Preferences and skills : four studies into unobserved human nature and its implications

• Main Author: Sandewall, Nils Orjan
• Published: London School of Economics and Political Science (University of London) 2009
• Subjects: 330.01
• Online Access: http://ethos.bl.uk/OrderDetails.do?uin=uk.bl.ethos.645889

This thesis contains four studies of generally unobservable skills and preferences of relevance to economic behaviour. The first chapter examines the validity of the assumption of equal latent ability among monozygotic twins. An influential literature has employed the schooling decisions of twins to estimate returns to schooling. Using a unique dataset including IQ test scores, income and two measures of schooling for 1780 monozygotic twins, within-pair differences in measured IQ are found to be a significant predictor of both income and schooling differences. In the second chapter, the effect of general cognitive ability on proposer and responder behaviour in the ultimatum game is examined, using a large and representative sample of 895 individuals. No effects are found on proposer behaviour, and only small, but statistically significant, effects are found on responder behaviour. In the third chapter, a sample of almost 30,000 mono- and dizygotic twins is used to study the heritability of financial risk-taking. Investment decisions made by virtually all Swedish adults regarding mandatory pension savings are taken as a field experiment to infer risk preferences. Standard techniques from behaviour genetics are used to partition variation in risk-taking into environmental and genetic components. The results suggest that genetic variation is an important source of individual heterogeneity in financial risk-taking and that the frequently reported parent-child associations in attitudes toward risk are, at least in part, genetically mediated. In the fourth and final chapter, the robustness of recent results indicating a large role for genes in determining variation in the propensity for self-employment is examined empirically using a novel dataset, and a large but imprecisely estimated difference between women and men is found.