A Note on the Effect of Religiosity on Fertility

The existence of such peer effects implies that nonreligious people in more religious societies may have more children than nonreligious people in less religious societies, whereas religious people in less religious societies may have fewer children than religious people in more religious societies.

The most plausible explanation for this is that the doctrine of the Catholic Church is pronatalist, whereas in Protestant religions, fertility is generally considered a matter of individual choice (e.g., Lehrer 1996).

In the absence of panel cointegration, conventional panel regressions involving nonstationary variables are spurious, often producing statistics that suggest significant relationships, when in fact none exist (e.g., Kao 1999).

The panel DOLS estimator corrects for endogeneity and serial correlation by including lead, lag, and current values of the differenced regressors in the regression.

Alternatively, the magnitude of the estimated effect can be evaluated by multiplying the DOLS coefficient on REL_it by the average change in the church attendance rate and dividing it by the average change in the birth rate. The resulting value is 0.222, implying that declining religiosity has been responsible for about 22.2 % of the fertility decline between 1925 and 2000.

The online appendix presents a sensitivity analysis demonstrating that the positive (average) religiosity-fertility coefficient is robust to the use of alternative estimation techniques, the inclusion of additional variables, the use of an alternative measure of fertility, and to splitting the sample period into two equal periods (1925–1960 and 1965–2000).

A precise definition of the subsamples is given in the online appendix.

This group is relatively heterogeneous, consisting of two Eastern Orthodox countries (Bulgaria and Cyprus), one Anglican country (Great Britain), and two countries whose majority population is a mix of Catholics, Protestants, and Anglicans (Australia and New Zealand). Therefore, the effect of religiosity on fertility may well differ across these subgroups. Unfortunately, the number of countries in these subgroups is too small to further subdivide the group of non-Catholic, non-Protestant Christian countries. Note that the Anglican Communion considers itself to be both Catholic and Protestant. Following the classification of the World Religion Dataset (available at http://​www.​correlatesofwar.​org/​data-sets/​world-religion-data), I therefore do not classify Anglicans as Protestants, as some studies have done, but I distinguish between Anglicans and Protestants.