Table 2.

Empirical Cost-Shift Literature (1996 to present)

Citation	Principal Data Source	Unit of Observation and Methods	Dependent Variable(s)	Independent Variables	Cost Shifting Results
Gowrisankaran and Town 1997	1991 Current Population Survey (CPS), Health Care Financing Administration (HCFA), and American Hospital Association (AHA) data.	Estimated a dynamic, structural model of the market for inpatient hospital services. Simulated welfare effects of 1984 change to Medicare hospital prospective payment system.	Model included the effects of entry, exit, investment, and multipayer pricing decisions.	Observable parameters include proportion of patients ill by payer, income threshold for free care, copayment, Medicare deductible, Medicare reimbursement rate, corporate tax rates, and discount rate.	Owing to more concentrated hospital markets, private payers experienced a 10% reduction in quality and a 1% decline in price.
Cross-Sectional Studies
Stensland, Gaumer, and Miller 2010	2002–2006 Medicare hospital cost reports and 2005 IRS filings.	Hospital-year level descriptive analysis.	Medicare margins.	Non-Medicare margins.	Hospitals with lower non-Medicare margins had higher Medicare margins. In turn, hospitals with higher Medicare margins had lower costs. Hospitals with higher market power had higher costs, lower Medicare margins, and higher private pay margins. Illuminates factors relevant to price discrimination.
Dobson, DaVanzo, and Sen 2006	2000 American Hospital Association annual survey data.	State-level OLS.	Private payers’ payment-to-cost ratio.	Medicare, Medicaid, and uncompensated care payment-to-cost ratio and HMO penetration.	Found statistically significant evidence of price discrimination. HMO penetration and private payment-to-cost ratio negatively correlated.
Fixed-Effects Studies
Zwanziger, Melnick, and Bamezai 2000	1983–1991 California Office of Statewide Health Planning and Development hospital discharge data.	Hospital-year OLS including hospital fixed effects and instrumental variables for costs.	Per-patient private payment.	Per-patient revenue for Medicare and Medicaid, measures of hospital competition, ownership status, average cost (instrumented), case mix, and hospital fixed effects.	Private prices increased in response to reductions in Medicare rates (elasticities from 0.58 to 0.17, depending on hospital market concentration); they had a small and generally insignificant response to changes in Medicaid reimbursement. Inclusion of separate markups for Medicare and Medicaid across multiple years complicates interpretation of a dynamic cost-shift rate.
Zwanziger and Bamezai 2006	1993–2001 California Office of Statewide Health Planning and Development hospital discharge data.	Hospital-year level OLS with hospital fixed effects and instrumental variables for cost.	Per-patient private revenue.	Per-patient Medicare and Medicaid revenue, average costs, level of hospital market competition (HHI), and HHI-year interactions.	A 1% decrease in Medicare (Medicaid) prices caused a 0.17% (0.04%) private price increase. Between 1997 and 2001, 12.3% of the total increase in private prices was caused by public payment decreases.
Difference Model Studies
Clement 1997/1998	1982–1983, 1985–1986, 1988–1989, 1991–1992 California Office of Statewide Health Planning and Development hospital discharge data.	Hospital-level OLS.	Change in logarithm of private revenue-cost margins.	Change in Medicare and Medicaid margins, total margin, other revenue, assets, hospital competition, HMO market strength, private occupancy rate, service mix, profit and ownership status, and other measures of case mix and hospital characteristics.	Negative correlations between public and private margins. Inclusion of separate margins for Medicare and Medicaid across multiple years complicates interpretation of a cost-shift rate. HMO market strength is negatively correlated with private prices.
Dranove and White 1998	1983, 1992 California Office of Statewide Health Planning and Development hospital discharge data.	Hospital-level OLS, SUR, and logit (for closings).	Changes in (1) private price/cost margin, (2) service levels, and (3) closings.	Public-payer caseload, hospital competition, hospital size, a high-tech hospital indicator, profit status, and drivers of demand.	Found no evidence of cost shifting. Service levels fell at Medicaid-dependent hospitals, and such hospitals were more likely to go out of business. Service level per admission positively correlated with hospital market concentration.
Friesner and Rosenman 2002	1995, 1998 California Office of Statewide Health Planning and Development hospital discharge data.	Hospital-level OLS.	Change in (1) private prices and (2) public and private-service intensity (length of stay).	Changes in Medicare or Medicaid charges and proportion unpaid, changes in number of beds, race, ethnicity, outpatient prices, and income.	Nonprofits cost shift, and for-profits do not. Both types lower service intensity for public payers.
Cutler 1998	1985–1995 data from Medicare cost reports and Interstudy.	Hospital-level OLS and logit (for closure and technology models).	(1) Changes in per-patient non-Medicare (includes Medicaid) private revenue, (2) hospital closure, (3) logarithm of number of hospital beds, (4) change in logarithm of FTE of nurses, and (5) indicators of acquisition of particular technologies.	Per-patient “Medicare bite”—growth difference between hospital market basket and Medicare payments—changes in cost, managed care enrollment, for-profit and ownership status, number of beds, and MSA size.	1985–1990: at least a dollar-for-dollar cost shift, no evidence of an effect on hospital closure; 1990–1995: no evidence of cost shifting, a small effect indicating increased closures. In both periods, nursing input was reduced. Little evidence that payment changes affected hospital size or diffusion of technology. Rise in managed care explains differences over two time periods.
Wu 2009	1996, 2000 Medicare hospital cost reports.	Hospital-level OLS with hospital fixed effects and instrumental variables for Medicare price and HMO market penetration.	Change in non-Medicare price.	Change in Medicare price and revenue (instrumented), private/public payer mix, hospital ownership type, level and change in HMO market penetration (instrumented), change in case mix, hospital occupancy rate, level and change in Medicaid-to-Medicare physician fee ratio, share of for-profit hospitals, and hospital market concentration.	On average, about 20% of Medicare payment reductions are shifted to private payers. Degree of cost shifting is lower for hospitals in more competitive markets or markets with a higher share of for-profit hospitals.

Note: Excludes any literature reviewed in Morrisey 1996.