Analysis of sex disparities in under-five mortality rates in Ghana: Insights from vector autoregressive modeling

Abstract

International monitoring organizations call for child mortality indicators to be disaggregated by gender. However, there remains a paucity of studies, especially, from the sub-Saharan region aimed at producing accurate forecasts of child mortality indicators with their sex variations. This study aims at investigating disparities in indicators of childhood mortality rates by sex in Ghana by employing vector autoregressive (VAR) model to analyze jointly annual recorded data on total, male and female under-five mortality rates (TU5MR, MU5MR, FU5MR, respectively). The results show gradual declining under-five mortality trends among sexes in both the historical and forecasted rates. The trivariate traditional and instantaneous Granger causality analyses found that any of the mortality indicators Granger causes the other two combinations, except TU5MR to MU5MR and FU5MR. The forecast error variance decomposition analyses revealed that FU5MR was the most exogenous variable while long-term impulse response function analyses indicated that unit shocks in FU5MR significantly increased TU5MR. The VAR(2) model forecast constructed revealed that contrary to recent predictions based on wider interval data derived from demographic health surveys, Ghana may meet the SDG 3.2.2 if ongoing efforts are sustained and that focusing policies and interventions on reducing FU5MR would largely contribute to reducing TU5MR in Ghana.