Annual Forum Papers

The Macro Implications of HIV/AIDS in South Africa

  • Year: 2000
  • Author(s): Channing Arndt; Jeff Lewis

In this paper, we report on the preliminary results from an analysis of the macro impact of HIV/AIDS in South Africa. We have constructed an economywide simulation model that embodies the important structural features of the South African economy, into which we have added major impact channels of the HIV/AIDS epidemic. Using available demographic estimates for the impact of the epidemic (on labor supply, death rates, and HIV prevalence) along with assumptions about behavioral and policy responses (household and government spending on health, slower productivity growth), we use the model to generate and compare two scenarios: a hypothetical "no-AIDS" scenario in which the economy continues to perform as it has over the last several years, and an "AIDS" scenario in which the key AIDS-related factors affect economic performance. Focusing on the differential between the "no-AIDS" and "AIDS" scenarios, we find that the impact of the epidemic could be substantial. Over the 1997-2010 simulation period, GDP growth rates in the two scenarios diverge steadily, reaching a maximum differential of 2.6 percentage points. The result is a GDP level in 2010 that is 17 percent lower in the "AIDS" scenario; an alternative measure of "non-health, non-food absorption" is 22 percent lower by 2010. While some of this decline is due to the lower population associated with the "AIDS" scenario, per capita GDP does drop by around 8 percent. In fact, our simulations suggest that, despite the fact that AIDS impacts the high-unemployment unskilled labor category more than others, the net effect of higher AIDS-related mortality and slower growth is to leave the unemployment rate largely unchanged.

We also use the model to "decompose" the overall decline in growth performance into the contribution of the various channels. Given our current assumptions, the largest share (close to half) of the deterioration in growth is attributable to the shift in government current spending towards health expenses (which increases the budget deficit and reduces total investment), while an additional third stems from slower growth in total factor productivity (TFP). The decomposition illustrates the importance of considering the slow moving nature and hence long duration of the epidemic. If the epidemic imposes a drag on the rate of accumulation of knowledge (reduced TFP growth) or the rate of accumulation of capital (through a switch from savings to current expenditure), these effects become amplified over time. Over the course of a decade, the implications for macroeconomic performance are substantial.

Looking forward, our analysis suggests several avenues for further investigation. First, the parameters used in specifying the various AIDS effects are based on fairly limited empirical evidence, and it will be important where feasible to supplement these with additional data. For example, we have limited the impact of AIDS on household expenditure patterns to an assumed increase in health service spending, but there may well be other shifts that will occur and that could be incorporated, based on survey results. Second, there are important dynamic effects that are not yet included in the model: for example, lower private and government spending on education (because of higher AIDS spending) will slow down skills accumulation and change labor force growth rates. Third, consideration must be given to how to capture the impact of alternative "intervention" policies - for example, at present there is no feedback between possible government policies to slow the spread of AIDS, and the demographic (and subsequent economic) trajectory of the epidemic.

Finally, interactions between the epidemic and alternative growth and development strategies should be examined. We find that interactions with key economic features, such as the unemployment rate, do not necessarily conform to the results that one might expect from a casual analysis. And, key policy decisions, such as financing for AIDS related government expenditures, are shown to be very important. These results suggest that, while the human crisis appears to be practically unavoidable, appropriate economic policy measures have the potential to significantly palliate the negative economic effects of the epidemic. For the policy-making process, the slow moving nature of the epidemic needs to be borne firmly in mind. The AIDS crisis does not require the snap policy decisions of, for example, the Asia financial crisis. Instead, deliberate speed, careful planning, and competent execution by government and other actors could substantially ameliorate the economic aspects of the AIDS crisis.