A double-hurdle count model for completed fertility data from the developing world

I report on a study of the socioeconomic determinants of completed fertility in Mexico. An innovative Poisson double-hurdle count model is developed for the analysis. This methodological approach allows low- and high-order parities to be determined by two different data-generating mechanisms and explicitly accounts for potential endogenous switching between regimes. Unobserved heterogeneity is properly controlled. Special attention is given to study how socioeconomic characteristics such as religion and ethnic group affect the likelihood of transition from low- to high-order parities. Findings indicate that education and Catholicism are associated with reductions in the likelihood of transition from parities lower than four to high-order parities. However, being an indigenous language speaker increases the odds of a large family.