My coauthors and I posted a substantially revised draft of “Market Size and Trade in Medical Services” last week. As I wrote about the first draft:

There’s a long-running discussion in health policy about “spatial mismatch”: are doctors and clinics too concentrated in big cities? Our paper emphasizes that you need to quantify the trade-off between local increasing returns and trade costs to answer this question. If the division of labor is limited by the extent of the market, there’s an upside to geographically concentrating production. The downside depends on how difficult it is for patients to travel.

New to the June 2026 edition are some demographics-based instrumental variables that provide evidence of regional increasing returns in the production of medical services. For example:

Table 3… regresses quality at the region-by-procedure-group level on the interaction of the region’s elderly share and the procedure group’s elderly demand share. This regression includes region and procedure group fixed effects, so a positive coefficient on the interaction means that regions with a higher elderly share produce disproportionately higher quality for care that the elderly disproportionately demand… For supply shifters to explain these results, factors used to produce elderly-oriented care would have to be relatively cheaper in elderly-abundant regions. We think this is unlikely. Doctors who specialize in care disproportionately consumed by the elderly are not likely to have systematically different preferences for regional amenities from other doctors.

We also present a case study of mesothelioma care:

These regressions using demand variation caused by past asbestos exposure suggest scale economies similar in magnitude to our aggregate estimates. Because mesothelioma death rates are orthogonal to aggregate service quality when controlling for log population and state fixed effects, a peculiar alignment of supply shifters would be required to explain these estimates without scale economies. Namely, the doctors who treat mesothelioma would need to find regions with past asbestos exposure particularly attractive, conditional on state fixed effects and total population.

We’ve also overhauled our counterfactual scenarios to make dollar-for-dollar comparisons of travel subsidies and output subsidies. Check it out.