Is the fertility decline a consequence of the growth of the welfare state? Evidence from historical data

3. Some empirical observations
The decline of fertility over the very long time horizon is illustrated in Figure 1. The selected graphs (Figure 2 and Figure 3) and the computed pair-wise correlations (Table 1) may in turn help providing some flavour of the nature of the basic results. Thus we can observe that fertility is in every case strongly and negatively related to the level of income, education and the size of the government. The ‘only significant’ positive correlation is between infant mortality and fertility; higher infant mortality rate may force parents to have more children and thereby guarantee the availability of the old-age support4.
One can also see that the longer life expectancy (at least seemingly) diminishes fertility. But one has to be careful with this type of an influence. If this relationship is scrutinized so that the effect is made conditional on income, income growth, economic structure, education, mortality and so on, it turns out that the conditional relationship becomes quite weak and the sign turns out to be positive5. The reason is obvious. Life expectancy mainly reflects technical change and the growth of income and welfare. Thus, when these variables are controlled, it may well be that increased life expectancy works in an ‘expected way’. The longer the period for which support is needed, the more investment in children takes place, and thus fertility increases along with an increase in longevity.
The negative relationship between the (various indicators of) welfare state and fertility seems to be robust for all variable combinations. The bigger the role of government in securing old-age income and support, the smaller is the private incentive to invest in children or physical or financial assets. The result is particularly clear in the case of income tax rate as an indicator of the size of government. But as one can see from Figure 3, the result is basically the same when pension expenditures or social security expenditures are used as indicators of the welfare state. Obviously, as pointed out by Ehrlich and Lui (1998), this has powerful implications for the private saving behaviour and thereby for the aggregate investment (in a global context or with Horioka-Feldstein puzzle in a single country setting). The relationship between pension/social security, wealth and saving has been analysed empirically since the 1970s with somewhat mixed results. The fertility behaviour has been considered much less and the combined effects on saving and fertility even less. Clearly, there is a lot of room and obvious reasons for further analysis6.

Table 1: Correlations with the fertility variable
Notes: in the historical data, the number of data points is 194, in the WDI data 2533, and in the MZES data 825. Due to very large number of observations, almost all coefficients are statistically significant at the 5& level of significance (Insignificant coefficients are marked with an asterix *).

4 See e.g. Atella and Rosati (2000) for further elaboration of this effect and supporting empirical evidence from India.
5 We have elaborated this conditional effect in Puhakka and Viren (2006).
6 See, however, Cigno and Rosati (2003) for German evidence and Gabos, Gal and Kezdi (2005) for Hungarian evidence.

Pages: 1 2 3 4 5

Tags: ,