Abstract
After a swift decline during the 1990s following the collapse of the Soviet Union, period fertility rates have either been stagnating or increasing in all countries of Central Asia. In this paper, I investigate the role of data artifacts, population composition effect, economic context and shifting tempo effect in explaining fertility changes in Central Asia. The analysis is primarily based on comparison of fertility data from the vital registration system with estimates from other data sources. The results show that the recent changes to be real and not a result of data artifact. The most plausible explanations are to be found in the three other non-exclusive factors (population composition effect, economic context and shifting tempo effect) that contributed jointly and simultaneously to push up the period fertility rates in the region.
ACKNOWLEDGEMENT
The author wants to specially acknowledge Evgeny Andreev for kindly making available population data by age, sex and ethnicity from the 1989 Census of the Soviet Union and his unpublished materials on fertility by ethnicity from the same census.
Notes
1. It is to be noted that while emigration contributed to the largest changes in the ethnic population composition in Central Asia, other factors such as re-identification might also have played a role.
2. I use the expression reduction of childbearing postponement and not ‘recuperation process’ as in order to determine if the births that were postponed by women were recuperated later at older ages requires cohort complete fertility data that are currently not available for Central Asia.
3. While the Bongaarts and Feeney's tempo-adjustment has been applied almost exclusively on data by single year of age, the data for Central Asia are available by five-year age groups only, introducing a small approximation in the procedure.
4. For both countries, only incomplete time series of the number of births by birth order and age of the mother are available. For Kazakhstan, this information is missing for 1994, and 1999 to 2002; for Kyrgyzstan, the years 1994, 1999 and 2000 are missing. In order to fill these gaps, the age-specific fertility rates for each birth order were linearly interpolated. In addition, for Kazakhstan, given the problems in the data quality of the age distributions for birth order 2, 3 and 4+ in 1992 and 1993, birth rates for these years were also estimated through linear interpolation. As the application of the Bongaarts-Feeney tempo-adjustment is applied here to reveal broad trends and that adjusted values are averaged over three years, the interpolation made would not change the main conclusion of the analysis.