Estimation of a simultaneous equation model of American fertility determination from state economic areas data
Although demographers, sociologists and economists have made great efforts in fertility research in recent years, they still differ in their opinions about the important determinants of fertility and the magnitude of their effect. In fertility study, demographers and sociologists have usually used simple correlation or tabular and graphical presentations, and almost all economists have employed single equation regression models. The single equation model is not appropriate for fertility analysis since the explanatory variables and fertility are usually interdependent and hence a simultaneous equation model should be used in fertility analysis.
This study fits a simultaneous equation model of fertility to the l96o and l970 state economic areas data of the United States by the two-stage least-squares method. The model consists of three equations: the fertility rate equation, the income equation and the female labor force participation rate equation. The fertility determinants are hypothesized to be family income, female labor force participation rate, education, the unemployment rate, race, and community of residence. In the income equation, the explanatory variables are female labor force participation rate, the level of education, the unemployment rate, the percentage of nonwhite population, the percentage of urban population, and the proportion of total population aged 15 to 64, and the explanatory variables in the female labor force participation rate equation are the fertility rate, the percentage of nonwhite population, the percentage of urban population, and the age structure of female. In the simultaneous equation model, the endogeneous variables are the fertility rate, the family income, and the female labor force participation rate, and the exogeneous variables are education, the unemployment rate, the percentage of nonwhite population, the percentage of urban population, the proportion of total population aged 15 to 64, and the age structure of female.
The two-stage least-squares estimates of the fertility equation indicate that at the .05 level, none of the regression coefficients in l970 are significant, while in l960 all regression coefficients of the explanatory variables except the level of education and the percentage of urban population are significant. The signs of all fertility determinants are as expected. The order of importance of fertility determinants are assessed by "beta" coefficients. The most important determinant of fertility rate is the female labor force participation rate in l960 and the family income in l970. From the findings in this study, one may infer that the best policy to slow the fertility rate in the United States is to attract women into the labor force or to raise parents' educational level.
Thesis74.C483.pdf
2.54 MB
Adobe PDF
d517e572694d16f6f2b505967efda20d