Family Background, Education and Earnings in Kenya*

This paper uses data collected in 2000 to first estimate OLS and ordered probit models to measure correlation between family background and workers ’ education. Then, human capital earnings functions are estimated to examine to what extent family background accounts for observed correlation between workers ’ education and earnings. Subsequently, it estimates returns to education with education treated as endogenous. Having well-educated parents is associated with greater educational attainment and earnings. Returns to education decline slightly when parents ’ education controls are in the earnings function. Instrumental variable estimation or self-selection correction suggests that estimates of returns to education may be larger than conventional estimates suggest