From risk and time preferences to cultural models of causality: On the challenges and possibilities of field experiments, with examples from rural southwestern madagascar

Abstract

Risk preference and time preference experiments have been conducted in field conditions in many rural and urban populations throughout the world, to evaluate how non-Western people subjectively trade off the value of reward amount, probability, and delay. My thesis is that it is difficult to quantitatively evaluate risk and time preferences with precision and that choice experiments have dubious internal validity (it is unclear whether the experiments evaluate the same thing-preferences-across all subjects and studies) and external validity (it is unclear how they relate to actual behavior). Probability and elapsing days preceding a reward are not necessarily the meaningful components of choice under risk and intertemporal choice for many people. I argue that if the goal of such work is to understand the cultural differences in thought and action related to risk and time, it may be more productive to examine how people in different cultures understand causal relationships linking natural, social, and supernatural factors to successful and unsuccessful outcomes. I illustrate my thesis by describing 13 risk and time preference choice experiments that I conducted among Masikoro farmers, Mikea hunter-gatherers, and Vezo fishermen in southwestern Madagascar from 2003 to 2008, contrasted with published risk and time preference studies from throughout the world. I find inconsistency in subjects’ choices across experiments and time, inconsistency in the determinants of choice, and a poor relationship between risk and time preferences and amount of time spent foraging and fishing. Then, I discuss some preliminary ethnographic and experimental attempts to understand the people’s causal models for risky economic outcomes. Preliminary evidence suggests that southwestern Malagasy understand activity risk as being caused primarily by natural factors and personal risk by supernatural factors.