Guest Post by Arpit Gupta: Dartmouth Atlas vs. Teacher Salaries

by Arpit Gupta

The Dartmouth Atlas of Health Care, which tracks the variation in Medicare’s medical expenditures across different regions of the United States, has had a significant impact on the debate over health reform. Peter Orszag’s latest column at Bloomberg View offers a Dartmouth Atlas approach to thinking about health costs:

The evidence thus suggests that in both Medicare and in private-insurance markets, higher costs are not associated with better quality. That underscores the opportunity identified by the Institute of Medicine: By reducing the very high costs that are not generating better quality, the U.S. could reduce total spending without diminishing the quality of care people receive.

This view informs efforts to cut the rate of growth of programs like Medicare — if the geographical variation in health spending doesn’t seem to deliver additional health rewards, we should be able to painlessly cut large amounts from health care and distribute the savings towards expanded insurance coverage.

But is this necessarily the right way to understand geographical variation in health spending? Economists typically start with prices as given, and try to figure out the underlying reasons why prices differ. Health economists like Orszag, driven by the Dartmouth Atlas, seem to go in reverse. Their underlying assumption is that the only legitimate sources of health cost variation must be those which deliver better health outcomes. Another hidden assumption is that different geographical regions have more or less similar underlying demographics and health propensities, as well as comparable supply-side cost drivers.  

Here’s a graph from the Dartmouth Atlas that shows where Orszag is coming from — health costs tend to vary enormously across geographical areas:

 

Surely it’s not a coincidence that areas that Dartmouth identifies as high-cost areas have high poverty, are urban, or are otherwise located in the South (an area with sizably higher obesity rates and lower life expectancies than the rest of the country). Personal health differences in terms of genetics, diet, and exercise drive enormous variations in health status and health spending around the country. Different areas of the country are also more expensive to operate businesses in than others — particularly high-cost urban centers. 

As a comparison, here’s a map of teacher salaries (referenced by Matt Yglesias):

 

The correlation between the two graphs isn’t perfect, but there is some meaningful overlap. Urban areas tend to both experience higher costs for Medicare services, as well as higher teacher wage costs. Teacher wages also seem to be high in California (collective bargaining?), the Pacific Northwest, and the upper Midwest (higher local wages?).  

If you compare educational wages to outcomes, I suspect you’ll find little correlation. New Jersey spends something like three times the amount per pupil than Utah, but does not get three times better outcomes. Many high-wage urban school districts get poor educational results. Applying the Dartmouth logic — since geographical variation in teacher wages doesn’t deliver better test scores, there should be lots of easy waste and fraud in the system we can cut. Rahm Emanuel shouldn’t stop where he is now — he should aim to cut Chicago teacher salaries in half because plenty of other well-performing school districts get the job done with far lower teacher salaries.  

That logic seems wrong. Undoubtedly there’s room to reform both education and health care, but using geographical cost variation as the sole barometer of waste in the system is potentially misleading. Costs for many services vary dramatically in the U.S. for a variety of reasons. Health costs in particular will vary because of supply-side differences in the cost of providing medical services, as well as the varied demographic and health status mix of populations served by local hospital. Studies that more carefully control for these factors — I discussed some here — find that additional health spending often does tends to improve health.

The libertarian response is to allow for market entry and competition to reduce waste and inefficiency, while accepting that health spending will vary across regions. Center-left technocrats offer a differ prescription for health care — opt for a more complicated program of health cuts and subsidies. But if the structure of health costs are not amenable to those types of simple approaches, cutting Medicare may prove more difficult than Democrats and Republicans have led us to think. There may be important tradeoffs in how to balance fiscal and health concerns.