Causes of regional variation in health-care use
In Germany, health-care use varies significantly by geographical region; for example, the rates of coronary bypass surgery vary by a factor of more than eight between districts (Kumar and Schoenstein, 2013). Wide regional variation in health-care use is also well documented for other countries (Skinner, 2011). There are several potential causes of regional variation in health-care use. On the one hand, regional differences can be explained by the supply side of medical care which includes e.g. differing practice styles and preferences among health providers but also differences in regional supplier density. On the other hand, regional variation can also be explained by differences in the demand for medical care; which includes for example differences in health and health-care needs between regions but also regional differences of patients’ preferences for medical care use. These alternative explanations lead to very different policy conclusions. Differences in the supply of medical care can point to the inefficient allocation of resources, where oversupply in some areas can coexist with undersupply in others. In contrast, differences in the demand for medical care use may still reflect the efficient allocation of resources. This project aims to study the drivers of geographic variation in health care utilization in Germany.
Data
In this project we exploit patient migration in order to disentangle the effects of supply factors from those of demand factors. We follow patients as they move from one region to another and examine how their medical care use changes in the years after the move. By following movers over time, we can control for both observed and unobserved patient characteristics. In recent years there has been substantial migration between regions in Germany, e.g. from the former East Germany to West Germany and from rural regions to urban centres. This approach makes it possible to uncover what shares of regional differences in medical care use can be attributed to supply- and demand-side factors. Our empirical approach is similar to that used, for example, by Finkelstein, Gentzkow and Williams (2015) for the Medicare population in the US.
Methods
Our analysis is based on administrative data from a group of health insurers (Betriebskrankenkassen). The data contains information on around seven million individuals, including around 300,000 movers, over 2006–2012. It also includes detailed information on medical care use and health-care expenditure.
Results
We find that 10 percent of geographic variation in utilization is attributable to place-specific supply factors, with the remainder due to demand side factors. Demand variation is explained to a significant degree by differences in patient health.