Do Commercial Health Care Prices Influence Medicare Spending?

Extensive geographic variation in per beneficiary Medicare spending has been widely documented, and recent work has found considerable variation among the commercially insured population. However, there appears to be little correlation in market-level spending between the Medicare and commercially insured populations (Chernew et al. 2010), and different factors appear to be driving variation across the populations.

Whereas prices paid to providers are administratively set in the Medicare program, prices for treating the commercially insured result from bilateral negotiations between insurers and providers.  Recent research has documented the impact of local insurer and provider market structure on negotiated prices, generally finding higher prices among more concentrated provider markets and lower prices among more concentrated insurer markets (i.e., Moriya et al. 2010, Melnick et al. 2011, Dunn and Shapiro 2012).  Other work has demonstrated that geographic variation in commercial prices drives variation in spending on the commercially insured (Dunn et al. 2012, Institute of Medicine 2013).

Notwithstanding this evidence, limited research has focused on the potential spillover effect of commercial prices and resultant price-margins within markets on Medicare utilization and spending. We therefore analyze the effect of commercial health care prices on Medicare spending across metropolitan areas over 2007-2009. 

Specifically, we use measures of per beneficiary Medicare spending and commercial prices from the IOM’s recent study.  Medicare spending includes all inpatient, outpatient, and drug reimbursement (Parts A, B, and D) and adjusts for health status based on age, sex, gender, race, and claims history; adjustments are also made for input costs.  An area-level commercial price index is constructed from the MarketScan database by dividing adjusted commercial spending by aggregate utilization, measured by weighting service-level utilization by average national reimbursement levels.  We first regress per beneficiary Medicare spending within metro areas on the commercial price index and hospital market concentration (constructed at the MSA-system level from American Hospital Association Surveys), and find substantial negative relationships.  Next, we instrument for commercial prices and hospital concentration using total population and the under-age-65 share.  These factors may affect commercial prices through insurance and hospital market structure, and do have adequate power.  The instrumental variables analysis implies that a commercial price measure 10% below its average is associated with Medicare spending that is 3.7% above its average (p < 0.01).

These results suggest that providers respond to low commercial prices by shifting service volume out of the commercial sector into Medicare.  This interpretation, if correct, implies that changes in insurer and provider market structure that affect negotiated commercial payment rates may have important spillover effects on Medicare spending.  We are further investigating this hypothesis by analyzing recent provider consolidations using Medicare and commercial claims.