Do the Poor Overweigh Low Probability Events?

The literature in Development Economics documents that households in developing countries are vulnerable to considerable risk and shocks. The types of risk and shocks households face range from covariate shocks such as draught, flooding, conflict, and inflation to idiosyncratic ones, e.g., loss of job, sickness, and death of a family member.  Most households however lack the access to formal financial markets to insulate their welfare against these shocks. Instead, households engage in a variety of informal risk mitigation (ex-ante) and shocks coping (ex-post) strategies. An emerging body of literature in Development Economics is interested in investigating the long-term implications of the different types of risk management strategies. If households overestimate low-probability negative events they would engage in sub-optimal risk management strategies, which could perpetuate risk-induced poverty trap.  In this project, we implement carefully designed field experiments in urban Tanzania to investigate if poor households do overweigh low-probability events. We also investigate time preference behaviors of the same subjects.     

The motivation is two-fold: First, we want to understand whether poor households indeed overweigh low-probability events. As documented in previous studies, if households in developing countries indeed overweigh low-probability events, the adverse implications on long-term welfare would be substantial. A farmer, who has a large negative expectation about rainfall, would be less likely to apply productivity enhancing modern agricultural inputs. A household who overestimates the likelihood of a conflict would likely choose to keep its wealth in the form of passive assets (e.g., Jewelries) instead of investing it on productive assets. As a result, households would be forced to remain in poverty trap, an equilibrium level of poverty from which one would not be able to come out without an external intervention. Our work therefore would have significant implications for understanding the behaviors of the poor and poverty traps.

Second, we would like to contribute to the general debate in economics on whether it is the Expected Utility (EU) or the Prospect Theory (PT) that describes the risk taking behavior of poor households in Africa. A number of previous studies in both developed and developing countries used the EU theory to describe risk taking behavior of subjects. In 2010, in a study by Tanaka, Camerer and Nguyen, which we refer as TCN from now on suggest an experimental setup which demonstrates that risk preferences of poor households can be better captured using the PT framework, constituting three relevant parameters: risk aversion, loss aversion, and nonlinear probability weighting. An important advantage of the TCN approach is its simplicity and its flexibility in enabling one to test whether EU theory or PT theory explains risk preferences of the subjects. Our project therefore provides one of the most comprehensive evidences on risk preferences of individuals in developing countries. Our project also provides one of the most comprehensive measures of time preference of households in developing countries.   

Theme 1: Individual behaviour, cooperation and trust.

Project status
Project | 20 March 2015