuadorian economy (Larrea, 2004). Almost simultaneously, 38
many banana plantations were destroyed by floods resulting from “El Niño” climatic phenomenon. As a consequence, the exports of this commodity fell. Both incidences significantly contributed to the 1999 economic crisis, which itself was one of the main reasons for Ecuadorians to leave the country6. Figure 4.5 Share of exports value 2007 (in US $).
Source: BCE (2010)
4.2
Development indicators
According to the World Bank classification, Ecuador is a lower middle income country that has a human development index of 0.806 (UNDP, 2009), which situates Ecuador in the group of countries with a high human development index. However, 42.7% of the population lived under the poverty line in 2008. In rural areas, the poverty rate rose to 50% while 39% of the urban population was considered poor (CEPAL, 2009). The poverty gap coefficient amounts to 14.7% whilst the Gini coefficient adds up to 0.480 in urban areas and 0.458 in rural regions (CEPAL, 2009). Unemployment affected 7.9% of the population in 2009 (INEC, 2010b). Nevertheless, this comparatively low unemployment rate hides the fact that 50.5% of the population is underemployed.
6
The causes and effects of the 1999 Ecuadorian crisis will be independently analyzed in the next section. 39
According to CEPAL (2009) 7% of the population was illiterate in 2005. This percentage is higher for women (8.3%) than for men (5.6%). Table 4.1 displays urban illiteracy rates by sex and age group. Although the absolute numbers suggest that women are more likely to be illiterate than men, this trend seems to diminish and even to be reversed for younger age groups. Furthermore, the average illiteracy rate appears to be lower for younger generations. Table 4.2 shows the net enrollment rates at national level and by region. Enrollments for both primary and secondary education are lower in rural than in urban areas. This difference is even more evident for secondary enrollment in urban areas which is twice as high as that reported in rural areas. Figure 4.6 displays the net primary and secondary enrollment for Ecuador and other Latin-American countries. Ecuador shows one of the highest primary enrollment rates within the group. Nevertheless, the secondary enrollment rate appears as the lowest. Table 4.1 Urban illiteracy rates by sex and age group for Ecuador (%), 2008. Age groups Sex 15-24
25-34
35-44
45-59
60 and
Total
over Both sexes
3.6
6.3
7.9
11.4
19.5
10.6
Men
3.1
7.5
6.7
8.2
15.8
8.6
Women
3.3
5.3
8.8
13.9
22.6
12.4
Source: CEPAL (2009) Table 4.2 Net enrollment rates by region for Ecuador (%), 2006. ational
Urban
Rural
Pre-primary
60.4
67.4
50.7
Primary
89.4
91.1
86.8
Secondary
65.9
76.3
43.2
Source: INEC (2007b)
40
Figure 4.6 Net primary and secondary enrollment rates for Ecuador and other LatinAmerican countries in 2007 (%).
Source: CEPAL (2009) Table 4.3 compares some health indicators with those of other Latin-American countries. Ecuador exhibits a low dietary energy supply when compared with other Andean countries (Colombia and Peru). This difference is even bigger if compared with other South American countries such as Chile and Argentina. The average underweight for age and under height for age scores rank as the highest within the group of countries selected which reveals the prevalent malnutrition problems among Ecuadorian children. The maternal mortality rate appears as the highest when compared with the rest of the countries within the group. Table 4.4 shows some health indicators for Ecuador. Malnutrition as well as diarrheic and respiratory diseases are still affecting a considerable share of the population. The problem of malnutrition is more wide-spread in rural areas where 26.1% of the children suffer from stunted growth (low height for age) and 11.2% are underweight (low weight for age).
41
Table 4.3 Health indicators for Ecuador and other Latin-American countries. Country
Maternal mortality rate a (per 100,000 live births)
Under 5 mortality rate (per 1,000 live births)
Under weight for age
Under height for age
Dietary energy supply (kcal/day per person)
Ecuador
210
25
9
23
2,300
Colombia
130
20
7
12
2,670
Peru
240
24
5
30
2,450
Argentina
77
16
4
4
3,000
Chile
16
9
1
1
2,980
Source: CEPAL (2009) Table 4.4 Share of Ecuadorian children affected by malnutrition, and diarrheic and respiratory diseases (%), 2006. Indicator
ational
Urban
Rural
Chronic malnutrition (height/age)
18.1
12.7
26.1
Global malnutrition (weight/age)
8.6
6.8
11.2
Acute malnutrition (weight/height)
1.7
1.5
2.0
Children with diarrheic diseases
25.0
23.1
27.7
Children with respiratory diseases
56.0
56.2
55.7
Source: INEC (2007b) Figure 4.7 shows the public expenditure on education for Ecuador and other Latin-American countries. It can be seen that Ecuadorian expenditure on health was very low (less than 1% of the GDP) in 2000. This is associated with the severe economic crisis that affected Ecuador in the late 1990s and which drastically reduced the country’s income. Despite the efforts to reverse this trend, Ecuadorian expenditure on health still ranks among the lowest in the LatinAmerican context.
42
Figure 4.7 Public expenditure on education for Ecuador and other Latin-American countries, 2000-2008.
Source: CEPAL (2009)
4.3
The late 1990s Ecuadorian crisis
Carrying out a study of Ecuador’s international migration without referring first to the late 1990s economic crisis would result in an incomplete analysis. As already mentioned, Ecuador exhibits a frail economy, very dependent on exports of raw materials that have unstable prices on international markets. Hence, any reduction in the prices of Ecuador’s exportable products will strongly hit the Ecuadorian economy. Besides, a reduction in production of agricultural commodities due to e.g. natural causes will also affect the national budget. This happened during the period 1997-1998 when “El Niño”, a natural periodic phenomenon, caused the most intense floods of the century. The production of exportable and non-exportable agricultural commodities was seriously affected (Salgado, 2000). As a result of this natural disaster, banana exports were reduced and infrastructure was severely damaged. The social and economic costs were very high for a country with a minimally diversified economy highly dependent on exports of primary products. The situation became even worse considering that the price of the oil barrel, the main source of fiscal income, dropped from US $ 18 in 1996 to US $ 9.2 in 1998. Additionally, the weak financial system was affected by the South East Asian crisis, the consequence of which for Latin-America was an abrupt withdrawal of short-term capital from the region. Banks, already affected by the overdue payments resulting from the “El Niño” 43
losses, were also hit by the massive withdrawal of speculative capital and the cutting of international credit lines (Larrea, 2004). In 1999, the government of Jamil Mahuad injected private banks with government funds in order to avoid their fall. Moreover, the president Mahuad froze the bank accounts intending to stop the drain of capital and to stop the inflation rate which had reached 60%. In this sense, CEDATOS (1999) estimates that US $ 3,851 million were frozen by April 1999. These efforts were useless and the financial system finally collapsed. According to Salgado (2000) the costs of this economic crisis by 2000 would reach US $ 4,000 million, 25% of the GDP for that year. In January 2000, the president Jamil Mahuad announced the substitution of the “Sucre”, the prior national currency, with the US Dollar with the desperate aim of reducing the inflation that in 1999 reached 60%. Protests arose, Mahuad fell and was replaced by Gustavo Noboa who continued with Mahuad’s plan of dollarizing the economy. The real magnitude and dimension of the crisis for the population are shown by indicators such as growth, poverty, inequality and unemployment. Figure 4.8 displays the evolution of GDP and GDP growth over the period 1995-2003. The economic growth in 1999 dropped sharply, GDP fell by more than 7%. Additionally, the GDP per capita fell from US $ 2,035 in 1998 to US $ 1,429 in 1999. According to Acosta et al. (2006), during this period Ecuador experienced the fastest impoverishment process in Latin-America before the Argentinean crisis. Figure 4.8 GDP and GDP variation for the period 1995-2003.
Source: BCE (2006) 44
The number of poor increased from 3.9 to 9.1 million between 1990 and 2000, while extreme poverty affected 4.4 million people in the same year. In the three biggest cities of the country the poverty rate rose from 34 to 71% of the population as the extreme poverty jumped from 12 to 31% between 1995 and 2000 (Figure 4.9). According to Acosta et al. (2006) another distinctive feature of this period was an increase of inequality. Figure 4.10 shows that in 1990 the poorest 20% of the population received 4.6% of the national income while by 2000 this value was reduced to 2.5%. In contrast, the richest 20% of the population increased their share in the total income from 52 to 61%. Figure 4.9 Poverty and extreme poverty in Quito, Guayaquil and Cuenca, 1995-2000.
Source: INEC Figure 4.10
Income distribution, 1990-1999.
Source: INEC 45
In addition, the unemployment rate rose to levels without precedents. In 1999 unemployment affected 14.4% of the population, which was more than twice the number of unemployed in 1995 (Figure 4.11). Middle-class citizens including permanent workers, public employees, self-employed and the retired saw their income and standard of living seriously reduced with the crisis. Furthermore INEC reports that 57% of the population was underemployed in 1999. The percentage of urban jobless grew from 4.1% in 1995 to 9.1% in 1997. The unemployment rate for men rose from 7% in 1997 to 11% in 1999 (INEC, 2010b), while the share of unemployed women went up from 13 to 20% during the same period. Figure 4.11
Evolution of the unemployment rate, 1995-2003.
Source: INEC Altogether, at the end of the 1990s, Ecuador experienced one of the worst economic crisis throughout its history (Acosta et al., 2006). It was characterized by a dramatic reduction of income, and considerable increment of unemployment, poverty and inequality. Under such circumstances, many Ecuadorians saw international migration as the only opportunity to ensure their families’ survival.
46
5 International migration and remittances in Ecuador Several authors (Acosta et al., 2006; Gratton, 2006; Jokisch and Pribilsky, 2002) divide Ecuadorian international migration in two stages. The first started in the 1960s and lasted till 1998 whilst the second was triggered by the late 1990s economic crisis and continued till 2003. Although both processes were carried out by impoverished people who left the country looking to improve their economic situation and support their families from abroad, there are marked differences regarding the geographic and social origin of the migrants as well as the intensity and duration of the processes. In this chapter both migratory stages, as well as the volumes, uses and effects of remittances arriving in Ecuador will be analyzed.
5.1
First stage of Ecuadorian international migration: the “American dream” (1961-1997)
Some researchers (Altamirano, 2006; Zambrano, 1998) propose that the pioneers in migrating to the United States were former employees of multinational banana traders who took advantage of their connections to stay and to settle in the New York area. Gratton (2006) and Kyle (2000) state that large-scale Ecuadorian migration to the United States started in the 1960s, was performed by inhabitants of the southern provinces of Azuay and Cañar, and was prompted by the fall in the sales of the “Panama hat” which had become the main source of income in these regions. Gratton (2006) explains that after the Second World War the “Panama hat” was no longer considered in fashion by the “modern men”, resulting in a sharp decline of the sales. The same researcher mentions that the core of the distribution network of the “Panama hat” was located in New York. The collapse of the “Panama hat” market led to high levels of unemployment and poverty in Cañar and Azuay. After this crisis some wealthy exporters migrated to Chicago and New York where they had established links with importers. Similarly, peasants, who had become more dependent on “Panama hat” exports than on agriculture itself, also began migrating (Jokisch, 2001) . The trend for women was to migrate to big cities in Ecuador while men tended to go to the United States. At the beginning, this first migratory process was predominantly male. In connection to this, Jokisch (2001) found that in some communities of southern Ecuador the relationship between male and female inhabitants is 60:100. 47
Ecuadorian out-migration rose remarkably during the 1980s coinciding with the fall of oil prices and the foreign debt crisis (Ramírez Gallegos and Ramírez, 2005) (Figure 5.1) and tended to stabilize by the 1990s with an average of 30,000 migrants per year (Gratton, 2006). Data about the number of Ecuadorians living in the United States differ from one source to another. The information provided by the U.S. census is not reliable because of inaccuracies in the questions about the nationality of the migrants surveyed (Jokisch, 2001; Jokisch and Pribilsky, 2002) and the reluctance of irregular migrants to be registered by the government, facts that could have led to underestimations in the number of Ecuadorians living in the United States. Jokisch and Kyle (2006) estimate that as many as 600,000 Ecuadorians live in the United States. From this amount, as many as 150,000 migrants came from Cuenca, the capital of the province of Azuay and its surrounding areas. By 2000, 64.3% of the Ecuadorian migrants in the United States had settled in New York, being “Queens” the borough with the highest concentration of Ecuadorians. Figure 5.1 Number of Ecuadorian migrants, 1976-1995.
Source: Ramírez and Ramírez Gallegos (2005) Until 1985, irregular migrants usually made their way to the United States by flying to Mexico and then crossing the U.S.-Mexican border by land. By 1990, the trend was to buy false U.S. visas and to borrow US $ 7,000-9,000 from “chulqueros”7 to pay the services of the
7
Usurers who loan money at monthly interests ranging between 5 and 15% and open mortgages. In regions with migratory tradition, “chulqueros” make the contact between the potential migrant and the “coyote”. 48
so called “coyotes” who would arrange the trip from Ecuador to New York (Jokisch, 2001; Jokisch and Pribilsky, 2002). However, the increasingly strict migratory controls in the United States, Mexico and Guatemala have raised the cost of travelling to the United States. In the 2000s, the trend is to go by ship to Guatemala and continue to Mexico by land to finally cross the U.S. border. The price for this trip has reached US $ 13,000 (El Comercio, 2009). Although the trip is dangerous, ships can be overcrowded and several have sunk; many people still risk their lives in order to reach the United States. In 2005 the Ecuadorian Army intercepted 13 ships and rescued 994 travelers while in 2006 the number rose to 1,009 migrants rescued (El Universo, 2006). Migration and remittances are reported to have changed the landscape of the provinces of Azuay and Cañar. Jokisch and Kyle (2006) recount that luxurious houses built next to subsistence crops are common in small towns in the countryside. The same authors observe that transnational migration has created “ghost towns” containing large, neglected houses built in 1980s and 1990s. Borrero (2002) reports that migrants’ demand has tremendously raised the price of land to a point where it is unaffordable for non-migrants’ households. Remittances are also associated with an increase in inflation rates. To illustrate, Cuenca, the capital of the province of Azuay, ranks among the most expensive Ecuadorian cities in terms of daily living expenses (Acosta et al., 2006).
5.2
Second stage of Ecuadorian international migration: the “Exodus to Spain” (1998-2003)
According to Jokisch and Pribilsky (2002) the first migratory networks between Ecuador and Spain were established by indigenous peoples coming from the provinces of Imbabura and Loja. However, Pedone (2006) states that although migratory movements to Spain were started by indigenous Ecuadorians from those regions, migratory networks between Ecuador and Spain were started by “mestizo” populations. De Prada (2005) speculates that Ecuadorian migration to Spain was encouraged by Catholic priests and missionaries who traveled to Ecuador in the late 1950s. According to Gratton (2006) the economic collapse in the late 1990s hit middle-class educated citizens who saw their income and standard of living greatly affected. These people chose international migration as a strategy to cope with the crisis. To support this argument Gratton compared the education level of Ecuadorian migrants surveyed after a process of 49
regularization in Spain in 2000 with those reported by INEC (2001) for the Ecuadorian population (Table 5.1). The data suggest that on average Ecuadorian migrants in Spain had accomplished more years of schooling than their non-migrant counterparts. Table 5.1 Education level of the Ecuadorian population (25 years and more) and of Ecuadorian immigrants in Spain (%). Education level
Ecuador (2001)
Ecuadorians in Spain (2000)
Less than primary school
11.8
0.1
Primary school
44.9
27.0
Secondary school
27.7
46.3
College
15.6
20.4
Source: Gratton (2006) In contrast to the first migratory wave, the majority of migrants of the second wave came from urban areas. Quito and Guayaquil, the largest cities in the country, were the most affected by this phenomenon. Due to the fact that the economic crisis affected all Ecuadorians, international migration turned into an overall national phenomenon. As a result of the restrictive migratory controls in the United States, Mexico and other Middle-American countries, paying for the services of “coyotes” became increasingly expensive. This drove many Ecuadorians to change their migratory patterns. Within a short period of time Spain and other European countries turned into destination countries for a considerable number of Ecuadorians, in a trend referred by Goycoechea and Ramírez Gallegos (2002) as “seeing the future in Spain”. Figure 5.2 shows that the proportion of migrants travelling to Spain during the period 1995-2000 jumped from 15% to 53%. In contrast, migration to the United States fell by more than a half during the same period. However, it still accounted for 30% of the total number of migrants during the period 19952000.
50
Figure 5.2 Permanent Ecuadorian migrants by country of destination, 1995-2000.
Source: Acosta et al.(2006) Figure 5.3 shows that the number of Ecuadorians who left the country jumped from 45,332 in 1998 to 108,837 in 1999, reaching the highest peak in 2000 with 158,369. The number slightly declines in 2002 and 2003 to drastically fall in 2004 to 64,081 migrants. This drop is due to the restrictive steps taken by the European Union against Ecuadorian migration and the increasingly rigorous controls of U.S. borders. (Acosta et al., 2006; Ramírez Gallegos and Ramírez, 2005) Figure 5.3 Evolution in the number of Ecuadorian migrants, 1996-2004.
Source: Based on DNM 51
Acosta et al. (2006) report that as many as 700,000 Ecuadorians left the country during the second stage of international migration. From this amount, 350,000 migrated to Spain, 200,000 to the United States, 70,000 to Italy and 40,000 to the rest of Europe and America. The authors also observe that the real number of Ecuadorians who migrated to Spain during this period (1998-2003) was higher than 500,000. Another distinctive characteristic of the second stage of international migration is that it was largely started by women. In 1997, 67.4% of the Ecuadorians holding work permits in Spain were women. This tendency was caused by an increase in the supply of jobs traditionally considered for women such as: indoor servants, careers for the elderly, children, or disabled people. In later years, migration to Spain became predominantly male; in 2000 the percentage of women with work permits in Spain had fallen to 55.5%. However, in the region of Murcia where the most important industries, construction and agriculture, demand mainly male labor, the number of male migrants is reported to be larger (Acosta et al., 2006). In this context, Martínez (2004b) found that in Cuenca, where emigration is mostly directed to the United States through a long and risky journey, the majority of migrants are men. Instead, in Quito and Guayaquil, where female migration is directed mainly to Spain, there is gender equilibrium (Table 5.2). Table 5.3 shows the specific characteristics of the two stages of Ecuadorian migration. Table 5.2 Emigration by sex in the cities of Quito, Guayaquil and Cuenca (%), 2004. Quito
Guayaquil
Cuenca
Men
50.7
49.9
66.9
Women
49.3
50.1
33.1
Source: Martínez (2004b)
52
Table 5.3 Distinctive features of the first and second Ecuadorian migratory stages. Features
First stage of migration
Geographic origin
Mostly restrained to the National distribution southern provinces of Azuay and Cañar
Gender
Principally beginning
Destination country
The United States
Spain
Period of time
1961-1997
1998-2003
Number of emigrants8
700,000
700,000
male
at
Second stage of migration
the Mainly female beginning
at
the
Source: Author’s own elaboration
5.3 Remittances As an outcome of the massive international migration movements during the late 1990s, the volume of remittances from abroad arriving in Ecuador also grew tremendously. Figure 5.4 shows the evolution of remittances between 1994 and 2008. Between 1994 and 1997 remittances increased by US $ 120 million per year on average. Between 1998 and 2000 remittances grew US $ 262 million on average. This growing trend is attributable to the intense migratory movements over that period. There is a stagnation period between 2001 and 2003 when remittances grew by US $ 78 million per year on average. Between 2004 and 2007 a trend of sharp growth can be observed. During this period, remittances increased by 23.3% per year on average, which can be associated with the massive regularization processes that took place in Spain in 2005 which allowed many Ecuadorians to have access to better jobs and hence to remit larger sums. However, remittances diminished by 19.2% between 2007 and 2009 as a result of the global financial crisis.
8
Acosta et al. (2006) estimate the number of Ecuadorian living overseas in 1,400,000, but remark that the real number could reach 2,500,000. 53
Figure 5.4 Evolution of remittances sent to Ecuador, 1993-2008 (million US $).
Source: Based on data from BCE In fact, remittances from abroad became the second source of external income for Ecuador being only surpassed by oil exports. Figure 5.5 displays the evolution of remittances as share of the Ecuadorian GDP. A sustained growing trend can be distinguished between 1993 and 1998. The peak is reached in 2000, the worst year of the crisis, when remittances accounted for 8.3% of the GDP. The value decreases to 5.6% in 2004 which is attributable to a recovery in oil prices and a slight reactivation of the economy (Acosta et al., 2006). There is another growing trend between 2004 and 2006 which can be ascribed to a better work positioning of Ecuadorian migrants in Spain as well as to a depreciation of the US Dollar with respect to the Euro during this period (BCE, 2007). Finally, there is a decreasing trend in 2008 and 2009 as a consequence of the world’s economic crisis, which hit the economies of the United States and Spain, the major host countries for Ecuadorian migrants, especially hard.
54
Figure 5.5 Remittances as share of GDP, 1993-2008 (%).
Source: Based on data from BCE Figure 5.6 shows the shares of remittances by country of origin for the period 2005-2009. During these years the largest share of remittances came from the United States followed by Spain. Both together have never accounted for less than 86% of total remittances arriving in Ecuador. Figure 5.6 Remittances by country of origin, 2005-2008.
Source: BCE (2009) 55
Table 5.4 shows the evolution of remittances by geographical region in Ecuador. The “Costa” (the Coast) increased its share in 2008 and 2009 with respect to 2007 which suggests that migrants from this region remitted more during the world financial crisis. Instead, in the Austro, a region with longstanding tradition of migration to the United States, the amounts remitted decreased between 2007 and 2009. Remittances sent to the “Sierra” (the Highlands) maintained a relative equilibrium in spite of the crisis while remittances sent to the “Oriente” (the Amazon rainforest) also experienced a reduction. Table 5.4 Remittances by geographical region, 2006-2009 (million US $). 2006
2007
2008
2009
Costa
856.7
917.5
1,167.4
980.1
Austro
819.7
1310.1
899.7
745.0
Sierra
708.6
740.0
686.3
726.4
Oriente
66.6
120.2
67.8
43.2
Source: BCE (2009) According to a study presented by the IADB (2003), about one million Ecuadorians (14% of the population in 2003) were remittance receivers. Such money transfers would have allowed receiver households to complement their budget and to alleviate poverty (Acosta et al., 2006). The IADB (2003) concludes that 67% of the remittances arriving in Ecuador are used to cover regular expenditures such as food, rent, basic services, medicines, and so on (Figure 5.7). 8% is destined for business investments, another 8% is saved in banks, 4% is used to buy property, and 2% is spent on education. 17% is spent on what migrants’ relatives call “luxuries”; this group includes entertainment, cars, brand name clothes, and trips to visit relatives living abroad. According to IADB (2003) and Acosta et al. (2006) the observed aversion to invest is associated with the low amounts of the transfers, the uncertainty about the country’s future together with a generalized distrust of the financial system after the economic crisis.
56
Figure 5.7 Use of remittances arriving in Ecuador, 2003.
Source: IADB (2003) Studies carried out in specific Ecuadorian regions or cities report slightly different results. For instance, in a survey carried out in the province of Loja by the Department of Human Mobility of the Social Pastoral of Loja in 2002, only 53% of the money remitted was used to cover regular expenditures (Figure 5.8), while 15% is invested. The repayment of debts incurred in order to migrate demands 21% of remittances. According to Sánchez (2004) almost 86% of the migrants from Loja work in Spain; further on Jokisch (2001) states that migratory movements from Ecuador to Spain were started by people from Loja who went to Spain in the early 1990s. Therefore, migrants from Loja belong to a relatively late migration movement and hence they must still devote a considerable part of the money they earn in Spain to repay the debts they contracted in order to migrate. 5% of the remittances arriving in Loja are oriented to finance migration of another family member. López and Villamar (2004) present similar results for the southern part of Quito, the capital of the country. Figure 5.9 shows that 57% of the transfers are used to meet regular expenditures while 20 % is dedicated to repay the money borrowed to migrate. 17% of the money received is invested. Again, a majority (63%) of migrants migrated to Spain in the second wave of migration; hence they must still send money to pay off the travel debt incurred in Ecuador.
57
Figure 5.8 Use of remittances in the Ecuadorian province of Loja, 2004 (%).
Source: Sánchez (2004) Figure 5.9 Use of remittances in the southern part of Quito, 2004 (%).
Source: López and Villamar (2004) González Casares et al. (2009) study the use of remittances in the cities of Quito, Guayaquil, Cuenca, Loja and Azogues; the top remittance-receiving cities in Ecuador. The authors report that 54% of remittances are spent on current consumption, 21% is devoted to health and education, 8% is saved in banks, 7% is used for long term investments while 4% is still dedicated to repay debts (Figure 5.10). Table 5.5 compares the shares of expenditure presented by both the IADB (2003) and González Casares et al. (2009) regarding current consumption, education, savings and 58
investment. Between 2003 and 2008, current consumption expenditures (food, health, housing, basic services) increased. There is also a significant increase in the expenditures on education; they jumped from 2% in 2003 to 8% in 2008. However, this increase is less than that reported by Ponce et al. (2008) who find that 18.14% of remittances were spent on education in 2007. On the other hand, the share of remittances invested on businesses declined from 8 to 5% between 2003 and 2008. The share of remittances saved remained unchanged in both studies (8%). Figure 5.10
Use of remittances in Quito, Guayaquil, Cuenca, Loja and Azogues, 2008 (%).
Source: González Casares et al. (2009) Table 5.5 Use of remittances in 2003 and 2008 (%). 2003
2008
Current consumption
61
67
Education
2
8
Business investment
8
5
Savings
8
8
Source: IADB (2003) and González Casares et al. (2009) 59
A few quantitative analyses have been carried out in Ecuador in order to measure the impact of remittances on development indicators. In general they present mixed results. To illustrate, Acosta et al. (2007) find a positive effect of remittances on children’s accumulated years of schooling in urban areas. Calero et al. (2009) suggest that remittance inflows raise the probabilities for children to attend school and lower the likelihood for children to work, especially for girls in rural areas. Further on, the authors state that remittances are used to keep children at school when migrants’ households face economic shocks. Pacheco (2007) investigates the effects of remittances on rural children’s school performance. He finds no significant differences between the school performance of children from remittance recipient households and that of children from non-remittance receiver families. Using data from the LSMS 2005-2006 for Ecuador, Ponce et al. (2009) find no significant effects of remittances on prevalence of malnutrition, prevalence of respiratory diseases, prevalence of diarrhea and z scores among Ecuadorian children. The authors do not find any effect of remittances on access to health services provided a household member is sick. On the other hand, remittances are reported to have positive effects on health expenditures. An increment of US $ 10 in per capita remittances is reported to lead to an increase of 0.9% in health expenditures. Relying on the same data base, Antón (2010) determines that remittances positively affect weight for age and weight for height z scores but have no effect on height for age scores among children aged 30)
0.92
0.87
Total
Source: Author’s own calculations with data from the LSMS 2005-2006 Table 6.15 Average years of education of household members for migrants’ and nonmigrants’ households.
Migrants’ households
umber of households 307
Mean (years) 5.39
Std. deviation (years) 2.68
4,491
4.48
2.38
4,798
4.99
2.92
Non-migrants’ households Total
Source: Author’s own calculations with data from the LSMS 2005-2006
6.2.6 Access to services Table 6.16 shows that 96.4% of migrants’ households have access to electricity while this percentage is reduced to 86.3% for non-migrants’ households. Availability of piped water is more common among migrants’ households (47.7%) than among non-migrants’ homes (31.9%). Similarly, migrants’ households are more likely to have an indoor water connection than their counterparts without migrants. Taking services as proxies for wealth, these 73
descriptive statistics reinforce the statement that international migration is more common among better-off households who have the means to finance migration. Table 6.16 Access to electricity, piped water and indoor water system for migrants’ and nonmigrants’ households (%). Migrants’ households
on-migrants’ households
Electricity
96.4
86.3
Access to piped water
47.7
31.9
Indoor water system
41.2
21.9
Source: Author’s own calculations with data from the LSMS 2005-2006
6.2.7 Home ownership and credit Table 6.17 shows that most households within the sample, regardless if they have migrants abroad or not, have a home of their own. Access to credit seems to be a drawback in the rural Ecuadorian context. 9.2% of the migrants’ and 9.5% of non-migrants’ households report having received credit. Table 6.17 Home ownership and access to credit for migrants’ and non-migrants’ households (%). Migrants’ households
on-migrants’ households
Home ownership
82.8
84.4
Access to credit
9.2
9.5
Source: Author’s own calculations with data from the LSMS 2005-2006 In sum, the descriptive statistics suggest that international migration from rural Ecuador is mainly directed to the United States. Nevertheless migration to Spain is also important. Migration to the United States is mainly concentrated in two provinces (Azuay and Cañar) while migration to Spain is nationally dispersed.
74
The number of non-migrants’ households receiving remittances is higher than that of migrants’ households which are remittance recipients. The probabilities of cropping or having livestock seem not to be affected by the household’s migratory status. In contrast, migrants’ households seem to use less fertilizer than their counterparts without migrants. Concerning livestock assets, migrants’ households appear to have more cattle than their non-migrants’ counterparts. Migrants’ households seem to demand more wage labor than their counterparts without migrants. Surprisingly, migrants’ households appear to be more participative in community works. Instead, international migration seems not to affect households’ participation in labor exchange agreements. Regarding entrepreneurship, migrants’ households appear to be more likely to own businesses than non-migrants’ households. However, migrants’ households seem to be more likely to use household (unpaid) labor. The heads of households with migrants abroad are slightly older than their counterparts without migrants. The education of the head is higher for migrants’ than for non-migrants’ households. The average education of the household members is higher for migrants’ homes. Families headed by females are more common among migrants’ homes than among nonmigrants’ households. In general, migrants’ households present a higher share of female members than their counterparts without migrants. Migrants’ households are more probable to have access to electricity, piped water and indoor water systems than are non-migrants’ homes. Finally, the probabilities of having a home of one’s own and having access to credit seem not to be affected by household’s migratory status.
75
7 Methodology In order to estimate the likelihood for a household to acquire livestock, to participate in community work, to engage in labor exchange agreements, to hire wage agricultural labor, and to own businesses, this study relied on probit models while for the number of household and non-household members employed in a business as well as for the spending on fertilizers; tobit models were used. This chapter offers an overview of these econometric methods as well as of the potential endogeneity problems that may arise when using migration and remittances as predictors.
7.1
The probit model
Maybe the simplest way to estimate binary response models is the linear probability model. It consists of running a simple OLS regression with a dummy as the left-hand side variable. When using this methodology the assumption of homoskedasticity would be violated and the predicted probabilities could exceed the bounds of 0 and 1 (Carter Hill et al., 2008; Gujarati, 2004; Studenmund, 2001). To cope with these limitations a more elaborated binary response model such as the probit model can be used. Following Woolridge (2002b) a binary response model is given by: P(y =1 | x) = G (β0+β1x1+…+βkxk) = G (β0+xβ)
(1)
where P is the probability for y to be 1, G is a function with values strictly between 0 and 1, 00) observations. y is a variable taking the value of zero for a non-irrelevant number of observations but is continuous with strictly positive values. Modeling this kind of corner solutions outcomes with OLS methods would result unsuitable (Wooldridge, 2002a). The main drawback of using OLS estimation to model corner solutions outcomes is the possibility of yielding negative fitted values which would bring in negative predictions of y (Wooldridge, 2002b).
77
Likewise the probit, the tobit model can be defined in terms of a latent or index variable. Following Greene (2003): yi* = xiβ + εi
(9)
yi = 0 if yi* ≤ 0 yi = yi* if yi* > 0 The latent variable yi* must have a normal and homoskedastic distribution with linear conditional mean. The observable output variable y will equal 0 if the unobservable variable y*≤0 and will take on the value of y* provided y*>0. Since y* is normally distributed, y will also be normally distributed over strictly positive values. Like in a probit analysis, the results of a tobit model cannot be directly inferred. The coefficients of a tobit analysis indicate to what extent a change of one unit in x can affect the latent variable y*. However, the variable under analysis is generally y. The marginal effects for the tobit model can be estimated for the latent dependent variable y*, the expected value of y for uncensored observations, and the unconditional value of y. In this study, the marginal effects will be calculated for the unconditional value of y by means of the following expression:
∂E ( y ) X β = Φ i β k ∂xk σ
(10)
Xβ Where Φ i β k is the estimated probability of observing an uncensored observation at σ such values of X. Table 7.1 summarizes the objectives, the indicators, the methodology and the sample sizes. The sample size for the estimation of the number of household and non-household members working in a rural business is of 1,425 households which report owning a household business. All calculations were carried out using the statistical software STATA 1013.
13
For a complete description of the software see Baum (2006). 78
Table 7.1 Objectives, indicators, methodologies, and sample sizes. Objective
Indicator
Methodology
To estimate the effects of migration and remittances on agricultural production patterns
Log of the households’ fertilizer expenditure
Tobit analysis
Sample size (households) 4,720
Household has acquired livestock in the year preceding the survey
Probit analysis
4,720
Household participates in community work
Probit analysis
4,720
Household takes part in labor exchange agreements
Probit analysis
4,720
Household has hired agricultural wage labor
Probit analysis
4,720
Household owns a rural business
Probit analysis
4,753
Number of household members employed in a household business
Tobit analysis
1,425
Number of nonhousehold members employed in a household business
Tobit analysis
1,425
To estimate the effects of migration and remittances on labor relationships
To estimate the impact of migration and remittances on rural entrepreneurship
79
7.3
Endogeneity
This part gives an overview of endogeneity and the use of instrumental variables in the context of limited dependent variable models. Identification strategies as well explanatory power of instruments and endogeneity tests will be presented in the following chapters.
7.3.1 The potential endogeneity of migration and remittances A growing number of empirical papers studying the effects of migration and remittances have addressed the endogeneity of the latter with respect to educational attainment (Calero et al., 2009; Hanson and Woodruff, 2003; López-Córdova, 2006; Mansuri, 2006; McKenzie and Rapoport, 2006), health outcomes (Antón, 2010; Hildebrandt and McKenzie, 2005; LópezCórdova, 2006; Ponce et al., 2009), entrepreneurship (Amuedo-Dorantes and Pozo, 2006; Kilic et al., 2007; Wahba and Zenou, 2009), rural income (de Brauw et al., 1999; McCarthy et al., 2006; Miluka et al., 2007), technology adoption (Mendola, 2008; Quinn, 2009) among other outcomes that are affected by decisions made at household level. Further on, Taylor and Mora (2006) warn about the endogenous nature of migration and remittances and conclude that studies ignoring such threats take the risk of yielding biased estimators. Although most contemporary studies rely on the use of instrumental variables, it is not the only way to deal with the potential endogeneity of migration that can be found in the literature. In this sense, a group of authors (Gray, 2008a; Gray, 2008b; Gray, 2009; Wouterse and Taylor, 2008) imply that the extent to which endogeneity can become a source of bias depends on the particular characteristics of each case of study14 . The authors argue that by removing from the model variables that affect both migration and the outcome variables, including control variables that account for household characteristics (Gray, 2009), lagged household assets (Wouterse and Taylor, 2008), and being careful when interpreting the results, the effects of endogeneity can be counteracted.
14
When modeling for activity choice in Burkina Faso Wouterse and Taylor (2008) claim that the absence of land markets limits the scope to which migration could affect land accumulation and hence the probability for their models to be endogenous. Similarly, Gray (2008a) argues that the rareness of land sales in southern rural Ecuador reduces the chances of reverse causality between migration and land ownership. 80
7.3.2 Causes of endogeneity and instrumental variables The meaning of the term endogeneity is: “determined within the system” (Carter Hill et al., 2008). Basically, endogeneity arises when one or more explanatory variables are correlated with the error term. Violating the OLS assumption that Cov(x,ε)=0 would bring in biased and inconsistent estimators. The problem of endogeneity can be caused by omitted variables, sample selection bias, causality problems, simultaneity and measurement errors (Carter Hill et al., 2008; Greene, 2003; Wooldridge, 2002b). An alternative to cope with endogeneity are instrumental variables. By definition an instrument is a variable that itself does not belong to the model but is correlated with the endogenous variable. Following Wooldridge (2002b) and given the model: y = β0 + β1 x + ε
(11)
The error term ε is suspected to be correlated with x: Cov(x,ε)≠0. A variable z is a suitable instrumental variable for x provided it meets two conditions: Cov(z,ε)=0 and Cov(z,x)≠0. Normally, one must rely on economic theory and common sense to determine whether Cov(z,ε)=0 or not. Instead, Cov(z,x)≠0 can be tested using an equation of the following form: x = δ0 + δ1z + v
(12)
If the null hypothesis that δ1=0 can be rejected at sufficient significance level, it is possible to conclude that z has explanatory power with regard to x. For equation (11): Cov(z y) = β1*Cov(z,x) + Cov(z,ε)
(13)
Assuming that: Cov(z,x)≠0, and Cov(z,ε)=0:
β1 =
Cov( z , y ) Cov( z , ε )
and the instrumental variable (IV) estimator:
81
(14)
n
∑(z - z)(y - y) i
βˆ1 =
i
i=1 n
(15)
∑(z - z)(x - x ) i
i
i=1
The standard errors for instrumental variables is σ2/SSTxR2x,z. As R2x,z0] In order to estimate a bivariate model with an endogenous predictor, it must be assumed that (ε1, v2) has 0 mean, bivariate normal distribution and is independent from z. For tobit models the model would be: y1*=max(0, z1δ1 + α1y2 + ε1)
(24)
y2=z1δ21 + z2δ22 + v2 = z2δ2 + v2
(25)
and
7.4
Testing for endogeneity and the validity of instruments
Nowadays most quantitative studies on migration and remittances rely on the use of instrumental variables. Nevertheless, the use of instrumental variables entails a cost in terms of efficiency of estimators. It is advisable to use a test of exogeneity to determine whether or not instrumental variables are needed (Wooldridge, 2002b). A number of tests (Hausman-Wu, Smith-Blundell, Wald) have been proposed to test for endogeneity. Nevertheless, such methodologies are based on the assumption that instruments are valid. For an instrument to be 83
valid, it must fulfill two conditions: it must have explanatory power with respect to the suspected endogenous variable (migration, remittances and town remittances in this case) and not directly influence the outcome variable (exclusion restriction). This study relies on the Smith-Blundell test (Smith and Blundell, 1986) to detect if the treatment variables are endogenous or not. It focuses on testing endogenity in limited dependent variable models and is available in STATA 10 under the commands probexog and tobexog for probit and tobit models respectively. In order to avoid the pernicious effects of weak instruments15, the explanatory power and the exclusion restriction of instruments were tested before applying the Smith-Blundell test of exogeneity. In the first case, instruments were included as predictors in the first stage regression in order to estimate their explanatory power with regards to the suspected endogenous variable. Concerning the exclusion restriction, instruments were included as regressors in the original model to test if they significantly affect the output variable under study. Another drawback of exogeneity tests may be their sensitivity to specification. To explain, the test could reject the null hypothesis of endogeneity under one specification and fail to reject it if a different specification is used. In order to cope with this threat, the exogeneity test was run with three different specifications; the first included household and household head characteristics, the second also considered land and home ownership as well as credit, services and road infrastructure while the third added a set of provincial dummies to the model. STATA 10 allows for the use of instrumental variables in probit and tobit models with the commands ivprobit and ivtobit respectively. However, the use of these commands is restricted to continuous endogenous explanatory variables and is not appropriate for discrete endogenous regressors (StataCorp, 2007). Provided that the null hypothesis of exogeneity is rejected and considering that this study is particularly interested in the effects of a dummy variable taking the value of 1 if the household has international migrants and 0 otherwise;
15
The smaller the correlation between the instruments and the endogenous variable, the larger the standard errors of the instrumental variable estimators will be. Furthermore, low correlations between the instruments and the endogenous variable can drive to asymptotic biased estimators (Wooldridge, 2002b). 84
estimation of endogenous systems was carried out using the command cmp (Roodman, 2007) which allows for the use of discrete endogenous predictors.
85
8
Effects of migration and remittances on fertilizer expenditure and cattle acquisition
This chapter focuses on analyzing the effects of migration and remittances on fertilizer expenditure and livestock acquisition in rural Ecuador. It begins by presenting the literature background to then explains the empirical strategy used. Next, the variables and the descriptive statistics are presented. Finally the results of the regression analyses are presented and discussed.
8.1
Literature background
As already pointed out several quantitative studies have focused on assessing the impact of overseas migration on agricultural activities. Researchers have paid special attention the effects of migration and remittances on agricultural expenditure and farm activity choice. To illustrate, Gray (2009) is able to determine that, ceteris paribus, remittance recipient households in southern Ecuador spent more on agricultural inputs than their counterparts without migrants. Instead, migration does not affect input expenditure. Gray points out that remittances are used to enhance yields and reduce labor demands on household members who stayed. Miluka et al. (2007) find that Albanian migrants’ households spend less on agricultural inputs and equipment rental than their counterparts without migrants. The authors imply that instead of spending remittances on labor saving technologies, migrants’ households prefer investing in the less labor demanding livestock production. Similar conclusions are drawn by Wouterse and Taylor (2008) who find out that intercontinental migration from Burkina Faso is associated with smaller net income from staple cropping and larger income from livestock production. According to the authors these findings reflect an imperfect labor market that averts migrants’ households from using remittances for hiring wage labor and rather encourages them to invest remittances in livestock. McCarthy et al. (2006) report a shifting pattern from crop to livestock production in Albanian migrants’ households. Rather than endorsing it to labor scarcity, the authors suggest that migrants’ households switch to livestock production due to the fact that it is more profitable than crop production. Pfeiffer and Taylor (2007) report that migration has no effect on livestock production in Mexico. The authors explain this finding by addressing that livestock production requires little labor which can be provided by marginal labor force, e.g. children. These findings are consistent with several qualitative studies (Caguana, 2008b; 86
Jokisch and Lair, 2002; Kyle, 2000; Martínez, 2004a; Pribilsky, 2007) carried out in Ecuador which hold that labor losses resulting from international migration have driven migrants’ households to switch from subsistence cropping to cattle production. Literature reports a tendency for migrants’ households to leave cropping in favor of livestock production. A drawback for the analysis is that the cross-sectional data set of the LSMS 20052006 does not allow controlling for the number of cattle owned by a household before migration. However, the LSMS 2005-2006 includes a question asking whether the household has acquired livestock during the year preceding the survey or not. By using this binary variable as the regressand, it is expected to estimate the extent to which migration and/or remittances have influenced cattle acquisition.
8.2
Specification and descriptive statistics
In order to estimate the impact of migration and remittances on the log of expenditure on fertilizers, one should first consider that this variable has a value of zero for a considerable number of observations (55.78% of the total sample) but is still continuous with strictly positive values. Modeling this kind of corner solution outcomes with OLS methods result inappropriate and rather the use of tobit models is recommended (Wooldridge, 2002a). Labeling the number of household members working in a business as EI and supposing that there is an unobservable variable EI* which is normally distributed and homoskedastic with 0 conditional mean: EI*= Miβ1 + Riβ2 + xiβ3 + εi
(26)
where Mi is a dichotomous variable taking the value of 1 if the household has migrants abroad, Ri is the monthly amount of remittances, xi is a vector of explanatory variables that will be described later on and εi stands for the error term; EI= 0 if EI*=≤0
(27)
EI= EI* if EI*=>0
(28)
For estimating the impact of migration and remittances on cattle acquisition, this study relied on a probit model of the following form:
87
Pr (CAi = 1|Mi, Ri, xi) = ϕ(Mi·β1,Ri·β2,xβ3)
(29)
where CAi is a binary variable that takes the value of 1 if the household has bought cattle during the year preceding the survey, Mi is the dummy for migration, Ri stands for the monthly amount of remittances, xi is the vector of control variables and ϕ stands for the cumulative density function. Table 8.1 displays the variables to be used in the analysis and their description together with the descriptive statistics. The dependent variables of interest are the log of the expenditure on fertilizers and the probability of acquiring cattle. This study includes migration and remittances as separate predictors. This decision is consistent with several other studies (Gray, 2008a; Gray, 2008b; Gray, 2009; Quinn, 2009) that also try to separate the effects of migration and remittances. However, others like MacKenzie and Sasin (2007) state that the effects of migration and remittances cannot be disentangled. Although migration and remittances are closely related, there are at least two reasons to analyze their effects separately. The first reason is connected to the structure of the survey itself. The questionnaire asks whether a household member has migrated in the last five years, independently from the answer if that household has received remittances during the last twelve months. Consequently, it could be the case that a household receives remittances from members that migrated before 2000. It is also possible that a household receives remittances from distant relatives or friends who were not household members before migrating. In any case, about 62% of the households claiming to have received remittances do not report having any household member abroad, which demonstrates that the decision to treat migration and remittances as separate covariates is prudent. The impact of labor losses resulting from rural out-migration deserves special attention. For instance, migration could drive migrants’ households to invest more in fertilizers to cope with increased labor scarcity and to turn to less labor intensive cattle farming. In the case of remittances a number of studies (Lucas, 1987; Lucas and Stark, 1985; Taylor et al., 1996; Taylor and Martin, 2001) demonstrate that remittances allow recipient households to overcome liquidity constraints that otherwise would prevent them from carrying out investments or adopting new technologies. The effect of remittances on fertilizer expenditure and cattle acquisition is expected to be captured by adding the monthly amount of remittances received by a household to the model. 88
Table 8.1 Definitions and descriptive statistics of the variables. Variable
Description
Mean
Std. Dev.
Dependent variables Fertilizer expenditure
Log of the expenditure on fertilizers
1.707
2.134
Cattle acquisition
Household bought cattle in the preceding year (0/1)
0.089
0.286
Treatment variables Migrant Household
At least one household member abroad (0/1)
Remittances
Monthly amount of remittances (US $)
0.064
0.245
17.684
94.934
Control variables Household head Age
Age of household head
Age squared
Squared age of household head
Sex Indigenous
50.885
16.061
2,847.218
1,712.567
Female household head (0/1)
0.163
0.370
Indigenous household head (0/1)
0.212
0.409
Education
Years of education of household head
Education squared
Squared years of education of household head
4.992
3.860
39.827
58.133
Household composition Children
Number individuals younger 30
0.855
0.559
Adult women
Number of females >30
0.875
0.572
HH education
Average education of household members
4.992
2.924
Assets Owned land
Number of hectares of owned land
9.832
85.191
Owned land squared
Squared number of hectares of owned land
7,352.802
363,455.5
Number of parcels
Total numbers of parcels owned by a household
1.629
0.962
Owned home
Household owns home (0/1)
0.843
0.363
Electricity
Household has electricity (0/1)
0.869
0.337
Piped water
Household has piped water system (0/1)
0.331
0.470
Credit
Household has received credit (0/1)
0.185
0.389
Road infrastructure Distance to the closest road Time to the closest market
Median of the distance to the closest road at provincial level in 2000 (km) Median of the time to the closest market at provincial level in 2000 (minutes)
0.432
0.721
49.058
12.851
Notes: The models also include provincial dummies. (0/1) identifies dummy variables.
Source: Author’s own calculations Other control variables include household and household head characteristics like age, sex, ethnicity, and education of the household head as well as household composition and the average education of household members. In this sense, Makokha et al. (2001) find that the 89
household head’s age is negatively correlated with the use of fertilizers while Demeke et al. (1998) report that the use of fertilizers is more common among households with a female head. A dummy variable taking the value of 1 if the household head considers himself/herself as indigenous is incorporated in order to account for the effect of ethnicity on the output variables. Household head’s education is reported to be positively correlated with the adoption of fertilizers in Ethiopia (Bacha et al., 2001; Demeke et al., 1998). In the case of Ecuador, Gray (2009) finds that agricultural input expenditure is negatively influenced by the household head’s age and the average education among household members. Considering that the division of labor in rural Ecuador could be influenced by gender and age (Martínez, 2000a; Martínez, 2004a; Martínez, 2000b), the number of children, young men, young women, adult men and adult women (see Table 8.1 for definitions) are included as separate predictors. Perz (2003) in his study of the Brazilian Amazon reports that the likelihood of using chemical fertilizers marginally increases with the number of children and adult household members. Instead, Gray (2009) establishes that the number of children and adult women in a household is negatively associated the spending on agricultural inputs. Adams (1998) determines that livestock accumulation is negatively correlated with the household head’s age and positively associated with the number of men and children in a household. Martínez (2004a) observes that female headed migrants’ households rely on livestock production as an alternative to cope with financial constraints and labor shortages caused by male migration. Control variables accounting for household assets include the extension of owned land, the number of parcels and home ownership. Availability of electricity and piped water are used as proxies for wealth. Besides, specification includes a dummy taking the value of 1 if the household has received credit. Gray (2009) reports that input expenditure is positively correlated with the extension of owned land but is not affected by the number of owned plots. In order to account for access infrastructure, specification includes the median of distance to the closest road and the median of the time needed to reach the closest market at provincial level both taken from the National Agricultural Census 2000. In this sense, the results of previous research are mixed. Perz (2003) determines that families living nearer the closest 90
town are more likely to use fertilizers whereas Gray (2009) finds no effect of distance to road on households’ input expenditures. In the case of cattle holding, Caviglia-Harris (2005) concludes that the larger distance to the closest market the fewer the heads of cattle that a household can accumulate. Finally, in order to account for regional differences, specification includes a set of 20 dummies, each one accounting for an Ecuadorian province. This strategy is consistent with other studies (Lin, 1991; Mukhopadhyay, 1994; Quinn, 2009) that use state or community dummies to control for regional variability.
8.3
Selection of instruments
As explained in Chapter 7, endogeneity is reported to be a serious threat for studies estimating the effects of migration and remittances on household decisions. As a first step to address this potential problem, this section describes the instruments used to test for endogeneity of migration and remittances with respect to fertilizer expenditure and cattle acquisition. Taking into account that for a variable to become a suitable instrument it must be highly correlated with the suspected endogenous variable and not directly influence the output variable. The variables chosen to instrument migration are: the number of cell phones available at household level and a dummy taking the value of 1 if at least one household member studies in a private school. Communication technologies, e.g. internet and cell phones are reported to be tools that have allowed Ecuadorian international migrants to be in permanent contact with their relatives back home (Mejía Estévez, 2006; Ramírez Gallegos and Ramírez, 2006). Therefore, the number of cell phones is expected to be higher for migrants’ households but it is not expected to be a factor affecting fertilizer expenditure. Ponce et al. (2008) report that children from migrants’ households are more likely to study in private schools. The explanatory power and the exclusion restriction of these instruments will be reported in the section 8.2.1. In order to instrument remittances, this study relies upon a dummy variable taking the value of 1 if a household receives remittances from Spain and a dummy taking the value of 1 if a household has received gifts in the form of clothes during the year preceding the survey. Calero et al. (2008) report having successfully used the country of origin of remittances as a 91
valid instrument for the monthly amount remitted16 while clothes are expected to be a common gift given by migrants to their relatives in Ecuador. In the case of cattle acquisition, the instruments for migration are a dummy taking the value of 1 if a household has a cell phone at its disposal17 and a dummy taking the value of 1 if the household is mono-parental. In many cases the absence of one of the parents might mean that he/she is an international migrant (Aguirre Vidal, 2009; Pedone, 2006). The instruments for remittances are the same than those used for fertilizer expenditure (remittances from Spain and clothes as a gift). Table 8.2 summarizes the instruments used in this section. Table 8.2 Instruments for migration and remittances having fertilizer expenditure and cattle acquisition as outcome variables. Outcome variable
Suspected
endogenous Instrumental variable
variable Fertilizer expenditure
Migration (0/1)
Number of cell phones Private school (0/1)
Remittances
Remittances from Spain (0/1) Clothes as gift (0/1)
Cattle acquisition (0/1)
Migration (0/1)
Cell phone availability (0/1) Mono-parental
household
(0/1) Remittances
Remittances from Spain (0/1) Clothes as gift (0/1)
16
Calero et al. (2008) exploit the country of origin of remittances (EEUU, Spain or Italy) as instruments for remittances. In this study, the dummy indicating if remittances come from EEUU is not taken into account because it influences the expenditure on fertilizers (t=1.87 and p-value=0.062).
17
The number of cell phones at home is not used in this case because it significantly influences the likelihood of cattle acquisition. 92
8.4
Results for fertilizer expenditure and cattle acquisition
8.4.1 Endogeneity tests In order to test the validity of the instruments selected, their explanatory power as well as the exclusion restriction were estimated. In the first case the joint significance of instruments when regressed as explanatory variables of migration resulted in χ2 statistics of 44.76 (pvalue=0.000) (Table A.1 in Annex A). In the case of remittances, the joint significance of instruments yields F statistics of 45.42 (p-value=0.000) (Table A.3). In both cases, the joint significance of instruments is well above the rule of thumb proposed by Staiger and Stock (1997) that instruments can be considered as valid if their joint significance in the first stage regression has a F value larger than ten. When instruments are included as explanatory variables of fertilizer expenditure, the joint significance yields χ2 of 0.92 (p-value=0.398) for migration and F of 0.84 (p-value=0.432) for remittances (Tables A.2 and A.4). The instruments also pass the Amemiya-Lee-Newey overidentification test with values of χ2=0.163 (p-value=0.660) for remittances and χ2=1.327 (p-value=0.249) for remittances. These results together suggest that the instruments proposed have explanatory power with respect to the suspected endogenous variables and do not influence the outcome variables proposed. Hence, they can reliably be used to test for endogeneity with the Smith-Blundell test. The results of this test are displayed in Table A.5. Although it fails to reject the null hypothesis of exogeneity with the third specification, it reports endogeneity in the model with specifications one and two. For the sake of prudence migration will be treated as an endogenous covariate. However, the null hypothesis of exogeneity of remittances cannot be rejected in any of the three specifications used and hence will be treated as exogenous. Regarding cattle acquisition, the joint significance of instruments for migration results in a χ2 statistics of 96.53 (p-value=0.000) (Table A.6) whilst the instruments for remittances have a joint significance of F=45.42 (p-value=0.000) (Table A.3). The second condition that instruments must not directly influence cattle acquisition is also met for migration (χ2=0.41, p-value=0.814) and remittances (χ2=0.70, p-value=0.704) (Tables A.7 and A.8). In this case, the Smith-Blundell test does not reject the null hypothesis of exogeneity, neither for migration nor for remittances with any of the three specifications used (Table A.9). Under these conditions, both migration and remittances will be considered as exogenous predictors. 93
8.4.2 Regression analysis Table 8.3 shows the results for both the tobit and the IV tobit obtained with the cmp Stata command. In both cases migration of one or more household members appears to increment expenditures on fertilizers. The coefficient in the IV tobit regression is almost twice as large as that of the tobit regression suggesting that endogeneity of migration produces a bias towards zero. The level of significance in the IV tobit (p-value=0.05) is lower than that of the tobit (p-value=0.01) which is, to some extent, normal considering the efficiency costs that the use of instrumental variables entail (Wooldridge, 2002b). The effect of remittances is not significant in both IV tobit and tobit models. Age, ethnicity, education and household composition do not appear to be determinant factors for households’ fertilizer expenditures. Instead, ceteris paribus female headed households seem to invest less in fertilizers than maleheaded ones. Cropping area does not appear to play any role in fertilizer expenditure; nevertheless the number of parcels owned by a household positively influences the outcome variable. Other predictors positively affecting expenditure on fertilizers are: availability of electricity and access to credit. In opposition households located further away from roads spend less on fertilizers. The results of the probit model for cattle acquisition are shown in Table 8.4. Migrants’ households appear to be 4% points more likely to have bought cattle during the year preceding the survey than their counterparts without migrants. Female headed households appear to be less likely to buy cattle than their male headed equivalents. The number of plots owned by a household is positively associated with cattle acquisition. Home ownership is another factor that marginally increases the likelihood of buying cattle. Households that have access to credit have a 7% higher probability of buying cattle than credit constrained households. In short, besides credit, migration is the factor most increasing the odds for a household to buy cattle.
94
Table 8.3 Determinants of fertilizer expenditures. IV Tobit
Tobit
Migrant Household
1.544** 0.890*** (0.776) (0.276) Remittances 0.0002 0.0003 (0.0006) (0.0007) Age 0.011 0.014 (0.025) (0.025) Age squared -0.0002 -0.0002 (0.0002) (0.0002) Sex -0.563*** -0.522** (0.216) (0.215) Indigenous -0.079 -0.075 (0.178) (0.173) Education 0.028 0.031 (0.048) (0.050) Education squared -0.0007 -0.0005 (0.002) (0.002) Children -0.055 -0.052 (0.040) (0.039) Young men -0.033 -0.036 (0.087) (0.088) Young women -0.046 -0.044 (0.093) (0.091) Adult men 0.208 0.207 (0.145) (0.145) Adult women -0.044 -0.056 (0.125) (0.130) HH education 0.024 0.026 (0.035) (0.036) Owned land -0.0001 -0.0002 (0.001) (0.002) Owned land squared 0.0000 0.0000 (0.0000) (0.0000) Number of parcels 0.688*** 0.694*** (0.063) (0.057) Owned home 0.090 0.089 (0.168) (0.166) Electricity 0.854*** 0.865*** (0.207) (0.207) Piped water -0.063 -0.052 (0.144) (0.149) Credit 1.839*** 1.846*** (0.187) (0.176) Distance to the closest road -1.766*** -1.773*** (0.498) (0.553) Time to the closest market -0.010 -0.011 (0.018) (0.021) Number of observations 4,720 4,720 Log-likelihood -7,952.519 -7,035.721 Notes: Coefficients are shown with standard errors in parentheses. *, ** and *** stand for significance at the 10, 5, and 1% levels, respectively. Specifications also include provincial dummies.
Source: Author’s own calculations 95
Table 8.4 Determinants of cattle acquisition. Probit
Robust S.E.
Marginal effects
Migrant Household
0.254**
0.113
0.038
Remittances
0.0001
0.0002
0.0000
Age
0.011
0.012
0.0014
Age squared
-0.0001
0.0001
-0.0000
Sex
-0.258***
0.097
-0.029
Indigenous
-0.038
0.077
-0.004
Education
0.0000
0.022
0.0000
Education squared
-0.0006
0.012
-0.0000
Children
-0.003
0.017
-0.0005
Young men
0.0008
0.041
0.0001
Young women
-0.022
0.041
-0.002
Adult men
-0.122*
0.070
-0.015
Adult women
0.036
0.055
0.004
HH education
0.004
0.016
0.0006
Owned land
0.002**
0.001
0.0003
Owned land squared
0.0000
0.0000
-0.0000
Number of parcels
0.093***
0.027
0.012
Owned home
0.161*
0.082
0.019
Electricity
0.140
0.099
0.016
Piped water
-0.092
0.069
-0.011
Credit
0.426***
0.088
0.070
Distance to the closest road
-0.003
0.007
-0.001
Time to the closest market
-0.037
0.180
-0.0004
Number of observations
4,720
Wald χ²
227.14***
Pseudo R²
0.09
Notes: *, ** and *** stand for significance at the 10, 5, and 1% levels, respectively. Specification also includes provincial dummies
Source: Author’s own calculations
8.5
Discussion
As explained above, the migration section of the LSMS 2005-2006 takes into account only those migrants who left after 2000. However, as addressed in Chapter 5 international migration has affected rural Ecuador for more than 40 years. Partly because of this drawback, 62% of the remittance receiving households within the sample report not having migrants abroad. The Smith-Blundell test fails to reject the null hypothesis of exogeneity of remittances with respect to fertilizer expenditures but rejects it for migration. A possible explanation for 96
these results is that fertilizer spending could be influenced by factors that also influenced migration decisions in the past like, for instance, wealth. Wealthier households may have had more chances to send members abroad. For remittance-recipient households this influence may be of smaller scope maybe because they do not have migrants abroad and thus have similar characteristics than those exhibited by non-migrants’ households; or because they sent migrants abroad a long enough time ago for the factors affecting migration at that time not to influence fertilizer expenditure at the present time. The case of cattle acquisition is significantly different. Using migration and/or remittances as predictors to explain the number of cattle owned by a household without using time lags would probably result in biased estimators. This could happen because factors affecting migration decisions in the past could also affect cattle assets at present time. The IADB (2003) reports that many Ecuadorian migrants were able to raise the funds to migrate by selling livestock or land which would also bring reverse causality into the model. Alternatively, this study used as outcome variable a dummy indicating whether or not a household has bought cattle during the year preceding the survey. As explained before, the number of Ecuadorians leaving the country sharply fell from 2003 onwards as a consequence of Spain’s visa imposition on Ecuadorian citizens and the increasingly strict controls at the US-Mexican border. Hence, cattle purchases one year before the survey (2006) are not expected to be influenced by the same factors affecting migration decisions. This statement is supported by the Smith-Blundell test that fails to reject the null hypothesis of exogeneity for both migration and remittances. Fertilizer expenditure seems to be raised by international migration but is not affected by remittances. Looking at these results, it is possible to conclude that households with migrants abroad buy fertilizers to cope with labor losses caused by migration. These results are not consistent with those reported by Gray (2008b) and Gray (2009) who found that input expenditures are positively influenced by remittances but not affected by out-migration. Such opposite results could be attributed to the differences of the databases used for each study18, to the fact that Gray does not test for endogeneity, or to both.
18
Gray (2008b) and Gray (2009) uses a data base of 385 observations collected in 5 cantons in the southern Ecuadorian province of Loja while this study relies on a data set of 4,720 households with national distribution. 97
Similarly, Gray reports that input expenditure is positively affected by the number of young men in a household and the cropping area, and negatively influenced by the number of adult women and the mean household education, predictors that appear as not significant in this study. Again, one could attribute those differences to specific characteristics of Gray’s research area that do not apply to other migrant-sending regions. Consistent with Gray’s findings, this study finds that ceteris paribus female headed households spend less on fertilizers. This could reflect that male labor losses prevent female headed households from keeping on cropping as referred by Martínez (2004a) and Martínez (2006a). The number of land parcels owned by a household seems to positively affect fertilizer expenditure reflecting that households with spatially distributed land invest more in fertilizers. Access to electricity is also positively correlated with spending on fertilizers suggesting that households with access to electricity are more aware of agriculture technologies to improve yields, or that poorer households cannot afford fertilization of their crops. Ceteris paribus, households living closer to roads spend more on fertilizers because transaction and transportation costs for fertilizers grow with the distance to drivable roads and also because wealthier households, for which fertilizers are affordable, live next to drivable ways. This finding contradicts those of Gray (2009) who found that distance to road has no effect on fertilizer expenditure in southern Ecuador, but is consistent with Perz (2003) who reports that households living closer to towns in the Brazilian Amazon are more likely to fertilize their crops than those located further away. The results of this study are consistent with the strand of literature stating that migration impels livestock production in Ecuador (Caguana, 2008a; Jokisch and Lair, 2002; Kyle, 2000; Martínez, 2006b; Pribilsky, 2007) and around the world (McCarthy et al., 2006; Miluka et al., 2007; Salas Alfaro and Pérez Morales, 2007; Wouterse and Taylor, 2008). The main argument to explain this trend is that migrants’ households switch from crop to cattle production in an effort to cope with labor losses caused by out-migration. The argument of labor shortage is reinforced if one takes into account that the likelihood of cattle acquisition is negatively affected by the number of adult males in a household. However, the results also show that female headed households are less likely to acquire cattle than their male headed counterparts. This finding suggests that female headed households without male support are less likely to have sufficient money to buy cattle. 98
Remittances appear not to influence cattle acquisition. In this regard, it may be argued that non-migrants’ households that receive remittances have not experienced labor losses and hence are still able to crop, or that the amounts they received are not sufficient to buy cattle. Another predictor significantly increasing the odds of acquiring cattle is credit. Financially constrained households are less likely to raise funds for buying cattle. Finally, cattle acquisition is also positively affected by available land and the number of parcels. However, the magnitude of the coefficients is small.
99
9 Effects of migration and remittances on labor relationships This chapter concentrates on analyzing the impact of international migration and remittances on community work participation, labor exchange and use of wage labor in rural Ecuador. After presenting a literature background, the specification and the variables used for the regression analysis are described. In addition, the instruments for the endogeneity test as well as the criteria used for their selection are explained. Finally, the results are presented and discussed.
9.1
Literature background
The impact of international migration on labor relationships has been continuously referred in qualitative studies analyzing the effects of migration and remittances on rural Ecuador. However, the conclusions drawn are mixed. On one side, several researchers (Caguana, 2008b; Camacho and Hernández, 2009; Martínez, 2002; Martínez, 2003; Martínez, 2004a; Martínez, 2006a) point out that migration and remittances inflows have led to the abandonment of ancient traditions of labor exchange and community work. The arguments presented are that migrants’ households have lost an important share of their labor pool due to international migration and therefore are not able to reciprocate the labor they would receive. It is also put forward that migrants’ households have no interest on keeping on cropping and hence are not willing to work in someone else’s fields provided they do not need labor in exchange. In addition, it is stated that migrants’ households that still crop prefer hiring wage labor instead of participating in labor exchange agreements. In contrast, Carpio (1992) states that rather than undermining labor exchange traditions, the labor losses caused by international migration have impelled labor exchange among indigenous towns. The author observes that inhabitants of these regions are reluctant to hire labor for agricultural chores and rather rely on labor exchange schemes to meet their labor requirements. Other authors (Jokisch, 2002; Pribilsky, 2007) report that labor exchange coexists with wage labor as strategies that migrant household use in order to cope with labor shortages in regions affected by international migration. Nevertheless, non-migrants’ households which are not able to afford wage labor may experience difficulties to meet the extra labor needed to carry out labor demanding agricultural tasks.
100
In a quantitative analysis, Gray (2009) is able to determine that labor reciprocity is positively influenced by female migration. Hired labor is reported to be negatively affected by female migration but positively associated with international remittances. Gray argues that households with female migrants tend to participate more in labor exchange agreements in order to substitute the part time agricultural labor provided by women. With regards to wage labor, the author speculates that female migration reduces labor household requirements owing to lower subsistence demands, and that the positive effect of remittances on wage labor reflects that migrants’ households use remittances to hire labor in order to improve yields and lighten labor demands on the household members who stayed. This chapter contributes to the debate about the impact of migration on labor relationships in rural Ecuador by estimating the effects of migration and remittances on: community work, labor exchange and use of hired labor force.
9.2
Specification and descriptive statistics
In order to assess the impact of migration and remittances on community work participation, this study relied on a probit model of the following form: Pr (CWi = 1|Mi, Ri, xi) = ϕ(Mi·β1,Ri·β2,xβ3)
(30)
where CWi is a dummy variable taking the value of 1 if a household has participated in community work during the year preceding the survey, Mi is a dummy indicating if a household has one or more of its members abroad, Ri, stands for the monthly amount of remittances that a household receives, xi accounts for a set of explanatory variables to be described later on and ϕ represents the cumulative density function. The same methodology was used to estimate the effects of migration and remittances on the likelihood for a household to engage in labor exchange agreements (LEi): Pr (LEi = 1|Mi, Ri, xi) = ϕ(Mi·β1,Ri·β2,xβ3)
(31)
and to hire wage labor (WLi): Pr (WLi = 1|Mi, Ri, xi) = ϕ(Mi·β1,Ri·β2,xβ3) This study is particularly interested in the size and direction of β1 and β2. 101
(32)
Table 9.1 shows the variables to be used together with definitions and descriptive statistics. The dependent variables are three dummies taking the value of 1 if a household has participated in community work, engaged in labor exchange agreements and hired wage labor, respectively. The explanatory variables of interest are: a dummy variable indicating if the household has at least one migrant abroad and the monthly amount of remittances that a household receives. Households with migrants abroad are expected to be less likely to participate in community work and labor exchange schemes because their labor pools have been shrunk as a consequence of migration. Instead, migrants’ households should be more likely to hire wage labor in order to equilibrate labor losses. Remittances should reduce the likelihood of labor exchange participation if recipient households prefer hiring wage laborers instead of exchanging their own labor with other households. Among other control variables, the model includes household head characteristics such as age, sex, ethnicity and education. In this sense, former studies find that households with older heads are less likely to participate in reciprocal labor agreements (Gilligan, 2004; Gray, 2008b; Gray, 2009) and that male headed households are less likely to hire wage labor (Gray, 2008b; Gray, 2009). Carpio (1992) reports that international migration has boosted reciprocal labor in Ecuadorian indigenous communities. In order to disentangle the effect of ethnicity from that of migration, a dummy variable taking the value of 1 if the household head considers himself/herself as indigenous is added to the model. To account for household composition, the specification includes the number of children, young men, young women, adult men and adult women (see Table 9.1 for definitions). In this regard, Gray (2009) finds that participation in reciprocal labor is positively affected by the number of young men, adult women and adult men while the use of hired labor is negatively correlated with the number of adult males. The model also considers the average education of household members which is reported to positively affect reciprocal labor and negatively influence use of wage labor (Gray, 2008b; Gray, 2009). Household asset predictors include the cropping area, the number of land parcels owned by a household, and a dummy accounting for home ownership. Gilligan (2004) determines that plot area has a positive effect on the likelihood of exchanging labor while Gray (2009) finds that the number of parcels is positively correlated with the probability of exchanging work. Availability of electricity and piped water control for the effect of service infrastructure. A 102
dummy indicating if a household has received a loan, accounts for the effects of credit. Additionally, specification includes the average per capita income and the average education both at town level. These variables account for the economic environment and for off-farm job opportunities at town level. The median of the distance to the closest road and the median of the time needed to reach the closest market are added to the model to control for road infrastructure. Finally, specification includes a set of provincial dummies accounting for regional differences.
9.3
Selection of instruments
The variables used to instrument migration in the model of community work are a dummy taking the value of 1 if the household has a cell phone at disposal and the number of children under grandparental care. In this regard, letting migrants’ children under relatives (mainly grandparental) surveillance has been a distinctive feature of Ecuadorian out-migration (Aguirre Vidal, 2009; Carrillo, 2006; Fresneda Sierra, 2001; Pedone, 2006). Remittances are instrumented with a dummy indicating if a household receives remittances from Spain and another dichotomous variable taking the value of 1 if any household member has been treated in a private medical institution. In this sense, Ponce et al. (2008) and Ponce et al. (2009) point out that migrants’ households are more likely to use private medical services. In the case of labor exchange, migration is instrumented with a dummy taking the value of 1 if the household is mono-parental and 0 otherwise and the number of children under grandparental care in a household while remittances are instrumented with the same variables used for community work (remittances from Spain and private medical attention). For the likelihood of hiring wage labor, the instruments for migration are the number of children under grandparental care and the dummy indicating whether a household is monoparental or not. Remittances are instrumented with a dummy taking the value of 1 if a household receives remittances from Spain and another dichotomous variable indicating if the household has received gifts in form of clothes. The validity of these instruments will be tested in section 9.4.1. Table 9.2 displays the instruments used in this chapter.
103
Table 9.1 Definitions and descriptive statistics of the variables. Variable
Description
Mean
Std. Dev.
Dependent variables Labor exchange
Household participated in labor exchange (0/1)
0.193
0.365
Community work
Household participated in labor exchange (0/1)
0.280
0.449
Wage labor
Household hired laborers (0/1)
0.177
0.382
Treatment variables Migrant household
At least one household member abroad (0/1)
0.064
0.245
Remittances
Monthly amount of remittances (US $)
17.684
94.934
Household head Age
Age of household head
50.885
16.061
Age squared
Squared age of household head
2,847.218
1,712.567
Sex
Female household head (0/1)
0.163
0.370
Control variables
Indigenous
Indigenous household head (0/1)
0.212
0.409
Education
Years of education of household head
4.992
3.860
Education squared
Squared years of education of household head
39.827
58.133
Household composition Children
Number individuals younger 30
0.855
0.559
Adult women
Number of females >30
0.875
0.572
HH education
Average education of household members
4.992
2.924
Assets Owned land
Number of hectares of owned land
9.832
85.191
Owned land squared
Squared number of hectares of owned land
7,352.802
363,455.5
Number of parcels
Total numbers of parcels owned by a household
1.629
0.962
Owned home
Household owns home (0/1)
0.843
0.363
Electricity
Household has electricity (0/1)
0.869
0.337
Piped water
Household has piped water system (0/1)
0.331
0.470
Credit
Household has received credit (0/1)
0.185
0.389
Town indicators Per capita income
Per capita income at town level
113.04
59.417
Town education
Average education at town level
5.408
1.623
Road infrastructure Distance to the closest road Time to the closest market
Median of the distance to the closest road at provincial level in 2000 (km) Median of the time to the closest market at provincial level in 2000 (minutes)
0.432
0.721
49.058
12.851
Notes: The models also include provincial dummies. (0/1) identifies dummy variables.
Source: Author’s own calculations 104
Table 9.2 Instruments for migration and remittances having community work, labor exchange and use of wage labor as outcome variables. Outcome variable
Suspected endogenous
Instrumental variable
variable Community work (0/1)
Migration (0/1)
Cell phone availability (0/1) Children under grandparental care
Remittances
Remittances from Spain (0/1) Private medical care (0/1)
Labor exchange (0/1)
Migration (0/1)
Children under grandparental care Mono-parental
household
(0/1) Remittances
Remittances from Spain (0/1) Private medical care (0/1)
Wage labor (0/1)
Migration (0/1)
Children under grandparental care Mono-parental
household
(0/1) Remittances
Remittances from Spain (0/1) Clothes as gift
9.4
Results for labor relationships
9.4.1 Endogeneity tests Table B.1 (Annex B) displays the first stage regression for migration in the model of community work participation. The joint significance of instruments yields a χ2 value of 51.96. Hence, the null hypothesis that instruments do not explain migration can be rejected at p-value=0.0000 probability. Regarding exclusion restriction of instruments, Table B.2 shows 105
that the joint significance of instruments when they are included as explanatory variables for the likelihood of community work participation results in a χ2=0.95 (p-value=0.811) which indicates that instruments do not directly influence the outcome variable. These results are supported by the Amemiya-Lee-Newey overidentification test which cannot reject the null hypothesis that instruments affect community work participation (χ2=0.499, p-value=0.480). In the case of remittances, Table B.3 shows that the null hypothesis that the true value of instruments is 0 can be rejected at 99.99% probability (F=42.90, p-value=0.0000). When included as predictors for community work participation (Table B.4), the instruments for remittances yield a joint significance of χ2=1.02 (p-value=0.600) suggesting that the instruments proposed do not influence the likelihood for a household to participate in community work. Instruments also pass the Amemiya-Lee-Newey overidentification test (χ2=0.227, p-value=0.633). Together these results suggest that the selected instruments for both migration and remittances are valid and can be used to test for endogeneity. The SmithBlundell test of exogeneity fails to reject the null hypothesis of exogeneity for migration and remittances under the three specifications utilized (see Table B.5). In the labor exchange model, Table B.6 shows that the joint significance of instruments for migration yield a χ2 statistics of 80.06 meaning that the null hypothesis that the explanatory power of instruments is zero, can be rejected at 99.99% probability. Moreover the joint significance of instruments when included as explanatory variables in the model of labor exchange (Table B.7) yields a χ2=1.12 (p-value=0.572) indicating that instruments do not significantly affect labor exchange participation decisions. Similarly, the null hypothesis that instruments do not influence labor exchange participation cannot be rejected by the Amemiya-Lee-Newey overidentification test (χ2=0.821, p-value=0.756). Since the variables used to instrument remittances in the model of labor exchange are the same as those used in for community work, the results of Table B.3 also apply in this case. Table B.8 shows that instruments for remittances also meet the exclusion restriction (χ2=1.98, p-value=0.370). Orthogonality of instruments with respect to the outcome variable is also supported by the Amemiya-Lee-Newey overidentification test (χ2=2.041, p-value=0.153). After testing for validity of instruments, the Smith-Blundell test of exogeneity was run. The null hypothesis of exogeneity of migration and remittances could not be rejected with any of the three specifications proposed (Table B.9). 106
The variables used to instrument migration in the model of wage labor are the same as those used for labor exchange (mono-parental households and the number of children under grandparental care). Therefore, the results presented in Table B.6 are also valid for the model of wage labor. Table B.10 shows the results for the test of exclusion restriction of instruments. The joint significance of instruments yields a χ2 statistics of 2.81 (p-value=0.245) which means that instruments fulfill the exclusion condition. The instruments also pass the Amemiya-Lee-Newey
overidentification
test
(χ2=1.029,
p-value=0.310).
Regarding
remittances, Table B.11 shows that the joint significance of instruments yields an F statistics of 45.74 indicating that the null hypothesis that the explanatory power of instruments is zero, can be rejected at 99.99% probability (p-value=0.0000). Instead Table B.12 shows that the explanatory power of instruments if included as regressors in the model of wage labor is not significantly different from zero (χ2=0.015, p-value=0.926). The Amemiya-Lee-Newey overidentification test also rejects the null hypothesis that instruments affect decisions of hiring wage labor (χ2=0.000, p-value=0.992). After having tested the validity of instruments, they were used to run the Smith-Blundell test of exogeneity. The results suggest that neither migration nor remittances are endogenous to the system.
9.4.2 Regression analysis Table 9.3 shows the results of the probit analysis for community work participation together with robust standard errors and marginal effects. Neither migration nor remittances have any significant effect on the probability for a household to participate in community work. With exception of ethnicity, household head characteristics appear not to play any role on the decisions concerning participation in community work. Having an indigenous head increases the probability for a household to participate in community work by 10% points. The number of children appears to have significant and positive effects on the likelihood of community work participation. Having a home of one’s own augments the likelihood of community work participation by 8% points. Households having access to piped water are 11% less probable to join communal works. Participation in community work is also negatively affected by the mean of the distance to the closest road at provincial level. For every kilometer that this value grows, the likelihood of communal work participation decreases by 13%. As expected the average per capita income at town level has a negative sign, however the magnitude of the effect is very low. 107
Migration and remittances also appear not to have any effect on the likelihood for a household to engage in labor exchange agreements (Table 9.4). These findings are inconsistent with the results presented by Gray (2009) who found that reciprocal work is positively affected by female migration. Ethnicity seems to play a significant role on the likelihood of labor exchange participation. Ceteris paribus, indigenous households are 10% more likely to participate in reciprocal labor schemes. Table 9.3 Determinants of community work participation. Probit
Robust S.E.
Marginal effects
Migrant household
0.029
0.101
0.08
Remittances
-0.0001
0.0002
-0.0000
Age
0.071
0.099
0.002
Age squared
-0.00001
0.0000
-0.0000
Sex
-0.070
0.080
-0.019
Indigenous
0.347***
0.061
0.108
Education
-0.007
0.019
-0.002
Education squared
-0.0008
0.011
-0.0002
Children
0.049***
0.014
0.014
Young men
0.023
0.034
0.006
Young women
-0.047
0.035
-0.013
Adult men
0.072
0.057
0.020
Adult women
0.013
0.047
0.003
HH education
0.003
0.014
0.0009
Owned land
-0.0002
0.0008
-0.0000
Owned land squared
0.0000
0.0000
0.0000
Number of parcels
0.080***
0.023
0.023
Owned home
0.327***
0.066
0.085
Electricity
0.021
0.074
0.006
Piped water
-0.424***
0.057
-0.115
Credit
0.039
0.082
0.011
Per capita income
-0.001**
0.0007
-0.0004
Town education
0.027
0.027
0.007
Distance to the closest road
-0.451***
0.019
-0.130
Time to the closest market
0.003
0.005
0.001
Number of observations
4,720
Wald χ²
1,134.33***
Pseudo R²
0.253
Notes: *, ** and *** stand for significance at the 10, 5, and 1% levels, respectively. Specification also includes provincial dummies
Source: Author’s own calculations 108
Consistent with Gray’s (2009) findings, labor exchange participation is positively correlated with the number of plots owned by a household. Instead, land area is negatively correlated with reciprocal work although the size of the effect is marginal. Table 9.4 Determinants of labor exchange participation. Probit
Robust S.E.
Marginal effects
Migrant household
-0.043
0.105
-0.010
Remittances
-0.0002
0.0002
-0.0000
Age
0.005
0.010
0.001
Age squared
-0.0001*
0.0000
-0.0000
Sex
0.085
0.081
0.020
Indigenous
0.368***
0.063
0.095
Education
-0.002
0.019
-0.0004
Education squared
-0.001
0.001
-0.0004
Children
0.013
0.014
0.003
Young men
0.040
0.032
0.009
Young women
-0.006
0.037
-0.001
Adult men
0.055
0.059
0.012
Adult women
-0.072
0.051
-0.016
HH education
-0.004
0.015
-0.001
Owned land
-0.002***
0.008
-0.0005
Owned land squared
0.0000***
0.0000
0.0000
Number of parcels
0.115***
0.024
0.026
Owned home
0.079
0.065
0.020
Electricity
-0.086
0.070
0.018
Piped water
-0.088
0.059
-0.020
Credit
0.202***
0.078
0.051
Per capita income
-0.002***
0.0007
-0.0006
Town education
-0.055**
0.027
-0.012
Distance to the closest road
0.431***
0.116
0.1000
Time to the closest market
0.006
-0.001
Number of observations
-0.008 4,720
Wald χ²
560.02***
Pseudo R²
0.151
Notes: *, ** and *** stand for significance at the 10, 5, and 1% levels, respectively. Specification also includes provincial dummies
Source: Author’s own calculations Credit raises by 5% the likelihood for a household to exchange labor with other households. An increase of one kilometer in the median of the distance to the closest road raises the 109
likelihood for a household to engage in labor exchange agreements by 10%. This result is consistent with the work of Gilligan (2004) for Indonesia. An increase of one year in the average town education leads to a decrease of 1.2% in the likelihood for a household to exchange labor. The average per capita income at town level is also significant and has the expected negative sign; nevertheless the size of the effect is small. Table 9.5 shows that having migrants abroad increases the probability for a household to hire wage labor by 7% points. These results are inconsistent with those reported by Gray who found that the use of hired labor is negatively affected by female migration. The likelihood of using hired labor is diminished by almost 9% provided a household is headed by an indigenous individual. Ceteris paribus, educated households and households with educated heads are more likely to use wage labor. Instead, households with more children and young men are less probable to hire agricultural laborers. The odds of hiring wage labor increases with the number of plots owned by a household. A new parcel of land would lead to an increase of 4% points in the odds of using hired labor. Access to electricity and piped water raises the likelihood for a household to hire wage labor by 4 and 3%, respectively. As in the models of both community and labor exchange participation, the effect of per capita income at town level is negative and significant although negligible in magnitude.
9.5
Discussion
After having proved the validity of the instruments selected, the Smith-Blundell test of exogeneity failed to reject the null hypothesis of exogeneity of both migration and remittances for the three outcome variables under study. These results suggest that: neither migration nor remittances are jointly determined with the probability of either participating in community work, exchanging labor or hiring wage labor; or that unknown factors affecting migration decisions in the past are not significantly influencing labor relationships at present. In any case, it is necessary to consider that failing to reject the null hypothesis of exogeneity does not necessarily mean that endogeneity is not present in a model but rather that endogeneity is not a problem of considerable magnitude in this particular case.
110
Table 9.5 Determinants of wage labor use. Probit
Robust S.E.
Marginal effects
Migrant household
0.281***
0.103
0.071
Remittances
0.0002
0.0002
0.0000
Age
0.004
0.009
0.0009
Age squared
-0.0000
0.0000
0.0000
Sex
-0.021
0.083
-0.004
Indigenous
-0.459***
0.083
-0.088
Education
0.072***
0.020
0.016
Education squared
-0.002**
0.001
-0.0005
Children
-0.080***
0.017
-0.018
Young men
-0.150***
0.038
-0.033
Young women
-0.004
0.036
-0.0009
Adult men
0.003
0.056
0.0008
Adult women
0.014
0.048
0.003
HH education
0.052***
0.0142
0.011
Owned land
0.001*
0.0006
0.0002
Owned land squared
-0.0000
0.0000
-0.0000
Number of parcels
0.168***
0.025
0.037
Owned home
0.108
0.067
0.023
Electricity
0193*
0.086
0.040
Piped water
0.125**
0.059
0.028
Credit
0.531***
0.070
0.146
Per capita income
-0.001**
0.0007
-0.0003
Town education
-0.014
0.028
-0.003
Distance to the closest road
0.287
0.266
0.064
Time to the closest market
0.016
-0.003
Number of observations
-0.017 4,720
Wald χ²
556.55***
Pseudo R²
0.147
Notes: *, ** and *** stand for significance at the 10, 5, and 1% levels, respectively. Specification also includes provincial dummies
Source: Author’s own calculations The results of this quantitative analysis contradict the statement of a group of researchers (Caguana, 2008b; Camacho and Hernández, 2009; Martínez, 2002; Martínez, 2003; Martínez, 2004a; Martínez, 2006a) holding that international migration and remittances have led to abandonment of ancient traditions of reciprocal work in rural Ecuador. Instead, other factors affect labor relationships. Community work appears to occur more frequently among indigenous households, which is consistent with other studies (Erasmus, 1956; Kimball, 111
1949), holding that reciprocal labor is more likely to appear among individuals with similar ethnic background. Ceteris paribus, households with more children are more likely to participate in community works. This finding is consistent with other research works (Caguana, 2008b; Pribilsky, 2001) reporting that children’s labor is used to meet households’ labor quota for community works in migrant-sending regions. Since migration has no effect on the likelihood for a household to participate in communal work, this finding suggests that migrants’ households are able to keep participating in community work by using child labor. Households owning a home appear to be more likely to participate in communal work. This could reflect that families that have put down roots in a town are more committed to take part in communal activities than those which in an opposite scenario are not permanent settlers. On the other hand, the odds of community work participation are reduced for households having access to piped water. This may signify either that wealthier households, with access to services, are less willing to participate in communal activities or that households having access to piped water see no need to participate in community projects aiming at water provision or maintenance of common water systems. The likelihood of participation in community work is negatively affected by the median of the distance to the closest road at provincial level. An increment of 100 meters in this value reduces the odds for a household to work in communal activities by 1.3%. In this sense, Table B.14 in Annex B shows that the largest shares of community work participation are registered in provinces from “La Sierra” (The Highlands) which are precisely those with the lowest values for the distance to the closest road and landholding area. This suggests that community work is more common in regions with higher population densities where the spirit of community cohesion is stronger. Ceteris paribus, the likelihood of exchanging labor is 10% higher for households with a head defining himself/herself as indigenous. The odds of engaging in labor exchange agreements are negatively associated with land area; nevertheless the size of the effect is very small. The number of plots owned by a household significantly increases the propensity for a household to exchange labor. This finding is consistent with that of Gray (2009) and reflects that households with spatially distributed plots tend to rely more on reciprocal work. Further on, it suggests that the number of parcels, independently from their size, plays a more important role than landholding area on labor demands. 112
Availability of credit increases the odds for a household to engage in labor exchange agreements by 8% points. One possible explanation is that households that asked for loans are engaged in commercial agriculture and hence have higher labor requirements that cannot be met only with wage labor. This finding contradicts former literature (Erasmus, 1956; Mitchell, 1991) stating that reciprocal labor flourishes among credit constrained households unable to pay wage labor but is consistent with another strand of literature (Geschiere, 1995; Gilligan, 2004; Guillet, 1980) holding that reciprocal work does appear even in non-liquidity constrained areas. Per capita income and average education at town level are negatively correlated with the likelihood of exchanging labor which suggests that labor exchange is less likely to occur in areas with economic dynamism where more off-farm job opportunities are available. Nevertheless, it is necessary to recall that the magnitude of the coefficients is small. In contrast to community work participation, the likelihood for a household to exchange labor is positively affected by the median of the distance to the closest road. A possible explanation for this difference can be seen in Table B.14 in Annex B. While the share of community work participation is mainly concentrated in provinces from “La Sierra” and tends to decrease as the median of the distance to the closest road mounts, the share of households exchanging labor remains high outside “La Sierra” and does not decrease for provinces with larger distances to the closest road. This finding may reflect that more dispersed households tend to rely more on labor exchange agreements. Although migration and remittances appear not to have any effect on reciprocal work, it raises the odds for a household to hire wage labor. Having migrants abroad increases the likelihood for a household to use paid labor by 7% points. This suggests that migrants’ households use wage labor to cope with labor loses caused by international migration. Remittances have no effect on likelihood of hiring wage labor. One possible explanation is that non-migrants’ households, which receive remittances19, do not invest remittances on paid labor due to the fact that they have not experienced labor losses. These findings are not consistent with those of Gray (2009) who found that the use of wage labor is positively affected by remittances and negatively influenced by female migration. Again, this divergence in results is endorsable to differences in data sources, sample sizes and specifications.
19
Recall that about 62% of remittance recipient households have no migrants abroad. 113
Having a self-defined indigenous head reduces the probability for a household to hire wage labor by 9% points. This is consistent with the results obtained for community work and labor exchange participation and indicates that indigenous households meet their labor needs via reciprocal work. The positive sign of the coefficients for household head education and the average education of household members may indicate that members of more educated households have access to off-farm jobs and hence must rely on wage labor to carry out agricultural chores. Households with more children and young men are less likely to hire wage labor. These results suggest that these groups are the main source of household labor for agricultural tasks. Cropping area marginally increases the likelihood of hiring paid labor while the number of plots significantly increases the use of wage labor. Access to electricity and piped water also raises the likelihood of hiring agricultural laborers. This implicates that wealthier households, with access to services, rely on the use of paid labor to carry out agricultural tasks. Another factor positively influencing the use of wage labor is availability of credit. This finding is consistent with the results for labor exchange and with former research (Gilligan, 2004; Jokisch, 2002; Pribilsky, 2007) stating that labor exchange and wage labor coexist and that the use of one of this labor sources does not necessarily mean the exclusion of the other. Per capita income has the expected negative sign but is small in magnitude. Overall, the results of this chapter suggest that international migration has not affected ancient forms of reciprocal labor as stated by former studies. Instead, other factors such as ethnicity, spatial distribution of plots and access infrastructure appear to play a significant role on households’ decisions of participating in reciprocal work activities. Migrants’ households appear to rely on paid labor to meet their agricultural labor needs. Further on, this finding supports the Jokisch’s (2002) observation that migrants’ households have not quit cropping after migration.
114
10
Effects of migration and remittances on entrepreneurship
This chapter centers on estimating the effects of out-migration, remittances and average remittances at town level on the likelihood of business ownership and the number of household and non-household members working in a rural household business. Besides presenting the literature background on the topic, it explains the methodology as well as the variables used for the quantitative analysis. The instruments used to test for endogeneity are described and finally the results are presented and discussed.
10.1 Literature background For several researchers (Lanjouw, 1998; Lanjouw and Lanjouw, 1995; Petrin, 1990; Petrin, 1994), rural entrepreneurship can become a powerful weapon to fight poverty and inequality in rural regions of developing countries. In this sense, Lanjow (1998) argues that analyses of rural poverty should not only study the links of the latter with the agricultural, but also with the non-agricultural sector. Yet, entrepreneurs in rural areas of developing countries must overcome several limitations in order to start businesses. According to Petrin (1994) and Paulson and Townsend (2004) the main bottleneck for rural business formation is the lack of credit in rural areas. In the context of the NELM, remittances can be used by migrants’ households to finance new agricultural technologies, to carry out investments and at the same time serve as an alternative source of income in case of crop or investment failure. Extending the NELM to rural entrepreneurship, remittances would allow migrants’ households to diversify their income by investing in the formation of businesses (Kilic et al., 2007). Likewise remittances can act as income insurance in case of business failure. Following this approach, several authors (Amuedo-Dorantes and Pozo, 2006; Arif and Irfan, 1997; Kilic et al., 2007; Massey and Parrado, 1998; Wahba and Zenou, 2009) are able to determine that international migration is positively associated with business formation. Some authors (Massey and Parrado, 1998; Taylor, 1999) state that remittances benefit not only receiving households but also non-receiving ones due to the aggregate effect of remittances at community level. In this context, non-migrant businesses could benefit from the boost in the demand for goods and services that comes together with remittances. Supporting this position, Massey and Parrado (1998) find that the aggregate amount of remittances at community level is positively associated with business formation in Mexico. 115
Regarding the potential of migrant businesses to generate employment, the literature presents mixed results. On one side, authors like Kilic et al. (2007) observe that migrants’ households rely on non-household labor to compensate labor losses caused by migration. On the contrary, Canales and Montiel Armas (2004) report that migrant businesses are characterized by lack of investments and dependency on household labor. Massey and Parrado (1998) find no relationship between remittances and the number of both household and non-household business employees. This chapter builds up on the existing literature about migration and entrepreneurship by estimating the effects of migration, remittances and remittances at town level on the probability for a household to own a business, and the number of household and non-household members employed in a household business.
10.2 Specification and descriptive statistics In order to assess the effects of migration on rural entrepreneurship and the number of household and non-household employees in a business; two methodologies are used. To estimate the likelihood for a household to own a business, this study relies on a probit model of the following form: Pr (Bi = 1|Mi, Ri, TRi, Xi) = ϕ(Mi·β1,Ri·β2,TRi·β3,xi·β4)
(33)
where Bi is a dichotomous variable, which takes the value of 1 if the household owns a business and 0 otherwise, xi is a vector that includes the control variables to be described later on and ϕ is the standard cumulative normal distribution. The coefficients of migration (Mi), remittances (Ri) and remittances at town level (TRi) are of particular interest for this study. In the case of the number of household and non-household employees, let FW and 5FW stand for number of household and non-household business workers respectively. Assuming that FW* and 5FW* are unobservable variables which are homoskedastic with zero conditional mean. FW*= Miβ1 + Riβ2 + TRiβ3+ xiβ4 + εi
(34)
5FW*= Miβ1 + Riβ2 + TRiβ3+ xiβ4 + εi
(35)
116
where Mi is a dichotomous variable indicating whether a household has migrants abroad or not, Ri is the monthly amount of remittances, TRi stands for the average remittance per household at town level, xi is a vector of explanatory variables and εi is the error term: FW= 0 if FW*= ≤ 0
(36)
FW= FW* if FW*= > 0
(37)
5FW= 0 if 5FW*= ≤ 0
(38)
5FW= FW* if 5FW*= > 0
(39)
and
Table 10.1 displays the variables used for the analysis as well as the descriptive statistics. Special attention is paid to the effect of international migration and remittances on the propensity to own a business and the number of household and non-household workers. As already explained in Chapter 8, only 38% of the remittance receiver households report having migrants abroad. Taking into account this data set particularity, migration and remittances will be included as separate explanatory variables. Additionally, specifications include the average amount of remittances received by a household at town level. With this variable, it is expected to capture the indirect or multiplier effects of remittances at community level. The likelihood of business ownership increases for male household heads (Amuedo-Dorantes and Pozo, 2006) who are young and well educated (Massey and Parrado, 1998). A dummy variable taking the value of 1 if the household head considers himself/herself as indigenous is incorporated in order to account for the effect of ethnicity on the output variables proposed. The number of children, young men, young women, adult men and adult women (see Table 10.1 for definitions) account for household composition. Martínez (2004a) identifies land scarcity due to egalitarian inheritance as the main driving force for entrepreneurship in rural Ecuador. If such a statement is true, households owning more land should show lower propensity to run a business. Having a home of one’s own positively affects the probability of forming a business (Massey and Parrado, 1998), it can be used as a collateral for getting a loan and at the same time offer a physical space for manufacturing and retail activities. A dummy variable accounting for home ownership is incorporated as a control variable. 117
Table 10.1 Definitions and descriptive statistics of the variables. Variable
Description
Dependent variables Business
Business ownership (0/1)
0.299
0.458
Number of household members working in a business Number of non-household members working in a business
2.330
1.355
0.826
1.896
Household head Migrant household
At least one household member abroad (0/1)
0.064
0.245
Remittances
Monthly amount of remittances (US $)
17.684
94.934
Town remittances
Average remittances received by a household at town level (US $)
17.684
38.467
Age of household head
50.885
16.061
Household workers Non-household workers
Mean
Std. Dev.
Treatment variables
Control variables Household head Age Age squared
Squared age of household head
2,847.218
1,712.567
Sex
Female household head (0/1)
0.163
0.370
Indigenous
Indigenous household head (0/1)
0.212
0.409
Education
Years of education of household head
4.992
3.860
Education squared
Squared years of education of household head
39.827
58.133
Household composition Children
Number individuals younger 30
0.855
0.559
Adult women
Number of females >30
0.875
0.572
HH education
Average education of household members
4.992
2.924
Assets Owned land
Number of hectares of owned land
9.832
85.191
Owned land squared
Squared number of hectares of owned land
7,352.802
363,455.5
Owned home
Household owns home (0/1)
0.843
0.363
Electricity
Household has electricity (0/1)
0.869
0.337
Piped water
Household has piped water system (0/1)
0.331
0.470
Indoors water system
Household has indoors water system (0/1)
0.231
0.421
Credit
Household has received credit (0/1)
0.185
0.389
Road infrastructure Distance to the closest road Time to the closest market
Median of the distance to the closest road at provincial level in 2000 (km) Median of the time to the closest market at provincial level in 2000 (minutes)
0.432
0.721
49.058
12.851
Note: The models also include provincial dummies. (0/1) identifies dummy variables.
Source: Author’s own calculations 118
10.3 Selection of instruments In order to test for the potential endogeneity of the treatment variables, this study considers a set of instrumental variables that are chosen according to the suspected endogenous covariate and the output variable in each case20. For the likelihood of owning a business, the instruments for migration are: the number of children under grandparental care and the average unemployment rate21 at town level in 2001. Letting migrants’ children under relatives (mainly grandparental) surveillance has been a distinctive feature of Ecuadorian out-migration (Aguirre Vidal, 2009; Carrillo, 2006; Fresneda Sierra, 2001; Pedone, 2006). As for unemployment, it is considered as one of the main factors that triggered migration in the late 1990s (Acosta et al., 2006; Ramírez Gallegos and Ramírez, 2005). To instrument for remittances, this study relies on two dummies indicating whether or not the household has received remittances from Spain and whether or not the household has received clothes22 as gifts during the twelve months preceding the survey. Instruments for average per household town remittances are taken from the National Census 2001 and include the number of people with internal migration experience and the average number of women per household, both at town level. In rural Andean sending regions, internal migration is seen as a first step before
20
As mentioned above, instrumental variables must not have any effect on the output variables. To be sure that
this condition is met, all the instruments were included in the models as regressors and were not used if they were correlated with the output variable. For these reasons, the instruments used for business ownership may differ from those used for the number of household and non-household members employed in a business. To illustrate, cell phone availability is highly correlated with business ownership and the number of non-household members employed in a business, the migration rate in 2001 significantly affect the likelihood of business ownership and the number of children under grandparental care influences the number of household members in a business. 21
It could be argued that the unemployment rate in 2001 might have influenced the decision of starting a
business in future years. However when this value is included as a covariate in the model estimating the propensity of owning a business, it appears not to have any effect on the outcome variable (z=0.45, pvalue=0.651). 22
It could be argued that clothes given by a relative abroad could be used to start a clothes store in Ecuador,
however this dummy variable does not explain the likelihood of business ownership (z=1.01, p-value=0.311). 119
international migration (Carpio, 1992) while the gender equilibrium is reported to be changed in such regions as a consequence of persistent male migration (Jokisch, 2001). In the case of the number of household members working in a family business, the variables chosen to instrument migration are dummies indicating whether or not the household is mono-parental and whether or not the household has a cell phone at its disposal. The instruments to test endogeneity of remittances with respect to the number of household workers are the same as those used for the likelihood of business ownership (remittances from Spain and clothes as a gift) while the average number of internet users at town level and the average number of women per household in 2001 served as instruments for average town remittances. For the number of non-household members working in a business, the dummy variable for mono-parental households and the migration rate at parish level from 2001 were used as instruments for migration. In this model, remittances were instrumented with dummies indicating whether or not a household receives remittances from Spain and whether or not the household receives remittances from the United States. Finally, the instruments for average town remittances are: the number of individuals with internal migration experience at town level and the average number of absent household members at town level, both taken from the National Census 2001. Table 10.2 summarizes the instruments used in this chapter.
10.4 Results for entrepreneurship 10.4.1
Endogeneity tests
In the case of business ownership, the joint significance of the instruments for migration yields a χ² statistics of 21.21 (Table C.1 in Annex C) while in cases of remittances and average town remittances (Tables C.3 and C.5) the joint significance test results in F values of 46.63 and 43.13, respectively. Overall the joint significance tests indicate that the hypothesis that the true value of instruments is zero can be rejected at 99.99% probability. The second condition that instruments must fulfill is that they do not influence the likelihood of business formation directly. When regressed as explanatory variables for business ownership, the joint significance of instruments for migration, remittances and average town remittances (Tables C.2, C.4 and C.5) yield χ² statistics of 0.46 (p-values = 0.794), 1.91 (p-value = 0.384) and 0.63 (p-value = 0.731) correspondingly indicating that instruments pass the second condition. 120
Table 10.2 Instruments for migration, remittances and average remittances at town level having business ownership and number of household and non-household members working in a business as output variables. Outcome variable
Suspected endogenous
Instrumental variable
variable Business ownership (0/1)
Migration (0/1)
Children under grandparental care Unemployment rate 2001
Remittances
Remittances from Spain (0/1) Clothes as gift (0/1)
Average town remittances
Number of internal migrants 2001 Average number of women 2001
Number of household
Migration (0/1)
Mono-parental household
members employed in a
(0/1)
business
Cell phone availability (0/1) Remittances
Remittances from Spain (0/1) Clothes as gift (0/1)
Average town remittances
Average number of women 2001 Average internet users 2001
Number of non-household
Migration (0/1)
members employed in a
Cell phone availability (0/1) Migration rate 2001
business Remittances
Remittances from Spain (0/1) Remittances from U.S. (0/1)
Average town remittances
Number of internal migrants 2001 Average number of absent household members 2001
121
The Amemiya-Lee-Newey overidentification test also supports orthogonality of instruments with χ² values of 0.018 (p-value = 0.893), 1.557 (p-value = 0.212) and 0.663 (p-value = 0.415) respectively. Regarding the number of household members working in a business, the joint significance of instruments for migration yields a χ²=29.49 (p-value=0.0000) (Table C.8). The joint significance of instruments for remittances and average town remittances yield F statistics of 17.58 (p-value=0.0000) and 39.03 (p-value=0.0000) (Table C.10 and C.12). These results suggest that the null hypothesis that the explanatory power of instruments is zero can be rejected. Tables C.9, C.11 and C.13 show the exclusion restriction tests for the suspected endogenous variables. The joint significance of instruments for migration, remittances and town remittances when included as predictors for the number of household members results in F statistics of 0.12 (p-value=0.887), 0.45 (p-value=0.636) and 0.63 (p-value=0.731), respectively. Hence, the null hypothesis that the joint significance of instruments is zero cannot be rejected. The Amemiya-Lee-Newey overidentification test also fails to reject the null hypothesis that instruments influence the number of household employees. The test yield χ² statistics of 0.049 (p-value=0.825), 0.358 (p-value=0.549) and 0.008 (p-value=0.927) for migration, remittances and town remittances, respectively. Table C.15 shows the first stage equation for migration in the model for the number of nonhousehold members working in a business. The joint significance of the instruments for migration results in a χ²=40.97 with a p-value of 0.0000. In the case of remittances and average town remittances (Tables C.17 and C.19), the joint significance of instruments yield F values of 17.10 (p-value=0.0000) and 42.82 (p-value=0.0000) correspondingly. Instruments also met the exclusion restriction condition with F statistics of 0.37 (p-value=0.693), 0.29 (pvalue=0.750) and 1.17 (p-value=0.312) for migration, remittances and average town remittances, respectively (Tables C.16, C.18 and C.20). These results are supported by the Amemiya-Lee-Newey overidentification test which yields χ² values of 0.057 (p-value=0.811), 0.009 (p-value=0.923) and 1.381 (p-value=0.239), respectively. After having tested the validity of instruments, a Smith-Blundell test of exogeneity was carried out for each of the outcome variables proposed (Tables C.7, C.14 and C.21). The test fails to reject the null hypothesis of exogeneity of migration, remittances and average town remittances for the three output variables under study with the three specifications utilized. 122
Therefore, migration, remittances and average town remittances will be considered as exogenous covariates.
10.4.2
Regression analysis
Table 10.3 shows the results of the probit analysis estimating the likelihood for a household to own a business next to robust standard errors and marginal effects. Neither migration nor remittances significantly affect the probability for a household to own a business. Instead business ownership appears to be positively affected by the education of the household head and average education of household members. The odds of business ownership are increased by the number of young women, adult men and adult women. Instead, the number of young men negatively influences the likelihood for a household to have a business. Owned land is negatively associated with business ownership although the magnitude of the effect is small. In contrast, households having a home of their own are more likely to have a household business. Having access to electricity, piped water and indoor water system raises the probability of having a business by 7, 7 and 11% correspondingly. Credit also seems to be a determinant for business ownership; it increases the odds of having a rural enterprise by 6% points. In the case of the number of household employees in a household business, it is positively affected by migration (Table 10.4) but not influenced by remittances and average town remittances. Other variables increasing the number of household members working in a business are the age of the household head as well as the number of young men, young women and adult women. Instead, the number of household members employed in a household business is lower for female-headed households. Table 10.5 displays the results of the tobit analysis estimating the effects of migration, remittances and average town remittances on the number of non-household members working in a business. In this case, neither migration nor remittances influence this variable. It appears to be positively affected by the average education of household members and availability of indoor water. In contrast, the number of non-household members employed is lower for female headed indigenous households.
123
Table 10.3 Determinants of business ownership. Probit
Robust S.E.
Marginal effects
Migrant household
0.0574
0.087
0.019
Remittances
-0.0003
0.0002
-0.0001
Town remittances
0.001
0.0006
0.0003
Age
0.012
0.008
0.004
Age squared
-0.0001**
0.00008
-0.00005
Sex
-0.104
0.072
-0.034
Indigenous
0.020
0.060
0.007
Education
0.033**
0.017
0.011
Education squared
-0.002**
0.0009
-0.0007
Children
0.017
0.013
0.005
Young men
-0.070**
0.029
-0.023
Young women
0.118***
0.031
0.040
Adult men
0.099**
0,050
0.033
Adult women
0.290***
0.042
0.098
HH education
0.066***
0.012
0.022
Owned land
-0.003***
0.001
-0.001
Owned land squared
0.0000***
0.0000
0.0000
Owned home
0.111*
0.058
0.036
Electricity
0.237***
0.077
0.075
Piped water
0.215***
0.050
0.074
Indoors water system
0.325***
0.055
0.114
Credit
0.187***
0.051
0.065
Distance to the closest road
-0.087
0.096
-0.029
Time to the closest market
0.005
0.004
0.001
Number of observations
4,753
Wald χ²
555.18
Pseudo R²
0.11
Notes: *, ** and *** stand for significance at the 10, 5, and 1% levels, respectively. Specification also includes provincial dummies
Source: Author’s own calculations
10.5 Discussion Neither migration nor remittances, whether at household or community level, significantly affect the probability for a household to own a business. One possible explanation for these results is that migrants tend to allocate their earnings in less risky investments, such as purchasing land or building houses. Land is seen as a safe investment the price of which tends to be higher than inflation rates (Adams, 1991; Jokisch, 2002). Hence, migrants may prefer 124
purchasing land rather than investing in entrepreneurial activities, the success of which is uncertain. Another possibility is that given the relative recentness of the massive Ecuadorian out-migration, a considerable share of remittances must still be dedicated to the repayment of loans asked for in order to migrate, as well as to cover the basic needs of households; therefore the amounts available for entrepreneurship may be low. It is also possible that the Ecuador’s economic and political instability post-crisis is still fresh in migrants’ minds. Consequently they do not want to risk their savings in entrepreneurial adventures as argued by the IADB (2003). Unfortunately, the analysis of these factors is out of the scope of this research work. Consistent with the findings of other studies (Amuedo-Dorantes and Pozo, 2006; Lanjouw, 1998; Massey and Parrado, 1998) both education of the household head as well as the average education of household members positively influence the likelihood of business ownership. The number of young women, adult men and adult women significantly raise the probability of owning a business. Especially important is the role of adult women, for every woman older than 30 the likelihood of owning a household business is augmented by 10%. In contrast, the likelihood decreases by 2% for every young man in the household. One possible explanation for this is given by Martínez (2000a) who argues that off-farm and salaried activities are mainly carried out by men. Land has a negative impact on the probability of owning a business which is consistent with Martínez (2004a) who reports that in the rural Ecuadorian context, entrepreneurship is a response to land fragmentation due to equalitarian inheritance. This process has considerably reduced the size of plots to a point where farmers can no longer earn their livelihood from cropping. Another possibility is that those who possess enough capital to invest in entrepreneurial activities prefer buying land, considered as one of the safest investment allocations in rural Ecuador. Home ownership raises the probability of having a business by almost 4%. Having access to credit increases the odds of owning a home enterprise by almost 6% points. Availability of electricity, piped water and indoor piped water increase the probability of business ownership by 7, 7 and 11% respectively. These results are consistent with those reported by Lanjouw (1998) who found that services and access infrastructure are key determinants in the likelihood of owning an enterprise in rural Ecuador. In this study though, road infrastructure, in form of the median of the distance to the closest road and the median of 125
the time needed to reach the closest market, have no effect on the probability of owning a business. In sum, remittances have not stimulated entrepreneurship in rural Ecuador, neither at household nor at community level. Instead, factors such as education, credit availability and good infrastructure are still key determinants of business ownership. Table 10.4 Determinants of the number of household members employed in a business. Tobit
Robust S.E.
Marginal effects
Migrant household
0.577*
0.306
0.279
Remittances
0.001
0.001
0.0005
Town remittances
-0.001
0.002
-0.0008
Age
0.073**
0.037
0.033
Age squared
-0.0008**
0.0003
-0.0003
Sex
-1.161***
0.321
-.450
Indigenous
-0.076
0.251
-0.033
Education
0.058
0.069
0.026
Education squared
-0.001
0.003
-0.0008
Children
0.052
0.065
0.023
Young men
0.383***
0.133
0.171
Young women
0.291***
0.111
0.130
Adult men
0.249
0.202
0.111
Adult women
0.557***
0.161
0.249
HH education
0.054
0.045
0.024
Owned land
0.003
0.003
0.001
Owned land squared
-0.0000
0.0000
0.0000
Owned home
0.149
0.259
0.065
Electricity
-0.097
0.030
-0.044
Piped water
0.185
0.196
0.083
Indoors water system
0.351*
0.202
0.159
Credit
0.174
0.189
0.079
Distance to the closest road
-0.075
0.333
-0.034
Time to the closest market
0.011
0.016
0.005
Number of observations
1,425
Log-likelihood
-1,995.51
σ
2.629
Notes: Coefficients are shown with standard errors in parentheses. *, ** and *** stand for significance at the 10, 5, and 1% levels, respectively. Specifications also include provincial dummies. Marginal effects are calculated for the unconditional expected value of the output variable.
Source: Author’s own calculations
126
Table 10.5 Determinants of the number of non-household members employed in a business. Tobit
Robust S.E.
Migrant household
0.815
0.620
Marginal effects 0.113
Remittances
-0.001
0.001
-0.0001
Town remittances
0.0008
0.004
0.0001
Age
-0.014
0.078
-0.001
Age squared
-0.0002
0.0007
-0.0002
Sex
-2.172***
0.722
-0.198
Indigenous
-1.284**
0.608
-0.134
Education
-0.126
0.175
-0.015
Education squared
0.011
0.009
0.001
Children
-0.025
0.144
-0.003
Young men
-0.243
0.250
-0.029
Young women
0.318
0.234
0.038
Adult men
0.124
0.413
0.015
Adult women
-0.367
0.331
-0.044
HH education
0.245**
0.110
0.029
Owned land
0.009
0.006
0.001
Owned land squared
-0.0000
0.0000
-0.0000
Owned home
0.718
0.549
0.079
Electricity
1.057
0.909
0.107
Piped water
0.047
0.453
0.005
Indoors water system
1.696***
0.478
0.226
Credit
0.445
0.449
0.056
Distance to the closest road
0.117
0.703
0.014
Time to the closest market
0.016
0.034
0.002
Number of observations
1425
Log-likelihood
-984.08
σ
4.218
Notes: Coefficients are shown with standard errors in parentheses. *, ** and *** stand for significance at the 10, 5, and 1% levels, respectively. Specifications also include provincial dummies. Marginal effects are calculated for the unconditional expected value of the output variable.
Source: Author’s own calculations Having a migrant abroad is positively correlated with the number of household members working in a business. A possible explanation is that more household labor is needed to cover the labor gap left by household members who migrated. Besides this consideration, this finding is consistent with Canales and Montiel Armas (2004) who argue that rather than entrepreneurial initiatives, migrant businesses are livelihood strategies the success of which is based on the over exploitation of household labor force. 127
The number of household members employed in a business is also positively influenced by the age of the household head which could reflect that more household labor is needed as the household head gets older. Instead, having a woman as household head is negatively correlated with the number of household members working in a business. According to Martínez (2000a) the typical non-agricultural activities carried out by women in rural Ecuador are retailing trade and handicraft manufacturing. These activities probably do not demand high amounts of labor. The number of young men, young women and adult women are positively correlated with the number of household members working in a business while the number of males older that 30 does not have any influence on the output variable. Among the covariates accounting for services, only the dummy for indoor piped water has a positive effect on the number of household members employed in a business. Only two variables (the average education of the household and the dummy for indoor water system) positively explain the number of non-household members working in a household business. These results may indicate that there are other variables besides those included in the model that affect the number of non-household members employed in a business. A similar drawback is reported by Massey and Parrado (1998), this time for both the number of household and non-household members working in a rural business. In any case, neither migration nor remittances have any impact on the outcome variable.
128
11
Summary and Conclusions
This research work has investigated the effects of migration and remittances on agricultural production patterns, labor relationships and entrepreneurship in rural Ecuador. It arrived at new findings which are summarized below. The conclusions drawn from the empirical analyses are followed by a section which provides some recommendations for policy makers and researchers.
11.1 Summary A large number of studies about international migration have focused on its effects on agricultural production and farm activities. According to the empirical findings of Taylor (1999) and Taylor et al. (1996) when testing the NELM, out-migration may be associated with a decline of agricultural production in an initial stage. This effect would be reverted, however, as soon as households invest remittances in new labor-saving technologies which yield higher profits. Similarly it is economically useful to invest remittances for switching from crop to livestock production provided this implies profit maximization (McCarthy et al., 2006; Miluka et al., 2007). In the particular case of this research work, the results support Taylor’s (1999) statement by rejecting the following null hypotheses: •
H1: Migration and remittances do not have any effect on households’ fertilizer expenditure.
Irrespective of the methodology used, tobit or instrumental variable tobit, this hypothesis can be rejected at 99 and 95% probability, respectively. Migration appears to increase households’ fertilizer expenditure which is consistent with the empirical work of McCarthy et al. (2006) and Miluka et al. (2007). These results indicate that migrants’ households try to cope with labor losses caused by out-migration. Taking into account that only 38% of the remittance receiving households has migrants abroad, the effect of migration could be considered as the effect of having lost household members due to international migration while the effect of remittances alone corresponds to the money that both migrant and non-migrants’ households receive from international migrants. In this sense, remittances alone do not affect the spending on fertilizers, suggesting that remittances received by households without migrants do not stimulate investment in fertilizer because their labor stock has remained unchanged. 129
• H2: Migration and remittances have no influence on the probability for a household to acquire cattle. Caution is required when establishing a link between international migration and livestock accumulation without time lags. There is the risk that wealthier migrants with more cattle could afford migration in the past which would bring endogeneity in the model. To cope with this problem this study has taken advantage of a question in the LSMS 2005-2006 which asks if the household has acquired cattle during the year preceding the survey. In this way, it was expected to avoid the effects of reverse causality. The results of endogeneity tests support this strategy. As in the case of fertilizer use, the results of a probit model indicate that migration increases the probability for a household to acquire cattle. Still this tendency may be a strategy to cope with labor losses rather than a way to increase incomes as described by Martínez (2006a) and Martínez (2004a). Overall, the results suggest that migrants’ households manage labor losses by spending more on inputs and investing in less labordemanding activities such as cattle holding. Traditions of community work and labor force exchange are repeatedly reported to have been undermined by international migration. Instead, it is stated that migrants’ households exhibit higher odds of hiring wage labor to replace their labor losses. Through quantitative methods, this study tested the related hypotheses and obtained the following results. • H3: Migration and remittances have no effect on the probability for a household to participate in community work. This hypothesis could not be rejected. Instead, it was shown that the likelihood for a household to participate in communal work is strongly determined by other factors such as ethnicity, spatial distribution of plots, community commitment and road infrastructure. • H4: Migration and remittances have no effect on the likelihood for a household to take part in labor force exchange arrangements. This hypothesis could not be rejected. Instead, factors such as ethnicity, distribution of parcels and distance to passable roads are determining in households’ engagement in labor exchange agreements. Availability of credit significantly increases the likelihood of exchanging labor indicating that labor exchange is not restricted to capital constrained households. Overall, the 130
rejection of H3 and H4 indicates that international migration has not undermined labor reciprocity in rural Ecuador. • H5: Migration and remittances have no effect on the demand for wage labor. The results of a probit analysis revealed that migrants’ households exhibit a higher propensity to hire wage labor than their counterparts without migrants. Hence, this hypothesis is rejected suggesting that migrants’ households rely on paid labor to compensate labor losses resulting from international migration. However, remittances alone have no effect on the likelihood of hiring paid labor. This indicates that remittance recipient households without migrants do not invest in wage labor as their labor stock remains unchanged. Altogether, the results of this part of the research suggest that migration did not lead to the abandonment of reciprocal work activities in rural Ecuador. Hence, if reciprocal labor practices have been weakened, it has occurred because of other factors. However, migrants’ households show a higher tendency to hire wage labor probably to balance their farm labor needs. The potential of rural entrepreneurship for enhancing income diversification and employment generation in rural areas has been repeatedly highlighted. However, small scale businesses in developing countries face inefficient credit and insurance schemes. In this context, remittances from abroad can become an important source of capital for business formation. Remittances are reported to benefit not only receiving households but also non-receiving ones through their multiplier effects on the demand of goods and services. This study has analyzed such statements by testing the following null hypotheses: • H6: Migration and remittances have no effect on the propensity for a household to own a rural enterprise. This null hypothesis could not be rejected for this study. Instead, the results of a probit analysis indicate that other factors such as credit, education, and service infrastructure play a significant role for the likelihood of households’ businesses formation. As argued in Chapter 10, there are a number of reasons that could explain the aversion exhibited by migrants’ households to invest remittances in entrepreneurial activities: market imperfections that turn entrepreneurship into a too risky activity; distrust of the financial system after the late 1990s 131
economic crisis; and the small magnitude of the transfers. The analysis of such factors is beyond the scope of this research work. In any case, the results of this study are clear in determining that neither migration nor remittances are associated with business ownership. • H7: The aggregate effect of remittances at town level does not influence the likelihood for rural households to start businesses. The explanatory variable accounting for the average monthly transfers received by a household at town level appears not to influence the likelihood of having a business in rural Ecuador. • H8: Migration and remittances do not influence the number of household members employed in a rural business. The analysis showed that the number of household members employed in a business is positively associated with international migration. Hence, this hypothesis was rejected at 90% probability. The most logical explanation for this effect is that more household labor is needed to fill the labor gap resulting from out-migration. Beyond this explanation, this finding reinforces the argument presented by Canales and Montiel Armas (2004) that businesses owned by migrants’ households must rely on household labor, both to be feasible and to survive in the market. • H9: Migration and remittances do not influence the number of non-household members working in a rural business. The results of a tobit model indicate that neither migration nor remittances, whether at household or at town level, have any effect on the number of non-household members working in a business. Caution is needed when interpreting these results because only two predictors (average education of the household and the dummy for indoor water system) positively explain the outcome variable and hence, there may be other factors not considered in the model influencing the number of non-household members working in a business. However, the model does indicate that neither migration nor remittances have incidence on the output variable studied.
132
11.2 Recommendations for policy makers and researchers One limitation for the empirical analysis of this study was the fact that 62% of the remittance receiving households within the sample reported not having any migrant abroad. This situation is imputable to limitations in the questionnaire used for the LSMS 2005-2006, which in the question regarding migration only asks about migrants who left after 2000 but not before, or to the fact that migrants also send money to households different from their own. In any case, using only migration as the treatment variable would imply neglecting 62% of the households that indirectly benefit from migration. On the other hand, using only remittances as the treatment variable would implicate ignoring the effects of household labor losses. These considerations should be taken into account by those using the LSMS 2005-2006 to measure the impacts of migration in Ecuador. In addition, the National Institute of Statistics (INEC) should pay attention to this shortcoming and try to avoid it in future surveys. This study has demonstrated that international migration is associated with higher households’ expenditure on fertilizers and, supported by other qualitative and quantitative studies, has concluded that this behavior is imputable to households’ strategies to cope with labor losses caused by international migration. Future research on this topic should be oriented to measure the productivity and income generation of migrants’ and non-migrants’ farms in order to determine if the increment in the use of fertilizers has improved the yields and income of migrants’ households as compared to those of their counterparts without migrants. A limitation in this context is that the LSMS 2005-2006 does not provide information about the extension of land dedicated to each specific crop which makes the estimation of yields and agricultural incomes difficult. In this regard, INEC should take this drawback into account and correct it for future household surveys carried out in rural areas. For a more exact estimation of the effects of out-migration on cattle accumulation, further research should estimate the change in the size of herds over time. For this purpose longitudinal data are required. In this regard, INEC should keep the same sample of the LSMS 2005-2006 for allowing the use of time lags. Overall, the results of this study show that the major constraint that migrants’ households face after a migrant has left is the loss of labor. This problem spreads to the community level increasing general labor shortages. In this sense, policy interventions should be oriented to provide not only migrant households but the whole rural population with training, extension and credit for the use of labor-saving technologies. 133
Reciprocal labor is a topic that should be studied beyond the sense of an ancient tradition practiced by isolated populations in the Andean regions and which is in danger of disappearing. Instead it should be analyzed as a strategy used by economically rational peasants to deal with high labor transaction costs and labor scarcity. In this sense, the returns to team and reciprocal work, their advantages and limitations, as well as their potential for labor supply in regions with labor shortages, e.g. migrant-sending regions, are still to be quantitatively assessed in the Ecuadorian case. This study found that international migration and remittances, both at household and community levels, have no influence on the probability for a household to own a business in rural Ecuador. This result contradicts the main stream of literature regarding migration and development which finds a positive correlation between migration and entrepreneurship. Instead, factors such as education, credit, and services play a significant role on the likelihood of business ownership. There is evidence that remittances have allowed Ecuadorian migrants’ households to smooth consumption levels and to improve the health status and educational attainment of their children. However, this alone cannot correct the effects of decades of lack of investment in education, credit, and infrastructure in rural regions. As stated by Taylor et al. (1996) cooperatives, banks, and workers’ associations specially conceived to fulfill migrants’ requirements as well as to direct remittances into productive activities would probably fail if the conditions that prevent migrants from investing do not change in the first place. Under these circumstances, should governmental and non-governmental organizations not concentrate on providing rural communities with infrastructure, credit, and schools before starting projects of co-development or establishing banks or cooperatives specially designed for migrants? These considerations should be analyzed before turning remittances into the spearhead of any development strategy.
134
References Acosta, A., D. Villamar, and S. López (2006), La migración en el Ecuador: Oportunidades y amenazas. Centro Andino de Estudios Sociales: Quito-Ecuador. Acosta, P., C. Calderón, P. Fajnzylber, and H. López. 2008 "What is the Impact of International Remittances on Poverty and Inequality in Latin America?" World Development, 36 (1): 89-114. Acosta, P., P. Fajnzylber, and H. López. 2007 The Impact of Remittances on Poverty and Human Capital: Evidence from Latin American Household Surveys. Wahington D.C.: World Bank Policy Research Working Paper 4247. Adams, R. H. 1991 The Effects of Remittances on Poverty, Inequality and Development in Rural Egypt: International Food Policy Research Institute. Adams, R. H. 1998 "Remittances, Income Distribution, and Rural Asset Accumulation." Economic Development and Cultural Change, 41 (1): 155-173. Adams, V. 1992 "Tourism and Sherpas, Nepal Reconstruction of Reciprocity " Annals of Tourism Research, 19: 534-554. Aguirre Vidal, G. 2009 "Cuidado y lazos familiares en torno a la (in)movilidad de adolescentes en familias transnacionales" In Miradas transnacionales. Visiones de la migración ecuatoriana desde España y Ecuador. ed. G. C. Zambrano and K. H. Basante. Quito-Ecuador: CEPLAES-SENAMI. Pp. 17-52. Altamirano, T. (2006), Remesas y nueva “fuga de cerebros”. Impactos transnacionales. Pontificia Universidad Católica del Perú: Lima-Perú. Amuedo-Dorantes, C., and S. Pozo. 2006 "Remittance Receipt and Business Ownership in the Dominican Republic." The World Economy, 29 (7): 939-956. Antón, J. I. 2010 "The Impact of Remittances on Nutritional Status of Children in Ecuador." International Migration Review, 44 (2): 269-299. Arif, G. M., and M. Irfan. 1997 "Return Migration and Occupational Change: The Case of Pakistani Migrants Returned from the Middle East." The Pakistan Development Review, 36 (1): 1-37. Aw-Hassan, A., M. Alsanabani, and A. Rahman Bamatraf. 2000 Impact of Land Tenure and Other Socioeconomic Factors on Mountain Terrace Maintenance in Yemen. Washington, D.C.: CAPRi Working Paper 3. Bacha, D., G. Aboma, A. Gemeda, and H. d. Groote. 2001, The Determinants of Fertlizer and Manure use in Maize Production in Western Oromiya, Ethiopia. Paper presented at the Seventh Eastern and Southern Africa Regional Maize Conference. 135
Baum, C. F. (2006), An Introduction to Modern Econometrics Using Stata. STATA PRESS. BCE 2006 Several issues. . BCE 2007 Evolución de las Remesas-2007. . BCE 2009 Evolución de las Remesas-2009. . BCE 2010 Sector Externo. . Bertoli, S. 2008, The impact of migration on poverty in Ecuador: Erring on the side of excessive optimism? Paper presented at the Migration and Development Conference, Lille. Borrero, A. L. 2002 "La migración: estudio sobre las remesas de divisa que ingresan al Ecuador." La Migración, 1 (1): 79-87. Caguana, M. 2008a "Diáspora de kichwa kañaris: islotes de prosperidad en el mar de pobreza" In Al filo de la identidad: La migración indígena en América Latina. ed. A. Torres and J. Carrasco. Quito-Ecuador: FLACSO-UNICEF. Pp. 127-146. Caguana, M. 2008b Impactos de la migración sobre el sistema andino tradicional, expresión de un capital social: El caso de las parroquias de Juncal, Ingapirca y el cantón El Tambo. Master Thesis, FLACSO, Cañar-Ecuador. Caguana, M., M. Pinguil, Q. Tenezaca, P. Peñafiel, and L. Zaruma. 2008 Migración internacional en la comuna Sisid: Cambios y adaptaciones en el territorio rural y economía local. Calero, C., A. S. Bedi, and R. Sparrow. 2008 "Remittances, Liquidity Constraints and Human Capital Investments in Ecuador." World Development. Calero, C., A. S. Bedi, and R. Sparrow. 2009 "Remittances, Liquidity Constraints and Human Capital Investments in Ecuador." World Development, 37 (6): 1143–1154. Camacho, G., and K. Hernández 2009 "Territorios en movimiento. Suscal: migración y ¿desarrollo?" In Miradas transnacionales. Visiones de la migración ecuatoriana desde España y Ecuador. ed. G. C. Zambrano and K. H. Basante. Quito-Ecuador: CEPLAES-SENAMI. Pp. 177-202. Canales, A. I. 2008 "Remesas, Desarrollo y Pobreza. Una Visión Crítica Desde América Latina" In Nuevas Migraciones Latinoamericanas a Europa. Balances y Desafíos. ed. I. Yépez Del Castillo and G. Herrera. Quito: FLACSO. Pp. 363-389.
136
Canales, A. I., and I. Montiel Armas. 2004 "Remesas e Inversión Productiva en Comunidades de Alta Migración a Estados Unidos: El Caso de Teocaltiche, Jalisco." Migraciones Internacionales, 2 (3): 142-172. Carpio, P. (1992), Entre Pueblos y Metrópolis: la Migración Internacional en Comunidades Austroandinas del Ecuador. ABYA-YALA: Cuenca-Ecuador. Carrillo, M. C. 2006 "El espejo distante. Construcciones de la migración en los jóvenes hijos e hijas de emigrantes ecuatorianos" In La Migración Ecuatoriana: Transnacionalismo, Redes e Identidades. ed. G. Herrera, M. C. Carrillo and A. Torres. Quito-Ecuador: FLACSO. Pp. 361-370. Carter Hill, R., W. E. Griffiths, and G. C. Lim (2008), Principles of Econometrics. John Wiley & Sons: New York. Caviglia-Harris, J. L. 2005 "Cattle Accumulation and Land Use Intensification by Households in the Brazilian Amazon." Agricultural and Resource Economics Review, 34 (2): 145162. CEDATOS (1999), Ecuador en Perspectiva. Quito-Ecuador. CEPAL. 2009 Anuario estadístico de América Latina y el Caribe. Santiago-Chile: CEPAL. Cohen, J. H. (1999), Cooperation and community: economy and society in Oaxaca. University of Texas: Austin. de Brauw, A., J. E. Taylor, and S. Rozelle 1999 "The Impact of Migration and Remittances on Rural Incomes in China." de Haas, H. 2006 "Migration, Remittances and Regional Development in Southern Morocco." Geoforum, 37 (4): 565–580. de Haas, H. 2007 Remittances, Migration and Social Development: A Conceptual Review of the Literature: Social Policy and Development Programme Paper Number 34. de Prada, M. A. 2005 "Flujos Migratorios Internacionales hacia España. Especificidad en la Región de Murcia" In La condición inmigrante. Exploraciones e investigaciones desde la Región de Murcia. ed. A. Pedreño Cánovas and M. H. Pedreño. Murcia-España: Universidad de Murcia. Pp. 61-74. Demeke, M., V. Kelly, T. S. Jayne, A. Said, J. C. L. Vallée, and H. Chen. 1998 Agricultural Market Performance and Determinants of Fertilizer Use in Ethiopia. Addis-Ababa: MINISTRY OF ECONOMIC DEVELOPMENT AND COOPERATION. Durand, J., W. Kandel, E. A. Parrado, and D. S. Massey. 1996a "International Migration and Development in Mexican Communities." Demography, 33 (2): 249-264. Durand, J., and D. S. Massey. 1992 "Mexican Migration to the United States: A Critical Review." Latin American Research Review, 27 (2): 3-42. 137
Durand, J., E. A. Parrado, and D. S. Massey. 1996b "Migradollars and Development: A Reconsideration of the Mexican Case." International Migration Review, 30 (2): 423444. El Comercio.2009 "Tres millones al exilio." 8 August El Universo.2006 "Naufragio motivó un solo juicio." 14 August Erasmus, C. J. 1956 "Culture Structure and Process: the Ocurrence and Dissapearance of Reciprocal Farm Labor." Southwestern Journal of Anthropology, 12 (4): 444-469. Ferraro, E. (2004), Reciprocidad, Don y Deuda. Formas y Relaciones de Intercambios en los Andes de Ecuador: La Comunidad de Pesillo. FLACSO-ABYA-YALA: QuitoEcuador. Fresneda Sierra, J. 2001 "Redefinición de las relaciones familiares en el proceso migratorio ecuatoriano a España." Migraciones Internacionales, 1 (1): 135-144. Geschiere, P. 1995 "Working Groups or Wage Labour? Cash-crops, Reciprocity and Money among the Maka of Southeastern Cameroon." Development and Change, 26: 503-523. Gilligan, D. O. 2004 The Economics of Agricultural Labor Exchange with Evidence from Indonesia. PhD Thesis, University of Maryland. González Casares, G. G., M. A. Viera Mendoza, and X. Ordeñana Rodríguez. 2009 "El Destino de las Remesas en Ecuador: un Análisis Microeconómico sobre los Factores que Determinan su Utilización en Actividades de Inversión." Revista de Economía del Caribe, 4: 76-108. Goycoechea, A., and F. Ramírez Gallegos. 2002 "Se fue, ¿a volver? Imaginarios, familia y redes sociales en la migración ecuatoriana a España (1997-2000)." ÍCO5OS, 14: 3245. Gratton, B. 2006 "Ecuador en la historia de la migración internacional ¿Modelo o aberración?" In La migración ecuatoriana: Transnacionalismos redes e identidades. ed. G. Herrera, M. C. Carrillo and A. Torres. Quito-Ecuador: FLACSO. Gray, C. L. 2008a "Environment, Land, and Rural Out-migration in the Southern Ecuadorian Andes." World Development, 37 (2): 457-468. Gray, C. L. 2008b Out-Migration and Rural Livelihoods in the Southern Ecuadorian Andes. PhD thesis, University of North Carolina at Chapel Hill, Chapel Hill. Gray, C. L. 2009 "Rural out-migration and smallholder agriculture in the southern Ecuadorian Andes." Population Environment, 30: 193-217. Greene, W. H. (2003), Econometric Analysis. Pearson Education: Upper Saddle River-New Jersey. 138
Guillet, D. 1980 "Reciprocal Labor and Peripheral Capitalism in the Central Andes." Ethnology, 19 (2): 151-167. Gujarati, D. N. (2004), Basic Econometrics. McGraw−Hill: New York. Hanson, G. H., and C. Woodruff 2003 "Emigration and Educational Attainment in Mexico." University of California, San Diego. Harris, J. R., and M. P. Todaro. 1970 "Migration, Unemployment and Development - 2-Sector Analysis." American Economic Review, 60 (1): 126-142. Herrera, G., A. Torres, A. Valle, A. Amezquita, and S. Rojas (2006), ECUADOR: Las Cifras de la Migración Internacional. UNFPA-FLACSo: Quito-Ecuador. Hildebrandt, N., and D. J. McKenzie. 2005 The Effects of Migration on Child Health in Mexico: World Bank Policy Research Working Paper 3573. IADB. 2003 Remittance Recipients In Ecuador A Market Research Study. Quito-Ecuador: Inter-American Development Bank-Multilateral Investment Fund. IADB. 2006 Las remesas como instrumento de desarrollo. Washington D.C.: Inter-American Development Bank-Multilateral Investment Fund. IMF 2010 World Economic Outlook Databases. . INEC 2001a Censo de Población y vivienda. Ficha Metodológica. . INEC. 2001b III Censo Nacional Agropecuario. Quito-Ecuador: INEC. INEC. 2006 Pobreza y Extrema Pobreza en el Ecuador. Quito-Ecuador: INEC. INEC. 2007a Homologación Metodológica del Cálculo de la Pobreza. Desigualdad e Indicadores Sociales a partir de la Encuesta de Condiciones de Vida (ECV). QuitoEcuador: INEC. INEC. 2007b Las Condiciones de Vida de los Ecuatorianos: INEC. INEC 2010a Indicadores Periódicos. . INEC 2010b Tasa trimestral de mercado laboral. . Inter-American Development Bank. 2003 Remittance Recipients In Ecuador A Market Research Study. Quito-Ecuador: Inter-American Development Bank-Multilateral Investment Fund.
139
Jokisch, B. 2001 "Desde Nueva York a Madrid: tendencias de la migración ecuatoriana." Ecuador Debate, 54: 59-84. Jokisch, B. 2002 "Migration and Agricultural Change: The Case of Smallholder Agriculture in Highland Ecuador." Human Ecology, 30 (4): 523-550. Jokisch, B., and D. Kyle 2006 "Las Transformaciones de la Migración Transnacional del Ecuador, 1993-2003" In La Migración Ecuatoriana: Transnacionalismos, Redes e Identidades. ed. G. Herrera, M. C. Carrillo and A. Torres. Quito- Ecuador: FLACSO. Pp. 58-69. Jokisch, B., and B. M. Lair. 2002 "One Last Stand? Forests and Change on Ecuador's Eastern Cordillera." Geographical Review,, 92 (2): 235-256. Jokisch, B., and J. Pribilsky. 2002 "The Panic to Leave: Economic Crisis and the "New Migration" from Ecuador"." International Migration, 40 (4): 75-101. Kilic, T., G. Carletto, B. Davis, and A. Zezza. 2007 Investing Back Home: Return Migration and Business Ownership in Albania: ESA Working Paper No. 07-08. Kimball, S. T. 1949 "Rural Social Organization and Co-operative Labor." The American Journal of Sociology, 55 (1): 38-49. Kyle, D. (2000), Transnational Peasants: Migration, 5etworks, and Ethnicity in Andean Ecuador. The Johns Hopkins University Press: Baltimore-Maryland. Lanjouw, P. 1998 Ecuador's Rural Nonfarm Sector as Route out of Poverty. Washington, D.C.: World Bank Policy Research Working Paper 1904. Lanjouw, P., and J. O. Lanjouw. 1995 Rural Nonfarm Employment. Washington D.C.: World Bank Policy Research Working Paper 1463. Larrea, C. (2004), Dolarización, Crisis y Pobreza. Abyayala: Quito-Ecuador. Lin, J. Y. 1991 "Education and Innovation Adoption in Agriculture: Evidence from Hybrid Rice in China." American Journal of Agricultural Economics, 79 (713-723). López-Córdova, E. 2006 Globalization, Migration and Development: The Role of Mexican Migrant Remittances. Buenos Aires: INTAL-ITD Working Paper 20. López, S., and D. Villamar 2004 "El proceso emigratorio en el sur de Quito" In Migraciones. Un juego con cartas marcadas. ed. F. Hidalgo. Quito-Ecuador: Abya-Yala. Pp. 367388. Lucas, R. E. B. 1987 "Emigration to South Africa's Mines." The American Economic Review, 77 (3): 313-330. Lucas, R. E. B., and O. Stark. 1985 "Motivations to Remit Evidence from Botswana." The Journal of Political Economy, 93 (5): 901-918. 140
Makokha, S., S. Kimani, W. Mwangi, H. Verkuijl, and F. Musembi. 2001 Determinants of Fertilizer and Manure Use for Maize Production in Kiambu District, Kenya. Mexico D.F.: CIMMYT. Mansuri, G. 2006 Migration, School Attainment and Child Labor: Evidence from Rural Pakistan. Washington D.C.: World Bank Policy Research Working Paper 3945. Martínez, L. 2000a "La especificidad del empleo rural" In Antología de estudios rurales. ed. A. Torres. Quito: FLACSO-ILDIS. Pp. 121-150. Martínez, L. 2002 "El Capital Social en la Tucayta (Tucuy Cañar Aiilupunapac Tantanacuy)" In Construyendo Capacidades Colectivas. ed. T. F. Carroll. Quito-Ecuador. Martínez, L. 2003 "Capital Social y Desarrollo Rural." ÍCO5OS, 16: 73-83. Martínez, L. 2004a "El campesino andino y la globalización a fines de siglo (una mirada sobre el caso ecuatoriano)." Revista Europea de Estudios Latinoamericanos y del Caribe, 77: 25-39. Martínez, L. 2004b "La emigración internacional en Quito, Guayaquil y Cuenca." Martínez, L. 2006a "Migración Internacional y Mercado de Trabajo Rural en Ecuador" In La Migración Ecuatoriana: Transnacionalismos, Redes e Identidades. ed. G. Herrera, M. C. Carrillo and A. Torres. Quito- Ecuador: FLACSO. Pp. 147-168. Martínez, L. 2006b "Migración Internacional y Mercado de Trabajo Rural en Ecuador" In La Migración Ecuatorian: Transnacionalismos, Redes e Identidades. ed. G. Herrera, M. C. Carrillo and A. Torres. Quito- Ecuador: FLACSO. Pp. 147-168. Martínez, L. ( 2000b), Economías Rurales: Actividades no Agrícolas. CAAP: Quito-Ecuador. Massey, D. S., R. Alarcón, J. Durand, and H. González. 1987 Return to Aztlan: The Social Process of International Migration from Western Mexico. Berkeley: University of California Press. Massey, D. S., and E. A. Parrado. 1998 "International Migration and Business Formation in Mexico." Social Sciences Quarterly, 79 (1): 1-20. Mayer, E. 1974 "Las reglas del juego en la reciprocidad andina" In Reciprocidad e Intercambio en los Andes Peruanos. ed. G. Alberti and E. Mayer. Lima-Peru: INSTITUTO DE ESTUDIOS PERUANOS. Pp. 34-65. Mayer, E., and C. Zamalloa 1974 "Reciprocidad en las relaciones de producción" In Reciprocidad e Intercambio en los Andes Peruanos. ed. G. Alberti and E. Mayer. Lima-Peru: INSTITUTO DE ESTUDIOS PERUANOS. Pp. 66-85. McCarthy, N., G. Carletto, B. Davis, and I. Maltsoglou. 2006 Assessing the Impact of Massive Out-Migration on Agriculture: ESA Working Paper No. 06-14. 141
McKenzie, D., and H. Rapoport. 2006 Can migration reduce educational attainment? Evidence from Mexico. Washington D.C.: World Bank Policy Research Working Paper 3952. McKenzie, D., and M. J. Sasin. 2007 Migration, Remittances, Poverty, and Human Capital: Conceptual and empirical challenges. World Bank: World Bank Policy Research Working Paper 4272. Mejía Estévez, S. 2006 "Transnacionalismo a la Ecuatoriana: Migración, Nostalgia y Nuevas Tecnologías" In La Migración Ecuatoriana: Transnacionalismo, Redes e Identidades. ed. G. Herrera, M. C. Carrillo and A. Torres. Quito-Ecuador: FLACSO. Pp. 481-492. Mendola, M. 2008 "Migration and technological change in rural households: Complements or substitutes?" Journal of Development Economics, 85: 150-175. Miluka, J., G. Carletto, B. Davis, and A. Zezza. 2007 The Vanishing Farms? The Impact of International Migration on Albanian Family Farming. Washington D.C.: World Bank Policy Research Working Paper 4367. Mitchell, W. P. 1991 "Some Are More Equal than Others: Labor Supply, Reciprocity, and Redistribution in the Andes." Research in Economic Antropology, 13: 191-219. Mochebelele, M. T., and A. Winter-Nelson. 2000 "Migrant Labor and Farm Technical Eficiency in Lesotho." World Development, 28 (1): 143-153. Mukhopadhyay, S. K. 1994 "Adapting Household Behavior to Agricultural Technology in West Bengal, India: Wage Labor, Fertility, and Child Schooling Determinants." Economic Development and Cultural Change, 43: 91-115. Mutersbaugh, T. 2002 "Migration, common property, and communal labor: cultural politics and agency in a Mexican village." Political Geography, 21: 473–494. OECD. 2003 Entrepreneurship and Local Economic Development. Paris: OECD. OECD. 2006 The new rural paradigm: policy and governance. Paris: OECD. Ortega, C. (2008), Finanzas Populares y Migración: tejiendo la red para el desarrollo local. FEPP: Quito-Ecuador. Pacheco, Á. 2007 Influencia de la Migración en el Rendimiento Escolar de Niños en Hogares Rurales Ecuatorianos Master Thesis, FLACSO, Quito-Ecuador. Paulson, A., and R. Townsend. 2004 "Entrepreneurship and Financial Constraints in Thailand." Journal of Corporate Finance, 10 (2): 229-262. Pedone, C. (2006), Estrategias Migratorias y Poder: Tú Siempre Jalas a los Tuyos. ABYAYALA: Quito- Ecuador.
142
Perz, S. G. 2003 "Social Determinants and Land Use Correlates of Agricultural Technology Adoption in a Forest Frontier: A Case Study in the Brazilian Amazon." Human Ecology, 31 (1): 133-165. Petrin, T. 1990, 2-5 October 1990 The Potential of Entreprenership to Create Income and New Jobs for Rural Women and Families. Paper presented at the Fifth Session of the FAO/ECA Working Party on Women and the Agricultural Family in Rural Development . Prague. Petrin, T. 1994, 8-14 September 1994 Entrepreneurship as an economic force in rural development. Paper presented at the Seventh FAO/REU International Rural Development Summer School, Herrsching, Germany. Pfeiffer, L., and J. E. Taylor 2007 "Gender and the Impacts of International Migration: Evidence from Rural Mexico" In Women in International Migration. ed. A. Morrion, M. Schiff and M. Sjoblom. Washington D.C.: World Bank. Pp. 99-123. Ponce, J., I. Olivié, and M. Onofa 2008 "Remittances for Development? A Case Study of the Impact of Remittances on Human Development in Ecuador." FLACSO. Ponce, J., I. Olivié, and M. Onofa 2009 "The Role of International Remittances in Health Outcomes in Ecuador: ¿Prevention or Response to Shocks?" Quito, Ecuador: FLACSO-Elcano Institute. Pribilsky, J. 2001 "Nervios and ‘Modern Childhood’. Migration and shifting contexts of child life in the Ecuadorian Andes." Childhood, 8 (2): 251-273. Pribilsky, J. (2007), La Chulla Vida: Gender, Migration, and the Family in Andean Ecuador and 5ew York City. Syracuse University Press: Syracuse. Quinn, M. A. 2009 "Estimating the Impact of Migration and Remittances on Agricultural Technology." The Journal of Developing Areas, 43 (1): 199-216. Ramírez Gallegos, F., and J. P. Ramírez (2005), La Estampida Migratoria Ecuatoriana, Crisis, Redes Transnacionales y Repertorios de Acción Migratoria. CIUDAD: QuitoEcuador. Ramírez Gallegos, F., and J. P. Ramírez 2006 "Redes Transnacionales y Repertorios de Acción Migratoria: de Quito y Guayaquil para las Ciudades del Primer Mundo" In La Migración Ecuatoriana: Transnacionalismo, Redes e Identidades. ed. G. Herrera, M. C. Carrillo and A. Torres. Quito-Ecuador: FLACSO. Ratha, D. 2003 "Workers’ Remittances: An Important and Stable Source of External Development Finance" In Global Development Finance. Striving for Stability in Development Finance. Washington D.C.: World Bank. Pp. 157-175. Reichert, J. S. 1981 "The Migrant Syndrome: Seasonal U.S. Wage Labor and Rural Development in Central Mexico." Human Organization, 40: 56-66. 143
Roodman, D. 2007 "CMP: Stata module to implement conditional (recursive) mixed process estimator." Boston: Statistical Software Components, Boston College Department of Economics. Salas Alfaro, R., and M. Pérez Morales. 2007 "Migración internacional, remesas y actividades agrícolas en una comunidad zapoteca oaxaqueña." CIMEXUS, 2 (2). Salgado, W. 2000 "La Crisis en el Ecuador en el Contexto de las Reformas Financieras." Ecuador Debate, 51. Sánchez, B. 2004 "El impacto de la emigración en Loja" In Migraciones. Un juego con cartas marcadas. ed. F. Hidalgo. Quito-Ecuador: Abya-Yala. Pp. 341-365. Sánchez Parga, J. 1984 "Estrategias de Supervivencia" In Estrategias de Supervivencia en la Comunidad Andina. ed. J. Sánchez Parga, M. Chiriboga, G. Ramón and A. Guerrero. Quito-Ecuador: CAAP. Pp. 9-58. Schrieder, G., and B. Knerr. 2000 "Labour Migration as a Social Security Mechanism for Smallholder Households in Sub-Saharan Africa: The Case of Cameroon." Oxford Development Studies, 28 (2): 223-236. Smith, R. J., and R. W. Blundell. 1986 "An Exogeneity Test for a Simultaneous Equation Tobit Model with an Application to Labor Supply." Econometrica, 54 (3): 679-685. Solimano, A. 2003, Workers Remittances to the Andean Region: Mechanisms, Costs and Development Impact. Paper presented at the Multilateral Investment Fund-IDB’s Conference on Remittances and Development, Quito-Ecuador. Staiger, D., and J. H. Stock. 1997 "Instrumental Variables Regression with Weak Instruments." Econometrica, 65: 557 – 586. Stark, N., and D. Markley. 2008 Rural Entrepreneurship Development II: Measuring Impact on the Triple Bottom Line: CFED & RIPRI. Stark, O. 1980 "On the Role of Urban-to-Rural Remittances in Rural Development " Journal of Development Studies, 16 (3): 369-374. Stark, O., and D. Bloom. 1985 "The new economics of labor migration." American Economic Review, 75 (2): 173-178. Stark, O., and D. Levhari. 1982 "On migration and risk in LDCs." Development and Cultural Change, 31: 191-196. StataCorp (2007), Stata Base Reference Manual I-P. Stata Press: Texas. Stuart, J., and M. Kearney. 1981 Causes and Effects of Agricultural Labor Migration from the Mixteca of Oaxaca to California: La Jolla: Program in United States Mexican Studies, University of California at San Diego. 144
Studenmund, A. H. (2001), Using Econometrics: A Practical Guide. Addison-Wesley: Boston. Taylor, J. E. 1992 "Remittances and Inequality Reconsidered: Direct, Indirect and Intertemporal Effects." Journal of Policy Modeling, 14 (2): 187-208. Taylor, J. E. 1999 "The New Economics of Labour Migration and the Role of Remittances in the Migration Process." International Migration, 37 (1): 63-88. Taylor, J. E., J. Arango, G. Hugo, A. Kouaouci, D. S. Massey, and A. Pellegrino. 1996 "International Migration and Community Development." Population Index, 62 (3): 397-418. Taylor, J. E., and P. L. Martin 2001 "Human Capital: Migration and Rural Population Change" In Handbook of Agricultural Economics. ed. B. Gardner and G. Rausser. New York: Elsevier Science. Taylor, J. E., and J. Mora. 2006 Does Migration Reshape Expenditures in Rural Households? Evidence from Mexico. Washington D.C.: World Bank Policy Research Working Paper 03415. Taylor, J. E., and T. J. Wyatt. 1996 "The Shadow Value of Migrant Remittances, Income and Inequality in a Household-farm Economy." The Journal of Development Studies, 32 (6): 899-912. Todaro, M. P. 1969 "A model of labor migration and urban unemployment in less-developed countries." American Economic Review, 59: 138-148. U.S. Census Bureau 1990 1990 Census of Population. . UNDP. 2009 Human Development Report 2009. Overcoming barriers: Human mobility and development. New York: UNDP. Vasco, C. 2007 Ecuadorian Out Migration to Spain: Causes and Economic Consequences. Master Thesis, Universität Kassel, Witzenhausen. Wahba, J., and Y. Zenou. 2009 Out of Sight, Out of Mind: Migration, Entrepreneurship and Social Capital: IZA Discussion Paper No. 4541. Wiest, R. E. 1979 "Implications of International Labor Migration for Mexican Rural Development" In Migration across Frontiers: Mexico and the United States. ed. F. Camara and R. Van Kemper. Institute for Mesoamerican Studies, State University of New York. Wooldridge, J. M. (2002a), Econometric Analysis of Cross Section and Panel Data. MIT. Press: Cambridge, MA.
145
Wooldridge, J. M. (2002b), Introductory Econometrics: a Modern Approach. South-Western Educational Publishing. World Bank. 1990 World Development Report 1990. Washington D.C.: World Bank. World Bank. 2001 World development report 2000/2001 - attacking poverty. Washington D.C.: World Bank. World Bank (2006), Global Economic Prospects. Economic Implications of Remittances and Migration. World Bank: Washington D.C. World Bank 2010 How we Classify Countries . Wouterse, F., and J. E. Taylor. 2008 "Migration and Income Diversification Evidence from Burkina Faso." World Development, 36: 625-640. Wouterse, F. S. 2008 "Migration and Technical Efficiency in Cereal Production: Evidence from Burkina Faso." IFPRI Discussion Paper 00815. Zambrano, G. (1998), El sueño americano. Los inmigrantes ecuatorianos en 5ueva York. Corporación de Investigación, Liderazgo y Desarrollo Ecuatoriano (CILDE): QuitoEcuador.
146
Annex A Table A.1 First stage regression for migration in the model of fertilizer expenditure (probit). Migration 0.187*** umber of cell phones 0.374*** Private school Age -0.061*** Age squared -0.0005*** Sex 0.495*** Indigenous 0.069 Education 0.065** Education squared -0.004*** Children 0.039* Young men -0.100* Young women -0.065 Adult men -0.055 Adult women -0.244*** HH education 0.004 Owned land 0.006** Owned land squared -0.0000* Number of parcels 0.085** Owned home -0.037 Electricity 0.389** Piped water 0.106 Credit 0.110 Distance to the closest road -0.057 Time to the closest market -0.002 Number of observations 4,720 Joint significance of instruments (χ²) 44.76*** p-value 0.000 Notes: *, ** and *** stand for significance at the 10, 5, and 1% levels, respectively. Specification also includes provincial dummies
147
Table A.2 Exclusion restriction of instruments for migration having fertilizer expenditure as the outcome variable (tobit). Fertilizer expenditure 0.097 umber of cell phones 0.709 Private school Age 0.017 Age squared -0.0003 Sex -0.448** Indigenous -0.069 Education 0.041 Education squared -0.0003 Children -0.047 Young men -0.063 Young women -0.070 Adult men 0-192 Adult women -0.096 HH education 0.020 Owned land -0.0002 Owned land squared 0.0000 Number of parcels 0.704*** Owned home 0.087 Electricity 0.865*** Piped water -0.048 Credit 1.860*** Distance to the closest road -0.057 Time to the closest market -0.002 Number of observations 4,720 Joint significance of instruments (χ²) 0.92 p-value 0.398 Notes: *, ** and *** stand for significance at the 10, 5, and 1% levels, respectively. Specification also includes provincial dummies
148
Table A.3 First stage regression for remittances in the model of fertilizer expenditure and cattle acquisition23 (OLS). Remittances 82.470*** Remittances Spain 9.882*** Clothes as gift Age -0.131 Age squared 0.0006 Sex 30.25*** Indigenous -0.922 Education 0.331 Education squared -0.058 Children 3.709*** Young men -3.077 Young women 4.000 Adult men 3.977 Adult women -1.859 HH education 2.152*** Owned land -0.011 Owned land squared 0.0000 Number of parcels 1.646 Owned home 4.651 Electricity 3.455* Piped water 11.035** Credit -1.543 Distance to the closest road 1.310 Time to the closest market -0.888*** Number of observations 4,720 Joint significance of instruments (F) 45.42*** p-value 0.0000 Notes: *, ** and *** stand for significance at the 10, 5, and 1% levels, respectively. Specification also includes provincial dummies
23
Since the specification used for fertilizer expenditure and cattle acquisition is the same and the instruments proposed for migration are also the same, the first stage regression for remittances is valid for both models. 149
Table A.4 Exclusion restriction of instruments for remittances having fertilizer expenditure as the outcome variable (tobit). Fertilizers expenditure -0.049 Remittances Spain -0.151 Clothes as gift Age -0.019 Age squared -0.0003 Sex -0.436** Indigenous -0.082 Education 0.039 Education squared 0.0000 Children -0.042 Young men -0.044 Young women -0.041 Adult men 0.199 Adult women -0.077 HH education 0.029 Owned land -0.0002 Owned land squared 0.0000 Number of parcels 0.704*** Owned home 0.078 Electricity 0.895*** Piped water -0.023 Credit 1.862*** Distance to the closest road -1.806*** Time to the closest market -0.011 Number of observations 4,720 Joint significance of instruments (F) 0.84 p-value 0.432 Notes: *, ** and *** stand for significance at the 10, 5, and 1% levels, respectively. Specification also includes provincial dummies.
150
Table A.5 Smith-Blundell test of exogeneity for migration and remittances for the fertilizer expenditure. Specification
1st
2nd
3rd
χ²
Prob.
χ²
Prob.
χ²
Prob.
Migration
7.100
0.077
3.334
0.067
0.790
0.373
Remittances
0.147
0.700
0.091
0.762
0.971
0.324
Suspected endogenous variables:
Notes: 1st specification: household and household head characteristics 2nd specification: household and household head characteristics, land and home ownership, credit, services, and road infrastructure 3rd specification: household and household head characteristics, land and home ownership, credit, services, road infrastructure and provincial dummies
151
Table A.6 First stage regression for migration in the model of cattle acquisition (probit). Migration 0.462*** Cell phone Availability 0.959*** Mono-parental household Age 0.066*** Age squared -0.0005*** Sex 0.247** Indigenous 0.079 Education 0.049* Education squared -0.003** Children 0.039* Young men -0.060 Young women -0.018 Adult men -0.061 Adult women -0.151** HH education -0.007 Owned land 0.005* Owned land squared -0.0000 Number of parcels 0.079*** Owned home -0.095 Electricity 0.300** Piped water 0.091 Credit 0.114 Distance to the closest road -0.064 Time to the closest market -0.001 Number of observations 4,720 Joint significance of instruments (χ²) 95.53*** p-value 0.0000 Notes: *, ** and *** stand for significance at the 10, 5, and 1% levels, respectively. Specification also includes provincial dummies.
152
Table A.7 Exclusion restriction of instruments for migration having cattle acquisition as the outcome variable (probit). Cattle acquisition 0.013 Cell phone Availability 0.091 Mono-parental household Age 0.013 Age squared -0.0001 Sex -250*** Indigenous -0.037 Education 0.001 Education squared -0.0008 Children -0.002 Young men -0.001 Young women -0.020 Adult men -0.125** Adult women 0.034 HH education 0.005 Owned land 0.002** Owned land squared -0.0000 Number of parcels 0.095*** Owned home 0.158* Electricity 0.145 Piped water 0.015 Credit 0.431*** Distance to the closest road 0.083 Time to the closest market -0.003 Number of observations 4,720 Joint significance of instruments (χ²) 0.41 p-value 0.814 Notes: *, ** and *** stand for significance at the 10, 5, and 1% levels, respectively. Specification also includes provincial dummies.
153
Table A.8 Exclusion restriction of instruments for remittances having cattle acquisition as the outcome variable (probit). Cattle acquisition 0.053 Remittances Spain 0.037 Clothes as gift Age 0.013 Age squared -0.0001 Sex -0.238** Indigenous -0.036 Education 0.001 Education squared -0.0008 Children -0.002 Young men -0.0001 Young women -0.020 Adult men -0.121* Adult women 0.032 HH education 0.006 Owned land 0.002** Owned land squared 0.0000 Number of parcels 0.095*** Owned home 0.163** Electricity 0.142 Piped water 0.015 Credit 0.429*** Distance to the closest road 0.086 Time to the closest market -0.003 Number of observations 4,720 Joint significance of instruments (χ²) 0.70 p-value 0.704 Notes: *, ** and *** stand for significance at the 10, 5, and 1% levels, respectively. Specification also includes provincial dummies.
154
Table A.9 Smith-Blundell test of exogeneity for migration and remittances for the fertilizer expenditure. Specification
1st
3rd
2nd
χ²
Prob.
χ²
Prob.
χ²
Prob.
Migration
0.013
0.907
0.027
0.868
0.007
0.932
Remittances
0.071
0.789
0.026
0.870
0.170
0.679
Suspected endogenous variables:
Notes: 1st specification: household and household head characteristics 2nd specification: household and household head characteristics, land and home ownership, credit, services, and road infrastructure 3rd specification: household and household head characteristics, land and home ownership, credit, services, road infrastructure and provincial dummies.
155
Annex B Table B.1 First stage regression for migration for the model of community work likelihood (probit). Migration 0.493*** Cell phone Availability 0.324*** Mono-parental household Age 0.063*** Age squared -0.0005*** Sex 0.487*** Indigenous 0.066 Education 0.065** Education squared -0.004*** Children 0.029 Young men -0.080 Young women -0.034 Adult men -0.037 Adult women -211*** HH education 0.002 Owned land 0.007** Owned land squared -0.0000* Number of parcels 0.074** Owned home -0.058 Electricity 0.325** Piped water 0.142* Credit 0.082 Per capita income 0.0004 Town education -0.017 Distance to the closest road -0.062 Time to the closest market -0.003 Number of observations 4,720 Joint significance of instruments (χ²) 51.96*** p-value 0.0000 Notes: *, ** and *** stand for significance at the 10, 5, and 1% levels, respectively. Specification also includes provincial dummies.
156
Table B.2 Exclusion restriction of instruments for migration having community work as the outcome variable (probit). Community work 0.018 Cell phone Availability 0.040 Mono-parental household Age 0.007 Age squared -0.0001** Sex -0.074 Indigenous 0.348*** Education -0.006 Education squared -0.0008 Children 0.048*** Young men 0.021 Young women -0.050 Adult men 0.071 Adult women 0.012 HH education 0.002 Owned land -0.0002 Owned land squared 0.0000 Number of parcels 0.080*** Owned home 0.325*** Electricity 0.018 Piped water -0.425*** Credit 0.039 Per capita income -0.014** Town education 0.027 Distance to the closest road -0.449*** Time to the closest market 0.003 Number of observations 4,720 Joint significance of instruments (χ²) 0.95 p-value 0.811 Notes: *, ** and *** stand for significance at the 10, 5, and 1% levels, respectively. Specification also includes provincial dummies.
157
Table B.3 First stage regression for Remittances in the model of community work and labor exchange participation24 (OLS). Remittances 82.053*** Remittances from Spain 40.745*** Private medical care Age -0.186 Age squared 0.001 Sex 31.136*** Indigenous -0.311 Education 0.513 Education squared -0.074 Children 4.138*** Young men -3.086 Young women 3.861 Adult men 3.520 Adult women -2.158 HH education 1.727** Owned land -0.011 Owned land squared 0.0000 Number of parcels 2.045 Owned home 4.893 Electricity 3.002 Piped water 7.854** Credit -0.760 Per capita income 0.140*** Town education -1.937 Distance to the closest road 1.247 Time to the closest market -0.780*** Number of observations 4,720 Joint significance of instruments (F) 42.90*** p-value 0.0000 Notes: *, ** and *** stand for significance at the 10, 5, and 1% levels, respectively. Specification also includes provincial dummies.
24
The instruments for remittances in the case of community work and labor exchange participation are the same. 158
Table B.4 Exclusion restriction of instruments for migration having community work as the outcome variable (probit). Community work 0.066 Remittances from Spain 0.036 Private medical care Age 0.007 Age squared 0.0001 Sex -0.075 Indigenous 0.351*** Education -0.007 Education squared -0.0008 Children 0.049*** Young men 0.022 Young women -0.049 Adult men 0.070 Adult women 0.009 HH education 0.002 Owned land -0.0003 Owned land squared 0.0000 Number of parcels 0.080*** Owned home 0.327*** Electricity 0.017 Piped water -0.426*** Credit 0.038 Per capita income -0.001*** Town education 0.027 Distance to the closest road -0.449*** Time to the closest market 0.003 Number of observations 4,720 Joint significance of instruments (χ²) 1.02 p-value 0.600 Notes: *, ** and *** stand for significance at the 10, 5, and 1% levels, respectively. Specification also includes provincial dummies.
159
Table B.5 Smith-Blundell test of exogeneity for migration and remittances for likelihood of community work participation. Specification
1st
3rd
2nd
χ²
Prob.
χ²
Prob.
χ²
Prob.
Migration
0.097
0.755
0.168
0.681
0.288
0.591
Remittances
1.393
0.237
1.446
0.229
1.016
0.313
Suspected endogenous variables:
Notes: 1st specification: household and household head characteristics 2nd specification: household and household head characteristics, land and home ownership, credit, services, and road infrastructure 3rd specification: household and household head characteristics, land and home ownership, credit, services, road infrastructure and provincial dummies.
160
Table B.6 First stage regression for migration for the model of labor exchange participation and use of wage labor25 (probit). Migration 0.355*** Children under grandparental care 1.028*** Mono-parental household Age 0.062*** Age squared -0.0005*** Sex 0.230** Indigenous 0.034 Education 0.045 Education squared -0.003** Children 0.030 Young men -0.030 Young women -0.0004 Adult men -0.026 Adult women -0.093 HH education 0.024 Owned land 0.006** Owned land squared -0.0000 Number of parcels 0.075** Owned home -0.079 Electricity 0.397** Piped water 0.150* Credit 0.095 Per capita income 0.0006 Town education -0.016 Distance to the closest road -0.061 Time to the closest market -0.002 Number of observations 4,720 Joint significance of instruments (χ²) 80.06*** p-value 0.0000 Notes: *, ** and *** stand for significance at the 10, 5, and 1% levels, respectively. Specification also includes provincial dummies.
25
The instruments for migration in the models estimating the likelihood of labor exchange participation and use of wage labor as well as the specifications are the same. 161
Table B.7 Exclusion restriction of instruments for migration having labor exchange as the outcome variable (probit). Labor exchange -0.091 Children under grandparental care -0.025 Mono-parental household Age 0.005 Age squared -0.0001 Sex 0.078 Indigenous 0.368*** Education -0.002 Education squared -0.001 Children 0.014 Young men 0.042 Young women -0.003 Adult men 0.055 Adult women -0.071 HH education -0.005 Owned land -0.002*** Owned land squared 0.0000** Number of parcels 0.114*** Owned home 0.079 Electricity -0.087 Piped water -0.090 Credit 0.202*** Per capita income -0.002*** Town education -0.055** Distance to the closest road 0.430 Time to the closest market -0.007 Number of observations 4,720 Joint significance of instruments (χ²) 1.12 p-value 0.572 Notes: *, ** and *** stand for significance at the 10, 5, and 1% levels, respectively. Specification also includes provincial dummies.
162
Table B.8 Exclusion restriction of instruments for remittances having labor exchange as the outcome variable (probit). Labor exchange 0.051 Remittances from Spain -0.072 Private medical care Age 0.004 Age squared -0.0001* Sex 0.073 Indigenous 0.363*** Education -0.002 Education squared -0.001 Children 0.012 Young men 0.043 Young women -0.004 Adult men 0.058 Adult women -0.065 HH education -0.005 Owned land -0.002*** Owned land squared 0.0000** Number of parcels 0.114*** Owned home 0.081 Electricity -0.085 Piped water -0.087 Credit 0.201*** Per capita income -0.002*** Town education -0.056** Distance to the closest road 0.433*** Time to the closest market -0.008 Number of observations 4,720 Joint significance of instruments (χ²) 1.98 p-value 0.370 Notes: *, ** and *** stand for significance at the 10, 5, and 1% levels, respectively. Specification also includes provincial dummies.
163
Table B.9 Smith-Blundell test of exogeneity for migration and remittances for likelihood of community work participation. Specification
1st
3rd
2nd
χ²
Prob.
χ²
Prob.
χ²
Prob.
Migration
0.057
0.810
0.027
0.867
0.096
0.756
Remittances
0.522
0.469
0.057
0.810
0.032
0.857
Suspected endogenous variables:
Notes: 1st specification: household and household head characteristics 2nd specification: household and household head characteristics, land and home ownership, credit, services, and road infrastructure 3rd specification: household and household head characteristics, land and home ownership, credit, services, road infrastructure and provincial dummies.
164
Table B.10 Exclusion restriction of instruments for migration having labor use of wage labor as the outcome variable (probit). Wage labor 0.218 Children under grandparental care -0.067 Mono-parental household Age 0.006 Age squared -0.0000 Sex -0.024 Indigenous -0.454*** Education 0.073*** Education squared -0.002** Children -0.075*** Young men -0.149*** Young women 0.002 Adult men -0.0003 Adult women 0.019 HH education 0.053*** Owned land 0.001* Owned land squared 0.0000 Number of parcels 0.170 Owned home 0.101 Electricity 0.202** Piped water 0.129** Credit 0.531*** Per capita income -0.001** Town education -0.014 Distance to the closest road 0.316 Time to the closest market -0.018 Number of observations 4,720 Joint significance of instruments (χ²) 2.81 p-value 0.245 Notes: *, ** and *** stand for significance at the 10, 5, and 1% levels, respectively. Specification also includes provincial dummies.
165
Table B.11 First stage regression for remittances in the model of use of wage labor (OLS). Remittances 82.10*** Remittances from Spain 9.51*** Clothes as gift Age -0.161 Age squared 0.0007 Sex 30.240*** Indigenous -0.322 Education 0.507 Education squared -0.076 Children 3.875*** Young men -2.839 Young women 3.941 Adult men 3.904 Adult women -1.792 HH education 1.780** Owned land -0.013 Owned land squared 0.0000 Number of parcels 2.045 Owned home 5.495 Electricity 2.474 Piped water 7.568** Credit -1.001 Per capita income 0.147*** Town education -1.524 Distance to the closest road 2.040 Time to the closest market -0.827*** Number of observations 4,720 Joint significance of instruments (F) 45.74*** p-value 0.0000 Notes: *, ** and *** stand for significance at the 10, 5, and 1% levels, respectively. Specification also includes provincial dummies.
166
Table B.12 Exclusion restriction of instruments for remittances having use of wage labor as the outcome variable (probit). Wage labor 0.036 Remittances from Spain 0.006 Clothes as gift Age 0.006 Age squared -0.0000 Sex 0.003 Indigenous -0.454*** Education 0.074*** Education squared -0.002** Children -0.077*** Young men -0.151*** Young women -0.002 Adult men 0.003 Adult women 0.009 HH education 0.053*** Owned land 0.001* Owned land squared 0.0000 Number of parcels 0.170*** Owned home 0.106 Electricity 0.198** Piped water 0.128** Credit 0.531*** Per capita income -0.001* Town education -0.016 Distance to the closest road 0.331 Time to the closest market -0.020 Number of observations 4,720 Joint significance of instruments (χ²) 0.15 p-value 0.926 Notes: *, ** and *** stand for significance at the 10, 5, and 1% levels, respectively. Specification also includes provincial dummies.
167
Table B.13 Smith-Blundell test of exogeneity for migration and remittances for likelihood of use of wage labor. Specification
1st
3rd
2nd
χ²
Prob.
χ²
Prob.
χ²
Prob.
Migration
0.017
0.893
0.083
0.772
0.523
0.469
Remittances
0.265
0.606
0.0001
0.989
0.001
0.970
Suspected endogenous variables:
Notes: 1st specification: household and household head characteristics 2nd specification: household and household head characteristics, land and home ownership, credit, services, and road infrastructure 3rd specification: household and household head characteristics, land and home ownership, credit, services, road infrastructure and provincial dummies.
168
Table B.14 Share of households participating in community work and share of households exchanging labor by province. Province
Region
Median of the distance to the closest road (km)
Mean of landholding (ha)
Share of households participating in community work (%)
Share of households exchanging labor (%)
Cañar
Sierra
0
0.50
26,1%
29,1%
Carchi
Sierra
0
1.50
34,6%
6,2%
Cotopaxi
Sierra
0
0.95
55,6%
22,3%
Chimborazo
Sierra
0
0.67
66,1%
27,3%
El Oro
Costa
0
2
2,5%
3,4%
Imbabura
Sierra
0
0.44
19,8%
8,2%
Los Ríos
Costa
0
1.75
1,7%
6,9%
Pastaza
Oriente
0
7.00
34,3%
14,3%
Pichincha
Sierra
0
0.84
15,1%
5,8%
Tungurahua
Sierra
0
0.35
48,8%
15,4%
Sucumbios
Oriente
0
7.00
22,6%
7,1%
Bolivar
Sierra
0.01
1.40
52,9%
35,3%
Azuay
Sierra
0.02
0.25
14,8%
16,3%
Guayas
Costa
0.5
0.70
3,2%
3,7%
Napo
Oriente
1
2.00
27,9%
25,0%
Orellana
Oriente
1
32.00
32,9%
27,6%
Zamora
Oriente
1
4.00
4,2%
14,1%
Loja
Sierra
1.1
1.00
24,2%
27,8%
Esmeraldas
Costa
2
7.00
18,0%
22,5%
Manabí
Costa
2
1.40
3,9%
21,8%
Morona
Oriente
2
10.60
8,6%
41,4%
169
Annex C Table C.1 First stage regression for migration in the model for business ownership (probit). Migration Children under grandparental care 0.350*** Unemployment rate 2001 -5.519*** Age .0071*** Age squared -0.0006*** Sex 0.470*** Indigenous 0.117 Education 0.053* Education squared -0.004* Children 0.036* Young men -0.069 Young women 0.001 Adult men 0.006 Adult women -0.190*** HH education 0.009 Owned land 0.007*** Owned land squared -0.0000** Owned home -0.040* Electricity 0.273 Piped water 0.124*** Indoors water system 0.312 Credit 0.198** Distance to the closest road -0.139 Time to the closest market -0.007 Number of observations 4,379 Joint significance of instruments (χ²) 21.21*** p-value 0.0000 Notes: *, ** and *** stand for significance at the 10, 5, and 1% levels, respectively. Specification also includes provincial dummies.
170
Table C.2 Exclusion restriction of instruments for migration having business ownership as the outcome variable (probit). Business ownership Children under grandparental care -0.052 Unemployment rate 2001 0.344 Age 0.016* Age squared -0.0002** Sex -0.119 Indigenous 0.004 Education 0.035** Education squared -0.002*** Children 0.017 Young men -0.067** Young women 0.120*** Adult men 0.115** Adult women 0.301*** HH education 0.066*** Owned land -0.003*** Owned land squared 0.0000*** Owned home 0.100* Electricity 0.244*** Piped water 0.236*** Indoors water system 0.293*** Credit 0.179*** Distance to the closest road -0.210 Time to the closest market 0.010 Number of observations 4,379 Joint significance of instruments (χ²) 0.46 p-value 0.794 Notes: *, ** and *** stand for significance at the 10, 5, and 1% levels, respectively. Specification also includes provincial dummies.
171
Table C.3 First stage regression for remittances for the model for business ownership (OLS). Remittances Remittances from Spain 81.404*** Clothes as gift 9.859*** Age -0.130 Age squared 0.0003 Sex 29.879*** Indigenous 0.501 Education 0.303 Education squared -0.068 Children 3.754*** Young men -2.694 Young women 4.169 Adult men 3.931 Adult women -2.318 HH education 1.437* Owned land -0.017 Owned land squared 0.0000 Owned home 5.308 Electricity 2.275 Piped water 6.417** Indoors water system 1.654*** Credit 1.206 Distance to the closest road 2.694 Time to the closest market -0.930*** Number of observations 4,753 Joint significance of instruments (F) 46.63 p-value 0.0000 Notes: *, ** and *** stand for significance at the 10, 5, and 1% levels, respectively. Specification also includes provincial dummies.
172
Table C.4 Exclusion restriction of instruments for remittances having business ownership as the outcome variable (probit). Business ownership Remittances from Spain -0.085 Clothes as gift 0.044 Age 0.013 Age squared -0.0001** Sex -0.111 Indigenous 0.022 Education 0.033* Education squared -0.002** Children 0.015 Young men -0.069** Young women 0.118*** Adult men 0.098** Adult women 0.291*** HH education 0.066*** Owned land -0.003*** Owned land squared 0.0000*** Owned home 0.107* Electricity 0.240*** Piped water 0.216*** Indoors water system 0.328*** Credit 0.190*** Distance to the closest road -0.084 Time to the closest market 0.004 Number of observations 4,753 Joint significance of instruments (χ²) 1.91 p-value 0.384 Notes: *, ** and *** stand for significance at the 10, 5, and 1% levels, respectively. Specification also includes provincial dummies.
173
Table C.5 First stage regression for average town remittances for the model for business ownership (OLS). Average town remittances umber of internal migrants 2001 0.0001*** Average number of women 2001 14.551*** Age 0.544*** Age squared -0.004*** Sex 0.700 Indigenous -3.644** Education -0.381 Education squared 0.005 Children -0.039 Young men -0.728 Young women 0.387 Adult men -0.413 Adult women -0.343 HH education 0.868*** Owned land -0.022** Owned land squared 0.0000* Owned home -3.775*** Electricity 5.218*** Piped water 3.7874*** Indoors water system 4.669*** Credit 2.586** Distance to the closest road 1.327 Time to the closest market -0.932*** Number of observations 4,753 Joint significance of instruments (F) 43.13*** p-value 0.0000 Notes: *, ** and *** stand for significance at the 10, 5, and 1% levels, respectively. Specification also includes provincial dummies.
174
Table C.6 Exclusion restriction of instruments for average town remittances having business ownership as the outcome variable (probit). Business ownership umber of internal migrants 2001 00000 Average number of women 2001 -0.068 Age 0.013 Age squared -0.0001** Sex -0.112 Indigenous 0.023 Education 0.033* Education squared -0.002** Children 0.016 Young men -0.070** Young women 0.118*** Adult men 0.097* Adult women 0.289*** HH education 0.066*** Owned land -0.003*** Owned land squared 0.0000*** Owned home 0.109* Electricity 0.238*** Piped water 0.214*** Indoors water system 0. 323*** Credit 0.189*** Distance to the closest road -0.083 Time to the closest market 0.005 Number of observations 4,753 Joint significance of instruments (χ²) 0.63 p-value 0.731 Notes: *, ** and *** stand for significance at the 10, 5, and 1% levels, respectively. Specification also includes provincial dummies.
175
Table C.7 Smith-Blundell test of exogeneity for migration and remittances for the likelihood of business ownership. Specification
1st
3rd
2nd
χ²
Prob.
χ²
Prob.
χ²
Prob.
Migration
0.009
0.922
0.864
0.352
0.543
0.461
Remittances
0.005
0.939
0.756
0.384
0.182
0.669
Average town remittances
0.552
0.457
0.072
0.788
0.023
0.879
Suspected endogenous variables:
Notes: 1st specification: household and household head characteristics 2nd specification: household and household head characteristics, land and home ownership, credit, services, and road infrastructure 3rd specification: household and household head characteristics, land and home ownership, credit, services, road infrastructure and provincial dummies.
176
Table C.8 First stage regression for migration for the model for household members employed in a business (probit). Migration Mono-parental household 0.961*** Cell phone availability 0.464*** Age 0.071*** Age squared -0.0005*** Sex 0.233** Indigenous 0.111 Education 0.055** Education squared -0.003** Children 0.037* Young men -0.073 Young women -0.020 Adult men -0.063 Adult women -0.147** HH education -0.013 Owned land 0.006** Owned land squared -0.0000* Owned home -0.083 Electricity 0.280* Piped water 0.085 Indoors water system 0.189** Credit 0.196** Distance to the closest road -0.052 Time to the closest market -0.002 Number of observations 1,425 Joint significance of instruments (χ²) 29.49*** p-value 0.0000 Notes: *, ** and *** stand for significance at the 10, 5, and 1% levels, respectively. Specification also includes provincial dummies.
177
Table C.9 Exclusion restriction of instruments for migration having the number of household members employed in a business as the outcome variable (tobit). umber of household members employed in a business Mono-parental household -0.306 Cell phone availability 0.002 Age 0.081** Age squared -0.0008*** Sex -1.050*** Indigenous -0.046 Education 0.064 Education squared -0.002 Children 0.074 Young men 0.371*** Young women 0.289** Adult men 0.234 Adult women 0.533*** HH education 0.059 Owned land 0.004 Owned land squared -0.0000 Owned home 0.190 Electricity -0.079 Piped water 0.198 Indoors water system -0.398 Credit 0.179 Distance to the closest road -0.075 Time to the closest market -0.009 Number of observations 1,425 Joint significance of instruments (F) 0.12 p-value 0.887 Notes: *, ** and *** stand for significance at the 10, 5, and 1% levels, respectively. Specification also includes provincial dummies.
178
Table C.10 First stage regression for remittances for the model for the number of household members employed in a business (OLS). Remittances Remittances from Spain 87.963*** Clothes as gift 20.572*** Age 0.140 Age squared -0.0000 Sex 30.209** Indigenous -4.114 Education -0.716 Education squared -0.037 Children 8.284*** Young men -4.052 Young women 4.969 Adult men 2.676 Adult women -7.027.0 HH education 2.596* Owned land 0.045 Owned land squared -0.0000 Owned home 16.492** Electricity 3.770 Piped water 5.882 Indoors water system 20.515*** Credit 2.797 Distance to the closest road 10.431 Time to the closest market -0.831** Number of observations 1,425 Joint significance of instruments (F) 17.58*** p-value 0.0000 Notes: *, ** and *** stand for significance at the 10, 5, and 1% levels, respectively. Specification also includes provincial dummies.
179
Table C.11
Exclusion restriction of instruments for remittances having the number of household members employed in a business as the outcome variable (tobit). umber of household members employed in a business Remittances from Spain 0.122 Clothes as gift 0.141 Age 0.080** Age squared -0.0008*** Sex -1.093*** Indigenous -0.046 Education 0.061 Education squared -0.002 Children 0.067 Young men 0.374*** Young women 0.290*** Adult men 0.241 Adult women 0.536*** HH education 0.058 Owned land 0.004 Owned land squared -0.0000 Owned home 0.201 Electricity -0.090 Piped water 0.193 Indoors water system 0.385* Credit 0.178 Distance to the closest road -0.059 Time to the closest market 0.010 Number of observations 1,425 Joint significance of instruments (F) 0.45 p-value 0.636 Notes: *, ** and *** stand for significance at the 10, 5, and 1% levels, respectively. Specification also includes provincial dummies.
180
Table C.12 First stage regression for average town remittances for the number of household members employed in a business (OLS). Average town remittances Average number of women 2001 241.673*** Average internet users 2001 31.865*** Age 1.217*** Age squared -0.010*** Sex 1.253 Indigenous -10.500*** Education -1.615** Education squared 0.061* Children 0.770 Young men -0.917 Young women 0.933 Adult men -3.072 Adult women -2.332 HH education 0.422 Owned land 0.003 Owned land squared 0.0000 Owned home -3.692 Electricity 3.282 Piped water 1.493 Indoors water system 4.211* Credit 1.496 Distance to the closest road 10.062 Time to the closest market -0.823** Number of observations 1,425 Joint significance of instruments (F) 39.03*** p-value 0.0000 Notes: *, ** and *** stand for significance at the 10, 5, and 1% levels, respectively. Specification also includes provincial dummies.
181
Table C.13 Exclusion restriction of instruments for average town remittances having the number of household members employed in a business as the outcome variable (tobit). umber of household members employed in a business umber of internal migrants 2001 00000 Average number of women 2001 -0.068 Age 0.013 Age squared -0.0001** Sex -0.112 Indigenous 0.023 Education 0.033* Education squared -0.002** Children 0.016 Young men -0.070** Young women 0.118*** Adult men 0.097* Adult women 0.289*** HH education 0.066*** Owned land -0.003*** Owned land squared 0.0000*** Owned home 0.109* Electricity 0.238*** Piped water 0.214*** Indoors water system 0. 323*** Credit 0.189*** Distance to the closest road -0.083 Time to the closest market 0.005 Number of observations 4,753 Joint significance of instruments (χ²) 0.63 p-value 0.731 Notes: *, ** and *** stand for significance at the 10, 5, and 1% levels, respectively. Specification also includes provincial dummies.
182
Table C.14 Smith-Blundell test of exogeneity for migration and remittances for the number of household members employed in a business. Specification
1st
3rd
2nd
χ²
Prob.
χ²
Prob.
χ²
Prob.
Migration
1.286
0.256
1.424
0.232
0.888
0.346
Remittances
0.303
0.581
0.269
0.603
0.067
0.795
Average town remittances
0.329
0.566
1.110
0.292
1.351
0.245
Suspected endogenous variables:
Notes: 1st specification: household and household head characteristics 2nd specification: household and household head characteristics, land and home ownership, credit, services, and road infrastructure 3rd specification: household and household head characteristics, land and home ownership, credit, services, road infrastructure and provincial dummies.
183
Table C.15 First stage regression for migration in the model for non- household members employed in a business (probit). Migration Cell phone availability 1.159*** Migration rate 2001 4.377*** Age 0.141*** Age squared -0.001*** Sex 0.188 Indigenous 0.250 Education 0.032 Education squared -0.002 Children 0.093*** Young men -0.067 Young women -0.061 Adult men -0.043 Adult women -0.114 HH education 0.021 Owned land 0.002 Owned land squared -0.0000 Owned home 0.108 Electricity -0.011 Piped water 0.286*** Indoors water system 0.073 Credit 0.027 Distance to the closest road -0.116 Time to the closest market -0.002 Number of observations 1,425 Joint significance of instruments (χ²) 40.97*** p-value 0.0000 Notes: *, ** and *** stand for significance at the 10, 5, and 1% levels, respectively. Specification also includes provincial dummies.
184
Table C.16 Exclusion restriction of instruments for migration having the number of nonhousehold members employed in a business as the outcome variable (tobit). umber of non-household members employed in a business Cell phone availability 0.604 Migration rate 2001 2.233 Age -0.003 Age squared -0.0003 Sex -2.248*** Indigenous -1.257** Education -0.129 Education squared 0.011 Children -0.028 Young men -0.239 Young women 0.307 Adult men 0.137 Adult women -0.377 HH education 0.244** Owned land 0.008 Owned land squared 0.0000 Owned home 1.011 Electricity 0.721 Piped water 1.680 Indoors water system 0.066*** Credit 0.427 Distance to the closest road -0.008 Time to the closest market 0.021 Number of observations 1,425 Joint significance of instruments (F) 0.37 p-value 0.693 Notes: *, ** and *** stand for significance at the 10, 5, and 1% levels, respectively. Specification also includes provincial dummies.
185
Table C.17 First stage regression for remittances for the model for the number of nonhousehold members employed in a business (OLS). Remittances Remittances from U.S. 614.375*** Remittances from Spain 550.913*** Age 5.663 Age squared -0.030 Sex 45.042 Indigenous -32.924 Education 13.357 Education squared -0.750 Children 31.895*** Young men -31.078 Young women 6.285 Adult men -42.024 Adult women -44.667* HH education 12.291** Owned land 0.730* Owned land squared -0.0005** Owned home 77.871** Electricity -0.565 Piped water 43.497 Indoors water system 68.842** Credit 8.510 Distance to the closest road 8.562 Time to the closest market -1.141 Number of observations 1,425 Joint significance of instruments (F) 17.10*** p-value 0.0000 Notes: *, ** and *** stand for significance at the 10, 5, and 1% levels, respectively. Specification also includes provincial dummies.
186
Table C.18 Exclusion restriction of instruments for migration having the number of nonhousehold members employed in a business as the outcome variable (tobit). umber of non-household members employed in a business Remittances from U.S. 0.400 Remittances from Spain 0.309 Age -0.0006 Age squared -0.0003 Sex -2.215*** Indigenous -1.258** Education -0.131 Education squared 0.011 Children -0.028 Young men -0.252 Young women 0.299 Adult men 0.108 Adult women -0.390 HH education 0.244** Owned land 0.008 Owned land squared -0.0000 Owned home 0.716 Electricity 1.035 Piped water 0.057 Indoors water system 1.688*** Credit 0.430 Distance to the closest road 0.103 Time to the closest market 0.016 Number of observations 1,425 Joint significance of instruments (F) 0.29 p-value 0.750 Notes: *, ** and *** stand for significance at the 10, 5, and 1% levels, respectively. Specification also includes provincial dummies.
187
Table C.19 First stage regression for average town remittances for the model for nonhousehold members employed in a business (OLS). Average town remittances umber of internal migrants 2001 0.0001*** Average number of absent household 98.754*** members 2001 Age 1.178*** Age squared -0.010*** Sex 2.080 Indigenous -7.037*** Education -1.486** Education squared 0.067* Children 0.702 Young men -0.349 Young women 1.674 Adult men -1.619 Adult women -2.428 HH education 0.583 Owned land -0.001 Owned land squared 0.0000 Owned home -1.635 Electricity 0.911 Piped water 4.733 Indoors water system 3.829** Credit 0.981 Distance to the closest road 1.027 Time to the closest market -0.417 Number of observations 1,425 Joint significance of instruments (F) 42.82*** p-value 0.0000 Notes: *, ** and *** stand for significance at the 10, 5, and 1% levels, respectively. Specification also includes provincial dummies.
188
Table C.20 Exclusion restriction of instruments for average town remittances having the number of non-household members employed in a business as the outcome variable (tobit). umber of non-household members employed in a business umber of internal migrants 2001 1.173 Average number of absent household 0.0000 members 2001 Age -0.0001 Age squared -0.00003 Sex -2.150 Indigenous -1.260 Education -0.110 Education squared 0.010 Children -0.030 Young men -0.226 Young women 0.299 Adult men 0.111 Adult women -0.389 HH education 0.235 Owned land 0.008 Owned land squared -0.0000 Owned home 0.752 Electricity 1.058 Piped water -0.002 Indoors water system 1.632 Credit 0.372 Distance to the closest road 0.024 Time to the closest market 0.021 Number of observations 1,425 Joint significance of instruments (χ²) 1.17 p-value 0.312 Notes: *, ** and *** stand for significance at the 10, 5, and 1% levels, respectively. Specification also includes provincial dummies.
189
Table C.21 Smith-Blundell test of exogeneity for migration and remittances for the number of non-household members employed in a business. Specification
1st
3rd
2nd
χ²
Prob.
χ²
Prob.
χ²
Prob.
Migration
0.119
0.729
0.071
0.789
0.240
0.623
Remittances
1.255
0.262
0.435
0.509
0.957
0.328
Average town remittances
0.034
0.852
0.176
0.674
0.970
0.324
Suspected endogenous variables:
Notes: 1st specification: household and household head characteristics 2nd specification: household and household head characteristics, land and home ownership, credit, services, and road infrastructure 3rd specification: household and household head characteristics, land and home ownership, credit, services, road infrastructure and provincial dummies.
190
Series International Labor Migration edited by Prof. Dr. Béatrice Knerr*
Vol. 1
Nadim Zaqqa (2006): Economic Development and Export of Human Capital. A Contradiction? The impact of human capital migration on the economy of sending countries. A case study of Jordan. Kassel, ISBN 978-3-89958-205-5
Vol. 2
Volker Hamann (2007): The Impact of International Labor Migration on Regional Development: The Example of Zacatecas, Mexico. Kassel, ISBN 978-3-89958-251-2
Vol. 3
Béatrice Knerr (Ed.) (2006): Vorweggenommene Wanderungsbewegungen aus Grenzgebieten in die EU. Kassel. ISBN 978-3-89958-281-9
Vol. 4
Germán A. Zárate-Hoyos (Ed.) (2007): New Perspectives on Remittances from Mexicans and Central Americans in the United States. Kassel, ISBN 978-3-89958-256-7
Vol. 6
Izhar Ahmad Khan Azhar (2008): Overseas Migration and its Socio-Economic Impacts on the Families Left Behind in Pakistan. A Case Study in the Province Punjab, Pakistan. Kassel, ISBN 978-3-89958-366-3
Vol. 7
Jörg Helmke (2010): Remittance-led development: Rebuilding old dependencies or a powerful source of human development? A view on Latin America. Kassel, ISBN 978-3-89958-968-9
Vol. 8
Béatrice Knerr (ed.) (2010): Multilevel Governance and Livelihood Strategies of International Migrants. Kassel, ISBN 978-3-86219-080-5
Vol. 9
Cristian Vasco (2011): The Impact of International Migration and Remittances on Agricultural Production Patterns, Labour Relationships and Entrepreneurship. The Case of Rural Ecuador. Kassel, ISBN 978-3-86219-086-7
Vol. 10
Rasha Istaiteyeh (2011): Economic Development and Highly Skilled Returnees: The Impact of Human Capital Circular Migration on the Economy of Origin Countries. The Case of Jordan. Kassel, ISBN 978-3-86219-084-3
Erweiterungen:
*) Department of Development Economics, Migration and Agricultural Policy (DEMAP), University of Kassel Steinstr. 19 D-37213 Witzenhausen
[email protected]
Cristian Vasco was born in Quito, Ecuador in 1978. He obtained his degree of agricultural engineer at the Army Polytechnic School in Ecuador and in 2005 joined the Masters‘ Program in International Ecological Agriculture at the University of Kassel, Germany. His Master thesis on “Ecuadorian out-migration to Spain: causes and economic consequences” was awarded with the annual “German Developing Countries Prize” for out-standing research accomplishments concerning Developing Countries at the University of Giessen, handed over by the German Federal Ministry of International Cooperation. It also received the prize for the best thesis written on non-European topics at the Faculty of Organic Agriculture, University of Kassel in the summer semester 2007. In 2008 he started his PhD studies at the Department of Development Economics, Migration and Agricultural Policy (DEMAP) at the University of Kassel. Over the following years he presented his research at several high-ranking international conferences. He obtained his PhD degree in early 2011.
The Impact of International Migration and Remittances on Agricultural Production Patterns, Labor Relationships and Entrepreneurship The Case of Rural Ecuador
ISBN 978-3-86219-086-7 kassel university press
/// Series edited by Béatrice Knerr ///
Cristian Vasco
Cristian Vasco
The research presented in this book quantitatively analyzes the effects of international migration and remittances on agricultural production patterns, labor relationships and entrepreneurship in rural Ecuador. The results show that migrants‘ households spent more on fertilizers and are more likely to accumulate cattle than their non-migrants‘ counterparts. They also demonstrate that neither international migration nor remittances influence the likelihood to participate in reciprocal work agreements in contrast to the conclusions of most sociological studies. Instead, migrants‘ households are more likely to hire wage laborers than their equivalents without migrants. Regarding entrepreneurship, neither migration nor remittances have any effect on households‘ likelihood to own a small-scale business and on the number of non-household members working in a business. Instead, the number of household members working in a business is positively influenced by international migration. These findings contribute to the debate about the effects of international migration on rural regions and provide policy makers as well as rural development and migration practitioners with information and facts to be taken into account for the design of projects linking international migration with rural development.
The Impact of International Migration and Remittances on Agricultural Production Patterns, Labor Relationships and Entrepreneurship
9
International Labor Migration 9
U_IntLabMig9_druck 22.02.11 09:40 Seite 1
