Individual preferences and farm-level decisions: experimental evidence from Costa Rica

This thesis contributes to the understanding of individual preferences and farm-level decision making, in interaction with the institutions and government policies that can influence these decisions. Decisions on investments and input used, decisions on whether or not to support local agricultural cooperatives, decisions the demand of credit with or without insurance, and decisions of voluntary certification schemes that have an impact on the livelihood of small farmers in developing countries. I used standard econometric methods to analyze the survey and data collected from three lab-in-the-field experiments, and a one quasi-experiment all implemented with real subjects: coffee farmers in rural Costa Rica.

Chapter 1 starts with an overview of the main topics: individual preferences and  farm-level decisions. It introduces the concept of risk preferences, social preferences, formal market-based strategies and certification schemes, that can be adopted by farmers to manage risks. More specifically, it presents the research questions, methodology, empirical strategy the study area and explains how the data was collected for the subsequent main chapters of the thesis.

In Chapter 2, we evaluate a survey-based method to other methods that elicit farmers’ risk attitudes.  For both researchers and practitioners, surveys can be easier to implement than field experiments in developing countries. We first assess correlations between a context-free survey estimate of risk-taking and context-specific risk preferences. Then, we test whether survey data predicts risk-taking behavior in an incentivized experiment. Finally, we show how the different estimates of survey risk preferences relate to real-life farming choices in a population of coffee farmers in Costa Rica. Our results indicate that one should be careful when extrapolating risk attitudes across contexts. Context-neutral and context-specific survey questions elicit different risk preferences. While the context-free survey estimate of risk preferences predicts risk-taking behavior in a context-free risk experiment, and context-specific estimates are associated with risk-taking in the same agricultural real-life context , the context-free survey estimate of risk-taking is not associated with actual risk-taking behavior in the agricultural setting. Connecting these methods to farm practices, we find that higher willingness to take risk is associated with the implementation of agricultural practices that require more farm investment. In contrast, farmers who report less willingness to take risks are more likely to have higher expenditures on fertilizer use. Researchers interested in using risk preferences as inputs into the design of policy instruments should make sure that preferences are elicited in the specific context targeted by the potential policy instrument.

Chapter 3 tests the internal and external validity of the typology of a conditional cooperator classified by using a public goods game together with the strategy method. Individuals categorized as conditional cooperators adapt their behavior to the group to which they belong. In Costa Rica, coffee farmers are traditionally organized in agricultural cooperatives, a setting very similar to the scenario presented to an individual facing the strategy method in a public goods game: how much to cooperate, given what others do. Our results show that conditional cooperators believe they contribute to the public good by matching the contribution of others in the experiment. However, we find no evidence that those classified as conditional cooperators in the experiment also behave this way when it comes to bringing coffee to the local cooperative in real life. We show supporting evidence to conclude that the typology of a conditional cooperator is internally consistent, but do not find evidence that the typology of conditional cooperators is externally valid. Our paper is a contribution to the external validity of context-free experiments and helps in understanding cooperative behavior relevant to the sustainability of agricultural cooperatives in the developing world.

In Chapter 4, we examine the effect of farmers’ liability on demand for credit with and without insurance. We test predictions of a theoretical model in a lab in the field experiment with coffee farmers in Costa Rica. Farmers choose how much to invest in six different settings, described on the one hand by whether the loan is insured or not, and on the other by their liability. Our results show that the uptake of loans bundled with insurance is significantly higher than uptake of loans without insurance, both when farmers are liable for sure for their debt, and interestingly when there is uncertainty about their liability. When farmers are not liable for their debt, i.e. under limited liability, the uptake of credit is high irrespective of whether the loans are insured or not. Our results suggest that in order to increase the uptake of insurance as a strategy to increase private investment and reduce the vulnerability of farmers to weather events which may lead to serious crop failure, it is important that farmers are liable with at least some probability. In terms of policy design, our results show that the “principle” of limited liability does not have be abandoned altogether in order to generate an increase in the uptake of insured credit.

Chapter 5 evaluates the environmental impacts of organic coffee certification. Eco-certification of coffee, timber and other high-value agricultural commodities is increasingly widespread. In principle, it can improve commodity producers' environmental performance, even in countries where state regulation is weak. But eco-certification will have limited environmental benefits if, as one would expect, it disproportionately selects for producers already meeting certification standards. Rigorous evaluations of the environmental effects of eco-certification in developing countries that control for selection bias are virtually nonexistent. To help fill this gap, we use detailed farm-level data to analyze the environmental impacts of organic coffee certification in central Costa Rica. We use propensity score matching to control for selection bias. We find that organic certification improves coffee growers' environmental performance. It significantly reduces chemical input use and increases adoption of some environmentally friendly management practices.

To conclude, Chapter 6 presents a synthesis and the previous core chapters. Conclusions, lessons learned and policy recommendations of individual chapters are reconsidered. Subsequently, results are placed in a thesis-wide perspective and finally, general conclusions made.