The Fertility Effect of Education: Regression Discontinuity for Counts and Exponential Models∗

Negative relations between women’s schooling and fertility have been observed by social scientists, leading some of them to propose education as a policy tool for family planning. Both fertility and schooling are, however, choice variables that may be affected by some common factors. Thiswould thenmake schooling endogenous, and consequently, the observed negative relationshipsmight not be causal. We take advantage of a compulsory education law change to estimate the impact of female education on fertility, using regression discontinuity estimators motivated by exponential regression models appropriate for counts or non-negative response variables. Our data utilizes the entire 1990 Population Census of Taiwan. We find that married women’s schooling has no impact on their number of children and the age at first-birth.