Healthy people are routinely ignored when analyzing curative health inputs. This practice overlooks people's long-term ability to affect their chances of falling sick, and may have perverse effects on welfare analyses. A dynamic model implies that input demand estimates conditioned on current illness can only be interpreted as short-run effects, in contrast to the long-run nature of unconditional estimates. In addition, conditional estimates may be biased from both sample-selection, and self-reporting of health status. By jointly modeling discrete choices for health inputs and health outcomes, a test for selection bias is derived using the multinomial probit. In data from Cote d'Ivoire, it is found that the usual short-run demand estimates do not suffer from selection bias. However, these conditional estimates differ from the easily estimated long-run unconditional effects, which are often the more relevant policy parameters.
Dow, William, Unconditional Demand for Curative Health Inputs: Does Selection on Health Status Matter in the Long Run?. Santa Monica, CA: RAND Corporation, 1995. https://www.rand.org/pubs/drafts/DRU1234.html.
Dow, William, Unconditional Demand for Curative Health Inputs: Does Selection on Health Status Matter in the Long Run?, Santa Monica, Calif.: RAND Corporation, DRU-1234-RC, 1995. As of May 11, 2022: https://www.rand.org/pubs/drafts/DRU1234.html