Risk-Aversion and Occupational Choices: Evidence from Matched Field Experiments and Survey Data in urban Ghana

This paper investigates the role of risk-aversion in explaining the allocation of urban workers between the formal and the informal sector of the Ghanaian economy. A theoretical model is proposed where every worker has an option between working in the free-entry informal sector and queuing for a formal job. Both alternatives are risky: in the informal sector the worker is exposed to the risk of unexpected income shocks, while if she chooses to search for a formal job she runs the risk of remaining unemployed. Conditional on identifying the riskier option by comparing these two sources of uncertainty, the model yields testable implications on the relationship between attitudes to risk and occupational choices. Using panel data on earnings matched with novel experimental data to elicit risk-aversion, I am able to test the theory in two steps. First, I estimate the effect of risk-aversion on sectoral allocation using a multinomial logit specification and I obtain that workers who are more riskaverse in the experimental setting are significantly more likely to queue for formal jobs and significantly less likely to work in the informal sector. Second, I estimate a comprehensive measure of income uncertainty and I find that working in the informal sector is indeed riskier than joining the queue of formal job-seekers. The implications of the model are therefore consistent with the empirical results and I conclude that attitudes to risk, like education and age, are a fundamental determinant of workers’ willingness to queue for formal occupations. As such, they should be taken into account when designing labour market policies in developing countries, especially if such policies affect the degree of income uncertainty (and ultimately of Risk) in different sectors of the economy. I am very grateful to my D.Phil supervisor, Dr Francis Teal, who made me part of the CSAE research project on the Ghanaian labour market and showed continued interest in my work on this paper. I would also like to thank Dr Abigail Barr, who designed the behavioural experiments which I conducted in the summer of 2007, Dr Justin Sandefur, Dr Glenn Harrison and Miss Sara Lenehan for all their advice. I am also grateful to the officers of the Ghana Statistical Office for their help with the data collection and to Mr. Camillo Falco for the useful discussions. Finally, I would like to thank all the survey respondents who participated in my experiments. D.Phil Candidate and Junior Research Officer at the Centre for the Study of African Economies, University of Oxford.