Sticking points: common‐agency problems and contracting in the US healthcare system

• Author: James B. Rebitzer,Michael Powell,Brigham Frandsen
• Date: 01 June 2019
• Published date: 01 June 2019
• DOI: http://doi.org/10.1111/1756-2171.12269

RAND Journal of Economics

Vol.50, No. 2, Summer 2019

pp. 251–285

Sticking points: common-agency problems

and contracting in the US healthcare system

Brigham Frandsen∗

Michael Powell∗∗

and

James B. Rebitzer∗∗∗

Wepropose a “common-agency” model for explaining inefficient contractingin the US healthcare

system. Common-agency problems arise when multiple payers seek to motivate a provider to

invest in improved care coordination. We highlight the possibility of “sticking points,” that

is, Pareto-dominated equilibria in which payers coordinate around contracts which give weak

incentives to the provider. Sticking points rationalize three hard-to-explain features of the US

healthcare system: widespread fee-for-service arrangements; problematic care coordination;

and the historical reliance on single-specialty practices to deliver care. The model also analyzes

the effects of policies promoting more efficient contracting between payers and providers.

1. Introduction

The US healthcare system is famously inefficient, but the causes are poorly understood

(Baicker and Chandra, 2011). One candidate explanation that has received considerable attention

from analysts and policy makers is inefficient contracting betweenpayers and providers, in partic-

ular, the heavyreliance on fee-for-ser vice payarrangements with physicians. This article proposes

a “common-agency” model for explaining the puzzling prevalence of inefficient contracting in

the US healthcare system.

∗Brigham Young University; frandsen@byu.edu.

∗∗Nor thwesternUniversity; mike-powell@kellogg.northwestern.edu.

∗∗∗Boston University and National Bureau of Economic Research; rebitzer@bu.edu.

The authors would liketo thank Dan Bar ron, Robert Gibbons, Jin Li, Thomas McGuire, Luigi Siciliani, Lars Stole, Jeroen

Swinkels, and Mike Whinston, and for helpful comments along with participants at the MIT Organizational Economics

Seminar (October 2015, Cambridge, Mass.); the NBER Personnel Economics Summer Institute meeting (July 2015,

Cambridge, Mass.); the BU, Harvard,MIT Health Economics Seminar (February 2015, Boston, Mass.); The Mar yland

Workshopon Health Infor mation Technology (October 2014, College Park, Maryland); The Society for Institutional and

Organizational Economics Conference (June 2016, Paris); and the European Health Economics Workshop (June 2014,

Lausanne, Switzerland). Weare, of course, responsible for any remaining errors. The views expressed herein are those of

the authors and do not necessarily reflect the views of the National Bureau of Economic Research.

C2019, The RAND Corporation. 251

252 / THE RAND JOURNAL OF ECONOMICS

Our analysis focuses on the common-agency problems that arise when multiple payers seek

to motivate a provider to take actions that benefit the payers. The provider in this case acts

as a “common agent” to the various payers. Strategic interactions among payers introduce two

distortions that shape the equilibrium contracts offered to the provider.The first distor tion, which

has been extensively analyzed in other contexts but only rarely in healthcare, is the free-rider

problem in which payersoffer weak incentives to the provider,because any given payer reaps only

a fraction of the marginal benefit of stronger incentives.1The second distortion is a coordination

failure between payers.Unlike the free-rider problem, coordination failures lead to the emergence

of what we refer to as “sticking-point equilibria,” that is, Pareto-dominated equilibria in which

all payers offer contracts that may entirely omit incentives for making efficient investments.

Previous analyses of common-agency problems have not emphasized the role of coordina-

tion failures among payers and the resulting sticking-point equilibria. A central contribution of

this article is to trace out the implications these equilibria have for the US healthcare system and

healthcare policy.In brief, we find that sticking-point equilibria offer a straightforward explanation

for three long-observed but difficult-to-explain features of the US healthcare system: the ubiquity

of fee-for-service contracting arrangements for physician services among private payers2,3; poor

care coordination across providers4–6; and the historical reliance on small, single-specialty prac-

tices rather than larger multispecialty group practices.7The common-agency model we propose

also provides insights on the effects of policies such as Accountable Care Organizations(ACOs),

that aim to promote more efficient forms of contracting between payers and providers.

To illustrate the logicof our model, imagine two private payers who wouldlike to encourage

a common provider (an independent physician practice, for example) to implement an electronic

medical record system. Implementing such a system requires effort from the provider, whereas

the benefits accrue mainly to the payers.8A natural way to motivate the provider to implement

the new system would be for payers to move away from traditional fee-for-service contracts and

offer the provider a share of the savings that the electronic medical records system generates for

the payer. In a common-agency setting, however, the shared-savings incentives offered by one

payer will also benefit the other payer. Not surprisingly, this externality results in an equilibrium

in which both payers offer weak incentives. As a result, the provider devotes low levels of effort

toward implementing the electronic medical records system.

If free-riding were the only market failure induced by common agency here, we would

expect to see an equilibrium in which private payers offered weak incentives for improved care

1For an important exception, see Glazer and McGuire (2002), which is, to the best of our knowledge, the first

application of common-agency models to the study of the US healthcare system.

2Zuvekas and Cohen (2016) report that 94.7% of physician office visits in 2013 werecovered under fee-for-service

arrangements, which reward volume without considering the cost to payers.

3Because of the “inherent inefficiencies and problematic financial incentives” (The National Commission on

Physician Payment Reform, 2013) fee-for-service contracts entail, the National Commission on Physician Payment

Reform recommended a rapid transition away from them.

4The Institute of Medicine’s (2001) assessment of care quality in the U.S. healthcare system found that care

delivery is often complex and poorly coordinated,leading to wasted resources, gaps in coverage, loss of information, and

reductions in the speed and safety with which care is delivered.

5Many of the problems of poor care coordination result either from mishandled referrals to specialists (Mehrotra,

Forrest, and Lin, 2011) or from fragmented care delivery (see Cebul et al., 2008; and Rebitzer and Votruba, 2011, for

a discussion of the problem of care fragmentation and Frandsen et al., 2015; Hussey et al., 2014; Agha, Frandsen, and

Rebitzer, 2017; and Romano, Segal, and Pollack, 2015, for estimates of its costs).

6Mishandled referrals and fragmented care delivery are exacerbated by relatively weakinvestments in technology

and process improvements that strengthen integration across providersand organizations (Milstein and Gilbertson, 2009;

Simon et al., 2017).

7Medical care in the United States has been historically deliveredby practitioners operating out of their own offices

or as attendings in hospitals (Starr, 1984; Robinson, 1999), and although it still commonly is (Burns, Goldsmith, and

Sen, 2013; Baker, Bundorf, and Royalty, 2014), a rapidly growing percentage of physicians are operating in largeg roups

organized by nonphysicianowners (see Burns and Pauly, 2018, for a reviewof physician models of practice organization).

8For example, the electronic medical record system could allow the payer to track and discourage duplicative

testing, treatments that are not cost-effective, excessiverefer rals to specialists, or unwarranted emergencyroom visits.

C

The RAND Corporation 2019.

FRANDSEN, POWELL AND REBITZER / 253

coordination, not one lacking such incentives altogether.Yet, fee-for-service contracts that do not

share with the provider any of the payer’s gains from improved care coordination are what has

been historically observed in most of the US healthcare system.

To explain the anomalous reliance on fee-for-service contracts within a common-agency

framework, we consider coordination failures among payers. We introduce coordination failures

into our previous example by adding the reasonable assumption that the transition to electronic

medical records also requires providers to purchase a new health information technology (HIT)

system. The fixed, up-front costs of purchasing such a system subtly alter the common-agency

problem and make it more severe.9Thisis because the two payers will only deviatefrom traditional

fee-for-service arrangements if they each believe that incentives jointly offered by both payers

fully compensate the provider for these fixed, up-front costs. Otherwise, neither payer will find it

optimal to shoulder the entire burden of motivating the provider to incur the fixedcost of the HIT

system. In this case, the payers will stay with the traditional fee-for-service contracts—even if

such contracts are Pareto dominated by higher-powered contracts each payer would offer if they

believed the other would as well.To the extent that many organizational innovationsthat improve

care coordination involve sizeable fixedcosts, sticking-point equilibria within a common-agency

model offer a plausible account for the persistence of both fee-for-service payments and poor

care coordination.10

In a sticking-point equilibrium, providers also face weakened incentives to form integrated,

multispecialty group practices, and so elect to deliver care through small, single-specialty prac-

tices. To see why, return again to the decision to invest in an HIT system. These systems enable

superior coordination and information handoffs in referrals, but they typically do not allow for

interoperability, that is, the easy exchange of information across organizations. In this setting,

the gains from investing in HIT systems are greater when providers operate within a multispe-

cialty group practice, and the gains from forming multispecialty group practices are similarly

enhanced by investments in HIT systems. Because of this complementarity, the failure to write

incentive contracts that encourage efficient investmentsin HIT systems also depresses the retur ns

to forming integrated multispecialty group practices.11

Many others have observed that fee-for-service pay structures suppress investments in the

technology and processes required for care coordination and integrated care delivery.12 Our key

contribution is to provide an explanation for the persistence of these pay practices. By rooting

the persistence of inefficient fee-for-service contracts in a specific market failure, our common-

agency model also provides insights on the effects of public policies that aim to promote more

efficient forms of contracting between payers and providers.

The common-agency approach wedevelop in this article has three policy implications that are

not apparent in more familiar principal-agent models. First, in traditional principal-agent models,

one might observe fee-for-service contracts, but such contracts would persist only when they

are efficient. Thus, in traditional principal-agent models—but not in our common-agency model

when there are sticking-point equilibria—policies aimed at promoting shared saving incentives

would be counterproductive, in the sense that they would move the market away from efficient

contracts. Second, common-agency problems become more severe as the number of payers

increases: simply increasing the competitiveness of insurance markets may therefore not lead to

more efficient contracting between insurers and providers. A subtle policy corollary, however,

9See Simon et al. (2007) for evidence showing the most cited barrier to adopting HIT is up-front costs.

10 A partial list of organizational innovations that improve the integration of care and plausibly involve a com-

bination of up-front fixed investments and ongoing expenditures of effort includes clinical decision-making support;

managerial and financial systems; new standards of care and protocols that focus more on primary care physicians and

nonphysician providers.

11 Burns, Goldsmith, and Sen (2013) and Burns and Pauly (2018) argue that the long persistence of small group

practices is due to the limited scale and scope economies in physician practices. Our model highlights a different

explanation: the complementarities between investmentsin care coordination and organizational for m.

12 See, for example, Crosson (2009), Burns and Pauly (2002), and Blumenthal (2011).

C

The RAND Corporation 2019.