In the Foreign Affairs November/December 2002 essay, we used demographic-epidemiological modeling techniques to imagine the repercussions of a variety of HIV trajectories for Russia, India, and China on health, mortality and economic output for the societies in question.

We used the SPECTRUM software package, developed by the Futures Group International for USAID, for all demographic-epidemiological projections. It is a very user-friendly package, and was selected for this reason.

SPECTRUM requires the programmer to posit a trajectory for the HIV pandemic. The trajectories we traced out were arbitrary, but, we argued in the Foreign Affairs essay, not unreasonable. Those paths were a "high", "middle" and "low" variant for each country--we called them "severe", "intermediate" and "mild". They do not in fact specify the only possible alternatives for these three Eurasian countries, but are instead a starting point for discussion. Other observers can use the same program to derive their own versions for their own assumptions if they wish, and can justify the paths chosen as they see fit.

With respect to the particular trajectories chosen, we then specified our model with particular assumptions about the dynamics of the epidemic. The most important assumptions were: 1) that the HIV epidemic in each country began around 1985; 2) that in each country the median time between infection with HIV and onset of AIDS was 9 years; 3 )that life expectancy with AIDS for the afflicted population was 2 years; and 4) that the distribution of HIV comported with the 'heterosexual' age-sex patterns seen already in low-income countries, especially those of the sub-Sahara. Our 'baseline' projections for demographic trends for all three countries were taken from US Census Bureau projections for Russia, China and India for the 2000-2025 period; none of those Census Bureau projections hypothesize any HIV impact on mortality or health, so we simply added those presumed effects. Again: all of these assumptions are arbitrary, but not unreasonable; others can alter these if they wish in their own recomputations if they so wish.

After modeling the demographic impact of HIV in these three countries under the assumptions described above, we then modeled the hypothetical economic impact of the pandemic through a "health based productivity model." This extremely simple model assumed that there is a direct correspondence between the life expectancy in a society and its economic potential. We used World Bank data on PPP-adjusted GNI (GNP) per capita and life expectancy to model the presumed international relationship between health and productivity. On the basis of the coefficients thus derived, and the Census Bureau estimates of the 'baseline' size of the population of 'potential working age' [the 15-64 group], we then simply computed the predicted output per person 15-64 years of age under our separate scenarios, and the GNP or GNI [the product of the output per person 15-64 and the number of people projected in the 15-64 group].

Readers may of course contest any of our assumptions, and may wish to produce their own alternative calculations. We offer these numbers in the spirit of humility, and with confidence that our method is transparent and, while necessarily arbitrary, not entirely unreasonable.