Estimating indicators of fertility timing from consecutive census data on children ever born
Kryštof Zeman, Wittgenstein Centre (IIASA, VID/ÖAW, WU) and Vienna Institute of Demography
This methodological paper develops the method of estimating indicators of fertility timing from time-static census data on children ever born. The idea is to compare consecutive census data and retrieve timing information from the difference in fertility levels recorded in them. We call this method and the derived indicators “intercensal”. The method was first introduced in the 1970s-1980s, but not used since that time. Our objective is to find method that can estimate fertility timing by highest educational attainment of mothers. Education of mothers has been argued to be the most important demographic dimension of fertility behaviour after age and gender. However, the timing of fertility by highest attained education level of mothers is usually not available from vital statistics or other direct demographic methods. Because the tempo of fertility is crucial for understanding the changes in fertility by education, we discuss several indirect methods of deriving it, and we further develop the intercensal method, which yields most stable results and is relative robust against the shortcomings addressed later on. Two variants of the method are presented, one looking on the fertility timing from the period perspective and another from the cohort perspective. We use census data from the Czech Republic, Austria, South Korea, and Brazil, always for three census data points (around 1991, 2001 and 2011). The regional variation helps us to evaluate the universal validity of the method in different socio-economic settings. The results show important shifts in the level and timing of fertility between the 1990s and the first decade of 21st century. We found broad differences in fertility level between educational groups, along with a strong postponement of births among better educated. Educational differences of fertility have even increased comparing the two intercensal periods, but also towards younger cohorts of women.
Presented in Session 58: Modelling fertility