Does Pension Coverage Predict Fertility? International Evidence From OLS, Fixed Effects and Machine Learning Random Forests

Author: Anika Ewald (Leibniz University Hannover) • Patrick A. Puhani (Leibniz University Hannover)
Posted: 13 November 2026

Abstract

We complement existing literature by estimating determinants of the total fertility rate for 183 economies between 2000 and 2023. Methodologically, we utilise pooled OLS and economy/country fixed-effects regressions and complement them with exploratory random forests, which take into account non-linearities and interaction effects. Variable importance measures suggest that after two indicators for life expectancy, GDP per capita, and schooling, pension coverage is among the more important predictors of fertility. Our results suggest that acknowledging the costs of childbearing and childrearing in pension systems may be a method to address declining fertility.
JEL codes: C53; H55; J13
Keywords: demography; birth rate; retirement; pay as you go