Abstract/Summary

We obtain rich measurements of risk preferences for 2939 subjects across 30 countries, and use the data to paint a picture of the distribution of risk preferences across the globe using structural equation models. Reference dependence and likelihood-dependence are found to be important everywhere. Model parameters in non-Western countries differ systematically from those in Western countries, with poorer countries substantially more risk tolerant than rich countries on average. We qualify previous findings on gender effects and cognitive ability by showing how they mainly impact likelihood-dependence. We further add novel evidence on the correlation between risk preferences and study major. Whereas we confirm previous results on observable characteristics of subjects explaining little of overall preference heterogeneity, a few macroeconomic indicators can explain a considerable part of the between-country heterogeneity.