Family background and returns to schooling in Taiwan

Estimated returns to schooling in Taiwan are sensitive to the inclusion of family background variables in the wage function. The effects of family background on worker's wage are significant in the private sector but not in the public sector, suggesting that wages are more sensitive to unobserved ability or family connections in the private sector than in the public sector. The estimated wage function is convex; returns to schooling increase with the level of education. The effect of father's schooling is larger than the effect of mother's schooling in the wage function; however, the effect of wife's schooling is even larger. These results suggest that estimates obtained without controlling for family background may be biased upward, reflecting unobservable worker characteristics and assortative mating behavior. Analysis of the potential role of measurement error bias suggests that inclusion of family background variables may overstate by one third the extent to which omission of family background variables affects the returns to schooling. [JEL J12, J24] (C) 1999 Elsevier Science Ltd. All rights reserved.