Exports & HIV in Africa

Emily Oster finds a downside to exports in Africa – transportation and trade spread epidemics (pdf):

I find a large and significant positive relationship between HIV and exports: a doubling of exports appears to lead to as much as a quadrupling in number of new HIV infections, which suggests that if exports overall had been 25% lower in Africa over the course of the epidemic only about half as many people would have become infected. In addition, this relationship seems to explain at least some of the very large decline in HIV prevalence in Uganda in the 1990s, which is typically attributed to the ABC (Abstain, Be Faithful and use Condoms) campaign, a widely replicated anti-HIV education effort. A decline in the coffee market accounts for between 30% and 60% of the decline in HIV incidence, suggesting the success of the ABC campaign may be overstated…

A central concern with interpretation is the possibility that is not the export-transit mechanism which drives the result but, rather, some omitted variable (for example, GDP) which drives both exports and HIV. Although this is potentially consistent with the primary results, I argue that there is evidence in favor of a causal interpretation of exports. First, the relationship between HIV and exports is stronger in areas with a greater density of roads and areas closer to ma jor cities, which is consistent with the transit mechanism. Second, instrumenting for export volume with world commodity prices also points to a positive and significant relationship. Third, the relationship between exports and HIV is stronger in countries where the major export is more closely linked to trucking. Finally, new HIV infections can be linked directly to truck imports, which strongly points to an effect of transit.

Of course, it’s not international trade per se that increases the spread of HIV. If economic activity within national borders spurred increased use of truckers to transport goods, that would result in the same effect. And the benefits of economic growth may outweigh the costs of higher HIV incidence. Nonetheless, this isn’t good news.

[HT: MR]