This Working Paper presents the methodology that has been applied since 2020 in the model used to draw up the population outlook for Belgium. The methodology considers that fertility is explained by both structural (education, labour market participation, etc.) and cyclical factors (the economic cycle, etc.). Those factors are taken into account through three elements: (1) expert opinion on long-term fertility trends, (2) an explicit trend in the fertility schedule and (3) the effect of cyclical and structural determinants of fertility using an error-correction model.
The methods for projecting fertility are numerous and varied. They are either based on probabilistic or deterministic methods using explanatory models or based on time series extrapolation. Expert judgment can also be taken into account. Some models simply project births by multiplying the total number of women by an average number of children per woman, but most project fertility by age of mother to take into account the fertility schedule and the age distribution of women of childbearing age. Finally, a choice has to be made on the ways to measure fertility: some models will include the total offspring (longitudinal or cohort approach), while others use the total fertility rate (cross-sectional approach).
Going back to 2008, the year from which the population projection model for Belgium was developed at the Federal Planning Bureau, the fertility projection methodology for national population projections was poorly developed, mainly due to data limitations. It was essentially based on age-specific fertility rates held constant in the projection, assuming no change in future fertility behaviours.
Fertility trends in recent years (described in the third section of the paper), and in particular since 2009, highlight the necessity to improve the methodology used so far, mainly on two issues. First, the structural determinants of fertility, in particular the increases in the number of years spent in education and in women's participation in the labour market, which influence both the level and schedule of fertility, were not taken into account. Second, cyclical factors, especially the economic situation, which can influence the fertility schedule in the short and medium term, were also barely taken into account in the projection model.
This Working Paper therefore aims to present a new methodology that has been applied since 2020 in the model used to draw up the population outlook for Belgium published every year by the Federal Planning Bureau and Statbel. The methodology considers that fertility is not static and is explained by both structural (education, labour market participation, etc.) and cyclical factors (the economic cycle, etc.). These factors are taken into account through three elements: (1) expert opinion on long-term fertility trends, (2) an explicit trend in the fertility schedule using Schmertmann's model (2003) and (3) the effect of cyclical and structural determinants of fertility using an error-correction model.
By considering the judgment of different experts, the uncertainty surrounding the future evolution of fertility can – to some extent – be taken into account. The inclusion of structural elements can help to determine a relatively stable long-term fertility level. Integrating the cyclical elements of fertility makes it possible to adjust fertility development in the short and medium term depending on the economic environment, in particular the evolution of the unemployment rate. Finally, by integrating demographic parameters (in this case the modal age) into the modelling of the fertility schedule, transparent choices on the future development of this schedule can be made, in particular the continued increase in age at childbirth.
The population projection model developed by the Federal Planning Bureau is deterministic, based on the (demographic) component method. Deterministic projections may be criticized for their lack of information on uncertainty around population projection. One way to address this issue is transparency in assumptions: the results of deterministic population projections are directly linked to the selected scenario. The user of the projections must be aware of this scenario to interpret correctly the projection results. This Working Paper, which describes in detail the methodology for projecting fertility in the national population projection, contributes to this objective.