Production risk and level of irrigation technology. Evidence from potato family farmers in Chile.
César Salazar Espinoza1 This version 15/05/2012
Abstract The adoption of modern irrigation can reduce farmers’ exposure to drought risk and counteract productive disadvantages in zones with limiting environmental conditions. This paper uses data with national coverage from family potato farmers in Chile to examine how production risk affects the adoption of modern irrigation. As it is common in developing countries, a large fraction of rural households depends exclusively on rainfall for water supply. We propose two approaches to tackle this problem. First, we treat nonirrigators data as one more level of technology by the estimation of an ordered probit model. Alternatively, we estimate models with sample selection when analyzing the shift from traditional to modern technology arguing that there may have a systematic trend for certain farmers to irrigate less than others. The results indicate that farmers with higher educational level, larger proportion of land under secure tenure arrangements, credit access, receiving extension services and those who reside in the plot are more likely to adopt new technology. On the other hand, the results for production risk and soil quality depend on the approach chosen. Whereas the results from the ordered probit confirm the virtues of irrigation technology as a risk decreasing-input and under more appropriate soil conditions, estimations from sample selection models show that modern technology seems to be perceived as a risk-increasing input and may be more probable to be adopted in zones with poorer soil quality. Even though the direction of the association in the first case is in line with that observed in previous studies, we argue that lower diffusion and knowledge on modern technology may be underlying these differences.
Key words: production risk, irrigation, technology adoption. JEL classification: D8, O13, Q15, Q55
1
Department of Economics, University of Copenhagen, Denmark and Department of Economics and Finance, University of Bio-Bio, Chile. e-mail:
[email protected],
[email protected].
1. Introduction Agricultural innovation is a dynamic process that responds to a changing environment and to a constant search for improved production methods aimed at both increasing the yield and adding value to products. Since the green revolution, agriculture in developing countries has experienced a transformation process principally driven by an increasing demand for agricultural products. It has triggered a massive adoption of technologies mainly targeted to increase productivity, as the use of chemical inputs, pesticides, traditional irrigation, and high-yield seed varieties. The relatively easy implementation and the low level of knowledge needed in the adoption has allowed a rapid spread among farmers, bringing productivity gains and lower prices that have resulted in increases in farmers’ incomes and poverty reduction (Evenson and Gollin, 2003) Even though the benefits derived from using those methods in terms of efficiency have been significant, concerns have been raised lately on the grounds of sustainability as the intensive use has consequences for environment and human health (Ruttan, 2002). In particular, the expansion of irrigation technology by traditional methods to face the inherent vulnerability of the agricultural economy to water shortage, has triggered pressures on water reservoirs and pollution problems2. In this context and under insufficient levels of investment in water supply infrastructure, the adoption of water-conserving irrigation technologies arises as one alternative to improve the use of water resources and minimize contamination problems (Dinar and Zilberman, 1991; Khana and Zilberman, 1996)3. Several studies have addressed the problem of adopting new irrigation methods by studying the effects of a series of institutional aspects, output, and input prices as well as intrinsic factors at the household level that describe farmers’ behavior and characteristics4.
2
Too much water poured on the ground by traditional irrigation systems may provoke a saturation process, making excess water flows over the land along with pollutants derived from remains of pesticides and fertilizers. 3
Improved technologies allow a more uniform irrigation which promotes saving and minimize the generation of residues without sizeable effects on yield. 4 The determinants of modern irrigation have been examined by Caswell and Zilberman (1985), Negri and Brooks (1990), Shrestha and Gopalakrishnan (1993). Kulshreshta and Brown (1993), Green et al. (1996), Dorfman (1996), Skaggs (2001), Foltz (2003), Daberkow and Mcbride (2003), Moreno and Sunding (2005), He et al. (2007), Schuck et al. (2007).
1
A key factor less studied in the literature of technology adoption is the role of risk in adopting modern irrigation methods. If we assume farmers are risk-averse and face constraints to mitigate the adverse consequences of income’s fluctuations, new technologies might not be adopted because this could raise production risk either by increasing yield variability or by pushing up the likelihood of crop failure. In order to avoid an increase in uncertainty, risk-averse farmers would prefer to maintain conventional agricultural practices, as ex-ante strategies to smooth income, at the cost of lower expected profits (Bardan and Udry, 1999). In contrast, if the new technology has the potential to attenuate variability in economic outcomes, then it would be more massively adopted among risk-averse individuals. The potential of distinct agro-ecological zones for growing crops, on the other hand, may affect the outcomes of adoption. Water scarcity and unfavorable soil conditions restrict the economic viability of farming activities, and water-saving irrigation methods may arise as one alternative to counteract these disadvantages by increasing the water effectiveness (Dinar and Yaron, 1990)5. Thus, farmers facing limiting growing conditions should react more quickly to the introduction of modern irrigation technologies as these substantially reduce impacts of quality soil and weather differences on profitability. The main objective of this work is to investigate the effect of production risk and on adoption of modern irrigation. A risk measure for each household is computed by the econometric estimation of the moments of a stochastic production function (Antle, 1983; 1987). Then, the estimated moments are incorporated along with other control variables in analyzing adoption decisions. This work utilizes data of potato family farmers, collected from the VII Agricultural and Forest Census 2007, Chile. Even though important progress has been observed in Chilean agriculture in terms of investment in more profitable crops and agricultural innovations in the last decade, the evidence shows that these achievements have mainly been driven by medium and large-scale farmers. Recent information, in fact, shows that only 6% of land held by family farmers is irrigated, and modern irrigation methods only represent 10.7% of total irrigation. This compares to 33.3% of improved irrigation observed among non-family farmers (FAO, 2009). This fact corroborates differences in
5
In particular, several studies have addressed the effects of land quality and climatic conditions on the choice of irrigation technology. For more information see: Caswell and Zilberman (1986), Lichtenberg (1989), Dinar and Yaron (1990), Dinar and Zilberman (1991), Dinar and Zilberman (1992), Dinar and Yaron (1992), Green and Sunding (1997), Negri et al. (2005).
2
patterns of technological adoption between small and large-scale farmers and motivates the study of the factors that could be lagging the adoption of irrigation in family agriculture, in spite of important efforts made to assist this segment. This is particularly important in the light of the recurrent natural phenomenon observed in Chile, namely Southern-Oscillation (El Niño and La Niña), which substantially increases the likelihood of drought risk. In fact, Chile has historically showed some extent of drought in one out of nine years, with a duration that ranges from one to six years and has extended to the 25% of the continental territory. Furthermore, studies on climate change in Chile project to the end of the XXI century an intensification of the aridity in the north zone, advances of the desert toward the south and less precipitation in the central zone. Thus, the augment in temperature and declination in precipitations along with an increasing demand for water resources foresee a higher drought risk for the next decades (FAO, 2010). Therefore, improvements in water efficiency by the adoption of modern technology become essential to reduce the vulnerability of agriculture at the face of this new scenario. Even though literature on adoption of improved irrigation technology is well-documented, there is little evidence regarding the effects of production risk on adoption decisions. Two exceptions are Koundouri et al. (2006) and Torkamani and Shajari (2008). Both studies find support of a positive and significant effect of production risk on adoption decisions, arguing that new irrigation technology is a risk-decreasing input. However, both studies base their results on samples of irrigators, allowing only to model changes from traditional to modern irrigations. Thus, none of them discuss potential biases that may arise when data of non-irrigators are either unobserved or not considered. In contrast, we use data with national coverage to analyze the adoption of modern irrigation, which allows us to include that part of the rural population that do not hold any irrigation method, and in consequence are totally exposed to climate fluctuations. Furthermore, studies of the effects of risk on adoption of modern irrigation in previous works have been carried out in settings where the new technology seems to show some degree of consolidation. On the contrary, and due to more regular precipitations in southern territories in Chile, rates of adoption in both traditional and modern irrigation are quite low, which makes our setting differ from prior contexts. In order to incorporate non-irrigators in the analysis of adoption of irrigation, we propose the estimation of two 3
models. First, we model irrigation decisions in levels, assuming that each category corresponds to a superior level of technology. Alternatively, we estimate sample selection models arguing the existence of a systematic tendency of certain types of farmers and locations to irrigate less than others. Foudi and Erdlenbruch (2011) applied a sample selection approach to study risk-decreasing properties of irrigation technology under a joint estimation of production function and production risk. However, to our knowledge, there is no evidence of previous works employing this approach to study the decision of shifting from conventional to water-saving technologies. Likewise, this work deals with a recurring problem observed in micro-studies in which a few villages or areas are surveyed and data lack sufficient variation capturing the agricultural potential (Doss, 2006)6. By taking advantage of a considerably larger sample taken from a national cross-section survey and the longitudinal extension (north to south) of Chilean territory, we are allowed to control for agro-ecological variations with more precision, and location variables will also pick up better the differences with respect to market access and infrastructure. The rest of the article is organized as follow: Section 2 reviews the main arguments enhanced by literature as determinants of technology adoption. Section 3 describes the empirical strategy and data used to carry out this study. Section 4 presents the main results, and section 5 concludes. 2. Conceptual framework in adoption of new technology There is an extensive literature aimed at explaining the process of technology adoption, and this can basically be classified in two dimensions. Firstly, literature enhances a series of institutional factors on the grounds of imperfections in credit, information, input and output markets as well as inadequate incentives associated with farm tenure arrangements (Feder et al., 1985). Secondly, differences in rates of adoption might be the result of heterogeneity at farmer level, which conditions the individual behavior.
6
Most previous studies are limited by geographic coverage and the level of data aggregation.
4
As regards the first set of factors, a key element relates to the reduced credit opportunities that farmers face. Lack of access to credit in rural areas arises as a consequence of asymmetric information and enforcement problems in the lender-borrower relation. This is particularly augmented among small farmers due to the lack of sufficient collateral7 (Udry, 1994). Thus, when innovations require significant levels of investment, credit constraints may turn out to be a limiting factor in adoption decisions. In addition, as the diffusion of innovation is scarce and fails to reach potential users, farmers may face trouble in obtaining the maximum benefit from shifting to modern technologies. The optimal use of new production methods entails realizing adaptations to the local conditions unknown at the time the shift is observed, which implies a costly learning process (Foster and Rosenzweig, 1995). Imperfect knowledge about the usage of new technology constitutes a barrier that delays adoption processes. Furthermore, it is expected that new technologies are more likely to be adopted in rural zones with better support infrastructure and localized near main commercial centers, in which the availability of both complementary inputs and maintenance services are more abundant (Sunding and Zilberman, 2001). Finally, tenurial arrangements that characterize the landlord-tenant relationship and land rent contracts can affect adoption decisions. For instance, small farmers without land of their own and facing short-term land contracts are less probable to adopt innovations since they may be unable to enjoy the long-term benefits of doing so (Bahduri, 1973). In relation to household characteristics, variables that control for differences in farm size, off-farm opportunities, human capital and gender play an important role to capture the intra-household dynamic in explaining adoption levels (Doss, 2006). A key factor less studied in technology adoption literature is the role of risk in investing in modern production methods. Rural households must deal with high income-fluctuations as exposed to a variety of uncontrollable factors such as climate conditions, economic fluctuations, policy uncertainty and individual-specific shock8 (Bardan
7
Consequently, most of the financing sources in agriculture are provided by informal lending channel at higher interest rates. First, climatic risk may trigger harvest failure as a result of occurrences of unexpected events from the nature such as drought, flood or frost. Second, the inherent volatility of agricultural market explained mainly by fluctuations in demand and supply generates another source of income variation via price uncertainty. Third, given the fact that farmers take most decision in advance far before final product is sold in the market, prices are not known at the time choices regarding production process are made. Fourth, the recurrent need that faces governments of changing the direction of the agricultural policy produces a further risk component that enhances the uncertainty in investment decisions. Lastly, individual shock may arise, for instance, when households are hit by an illness or unexpected death of one of its members whose contribution to the family’s budget is vital. 8
5
and Udry, 1999; Dercon, 2002). In this context, decision making under uncertainty is characterized by risk because some possible outcomes have an undesired effect. In this work, we focus on production uncertainty that comes from the fact that quantity of output is not known with certainty due to unsystematic factors such as weather affecting the production process. If we assume farmers are risk averse, they would adjust their behavior implementing actions aimed at smoothing income and/or consumption (Dercon, 2002). However, when credit and insurance markets are absent or incomplete, poor farmers characterized by limited landholdings, few assets and lower schooling, face serious constraints to smooth consumption9. Consequently, it is expected that farmers depend on ex-ante management strategies to smooth income, which would imply preferences for traditional technology with lower expected returns. That is true for new farming methods, that due to higher exigencies of knowledge in its use, adoption would arise the likelihood of failure in spite of higher expected benefits. However, irrigation technology is recognized for being a risk-decreasing input when reducing dependence on rainfall and water availability, which make yields more stable. Thus, we expect that probability of adopting irrigation will be greater among farmers facing higher production risk. In addition, agro-ecological factors that capture environmental heterogeneity are essentials to attain a better understanding of the process of technology adoption. In particular, modern irrigation methods have a higher relative advantage in zones with lower land quality. Thus, if we assume lower land quality implies lower water-holding capacity, gains of the transition from traditional to modern technologies will be greater in locations with relatively poorer soil quality because modern irrigation increase irrigation effectiveness, and in consequence, less water is needed to satisfy the nutritional requirements of crops10 (Caswell and Zilberman, 1986). In brief, the rate of increase in profit from shifting to modern technologies varies in accordance with environmental characteristics, as selection of irrigation technology influences the impacts of quality of environmental inputs on profitability (Dinar and Zilberman, 1991).
9
Alternatively, poor farmers get involve in informal risk–coping mechanisms based on agreements made by members of a same group or community to support each other in the case of shocks. In addition, they also copy with risk by generating incomes from off-farm activities. 10 Irrigation effectiveness is measured by the ratio between consumed water and applied water.
6
3. Empirical strategy and data Firstly, we discuss the estimation procedure for computing production risk measures along with two approaches proposed to estimate the adoption process. Secondly, we describe the data used in this study as well as the criteria taken into account when selecting the sample, given limitations of the data and the particular interest in family agriculture.
3.1 Estimation procedure In order to obtain estimates of risk attitudes, two approaches have been proposed to compute plausible risk measures: the econometric approach, which uses household production data and information on prices; and the experimental framework, which elicits preferences from choices made in hypothetical settings. This work adopts an econometric approach to estimate the risk structure as suggested by Antle (1983). The author proposes to compute the moments to approximate the distribution of outcomes, stating that optimal input choices depend on both the marginal effect of inputs on outputs and risk attitudes. The procedure to analyze the effect of production risk on adoption decisions follows two steps. In the first stage, we compute risk measures for each famer by estimating the first three moments of a stochastic production function11. The moments of the production distribution are estimated by following a sequential procedure in which, firstly, production is regressed against a set of inputs. The model is specified as follows: (1)
yi f ( xi , zi , hi ; ) i
where i=1...N denotes individual farmers, yi is the logarithm of output measured in kilos, xi is a vector of conventional inputs including land, labor, capital and irrigation. Land and irrigation are expressed in logarithm of hectares12, labor is the logarithm of the sum of both family and hired labor, and capital is the
11
Due to the lack of data on prices, we proxy production risk by the moments of the production function rather than the moments of the profit function as Koundouri et al. (2006) do. This assumption lies in the linear relationship between the moments of the profit and production function valid under price-taker individuals. 12 1 hectares=2.5 acres.
7
logarithm of the value of physical assets13. Zi is a vector of farmers’ characteristics including age, education14 and agricultural dependence15.
Additionally, we include a set of variables (hi) to capture
variations in soil quality, rainfall and regional differences. Soil quality is measured by the percentage of nonand slightly eroded soil in each locality16 and rainfall17 was constructed as the historical mean of precipitations. Unobserved regional differences are picked up by including dummy variables. These may capture differences in soil and climatic conditions not observed in the data, transport and processing infrastructure, marketing and access to input and output markets. Region VIII serves as a benchmark. i is the identically independently distributed error term, representing uncertainty in the production process18. Under expected profit maximization, explanatory variables are assumed to be exogenous, and thereby, Ordinary least squares (OLS) of (1) produces consistent and efficient estimates of the parameter vector β (Koundouri et al., 2006) . The s central moment of production conditioned on inputs about its mean defines as:
(2)
s (.) E y(.) 1
s
where 1 denotes the mean or first moment of production. Thereby, the estimated errors
ˆi yi f ( xi , zi , hi ; ) from (1) are estimates of the first moment of production distribution. In order to
13
The capital variable was built using information with respect to ownership of mechanized draught capital. These were weighted considering market prices. 14 It takes the value of 0 if the farmer has no formal education, 1 if s/he has partially completed elementary school, 2 if s/he has completed elementary education, 3 if s/he has partially completed high school, 4 if s/he graduated from high school, 5 if s/he has partially completed a technical program, 6 if s/he has completed a technical program, 7 if s/he has partially completed her/his university education and 8 if s/he graduated from college/university 15 It takes the value 3 if the agricultural income represents 75% or more of household income, 2 when the proportion is between 50% and 75%, 1 if it is between 25% and 50% and 0 if this percentage is less than 25% 16 Information of erosion was used as a proxy for land quality. It was extracted from a recent study conducted by the Center of information in natural resources (CIREN, 2010), addressed to determining the current and potential erosion of soils in Chile. The methodology for determining the level of erosion integrate a set of soil, topographic, climatic and biological characteristics. Thus, erosion will be more severe to the extent that soils are more porous and sandier, fields are more sloped and hold less vegetation as well as in those locations where precipitations are more “aggressive”. 17 The historical averages were calculated using agro-climatic information provided by the Chilean institute of meteorological information (CIMI, 2007) between the agricultural seasons 1999-2000 and 2005-2006. Climate measures per locality were obtained by matching localities with the nearest meteorological stations. Unfortunately, information for both the south-central zone and central zone are not available in the agro-climatic yearbooks with same level of detail as the rest of the zones. For these cases, we used information available in climatic yearbooks collected from meteorological stations situated at airports. 18 Literature recognizes other sources of uncertainty that are evenly important: price uncertainty, that relates to the volatility of agricultural market; technological uncertainty, associated with evolution in production techniques that may make past investment obsolete; and policy uncertainty, given by government interventions and the need for changing agricultural policy in response to new and conflicting objectives. However, we abstract from these types of uncertainty and focus on production uncertainty (Moschini and Hennessy, 2001)
8
compute estimates of the second moment, the estimated errors ˆ are squared and regressed on the same set of inputs as in (1):
(3)
ˆi2 g ( xi , zi , hi ; ) i
OLS provides consistent estimates of the parameter i and the predicted errors are consistent estimates of the second moment of production distribution (Antle, 1983). The third moment is computed following the same procedure, but now the estimated errors are raised to the power of three and regressed against the explanatory variables included in (1). In a second stage, the estimated moments are incorporated as explanatory variables in discrete choice model along with other control variables in order to analyze how production risk affects the adoption of modern irrigation. Control variables include a dummy denoting if the farmer is male; farmers’ age measured in years; farmers’ level of education; a categorical variable that captures the degree of dependence on agricultural activity; a dummy variable indicating if the farmer lives in the plot; the farm’s size measured in total hectares; secure tenure measured by the ratio between the sum of family-own hectares and rental land over total hectares; a dummy variable indicating if the farmer had access to at least one of following alternatives of credit during the last two years: loans from INDAP (National Institute of Agricultural Development)19, loan from the state bank, loan from private banks or credit line from either input providers or agro-industry; a dummy variable indicating participation in any agricultural organization; a dummy variable denoting if the farmer received extension services during the last two years; the percentage of non-and slightly eroded soil; and a set of dummy variables capturing regional differences. For purposes of estimation, previous works have relied on a sample of irrigators, distinguishing between adopter of new technology and non-adopters (traditional irrigation). The modeling of these decisions have been done by fitting a probit model to study the shift from traditional to modern technology. However, we
19
The National Institute of Agricultural Development is a public and decentralized organism with indefinite duration, created to combat the poverty and increase competitiveness among small farmers thought actions aimed at strengthening human, physical and financial capital.
9
argue that this model may fail to describe the true technology adoption process, especially in settings in which famers still depend on rainfall for irrigation. In this context, we propose two alternative models to tackle this problem. Firstly, we model irrigation decisions as a categorical variable such that each category corresponds to a superior level of irrigation technology. For these purposes, we estimate an ordered probit model, which fits better settings in which the response variable is ordered. This approach would represent better the shift underlying the use of water input, having as benchmark full reliance on rainfall.
Secondly, we estimate a binary response model with sample selection (Van de Ven et al, 1981), arguing that whether or not we observe modern irrigation may depend on a previous individual’s irrigation decision. Thus, when considering only irrigation data, we might lose track of some people who are eligible to adopt modern irrigation. If this characteristic is systematic, a probit estimation may lead to inconsistent estimates (Wooldridge, 2010). This fact might even be more problematic in contexts where the ratio of non-irrigated land is sizeable and the probability of irrigation follows a systematic tendency mainly explained by location factors. The latter requires defining a set of exogenous variables that identify the selection in irrigation. Defining Y as the irrigation decision observed with the value of 1 if the farmer irrigates and 0 otherwise, the selection equation can be expressed as follows:
(4)
Pr Yi 1 Pr Yi* 0 Pr w' i 0 Pr i w'2
where Yi * corresponds to a latent variable that depends on a set of variables w2, included to allow for identification of parameter . Basically, we argue that irrigation decisions respond to the availability of water source in the plot and moments of the rainfall distribution. Thus, we use historical data to compute the mean, standard deviation and skewness of rainfall. It is expected that the probability of becoming an irrigator is greater in those zones with lower levels of rainfall, larger variance and skewness. Furthermore, availability of water source in the plot should turn out a necessary pre-condition to adopt irrigation. We account for
10
water source by defining a dummy variable that takes the value of 1 if the farm holds at least one of the following sources for irrigation: well, spring, river, stream, reservoir, lake, lagoon or another.
3.2 Data description This study uses data from the VII National Agriculture and Forestry Census (INE, 2007), designed as a system of continuous statistics whose objective is to collect information on the state of agricultural and forestry resources in Chile. The farm level data is combined with location level data for soil and climatic characteristics. The physical variables defined by locality are assumed to proxy for farm characteristics. In particular, we are interested in explaining the low levels of adoption observed among family farmers. Family agricultural production represents one third of agricultural Gross Domestic Product (GDP) and contributes with 1.2% to total GDP. In addition, this segment controls 85% of the farms in the country, and generates 60,000 direct and indirect jobs, benefiting 1.2 million of people. Furthermore, family agricultural producers are one of the main food suppliers in the country, providing around 60% of the food consumed in domestic markets (INDAP, 2011; Campos, 2002). Family farmers are characterized by being mainly involved in traditional farming activities such as annual crops, horticulture and extensive cattle farming. They mainly employ the family members to meet the permanent demand for labor, and only hire workers to meet excess demand in both sowing and harvest seasons. Moreover, family farmers hold small plots localized in areas with reduced agricultural potential. They operate at low levels of working capital and face credit constraints, which evidently restrict the possibilities to reach levels of production consistent with the generation of surplus. Even though agricultural production is the main source of income, family farmers also obtain resources from alternative sources, including handicraft activities and remittances sent from relatives living in urban areas (Echeñique, 2006). Aware of the complexity of defining precisely a representative sample of family farmers that matches with the prior definition, we employ the current criterion based on size, used by the Ministry of Agriculture. Accordingly, a family farmer is defined as holding less or 12 hectares of basic irrigation (HBI). The latter 11
requires transforming information on irrigated and dry land into HBI by using coefficients of conversion that captures differences in agricultural potential across zones. For theses purposes, and following FAO (2009), we utilize the coefficients defined in the law 16.640, enacted in 1967 under the agrarian reform of Chile. We selected a sample of potato family producers to examine technology adoption patterns. The choice of this crop lies in its economic and social importance for agricultural income-dependent farmers. This crop is the third more relevant annual crop in terms of surface and the second most important in terms of production value. Moreover, potato farming is important from a social perspective when it is labor intensive. Furthermore, potato producers extend across the whole territory, and its distribution can be organized in four well-differentiated agro-climatic zones: north-central zone, comprising Regions IV and V; central zone, including Regions VI, VI and XIII; south-central zone, containing Regions VIII and IX; and south zone, consisting of Regions X and XIV. The later allows us to pick up agricultural potential with more precision. Finally, potato farming is highly sensitive to humidity changes. Lush foliage and difficulties to satisfy water needs from soil humidity because of the surface roots claim for the maintenance of an appropriate hydric status. Thereby, the provision of irrigation water as a complement to rainfall becomes essential to avoid diminishments in yield and quality. Given lack of precision in the data collection about inputs per crop, we need to define a criterion to ensure association between use of inputs and production. Our sample consists of 7,274 potato producers, mostly non-irrigated farms, whose 100% of the land is allocated to this crop. This clearly reflects that irrigation technology is in an initial stage of the adoption process and that still might respond to individual and location factors. Appendix 1 shows the number of farmers and hectares that fall into each category of irrigation. Not surprising, the adoption of modern irrigation is extremely low. Only 70 farmers report the usage of at least one of the improved technologies. The figures show that 17% of farmers hold some system of irrigation, which amount to 34.4% of the farming land. In addition, modern irrigation is adopted by 5.6% of irrigators, comprising 13.7% of total irrigated hectares. As regards traditional irrigation, farmers using furrow technology amounts to 70.7% of the total; it figures reaches 61.5% as measured by irrigated hectares. On the
12
other hand, sprinkler technology seems to be the system of modern irrigation more broadly used among potato producers since three out of four irrigators adopt this system. Table 1 depicts summary statistics for inputs, environmental and location variables per irrigation category. Table1: Mean and standard deviation for inputs, environmental and location variables per irrigation category. Modern Irrigation Traditional Irrigation Non-Irrigators Whole sample Mean Std Dev Mean Std Dev Mean Std Dev Mean Std Dev 114,114.30 351,117.10 18,721.30 41,402.10 9,033.70 48,555.60 11,739.50 61,987.70 Production (kilos) Yield (kilos/hectares) 13,636.14 10,693.16 11,791.30 8,734.11 11,227.95 8,531.67 11,341.29 8,591.53 24.93 36.70 4.44 12.73 15.81 25.25 14.08 24.21 Total surface (hectares) Cultivated land 4.76 9.94 1.47 3.09 0.66 1.89 0.83 2.54 4.71 9.65 1.55 2.04 0 0 0.28 1.43 Irrigated surface (hectares) 6.14 7.30 4.18 6.76 3.46 4.57 3.59 5.01 Total labor (N°) 4.24 7.52 3.11 6.76 1.68 4.40 1.93 4.93 Hired labor Permament 0.64 1.61 0.29 0.73 0.06 0.34 0.10 0.46 Temporal 3.60 7.01 3.11 6.76 1.62 4.33 1.83 4.80 1.80 1.66 0.91 1.08 1.69 1.35 1.56 1.34 Family labor Permanent 0.81 0.87 0.42 0.65 0.65 0.82 0.61 0.80 Temporal 0.99 1.55 0.49 0.88 1.04 1.21 0.95 1.19 6,862,143 8,778,120 3,394,974 7,209,798 2,408,747 4,253,057 2,608,270 4,932,854 Capital (Chilean pesos20) 1120.2 537.1 681.3 333.5 1445.9 280.6 1319.5 405.9 Rainfall (milimeters) Standard deviation 208.7 100.8 189.9 90.7 220.3 30.4 215.3 48.1 Skeewness -0.27 0.858 -0.043 0.818 0.033 0.794 0.018 0.79 43.0% 0.26 43.0% 0.23 44.0% 0.24 44.0% 0.24 Soild quality (ratio) Location variables North-central zone 17.0% 22.0% 0.0% 3.0% 17.0% Central zone 3.0% 56.0% 2.0% 11.0% 3.0% South-central zone 63.0% 22.0% 38.0% 36.0% 63.0% South zone 17.0% 0.0% 59.0% 49.0% 17.0% 70 1,153 6,051 7,274 Observations Source: Own elaboration based on Censual data, 2007. Category
As seen in table1, production yield increases insofar irrigation become more sophisticated. The same is true for the cultivated area, saying that farmer irrigating are larger on average. Higher sunk costs in irrigation gives rise economies to scale, feature that makes the adoption more profitable to the extent that it is extensively used in the plot. As commonly known, labor is mainly temporal and respond to the seasonal nature of agricultural activity. It is also observed that hired labor becomes more important at higher levels of irrigation technology. Because irrigation raises productivity and cultivated area, which require more labor, family members are no longer sufficient to meet an increase in production, and as a result, extra labor has to be hired. The value of mechanized capital is also greater with irrigation, as it presumes purchases of equipment and complementary inputs for its implementation. Apparently, there seems not to have sufficient 20
1 US$ corresponds to approximately 500 Chilean pesos.
13
evidence to state differences in technology adoption based on variation in soil conditions. Furthermore, irrigators mostly reside in zones characterized by fewer precipitations. In particular, whereas modern irrigators predominate in South-central zone, farmers using traditional irrigation mostly reside in Central zone. Moreover, non-irrigators are mainly located in Southern localities, which show the highest levels of precipitations. Finally, modern irrigators tend to concentrate in localities that historically have experienced the maximum level of rainfall. The latter seems to show that adoption of traditional irrigation may respond more strongly to rainfall availability than modern irrigation. Table 2 presents information on farmers’ characteristics and institutional aspects per irrigation category. Table 2: Farmers’ and institutional characteristics per irrigation category. Category
Modern Irrigation Traditional Irrigation N° 55 54
Mean 54.44 79.0% 77.0%
N°
Non-Irrigators N°
Mean 56.36 70.0% 79.0%
Whole simple
N° Mean 56.13 Age 876 4,236 5,165 71.0% Male=1 657 4,780 5,528 76.0% Reside in Plot =1 Education No education 2 2.9% 67 5.80% 163 2.70% 233 3.20% Incomplete elementary 29 41.4% 548 47.50% 3,685 60.90% 4,263 58.60% Complete elementary 8 11.4% 209 18.10% 938 15.50% 1,157 15.90% Incomplete secondary 9 12.9% 90 7.80% 411 6.80% 516 7.10% Complete secondary 6 8.6% 122 10.60% 381 6.30% 509 7% Incomple technical 0.0% 6 0.50% 24 0.40% 29 0.40% Complete technical 11 15.7% 55 4.80% 248 4.10% 313 4.30% Incomplete college 1 1.4% 12 1% 48 0.80% 58 0.80% Complete college 4 5.7% 44 3.80% 151 2.50% 196 2.70% Dependence >75 25 35.7% 205 17.80% 1,355 22.40% 1,586 21.80% >50 and 25 and
