Cross-Country Determinants of Declines in Infant Mortality: A Growth Regression Approach

This paper examines the determinants of infant mortality rates in a cross section of countries using a "growth regression" approach. The paper argues that infant mortality is a valid measure of living standards, worthy of study in its own right. It also shows that first order differential equation implicit in the standard growth regression is appropriate to studies of infant mortality rates, perhaps more so than those of GDP. From the perspective of identifying possible determinants of declines in infant mortality, these results are rather discouraging. Only two policy variables, primary school enrolments and DPT vaccination rates for infants, show any consistent correlation with declining infant mortality, and even those correlations are not robust to the inclusion of fixed effects, a simple way to pick up time invariant unobserved variables. From the perspective of the growth regression literature, a more interesting "non-result" is the fact that there is no evidence at all that the black market premium, the M2/GDP ratio, inflation, or the real exchange rate, all policy variables that typically explain economic growth, help to explain declining infant mortality, and only weak evidence that real GDP per capita itself is correlated with these declines. We have long known from microeconomic data that income is not a very good predictor of children's health status. These results confirm that in a growth regression context. They also suggest that the determinants of progress of nations in one welfare dimension, economic growth, are distinct from those in another, infant mortality.