Hospital systems and bargaining power: evidence from out‐of‐market acquisitions

• Date: 01 August 2017
• Author: Kevin E. Pflum,Matthew S. Lewis
• Published date: 01 August 2017
• DOI: http://doi.org/10.1111/1756-2171.12186

RAND Journal of Economics

Vol.48, No. 3, Fall 2017

pp. 579–610

Hospital systems and bargaining power:

evidence from out-of-market acquisitions

Matthew S. Lewis∗

and

Kevin E. Pflum∗∗

Analyses of hospital mergers typically focuson acquisitions that alter local market concentration.

However, as prices are negotiatedbetween hospital systems and insurers, this focus may overlook

the impact of cross-market interdependence in the bargaining outcome. Using data on out-of-

market acquisitions occurring acrossthe United States from 2000–2010, we investigate the impact

of cross-market dependencies on negotiated prices. We find that prices at hospitals acquired by

out-of-market systems increase by about 17% more than unacquired, stand-alone hospitals and

confirm that out-of-market mergers result in a relaxation of competition, the prices of nearby

competitors to acquired hospitals increase by around 8%.

1. Introduction

Over the last 20 years, there has been a transformation in the way researchers and antitrust

authorities assess the competitiveness of hospital markets and evaluate mergers. Theoretical and

empirical models of hospital competition now incorporate more directly the contract negotiations

between managed care organizations (MCOs) and hospitals. These models revealthat equilibrium

reimbursement rates depend heavily on the relativebargaining positions of the MCO and hospital,

which are determined by the profits each party would earn if they fail to agree to a contract that

includes the hospital in the MCO’s provider network. When twoor more potentially substitutable

hospitals merge and jointly negotiate prices with an MCO, the relative bargaining position of the

hospital system increases because the MCO now finds it increasingly difficult to offerenrollees a

sufficiently attractive provider network without this group of hospitals. Recent empirical studies

∗Clemson University; mslewis@clemson.edu.

∗∗Bates White, LLC; kevin.pflum@bateswhite.com.

The authors benefitted greatly from the comments and suggestions of David Balan, Samuel Kleiner, Aviv Nevo, and

three anonymous referees, as well as seminar participants at the University of North Carolina, Clemson University, the

University of Alabama, the Bureau of Economic Analysis, the Congressional Budget Office, the 2013 International

Industrial Organization Conference in Boston, the 2014 American Society of Health Economists Biannual Conference

in Los Angeles, and the 2015 American Economic Association Annual Meeting in Boston. We also thank Yaxuan He

for excellent research assistance. A web Appendix containing additional results tables and robustness checks is available

from the author’swebsite (kevinpflum.com/papers/Lewis_Pflum_RAND_online_appendix.pdf).

C2017, The RAND Corporation. 579

580 / THE RAND JOURNAL OF ECONOMICS

have consistently found that mergers between local rival hospitals result in significantly higher

reimbursement rates (e.g., Capps, Dranove, and Satterthwaite, 2003; Gowrisankaran, Nevo, and

Town, 2015), and antitrust authorities have adopted these new approaches to more successfully

challenge local mergers (Farrell et al., 2011).

Mergers between hospitals located in different markets are also fairly common and have

contributed significantly to the persistent expansion of largeregional and national hospital systems

in recent years. Despite this, relatively little is known about the potential competitive effects of

these out-of-market mergers and acquisitions. Existing theoretical and empirical approaches have

not incorporated the potential for cross-market dependencies, focusing only on the impact of

changes in local market structure. Yet there are mechanisms that can generate cross-market

interdependencies that may have an impact on negotiated reimbursement rates. For example,

Vistnes and Sarafidis (2013) consider situations in which employers attempt to contract with a

single MCO to cover employees in multiple patient markets and illustrate how a hospital in such a

situation might be more valuable to the MCO if it is a member of a system that operates hospitals

across many of these different markets. Alternatively, Gowrisankaran, Nevo, and Town (2015)

suggest that mergers involving hospitals in different markets could generate price changes by

allowing an MCO operating in both markets to arrange adjustments in relative reimbursement

rates across cities that are mutually beneficial to both the MCO and the hospitals. Finally, Lewis

and Pflum (2015) highlight that systems may also provide their member hospitals with additional

information and support during contract negotiations, allowing the hospital to extract from MCOs

a larger share of the profits generated by a successful contract (i.e., hospitals in systems may have

higher bargaining power).

Despite this growing awarenessof these additional competitive effects, there has been essen-

tially no empirical investigation of their relative importance or overall impact on hospital prices.

To begin to fill in this gap, we investigate the degree to which system membership may influence

negotiated reimbursement rates through channels unrelated to local market concentration. Rather

than attempting to specify a particular model or test a particular mechanism, we identify the

overall importance of these additional mechanisms by examining the impact of out-of-market

mergers that do not alter local market structure. Using a difference-in-differences approach to

study 81 such hospital acquisitions occurring across the United States during the years 2000–

2010, we find that the average net reimbursement rates at these hospitals increase by about 17%

after joining an out-of-market hospital system with some specifications suggesting even larger

effects.

We adopt a variety of strategies to control for unobservable market conditions that may

influence hospital pricing or be correlated with the likelihood of a hospital acquisition. To

ensure that treatment effects are not driven by differing price trends, we estimate specifications

that allow for separate sets of year fixed effects for acquired hospitals and control hospitals and

specifications that include leads and lags of the treatment variable to revealdiscrete jumps or kinks

in price trends at the time of acquisition. Wealso estimate a difference-in-difference-in-differences

(triple-differences) specification which compares the difference-in-differences estimate of the

price change experienced by acquired hospitals with the difference-in-differences estimate of the

change in price experienced by other hospitals in the markets where these hospitals are acquired.

Additional evidence is providedby examining the impact of out-of-market acquisitions on the

prices of nearby rival hospitals. Neighboring hospitals do not experience a change in ownership,

eliminating any possibility that observed changes in their price mistakenly reflect some type

of acquisition-related accounting or administrative change in hospital operations rather than an

actual price change. The findings confirm that out-of-market mergers result in a relaxation of

competition by revealing that prices rise at nearby rival hospitals in response to price increases

by acquired hospitals. As expected, the price increases at rival hospitals are smaller than at the

acquired hospital and decrease in magnitude as distance from the acquired hospital increases,

providing additional evidencethat the estimated price effects are directly caused by the acquisition

rather than reflecting some confounding change in the local market.

C

The RAND Corporation 2017.

LEWIS AND PFLUM / 581

We also consider the possibility that the identified increases in reimbursement rates are

associated with other cost-related changes that may occur at hospitals during acquisition. A

collection of auxiliary regressions show that the observed price increases do not appear to be

a result of changes in patient case-mix or hospital quality, or a change in the cost of providing

care more generally. Moreover, the results of the rivals analysis further contradict the possibility

that average prices at acquired hospitals increase because patients that require more intensive

treatment substitute to the hospital in response to some change in quality or service, as we would

then expect patient complexity and prices to fall at nearby rival hospitals.

Our findings represent the first direct empirical evidence of cross-market dependencies in

hospital competition and should serve as a warning sign to researchers and policy makers that have

previously focused their scrutiny exclusively on in-market mergers.1Although many potential

cross-market mergers might not involvecross-market dependencies, the acquisitions that systems

have chosen to undertake over the last decade substantiallyincreased prices at acquired hospitals

on average. It is crucial to better understand and account for these effects when analyzinghospital

competition more generally. The implications for competition policy depend on the nature of the

cross-market dependencies. For example, if prices increase because large employers have fewer

MCOs with which to negotiate coverage for employees spread across multiple markets, then

this reflects a standard “lessening of competition” due to the elimination of substitutes (similar

in many ways to the elimination of competitors within markets). On the other hand, if prices

increase because merged hospitals are simply able to extract from MCOs a larger share of the

profits generated from a successful contract (i.e., they benefit from an enhanced bargaining power

or bargaining weight), it is not entirely clear that a lessening of competition has occurred.

We conclude our analysis with an investigation of how acquisition price effects vary by the

size of the acquiring system, the size of the acquired hospitals, and the proximity to other markets

with system partners. The results reveal that acquisition-related price increases remain large even

when acquired hospitals are very far from existing system partners and are located in seemingly

unrelated geographic markets. Interestingly, however, prices do increase somewhat more when

a hospital is acquired by a larger system or when the acquired hospital is relatively small. We

cautiously interpret these patterns to be consistent with the idea that out-of-market mergers impact

prices by improving the acquired hospitals’ negotiating ability (or bargainingpower) and that the

observed price effects are not likely to be entirely explained by employers attempting to insure

employees in multiple markets. More research is necessary, however, to investigate the different

potential sources of cross-market pricing power and determine the extent to which such mergers

warrant greater antitrust scrutiny.2In addition, because some mechanisms capable of generating

price increases in out-of-market mergers have the potential to also impact in-market mergers,

existing models should be reevaluated to assure that assessments of local mergers are not biased

by the exclusion of other relevant factors.

In the following section we outline our empirical strategy and describe the data used in our

analysis. Our primary results are presented in Section 3 and results on the cost of care and profit

margins presented in Section 4. We consider robustness and alternative explanations in Section

5, discuss possible sources of cross-market dependencies in Section 6, and offer concluding

comments in Section 7.

2. Empirical strategy and data

Several existing studies havefound suggestive evidence that cross-market system affiliation

can lead to higher prices. Melnick and Keeler (2007) discuss this possibility and present reduced-

form evidence that system hospitals seem to receive higher prices than nonsystem hospitals, even

1Current antitrust evaluation of hospital mergers relies on establishing how substitutable hospitals are from the

perspective of patients (FTC and USDOJ, 2004), implying that cross-market mergers are unlikely to ever be evaluated

beyond the initial screening phase.

2Wediscuss the implications for antitr ust policyin more detail in Section 7.

C

The RAND Corporation 2017.