The impact of the managed care backlash on health care spending

• Published date: 01 March 2020
• DOI: http://doi.org/10.1111/1756-2171.12306
• Date: 01 March 2020

RAND Journal of Economics

Vol.51, No. 1, Spring 2020

pp. 59–108

The impact of the managed care backlash

on health care spending

Maxim L. Pinkovskiy∗

The health spending slowdown associated with the managedcare revolution in the 1990s suggests

that managed care may have been successful in controlling health care spending. I exploit the

passage of state regulation during the “managedcare backlash” as well as geographic variation

in managed care intensity to measure the impact of managed care on spending. I find that

restricting managed care causes a large and significant increase in hospital spending, which

cannot be explained by changes in hospital market concentration, other regulatory activity, and

multiple other possible explanations. I also do not find effects of the backlash on mortality.

1. Introduction

The growth of health care spending as a share of Gross Domestic Product (GDP) has been

one of the defining features of the US health care sector. Health care spending nearly doubled as

a share of GDP in 30 years, rising from 8.3% in 1977 to 15.9% in 2007, and often has grown

at a linear (and hence, unsustainable) rate for decades at a time. The problem of controlling

the growth in health care spending and ensuring its efficient use is particularly urgent because

health spending forms a significant part of US government expenditures, putting pressure on

other government priorities and tax rates, and making coverage expansion expensive.

The overalltrend of rising health care expenditures in the United States saw a temporary break

during the 1990s, when health care spending as a share of GDP held steady at 13.3% from 1993

to 2000. This stabilization of health care expenditures coincided with the peak of the so-called

∗Federal Reserve Bank of New York; maxim.pinkovskiy@ny.frb.org.

This paper was written as the first chapter of the author’s Ph.D. dissertation at the Massachusetts Institute of Technology.

I am immensely grateful to Daron Acemoglu, Amy Finkelstein, and Jerry Hausman for guidance and advice. I am very

grateful to David Autor, Manasi Deshpande, Nicole Gorton, TalGross, Jonathan Gruber, Andrew Haughwout, Katherine

Ho, Horacio Larreguy, Claire Lim, WilliamNober, Matthew Notowidigdo, Benjamin Olken, Christopher Palmer, James

Poterba,Adam Sacar ny,Joe Shapiro, James Snyder, Heidi Williams,seminar par ticipants at MIT,the American Enterprise

Institute, the Bureau of Economic Analysis, Mathematica, and the Federal Reserve Bank of New York, and conference

participants at the 2014 System Applied Microeconomics Conference (Minneapolis, MN), 2017 Society for Econometric

Measurement Conference (Cambridge, MA), and 2018 National TaxAssociation Annual Conference (New Orleans, LA)

for extremely valuable feedback, comments, and suggestions. I am very grateful to Jean Roth, Mohan Ramanujan and

the NBER for helping me with access to data from the AHA Annual Survey. I am very grateful to Chad Syverson and

two anonymous referees for very thoughtful comments. I am deeply indebted to Laurence Baker for sharing with me his

county-level HMO penetration measure and to Susan Laudicina for sending me her publication,“State Legislative Health

Care and Insurance Issues.” Finally,I am very grateful to the Paul and Daisy Soros Fellowship for intellectual stimulation,

and to the National Science Foundation Graduate Research FellowshipProgram (NSF GRFP) and to the Humane Studies

Fellowship (Institute for Humane Studies) for graduate study funding.

C2020, The RAND Corporation. 59

60 / THE RAND JOURNAL OF ECONOMICS

FIGURE 1

US NATIONAL HEALTH EXPENDITURES, % OF GDP [Color figure can be viewed at wileyonlinelibrary.com]

8

10

12

14

16

1977

1978

1979

1980

1981

1982

1983

1984

1985

1986

1987

1988

1989

1990

1991

1992

1993

1994

1995

1996

1997

1998

1999

2000

2001

2002

2003

2004

2005

2006

2007

Year

US National Health Expenditures, % of GDP

Source: Center for Medicare and Medicaid Services (CMS): National Health Expenditure Accounts.

managed care revolution, which saw the replacement of conventional insurers (who reimbursed

hospitals and physicians for services provided without regulating utilization) by health insurance

organizations that managed the medical care of their enrollees. The organizational innovation of

managed care firms was to integrate physicians and insurers partially or completely to align their

incentives and discourage physicians from inducing demand for medical care. The most well-

known type of managed care organization,the health maintenance organization (HMO), restricted

its patients to see a strictly delimited network of providers, who sometimes were its employees.

Although the growth of health insurance premiums slowed significantly, patients and physi-

cians chafed at managed care controls. At the end of the 1990s, there arose a widespread backlash

against managed care cost containment practices, with increasingly negative media coverage of

managed care. Ultimately, state governments passed “patients’ bills of rights” that limited the

ability of managed care firms to restrict care and shape the incentives of medical practitioners.

Health care spending resumed rising as a share of GDP in 2001, at the height of the managed care

backlash. Figure 1 presents the time series of the US health care share between 1977 and 2007,

with both the slowdown during the managed care revolutionand the speedup during the backlash

clearly visible. It remains an open question whether managed care succeeded in stabilizing US

health care spending or whether the slowdown in US health care spending growth in the 1990s

was a product of other factors (Glied, 2003).

This article finds that the managed care backlash, as captured by regulations passed to

restrict managed care cost containment practices (hereafter, backlash regulations), in fact caused

the subsequent increases in hospital spending. I show that passage of backlash regulations was

associated with increases in hospital spending shares in counties with high managed care intensity,

proxied bythe estimates of county HMO penetration from Baker and Phibbs (2002). Using county-

level data over time, I employ a research design that is robust to key potential confounders,

including the endogeneity of passage of the backlash regulations and trends in health spending

that depend on managed care intensity but are unrelated to the passage of the backlash regulations.

The key dependent variable of myanalysis is the hospital spending share of county personal

income, which I define as the hospital share. My results indicate that the managed care backlash

increased the growth rate of the hospital share by 1.4 percentage points per backlash regulation

in a county with maximum possible managed care intensity relative to a county with minimum

intensity.The magnitude of this estimate is consistent with the managed care backlash accounting

for over 90% of the overall increase in the nationwide growth rate of the hospital share observed

C

The RAND Corporation 2020.

PINKOVSKIY /61

FIGURE 2

EFFECTS OF MAJOR PASSAGEOF BACKLASH REGULATIONS ON HEALTHSHARE OF COUNTY

PERSONAL INCOME [Color figure can be viewed at wileyonlinelibrary.com]

-.05

-.04

-.03

-.02

-.01

0

.01

.02

.03

.04

.05

Percentage point increase in health share

-5

-4

-3

-2

-1

0

1

2

3

4

5

Years after major passage

Counties in top quartile by 1995 HMO penetration

Counties in bottom quartile by 1995 HMO penetration

Year of major passage is year in which the county's state passed the largest number of backlash regulations.

Effects of Major Passage of Backlash Regulations

Source: Author’scalculations. Blue and red lines show percentage point increases in AHA hospital spending as a share

of county personal income for counties in the top and bottom quartiles of HMO penetration in 1995.

between the 1995–2000 period and the 2001–2005 period. My findings are robust to alternative

spending measures. The rise in hospital spending growth appears to work through an increase

in spending per hospital employee, rather than through an increase in the number of hospital

employees, and likely, through an increase in hospital admissions rather than through an increase

in expenditures per admission. I detect no effect of backlash regulations on mortality, although

I cannot rule out mortality impacts large enough that, at typical values of a statistical life, the

willingness to pay for them would exceed the increase in health spending associated with the

managed care backlash. I provide a variety of robustness checks for my results by conducting

permutation tests on the data, showing that the passage of the backlash regulations predates the

resumption of health care spending growth, exploring the possibility that either HMO penetration

or the backlash regulations variable may be proxyingfor other measures of county heterogeneity

(such as urbanization or population density) or relevant developments in the health care sector

(such as hospital consolidation and other regulatory movements unrelated to managed care),

and varying the construction of my regulation measures. Viewing the managed care backlash

regulations as plausibly exogenous variation in dampening the intensity of managed care, my

results are consistent with the thesis that managed care reduced spending without measurably

sacrificing outcomes.

I illustrate my main result in Figure 2. I define the year of major passage for a county as

the year in which its state passed its largest number of backlash regulations, a proxy for the peak

of the political backlash in the state. I then plot the average growth rates of the hospital share

for high and low HMO penetration counties relative to the year of major passage. It is clear

that the growth of the hospital share is low (in fact, substantially negative) before the year of

major passage in the high penetration counties, whereas the growth rate of the hospital share is

higher (generally positive) before the year of major passage in the lowpenetration counties. Both

growth rates are evolving fairly continuously up to the year of major passage, with perhaps a

slight upward trend in the growth rate for the high penetration counties. However, two years after

the year of major passage, the growth of the hospital share shoots up to nearly 5% for the high

penentration counties, much higher than the growth rate for low penetration counties. The growth

rate for the high penetration counties then declines somewhat, but still remains positiveand much

C

The RAND Corporation 2020.