Introduction: the period TFR and its flaws

Demographers and policy makers rely heavily on the TFR for monitoring trends in fertility, and they do so for good reasons. First, the data requirements for the TFR are limited and it is easy to compute. Second, the TFR is a standardized measure of fertility and –unlike the CBR– independent of variability in the age-structure of populations. Third, the TFR is a contemporaneous measure of fertility and –unlike the CFS (or cohort TFR)– can be computed for the year just passed. The TFR has one important flaw, however, and that is its sensitivity to changes in the age pattern or timing of fertility. Such tempo changes are often contrasted with changes in the quantum of fertility, which refer to the fertility level irrespective of timing. The best known measure of (cohort) quantum is the CFS.

The age pattern of fertility can change for various reasons. One of the most extreme examples is the delay in (family formation and) childbearing in Europe and the US during WWII. Usually, perturbations in the age pattern of childbearing are more subtle and nowadays facilitated by effective contraceptive methods. John Hajnal was among the first to realize that period fertility measures may no longer be indicative of completed family size when couples have the means to carefully time the arrival of their children: “Under such circumstances a change in the rate at which people are having children in a given year can no longer be taken as an indication of a change in the number of children they will bear altogether in the course of their reproductive lives”. (Hajnal 1947:143 )

This quote illustrates that demographers have realized for quite a while that tempo changes can distort period demographic measures. The issue garnered renewed interest in the late 1990s because some observers doubted that the extremely low fertility rates registered in some European and Asian populations were genuine. Instead, they suspected that these period estimates of the TFR were biased (downward) by changes in the timing of fertility of successive birth cohorts, and set out to estimate the magnitude of that bias.

In this lecture, we discuss the limitations of age-based period measures of fertility, and discuss one method for correction tempo distortions in the period TFR in greater detail. In an epilogue, we extend the discussion about tempo distortions to the life expectancy in a synthetic cohort.