Cross-sectional analyses of the relationship between contraceptive prevalence and the total fertility rate of developing countries show the expected strong negative correlation. However, this correlation is much weaker in Sub-Saharan Africa than in the developing world as a whole.
This paper aims to explain the unexpected weak effect of contraceptive use on fertility in sub-Sharan African countries by using different regression models to obtain unbiased effects.
Using DHS survey data from 40 developing countries, the analysis consists of three steps: 1) examine the conventional cross-sectional TFR-CPR relationship by region at the time of the latest available surveys, 2) remove known technical flaws in the comparisons of fertility and contraceptive prevalence and 3) analyze multiple observations of TFR and CPR per country using pooled OLS and fixed effect regressions.
The conventional cross-sectional analyses produce biased results in part because technical factors, in particular post-partum overlap, create a downward bias in the effect of prevalence on fertility in sub-Saharan Africa. In addition, and more importantly, the cross-sectional regression OLS parameters have a bias due to confounding country fixed effects. Technical adjustments and the use of fixed effect models remove these biases.
A rise in contraceptive prevalence among fecund women has the same average effect on fertility in sub-Saharan Africa as in other regions of the developing world.