Do the HIV/AIDS Models Reproduce the Level and Shape of Mortality Being Experienced in Sub-Saharan Countries?

Rob Dorrington, University of Cape Town

Most models of the demographic impact of HIV/AIDS in most sub-Saharan countries rely on data drawn from an often unrepresentative sample of antenatal clinics to indicate the extent and trend of the epidemic in these countries. Where countries have undertaken household prevalence surveys these have often been interpreted to be contradicting the estimates from models based solely on antenatal clinic data. By and large macro modelers have not inspected the output from their models to see if they are consistent with the level of mortality being estimated by demographers for these countries. This paper examines the performance of a few models in this regard and finds that for a number of countries the international agencies exaggerate the impact on mortality but in others the reverse is the case, and concludes that the performance of these models could be improved if they were calibrated to be consistent with the observed mortality.