Sources of Geographic Variation in Health Care: Evidence from Patient Migration
By Amy Finkelstein, Matthew Gentzkow, and Heidi Williams
NBER Working Paper 20789, National Bureau of Economic Research, December 2014
We study the drivers of geographic variation in US health care utilization, using an empirical strategy that exploits migration of Medicare patients to separate the role of demand and supply factors. Our approach allows us to account for demand differences driven by both observable and unobservable patient characteristics. We find that 40-50 percent of geographic variation in utilization is attributable to patient demand, with the remainder due to place-specific supply factors. Demand variation does not appear to result from differences in past experiences, and is explained to a significant degree by differences in patient health.
From the Introduction
Health care utilization varies widely across the United States. Adjusting for regional differences in age, sex, and race, health care spending for the average Medicare enrollee in Miami, FL was $14,423 in 2010, but just $7,819 for the average enrollee in Minneapolis, MN. The average enrollee in McAllen, TX spent $13,648, compared to $8,714 in nearby and demographically similar El Paso, TX.
In this paper, we exploit patient migration to separate variation due to patient characteristics such as health or preferences from variation due to place-specific variables such as doctors’ incentives and beliefs, endowments of physical capital, and hospital market structure. As a shorthand, we refer to the former as “demand” factors and the latter as “supply” factors.
Like past decompositions, ours is not sufficient to draw strong conclusions about the efficiency of observed geographic variation. Though it may be tempting to see supply-driven heterogeneity as evidence of waste, such variation could reflect different allocations of physical or human capital, and so be consistent with efficiency. Conversely, demand-driven heterogeneity could reflect patient misinformation, and so contribute to inefficiency. We view our findings as both a first step toward a more welfare-relevant understanding and a clarification of an influential body of existing evidence.
We find robust evidence that 40 to 50 percent of geographic variation in the log of health care utilization is due to fixed characteristics of patients that they carry with them when they move. Our examination of mechanisms suggests that a large part of this demand-side heterogeneity may be due to patient health. The remaining 50 to 60 percent of variation is due to place-specific factors, possibly including doctor practice patterns and characteristics of health care organizations.
These results suggest that demand-side factors play a larger role in geographic variation than conventional wisdom might suggest. This does not translate immediately into conclusions about efficiency. The correlation of utilization with demand-side factors (and with patient health in particular) may reflect differences in the marginal impact of treatment or the marginal utility from a given impact, and so be consistent with efficiency. But it could also reflect differences in other demand drivers, such as patient information or beliefs. A more careful examination of the efficiency implications of the geographic variation is an important direction for further work.
Our findings have implications beyond our patient-place decomposition. The fact that habit formation seems limited implies that demand-side differences in utilization are unlikely to change quickly in response to policy, at least among the 65 and over population, a population that accounts for about a third of total annual health care spending. The fact that a large part of demand-side geographic variation reflects variation in patient health may also point to limits to the effectiveness of demand-side policies aimed at changing patient beliefs or preferences. At the same time, the sharp adjustment we observe around moves suggests policies that affect the supply-side can have immediate impacts.
While we have taken a first step toward understanding the origins of the patient component we measure, it remains for future work to better understand the mechanisms behind the place component. Particularly interesting questions concern the role of physicians’ training and practice patterns, and the role of health care organizations.
By Don McCanne, MD
Although this article is technically challenging to read and absorb, and the authors caution that more work is needed to understand better the implications of their initial results, nevertheless, this is a very important article because it improves our understanding of geographic variations in health care utilization as related to “patient characteristics such as health or preferences” (demand-side) and to “place-specific variables such as doctors’ incentives and beliefs, endowments of physical capital, and hospital market structure” (supply-side). Because of concerns over global health care spending, many important demand-side and supply-side policies are being put in place without an adequate regard of the policy science behind those decisions, both known and unknown.
On the demand side, it appears that the health of the patient is far more important in patient decisions on utilization than are patients’ beliefs, preferences and habits. This is as it should be. The health care system should be there to serve the health care needs of the patient. Yet the leading efforts to control demand-side over-utilization include measures that impair access, particularly financial barriers such as high deductibles and tiered layers of coinsurance, and provider barriers established through the use of narrow- and ultra-narrow networks. Policies that prevent patients from getting the care they need should be rejected. The claim that these policies control excess patient demand is based on the fiction that patients are demanding excessive care; they are not.
On the supply side, much has been written about excess capacity that results in excessive utilization, including the alleged excessive supply of high-tech specialized services. The authors of this study state that we need to learn more about the role of physicians’ training and practice patterns, and the role of health care organizations. Yet the leading efforts to control supply-side over-utilization include the use of accountable care organizations, in spite of the lack of evidence of their effectiveness, and the use of dubious pay-for-performance levers to guide physicians’ practice patterns. Many contend that about one-third of care provides marginal benefits that could be done away with, since the the overall negative impact would be small (but not zero). But it is very difficult to prospectively select the care that could be omitted. That is what the accountable care organizations are supposed to be doing, yet, to date, there has been very little reduction in spending under these programs, and often that reduction is erased by the performance rewards.
We already know of very effective policies that could control excesses on the supply side. Our prices are created on the supply side, and they are too high. They could be brought into line through public policies that would ensure adequate funding of services and products while reducing the waste of excessive pricing. The complexity of the supply-side administrative services wastes tremendous resources. That waste could be dramatically reduced through an administratively simplified single payer financing infrastructure. Excess capacity that invites over-utilization could be reduced through central planning and budgeting of capital improvements – a step that would also address our even greater need to improve capacity in regions wherein services are inadequate.
So let’s get to the basics. Demand side? Remove barriers to care. Supply side? Begin by replacing our dysfunctional health care financing system with a single payer national health program. Once we have those in place, then we can sit around and have intellectual discussions about physicians’ training and practice patterns, and the role of health care organizations. But we can’t waste our time with decades of esoteric policy studies when we have important work to do right now.