Risk preferences and economic behavior: Experimental investigations in the field
Empirical Social Research
Final Report Abstract
The poor populations of developing countries are extremely exposed to the vagaries of chance. Agricultural output is dependent on highly variable weather patterns, formal jobs are scarce, and the future is highly uncertain. Institutional mechanisms to shelter people from these risks are rare, and even where they exist, seldom reliable. Lives can be brief and brutal. This stands in stark contrast to the situation in Western countries, where even modest risks such as bicycle theft are often over-insured, and where many countries provide generous safety nets. Against this backdrop, it is of prime importance to better understand the behaviour of the poor, and the extent to which it may contribute to achieving economic success. At the same time - and for exactly the same reasons - the high-risk environment of developing countries constitutes an ideal context in which to investigate the determinants of risk and time preferences. Against this background we conducted a series of experiments in the field in Karnataka state, India, as well as some lab experiments in Europe to develop new measurement tasks and theory to provide foundations for the field work. The experiments were aimed at tackling issues of noise in measurement and decisions, rationality in decision making, and the interaction between decisions under risk and decisions concerning intertemporal tradeoffs. We develop a novel measurement task of time discounting in a risky context that allows for the non-parametric identification of the discount function under discounted expected utility. We show the ease of employment of the task by using it to compare discounting for money and the consumption of tea in the field. We further expanded on those insights by building a behaviourally accurate model for decision making under risk and over time, in which discounting elicited under risk is corrected for distortions deriving from nonlinear probability weighting and from differences in the utility curvature for risk and for time (difference between the parameter of constant relative risk aversion and the inverse of the coefficient of inter-temporal substitution). Another part of the research project focused on how measurement error can attenuate correlation and even systematically distort measured preferences in the presence of some respondents who respond randomly to choices between lotteries: I revisit recent evidence uncovering a ‘preference for certainty’ in violation of dominant normative and descriptive theories of decision-making under risk. I show that the empirical findings are potentially confounded by systematic noise. I then develop choice lists that allow to disentangle these different explanations. Experimental results obtained with these lists reject explanations based on a ‘preference for certainty' in favor of explanations based on random choice. From a theoretical point of view, the levels of risk aversion detected in the choice list involving certainty can be accounted for by prospect theory through reference dependence activated by salient outcomes. Several data sets based on the project remain to be analyzed.
Publications
- (2018). Certainty preference, random choice, and loss aversion. A comment on ‘Violence and Risk Preference: Experimental Evidence from Afghanistan'. American Economic Review
F.M. Vieider
(See online at https://doi.org/10.1257/aer.20160789)
