This document summarizes a study examining whether hospitals shift costs from uninsured patients to private payers. The study replicates previous analyses using Texas hospital data from 2000-2007. Regression models relate private payer prices to Medicare, Medicaid, and uninsured prices. The results provide little evidence that hospitals shift costs to private payers from government or uninsured sources. The authors suggest further research on subsidies for uninsured care and better measures of market competition.
1. DO HOSPITALS SHIFT THE
COSTS OF THE UNINSURED TO
PRIVATE PAYERS?
By
Vivian Ho, Jerome Dugan, and Meei-Hsiang Ku-Goto
Rice University
Baker Institute for Public Policy and Department of Economics
HIGHLY PRELIMINARY - DO NOT CITE
2. Introduction
Cost shifting occurs when providers raise prices to one
group of payers in response to another group’s paying
lower prices (M. A. Morrisey 1994).
Public perception that rising number of uninsured
patients is leading to higher premiums for those who
are privately insured (Washington Post, St. Louis Post-
Dispatch, Houston Chronicle)
Reductions in Medicaid and Medicare reimbursement
rates may have led to higher private insurance
premiums.
3. Previous Theoretical Literature
Cost shifting ruled out if hospitals maximize
profits (Morrisey 1994; Frakt 2010)
Ppublic payer↓ → Qpublic ↓ , Pprivate ↓
Cost shifting occurs if hospitals maximize utility.
U = U ( π , patient volume or quality)
ability to cost shift depends on market hospital
power wrt insurers and mix of private pay versus
public pay patients.
4. Previous Empirical Literature
6 studies employ regression methods: Are reductions in
Medicare/Medicaid payments, or increases in uninsured
care associated with increases in prices charged to
private pay patients?
4 of 6 studies find cuts in Medicare or Medicaid lead to
higher private pay payments.
4 of 6 studies are based on data from California, which
had data to estimate price per discharge.
5. Previous Empirical Literature
Past regression-based studies do not explicitly
consider hospital revenues from the uninsured and
self-pay patients.
Price estimates in previous studies may be subject to
measurement error.
2 of the 6 studies use data from Medicare cost
reports, which only distinguish Medicare revenue
from non-Medicare revenue.
"Private pay revenue" includes payments from Medicaid,
self-pay, and uninsured patients.
6. Previous Empirical Literature
Dranove & White (1998) estimate an overall private
price by applying a fixed market basket to prices from
10 different cost centers for each hospital.
But hospitals may vary widely in revenues by cost center (e.g.
surgical intensive care vs. diagnostic radiology).
Zwanziger et al (2000) account for both inpatient and
outpatient care when estimating prices, but must
aggregate over the 2.
Cutler (1998) and Wu (2010) instrument for reductions
in Medicare price using (reduced) updates to DRG
reimbursement rates.
7. Previous Empirical Literature
This paper replicates analyses from previous
papers using Texas data.
25.2% of population uninsured in TX, compared w/
18.5% in CA
Relating private pay price to prices of other payers.
Zwanziger et al (2000 & 2006), Wu (2010).
But we explicitly consider the uninsured.
8. Data
2000-2007 versions of the American Hospital
Association annual survey
Contains net revenue by payer type, which is not reported in
the national AHA survey.
Payer types: private pay, Medicare, Medicaid,
uninsured/self-pay.
The state and local governments provide substantial subsidies
to hospitals to care for uninsured and under-insured patients.
Upper Payment Limit (UPL) funds paid by federal gov’t to
bring Medicaid prices up to level of Medicare payments.
Both these payment categories included in uninsured/self-
pay.
9. Data
Medicaid DSH payments reported separately, but could
be allocated to either Medicaid or uncompensated care.
Hospital expenses are not reported by payer type.
Texas hospital discharge data is used to sum charges by payer
type and adjusted to reflect costs using the cost-to-charge
ratio for each hospital and year.
The DRG weight for each hospital discharge is
obtained from the state and averaged by hospital to
create a case mix index.
10. Empirical Models
Private pay price (revenue/discharges) regressed on:
Medicare price
Medicaid price
Uninsured/Self-pay price
Cost per patient
casemix (quadratic)
county HHI index (based on bedsize, quadratic)
county managed care penetration (quadratic)
year fixed effects
19. Bargaining Power
Wu: Hospitals w/ large private payer volumes may be
better able to cost shift onto private payers.
Wu’s measure of bargaining power
More appropriate measure may be
20. Table 5: Private Payer Price Regressions with
Bargaining Power
(1)
Coef. t
Medicare price 0.102 (1.45)
Medicaid price -0.058 (-1.25)
Medicaid*bargain -0.055 (-0.76)
Unins/self pay price -0.022 (-1.37)
Unins/self*bargain -0.032 (-1.21)
Bargain -182.067 (-1.68)
Average cost 0.586 (3.73)
N 1255
Price & Cost in logs
Hospital Fixed Effects
First Differences X
Regression also includes Casemix, HHI, and Managed care penetration in quadratic form, year dummies,
and a constant.
21. Conclusions & Future Work
Very little evidence of cost shifting to private
payers from either government sources or the
uninsured in Texas.
Further research on the DSH payments and
payments for uninsured/self-pay care.
Identify better measures of hospital and
insurance market competition.
22. Conclusions & Future Work
Cutler (1998) tests for effects of Medicare price
cuts on:
Hospital closures*
Change in beds
FTE RNs and LPNs*
Cardiac services (PCI, cardiac cath, CABG)
Diagnostic radiology (CT, MRI, PET, SPECT)
ERs
Notas do Editor
Portion of Table 3: Multivariate Regression Models for Private Payer Price
Portion of Table 3: Multivariate Regression Models for Private Payer Price
Portion of Table 4: Multivariate Regression Models for Private Payer Price Accounting for Ownership Status
Portion of Table 4: Multivariate Regression Models for Private Payer Price Accounting for Ownership Status