Regional Variation in Risk and Time Preferences: Evidence from a Large-scale Field Experiment in Rural Uganda

Abstract

Experiments measuring risk and time preferences in developing countries have tended to have relatively small samples and geographically concentrated sampling. This large-scale field experiment uses a Holt–Laury mechanism to elicit the preferences of 1,289 randomly selected subjects from 94 villages covering six out of seven agro-climatic zones across rural Uganda. As in previous studies, we find evidence of risk aversion and loss aversion among most subjects. In addition, we find significant heterogeneity in risk attitudes across agro-climatic zones. Especially, the farmers in the agro-climatically least favourable zone, the uni-modal rainfall zone, are the most risk-averse, loss-averse and impatient. We also find significant relationships between risk attitudes and village-level predictors such as the distance to town and the road conditions. After controlling for the village-level factors, we find that the level of schooling still positively correlates with the individual's level of loss tolerance and patience. The main results are not altered by allowing for probability weighting in subjects' choices. Overall the results provide clear evidence that within one country there may be significant regional variations in risk and time attitudes. We conjecture that the agro-climatic conditions that affect farmers' livelihoods may also affect their risk and time preferences, and village-level development in infrastructure could improve the household perception of investment-related policies.