The Effect of Urbanization on China’s Fertility

Abstract

The relationship between urbanization and fertility decline is known to be inverse in developed countries. However, the nature of this relationship in developing countries that already have relatively low fertilities is not well-understood. This study aims to illustrate how much urbanization contributed to China’s fertility decline between 1982 and 2008 and forecasts how much it can contribute to future reductions in fertility. The study examines changes in the total fertility rate (TFR) at both the national and provincial levels, given regional differences in the urbanization rate. The results show that changes in rural fertility behavior accounted for most of the decline in the national TFR between 1982 and 2008. This finding suggests that official birth control policies were instrumental in curbing China’s population growth. However, urbanization was responsible for about 22% of the decrease in TFR during this period, and its effect was especially important during the latter years (2001–2008). In most provinces, urbanization associated with a decline in provincial-level fertility. The forecasts indicate that urbanization will become the primary factor behind future declines in national fertility. Given the negative effect of urbanization on the TFR, it is possible to relax the one-child policy without having adverse implications for population growth.

Appendices

Appendix A: A Simplified Version of the Decomposition Equation

To derive a simplified version of decomposition equation, Eq. 1 requires three additional assumptions: (a) F _x is constant for all x, i.e., age specific fertility rates are constant in all ages; (b) k _x is constant for all x, meaning that the proportion of urban females at aged x in the total female population is constant; and (c) the sex composition of urbanites remains constant while urbanization rate (C _u ) increases such that C _u  = k _u . Under these assumptions, Eq. 1 can be re-written as

$$ {\text {TFR}}_{\text {asfr}} = F_{r} C_{r} + F_{u} C_{u} $$

(3)

and decomposing (3),

$$ \Updelta {\text {TFR}} = \left( {\overline{{F_{u} }} - \overline{{F_{r} }} } \right)\Updelta C_{u} + \overline{{C_{r} }} \Updelta F_{r} + \overline{{C_{u} }} \Updelta F_{u} $$

(4)

where F _r and F _u denote rural TFR and urban TFR, respectively; C _r and C _u denote the proportion of rural and urban population; and again we have C _r  + C _u  = 1. Table 4 shows that the difference in TFR between using Eqs. 1 and 3 is minimal (see the last column of Table 4), suggesting that it is not unreasonable to decompose TFR, rather than TFR_asfr, in the decomposition exercise and the forecasts of TFRs (see Appendix B).

Table 4 A simplified decomposition of TFRs

Appendix B: Forecast of TFRs in Urban and Rural Areas

To forecast future national fertility, we estimated a conventional time series model for the log-transformed rural and urban TFRs (F _r and F _u ), conditional upon that the TFRs are greater than 0 (e.g., Box et al. 2008) We used data from 1950 to 2008. The fitted models for F _u and F _r are given below (standard errors in parentheses):

$$ \ln F_{u,t} = 0.992\left( {0.0096} \right) \times \ln F_{u,t - 1} ,\;R^{2} = 0.904 $$

$$ \ln F_{r,t} = 0.995\left( {0.0065} \right) \times \ln F_{r,t - 1} ,R^{2} = 0.903 $$

Using these equations, it is straightforward to forecast TFRs in urban and rural area for the next 20 years (see Fig. 7). Figure 7 shows that rural TFRs in next 20 years are fairly stable at approximately 1.6, while urban TFRs are around 1.1.

Fig. 7

Forecast of TFRs in urban and rural areas. Sources: The 1950–1977 data came from 1/1,000 fertility survey conducted in 1982; the 1978–1987 data were from the 1988 2/1000 fertility and contraceptive use survey; the 1988–1992 data came from the 1992 National Fertility Survey; the 1993–2000 data were obtained from the 2001 National Fertility and Reproductive Health Survey; and finally the 2001–2008 data came from the Annual Population Monitoring Surveys