What do Twins Studies Reveal About the Economic Returns to Education? A Comparison of Australian and US Findings

Abstract

Conventional estimates of rates of return to education are constrained because they are unable to isolate the returns to from the contribution of individual ability and the influence of family background. Potentially, studies using a sample of twins may overcome these problems. Monozygotic twins are genetically identical and, if they have been reared together, share the same family background. Differences in income between identical twins can therefore be associated with differences in the amount of education they have undertaken, in order to estimate the independent influence of education. Comparisons of such estimates with estimates made from a sample of dizygotic twins, who are genetically similar but not identical, and who are also reared together, may permit evaluation of the biases associated with ability and family background in conventional estimates of the return to education. In an important paper, Jere Behrman et al. (1977) analyzed data from a sample of male World War II veteran twins and found that, of the overall rate of return to of 8 percent, 2.7 percentage points, could be attributed to per se, 3.2 percentage points could be attributed to genetic factors, and 2.1 percentage points could be attributed to shared family environment. In a recent paper, however, Orley Ashenfelter and Alan Krueger (1994) have estimated the economic returns to education using data on a new sample of twins which permitted them to adjust their estimate for omitted ability variables and measurement error. On this basis, they find that family and genetic effects make virtually no contribution to the returns to schooling. Measurement error, however, biases the estimated return to downward. The findings of Ashenfelter and Krueger (1994) are very different from previous work in this field, and there is now some confusion as to what analysis of twins data really reveals about the economic returns to schooling. Ashenfelter and Krueger's results are based on analysis of data for a sample of 298 individuals (149 pairs of identical twins) who attended a twins festival in 1991. They remark at the end of their paper Only additional data collection is likely to lead to better estimates of the returns to schooling (p. 1171).