Medicare Fees and the Volume of Physicians' Services

Article excerpt

This paper estimates the relationship between Medicare fees and quantities provided by physicians for eight specific services. It uses data for 13,707 physicians who responded to surveys in 2000/2001 and/or 2004/2005 and were linked to all Medicare claims for their Medicare patients. Results show that Medicare fees are positively related to quantity provided for all eight services, and are significantly different from zero and elastic for five of them. The findings are consistent with the general economic proposition that supply curves for metrical services are positively sloped, and provide no evidence of volume-offset behavior for the services examined. These results also imply that Medicare could influence volume growth for specific services by varying their fee changes, and that uniform fee changes will have differential effects on service volume because of variation in underlying supply elasticities.


With accelerating health care costs and a growing beneficiary population, the Medicare trustees predict the program will become insolvent in 2017, magnifying pressure to find ways to control spending (Boards of Trustees 2009). Most of the recent growth in program costs is due to increases in the volume of services provided to beneficiaries, rather than increases in the number of beneficiaries or greater disease burdens among beneficiaries (MedPAC 2008a; Maxwell, Zuckerman, and Berenson 2007).

A major conundrum in trying to understand the reasons behind volume growth is the role of the Medicare fee schedule (MFS). Medicare physician fees, on average, have been relatively flat in recent years, effectively declining after accounting for inflation. Since volume has continued to grow, many conclude that either fee changes do not affect service volume, physicians respond perversely to fee decreases by increasing the volume of services, or both. In economists' parlance, this implies the supply curve is backward bending. Since the fee schedule is one of the few policy tools that Medicare potentially has available to influence service volume, it is critical to have a better understanding of how changes in Medicare fees affect Medicare service volume.

An earlier study (Hadley and Reschovsky 2006) addressed this relationship by estimating models that measured Medicare volume as the total number of relative value units

(RVUs) provided, and defined an average Medicare fee as total allowed charges for all services divided by the total number of relative value units. (1) The analysis indicated that this average Medicare fee variable was subject to substantial endogeneity bias. (2) Adjusting for this bias by using instrumental variable estimation resulted in a supply elasticity estimate that was statistically significant, positive, and greater than one in magnitude. If the fee elasticity is positive, then the aggregate association observed between Medicare volume and Medicare fees must reflect the effects of other factors, such as the introduction of new services, RVU upgrades for selected services, physicians" responses to cost reductions not captured by RVU assignments, and changes in private demand that make Medicare's payment rates look relatively more generous.

This study extends that earlier research by estimating models of physicians' provision of specific individual services to their Medicare patients. Focusing on specific services allows the construction of an exogenous measure of variations in the Medicare fee for each service and avoids the statistical problems inherent in using instrumental variable methods to estimate an aggregate supply model. (3) If the fee elasticity for individual services were also to be positive, then this result would reinforce the inference drawn from the earlier study based on instrumental variable estimation. The methodology we employ also could provide policymakers with potentially valuable information about differences in fee elasticities across various services that could be used in reforming payment methods. …


An unknown error has occurred. Please click the button below to reload the page. If the problem persists, please try again in a little while.