• How unusual is Medicare’s geographic variation?

    CBO’s 2008 Geographic Variation in Health Care Spending is a fantastic, if ever so slightly dated, review of the health care spending geographic variation literature. The text itself covers all the important angles, and Table 1 lists many relevant papers. All in all, it’s a great way to get up to speed on this topic.

    The report also includes an interesting comparison of geographic variation of Medicare spending to that of other goods and services. I turned the key results from the table on page 7 (box 1) into a graph, below. The “coefficient of variation” (vertical axis) is the standard deviation divided by the mean. It’s a measure of how much variation there is, relative to the overall level. The blue bars are uncontrolled variation: just the raw data. The maroon bars are variation after controlling for income. The universe is 24 large metropolitan statistical areas (see report for details).

    What I find interesting about the chart is how unimportant income is in explaining variation in Medicare or transportation and how close in value the coefficients of variation between those two categories are. It makes me wonder, how unusual is observed geographic variation in Medicare spending? Are there other goods and services that vary as much? Are those variation explained by similar things (income, supply factors, demand factors, etc.) to the same degree? Has there been any other work on this?

    Note that the above does not imply that variations in Medicare spending are unimportant or unworthy of attention. As the authors of the CBO report write,

    Although the variation in Medicare spending is not completely out of line with that observed in other sectors of the economy, it does warrant closer examination. It is reasonable to assume that the value of the goods and services consumed in most other sectors is readily apparent. For example, if people in a given area spend a relatively large amount on housing, it is reasonable to assume that they are aware of the attendant benefits (for example, the area might provide better job opportunities, have a mild climate, offer access to cultural amenities, or have good public schools) and they choose to spend more as a result. That assumption does not necessarily hold for health care. Health care providers usually have a strong influence on the choice of treatment, and the quality or value of the benefits received from higher spending is much more difficult for patients to discern.

    In other words, to a larger degree than for other goods and services, variations in Medicare (or,  health care) spending should remain even after controlling for things that patients can readily observe and understand. The question is, is this empirically verifiable?

    More geographic variation posts here.

    Share
    Comments closed
     
    • I was not surprised when I originally looked at this chart a while back and not as I reacquainted myself with it this morning. The key issue of course is the income elasticity and the responsiveness of various categories. Most people are more surprised by the transportation result (though I’m not). A simple and not entirely wrong interpretation is that you should expect health care and transportation to continue to increase as a share of GDP. We all know that’s happening with health care, it’s less well known that it’s happening with transportation (and I would extend “transportation” to virtual transportation as well).

      Jim

      • The claim by CBO is that people know what they are getting for $ spent on transportation, less so for health care. Is it true? What about other goods and services? As far as I know, these questions are not answered quantitatively in any consistent, side-by-side analysis.

        • Right, Austin, as you well know from the types of “rooting around in the literature” that you do, our greatest weaknesses as economists are doing good decompositions of important effects for making sector comparisons. So let me answer your query three ways. First, the direct way:

          1) No, in some real sense CBO is right and we know less about what objectively we are getting for health care than we know on transportation. But this is by far the worst possible answer to that question since it leaves out all of the interesting issues.

          2) A slightly better answer (and a relativist economist retort that does resonate with me somewhat) is to say, revealed preferences on utility are all that really matter, so some objective truth is quite irrelevant, people know what they think they’re getting and if that drives their behavior, that’s enough.

          3) The best answer requires the quantitative consistent side-by-side decomposition analysis we really would want. I have intuition on that answer, but no good analysis. That analysis requires (at a minimum) decomposition of time preference effects, elasticity of income/income effects, risk preference effects (the uncertainty on whether we will get in an accident or stuck in traffic on transportation are pretty trivial compared to the issues in health care), “climate effects” (anything that varies by location and by time that essentially is exogenous from the economic actors, and expected value effects (I have the behavioral economics view of loss aversion and asymmetric loss functions, so you can’t solve the problem with standard expected utility analysis). So, coefficients of variation are a good approach to measurement and in the decomposition I think all of these effects I’ve listed are really important (which are more important and how important requires that empirical analysis). And I think both answer 1 and 2 are wrong and the complex real answer lies in between.

          Probably more than anyone wants to read on a hot summer day,
          Jim