Returns to education in Russia: Where there is risky sexual behaviour there is also an instrument

Finding an instrument that is orthogonal to the disturbance term in the wage equation has been a topic of great deal of debate. Recently, Chesson et al. (2006) proposed that higher discount rates are significantly associated with a range of sexual behaviours, including having sex before age of 16 years. Following their paper, we use unprotected sexual behaviour and earlier age of sexual intercourse as alternative instrumental variables to account for endogeneity in schooling. The Bound et al. (1995) F-test indicates that our instruments are strongly correlated with schooling. We fail to reject the test of over-identifying restrictions, which demonstrates that the proposed instruments are valid and not correlated with the current earnings. In line with previous studies, our results suggest that the IV estimates of the returns to schooling are higher than the OLS estimates. In addition to the conditional mean, the proposed instruments are applied over the wage distribution using Chernozhukov and Hansen (2008) instrumental variable quantile approach.