A paper by Cutler and Sheiner claim that if all areas in the United States had the level of spending prevailing at the 10th percentile from a ranking of 315 metropolitan statistical areas (MSAs) by Medicare spending, Medicare spending would fall by almost 30%. However, this may only be true in a static sense. If the high-cost areas one year are not likely to be high cost in another year, the regional differences in spending may be simply random noise. If this is the case, there are few policy interventions that could reduce effectively reduce spending in high cost areas. On the other hand, if certain regions are persistently high cost, then certain policies targeting high-cost areas may be useful.
Rettenmaier and Wang use a 30 year data sample to evaluate whether states with high per capita Medicare spending tend to remain so over time. Specifically, they use data from 1974-2003 from the Continuous Medicare History Sample (CMHS), a large longitudinal 5% sample of all Medicare beneficiaries from 1974 to 2003. They also control for various state-level factors such as average Medicare beneficiary age (65–69, 70–74, 75–79, 80–84, and 85+), education, income, average inequality level, differences in Medicare prices, share of blacks, share of men, deaths in the per Medicare beneficiary, and the supply of doctors and hospital beds in the area. The authors also examine cross-state variation in end-of-life spending for beneficiaries who died during the year and measure the impact of four important pecies of legislation: “the implementation of the PPS in 1984, the reinterpretation of home health care in 1989, the implementation of the fee schedule for physicians in 1992, and the implementation of the 1997 Balance Budget Act’s (BBA) reforms in 1998.
The authors major findings are as follows:
First, Medicare spending and utilization disparities are significant even at the aggregate state level. They remain significant even if we exclude state differences in various demographic, demand, and supply side factors that may affect the use of health care. Second, these variations have, in most cases, narrowed over time by differing magnitudes. The gap between the low and the high spending states has closed due to more rapid growth at the lower end of the distribution than at the upper end of the distribution. Further, the downward trend could be reversed and some changes in quantity of covered services as well as overall expenditure in state-to-state variations coincide with policy changes that affect medical utilization and the payment mechanism
The authors also find that the PPS, the re-interpretation of home health care and the Balance Budget Act all tended to reduce regional variation in spending. But the impact was statistically insignificant.
This paper has a number of drawbacks. First, the data only go until 2013, so no information on regional variation in spending is known for the last 10 years. Second, risk adjustment occurs at the state rather than the individual level. Since healthcare spending is typically right-skewed, this approach may mask differences in the distribution of case mix within a state. For instance, a state that has two individuals of average health may actually spend less than a state that has one very healthy and one very sick beneficiary; even though average health may be the same, the right skewed distribution means that a state-level risk adjustment approach may be inaccurate. This critique could apply to any of the covariates used. Third, although the paper use an extensive set of information on patient demographics, there is information used to measure patient comorbidities.
Despite these drawbacks, examining trends in Medicare spending over such a long time period does provide important information that researchers and policymakers can use.
- Cutler D, Sheiner L. 1999. The geography of Medicare. American Economic Review 89: 228–233.
- ANDREW J. RETTENMAIER and ZIJUN WANG. REGIONAL VARIATIONS IN MEDICAL SPENDING AND UTILIZATION: A LONGITUDINAL ANALYSIS OF US MEDICARE POPULATION. Health Econ. 21: 67–82 (2012)