Constraint or choice? Disentangling fertility determinants by switching regressions

In 1953, many poor countries had not yet approached the demographic transition. Accordingly, income generally had a positive impact on fertility in poor countries, while it has a negative impact today. Easterlin's supply-demand framework offers an explanation for this nonlinearity by attributing the positive relationship to Malthusian (or "supply") factors and the negative relationship to "demand" factors. This paper estimates Easterlin's supply-demand framework by switching regressions in a panel data set of 152 countries from 1953 to 1998. The technique allows the identification of several factors affecting the Malthusian constraint and the demand for children, such as income, source of income, urbanization, religion and medical environment. It is found that a combination of higher GDP per capita, a decrease in infant death rate and an increase in education explain a substantial part of the reversal of the relationship between income and net fertility over the sample period.