Motion-Secondary: UFCW & EMPLOYERS BENEFIT TRUST ON BEHALF OF VS. SUTTER HEALTH et al

• 
• 
• 
• 
• 
• 
• 
• 

						
                                
Preview
                                oOo em ND HW BF WY =

Xavier Becerra
Attorney General of California

Kathleen Foote
Senior Assistant Attorney General ite Ee ieee
Michael Jorgenson FILED
Supervising Deputy Attorney General ace Court of California,
eryl Lee Johnson (SBN 66321) — rranisee
Esther La (SBN 160706) 03/08/2019
Emilio Varanini (SBN 163952) Clerk of the Court
Deputy Attorneys General : Doputy Clork

455 Golden Gate Avenue, Suite 11000

San Francisco, CA 94102-7004

Tel 415.510.3541 / Fax 415.703.5480
E-mail: Emilio. Varanini@doj.ca.gov
Attorneys for Plaintiff, People of the State of
California

Richard L. Grossman (SBN 112841)
Philip L. Pillsbury Jr. (SBN 072261)
Pillsbury & Coleman, LLP

600 Montgomery Street, 31° Floor
San Francisco, CA 94111

Tel 415.433.8000 / Fax 415.433.4816
Email: UEBT@pillsburycoleman.com

Lead Counsel. ‘jor Plaintiff UFCW & Employers Benefit
Trust and the Class (Additional Counsel not listed)
SUPERIOR COURT OF THE STATE OF CALIFORNIA
COUNTY OF SAN FRANCISCO
UFCW & Employers Benefit Trust, on behalf | Case No. CGC 14-538451
of itself and all others similarly situated Consolidated with
Case No. CGC-18-565398
Plaintiffs,
DECLARATION OF RUSSELL TAYLOR
vs. IN SUPPORT OF PLAINTIFF’S MOTION
TO EXCLUDE EXPERT OPINION OF DR.
Sutter Health, et al., GOWRISANKARAN THAT KAISER AND
SUTTER COMPETE IN THE SAME
Defendants. ANTITRUST MARKET [REDACTED]
Date: April 30, 2019
People of the State of California, ex. rel. Time: 2:00 p.m.
Xavier Becerra, Dept.: 304
Judge: Hon. Anne-Christine Massullo
Plaintiff,
Action Filed: April 7, 2014
vs. Trial Date: August 12, 2019
Sutter Health,
Defendant.

DECLARATION OF RUSSELL TAYLOR IN SUPPORT OF PLAINTIFF’S MOTION TO EXCLUDE EXPERT
OPINION OF DR. GOWRISANKARAN - Case No. CGC 14-538451I, Russell Taylor, declare as follows:

1. lam an attorney admitted to practice before this Court. I am an associate with
Farella Braun + Martel LLP, attorneys of record for Plaintiff UFCW & Employers Benefit Trust
(“UEBT”) and the Plaintiff Class. I have personal knowledge of the facts set forth below and if
called as a witness, I could and would competently testify to the matters stated herein. I make this
declaration in support of Plaintiff's Motion To Exclude Expert Opinion of Dr. Gowrisankaran that
Kaiser and Sutter Compete in the Same Antitrust Market.

2. Attached as Exhibit 1 is a copy of Easterbrook, Gowrisankaran, et al., Accounting
| for Complementarities in Hospital Mergers: Is a Substitute Needed for Current Approaches?,
Forthcoming, 82 Antitrust L.J. No. 2 (2019).

3. Attached as Exhibit 2 is a copy of DOJ & FTC, Commentary on the Horizontal
Merger Guidelines, March 2006.

4, Attached as Exhibit 3 is a copy of DOJ & FTC, Horizontal Merger Guidelines,
Aug 19, 2010.

5. Attached as Exhibit 4 is a copy of What is Sutter Health, available at

https://www.sutterhealth.org/about/what-is-sutter-health, last visited 3/6/2019.

6. Attached as Exhibit 5 is copy of Kate Ho & Robin Lee, Jnsurer Competition in
Healthcare Markets, Econometrica, Vol. 85, No. 2 , March 2017, 379-417.

7. Attached as Exhibit 6 are excerpts from the deposition of Joan Kezic taken in this
matter on November 1, 2018.

8. Attached as Exhibit 7 are excerpts from the deposition of Patrick Fry taken in this
matter on June 26, 2018. This transcript has been designated Highly Confidential Attorneys’ Eyes
Only.

9 Attached as Exhibit 8 are excerpts from the Declaration of Gautam Gowrisankaran,
dated October 29, 2018. This declaration includes material designated by Sutter as Highly

Confidential -AEO-Strategic Competitive Data.

1
DECLARATION OF RUSSELL TAYLOR IN SUPPORT OF PLAINTIFF’S MOTION TO EXCLUDE EXPERT
OPINION OF DR. GOWRISANKARAN - Case No. CGC 14-53845110. Attached as Exhibit 9 are excerpts from the deposition of Gautam Gowrisankaran,
Ph.D. in this matter on December 17 and 18, 2018. This transcript has been designated by Sutter
as Highly Confidential Attorneys’ Eyes Only.

11. Attached as Exhibit 10 are excerpts from the Expert Report of Gregory S. Vistnes,
Ph.D., dated August 31, 2018. This document includes material designated Highly Confidential —
AEO-Strategic Competitive Data.

12. Attached as Exhibit 11 are excerpts from the Rebuttal Expert Report of Gregory S.
Vistnes, Ph.D., dated January 31,2019. This document includes material designated Highly
Confidential -AEO-Strategic Competitive Data.

13. Attached as Exhibit 12 is a copy of a document produced in this matter, numbered

BSC_UFCW00007609. This document includes material designated Highly Confidential -AEO-

Strategic Competitive Data.
14. Attached as Exhibit 13 is a copy of a document produced in this matter, numbered

ABCLH119031. This document has been designated Highly Confidential -AEO-Strategic

Competitive Data.

15. Attached as Exhibit 14 is a copy of a document produced in this matter, numbered
AET-ES10022876. This document has been designated Highly Confidential -AEO-Strategic
Competitive Data.

16. Attached as Exhibit 15 is a copy of Exhibit A1 to the Declaration of Gautam
Gowrisankaran, dated October 29, 2018. This document includes material designated Highly
Confidential -AEO-Strategic Competitive Data.

17... Attached as Exhibit 16 is a copy of Exhibit A2 to the Declaration of Gautam
Gowrisankaran, dated October 29, 2018. This document includes material designated Highly

Confidential -AEO-Strategic Competitive Data.

18. Attached as Exhibit 17 is a copy of Exhibit A3 to the Declaration of Gautam

Gowrisankaran, dated October 29, 2018. This document includes material designated Highly

Confidential -AEO-Strategic Competitive Data.

2

DECLARATION OF RUSSELL TAYLOR IN SUPPORT OF PLAINTIFF’S MOTION TO EXCLUDE EXPERT
OPINION OF DR. GOWRISANKARAN - Case No. CGC 14-53845119. Attached as Exhibit 18 is a copy of Exhibit 12B to the Rebuttal Expert Report of
Gregory S. Vistnes, Ph.D., dated January 31,2019; This document includes material designated
Highly Confidential -AEO-Strategic Competitive Data.
20. Attached as Exhibit 19 is a copy of Exhibit 12C to the Rebuttal Expert Report of
Gregory S. Vistnes, Ph.D., dated January 31, 2019; This document includes material designated
Highly Confidential -AEO-Strategic Competitive Data.
21. Attached as Exhibit 20 is a copy of Exhibit 12A to the Rebuttal Expert Report of

Gregory 8. Vistnes, Ph.D., dated January 31, 2019; This document includes material designated

Highly Confidential -AEO-Strategic Competitive Data.

22. Attached as Exhibit 21 are excerpts from the deposition of Maya Greenfield taken
in this matter on November 5, 2018. This transcript has been designated Highly Confidential
Attorneys’ Eyes Only Competitive Data.

23. Attached as Exhibit 22 are excerpts from the deposition of Douglas Sturnick taken
in this matter on November 12, 2018. This transcript has been designated Highly Confidential
Attorneys’ Eyes Only.

I declare under penalty of perjury under the laws of the State of California that the
foregoing is true and correct.

Executed on this 8th day of March, 2019, at San Francisco, California.

Vv T=

Russell Taylor

29709\12332076.1

3

DECLARATION OF RUSSELL TAYLOR IN SUPPORT OF PLAINTIFF’S MOTION TO EXCLUDE EXPERT
OPINION OF DR. GOWRISANKARAN - Case No. CGC 14-538451EXHIBIT 1

DECLARATION OF RUSSELL TAYLOR IN SUPPORT
OF PLAINTIFF’S MOTION TO EXCLUDE EXPERT
OPINION OF DR. GOWRISANKARAN THAT KAISER
AND SUTTER COMPETE IN THE SAME ANTITRUST
MARKETForthcoming, 82 Antitrust L.J. Issue no. 2 (2019); subject to further revision
prior to publication; not for circulation, attribution, or quotation.

ACCOUNTING FOR COMPLEMENTARITIES
IN HOSPITAL MERGERS:
IS A SUBSTITUTE NEEDED FOR
CURRENT APPROACHES?

KATHLEEN F, EASTERBROOK
Gautam GowRISANKARAN
Dina OLDER AGUILAR
Yurer Wu*

Hospitals mergers are common and increasingly frequent occurrences in the
United States, Between 2010 and 2015, there were an average of 93 hospital
mergers announced per year, substantially more than the annual average of 58
between 2004 and 2009.! Policymakers have challenged several recent pro-
posed mergers with the concer that they would lead to a loss of competition,

* Kathleen F, Easterbrook is a data scientist at Airbnb. Gautam Gowrisankaran is the Arizona
Public Service Professor at the University of Arizona, an Affiliated Professor at HEC Montre:
Research Associate at the National Bureau of Economic Research, and a Senior Advisor at Co
nerstone Research. Dina Older Aguilar, and Yufei Wu are economists at Cornerstone Research.
Gowrisankaran also acknowledges funding from the National nce Foundation (Grant SES-
1425063). The authors did not receive any other compensation for their work on this article. We
thank Kostis Hatzitaskos, Cagatay Kog, Aviv Nevo, Ted Rosenbaum, and Andrea Shepard for
helpful comments. We are grateful to Per Axelson, Yuan Fei, Matt Lord, Stacy Ni, William
Portnof, Josh Rains, and Matthew Rosenthal for excellent research assistance. This article dis-
cusses the proposed merger in United States v. Cabell Huntington Hospital. Gowrisankaran and
staff at Comerstone Research consulted on behalf of the parties in this proposed merger. The
discussion of the parties and their proposed merger is based on public materials and docs not
necessarily reflect the opinions of the parties. This article also discusses the acquisition of Tuc-
son Heart Hospital by Carondelet Health Network. Gowrisankaran has taught executive educa-
tion classes to the senior management at Carondelet Health Network. The discussion of the
parties here is also based on public materials and does not necessarily reflect the opinions of the
parties. In general, the views expressed herein are solely those of the authors, who are responsi-
ble for the content, and do not necessarily represent the views of Cornerstone Research, Airbnb,
or any other entity.

‘Am. Hosp. Ass’n, TrendWatch Chartbook 2016, at 27 Chart 2.9, www.aha.org/tesearch/
reports/tw/chartbook/2016/20 l6chartbook.padf.

501502 ANTITRUST LAW JouRNAL [Vol. 82

by increasing prices and/or reducing quality.? Such concerns are consistent
with certain theoretical economic models, which predict that mergers between
close competitors increase prices,> and with empirical evidence on some hos-
pital mergers.*

Despite this presumption that a merger between two hospitals serving the
same market will reduce competition and increase prices (in the absence of
other effective competition), the empirical evidence on hospital mergers is
mixed. For instance, Christopher Garmon estimates the impact of
pleted hospital mergers on prices, concluding that only nine led to sta
significant price increases.* Deborah Haas-Wilson and Garmon find that the
merger of Evanston and Highland Hospitals in Illinois led to higher prices,
while the merger of two community hospitals in the same area did not.® Price
effects can vary for different insurers even within a single merger.”

Courts have also reached different opinions in their prospective assess-
ments of the potential competitive effects of proposed hospital mergers. For
example, in two recent proposed mergers, the lower courts allowed the merg-
ers to proceed but the appellate courts overturned these decisions.*

2 John J, Miles, Ober Kaler, The FTC’s Three Current Hospital Merger Challenges: Will the
FTC Ever Lose?, Lexoocy (Mar. 29, 2016), www lexology.comflibrary/detail aspx?g=4988dc
2d-£913-4eef-acc7-177fa56d | 124.

4 For example, Steven Berry and Ariel Pakes characterize certain cost and demand conditions
under which mergers can cause prices to increase, though they also describe the limitations of
theoretical models in predicting the price effect of mergers for firms facing nonlinear demand,
economies of scale, or asymmetrically heterogeneous products. Steven Berry & Ariel Pakes,
Some Applications and Limitations of Recent Advances in Empirical Industrial Organization:
Merger Analysis, 83 Am. Econ. Rev. (Papres & Proc.) 247, 248 (1993).

+ In the view of Martin Gaynor, Kate Ho, and Robert Town, “Mergers between rival hospitals
are likely to raise the price of inpatient care and these effects are larger in concentrated markets.
The estimated magnitudes are heterogeneous and differ across market settings, hospitals, and
insurers.” Martin Gaynor, Kate Ho & Robert J. Town, The Industrial Organization of Health-
Care Markets, 53 J. Econ. Lireraturr 235, 262 (2015). See also Cory Capps, David Dranove &
Mark Satterthwaite, Competition and Market Power in Option Demand Markets, 34 RAND J.
Econ. 737 (2003); Gautam Gowrisankaran, Aviv Nevo & Robert Town, Mergers When Prices
Are Negotiated: Evidence from the Hospital Industry, 105 Am. Econ. Rev. 172 (2015).

$ Christopher Garmon, The Accuracy of Hospital Merger Screening Methods, 48 RAND J.
Econ. 1068, 1086 (2017). Six led to statistically significant price decreases, and 13 had no
statistically significant effect on prices.

Deborah Haas-Wilson & Christopher Garmon, Hospital Mergers and Competitive Effects:
Two Retrospective Analyses, 18 Int’ J. Econ. Bus, 17 2011.

? See Aileen Thompson, The Effect of Hospital Mergers on Inpatient Prices: A Case Study of
the New Hanover-Cape Fear Transaction, 18 Ist’. J, Econ. Bus. 91 (2011). The article finds
that afier the Hanover/Cape Fear hospital merger in New England, prices increased for two insur-
ets, remained similar for a third, and decreased for a fourth

8In 2016, in a preliminary injunction hearing, the U.S. District Court for the Northern Dis-
trict of Illinois refused to block the proposed merger between the healthcare systems NorthShore
and Advocate. However, the Seventh Circuit Court of Appeals disagreed with the lower court’s
analysis of the set of relevant competitors and ordered that the case be retried. Similarly, the U.S.
District Court for the Middle District of Pennsylvania initially allowed the Penn State Hershey2018] ACCOUNTING FOR COMPLEMENTARITIES IN HosprraL MERGERS 503

This article advances one reason why different hospital mergers can have
different effects on prices that has not been widely discussed in academic
literature or in merger enforcement: complementarities across hospitals in
their value to insurers. Products that are complements deliver more value
when consumed together than the sum of the values that they deliver alone.
Complementarities are important to merger analysis in general because the
merger of complementary products can lead to price decreases.° For hospitals,
prices are determined by negotiation between hospitals and insurers. Thus, if
two hospitals that are complements to an insurer merge, the insurer may actu-
ally be able to negotiate lower prices with the newly merged entity than it
could when negotiating with the hospitals individually.

Why might hospitals be complements to insurers? A key role that insurers
perform is in the construction of networks of hospitals and other providers.
The attractiveness of a provider network, and the marketability of an insurer’s
plan, will depend on how well that network can deliver the medical care re-
quired by plan enrollees, who often will not know in advance what healthcare
services they and their families will require. For this reason, insurers generally
need a broad network of hospital providers, offering the range of health ser-
vices most frequently needed, in a geographic area to successfully sell insur-
ance products in that area. Therefore, two hospitals can be complements to
insurers when each offers a critical service in a geographic area that the other
does not. In the extreme, if both hospitals are required for a plan’s marketabil-
ity, each hospital may offer no value to the insurer on its own, but together
would create a marketable plan.

Medical Center and PinnacleHealth System merger to proceed, but the Third Circuit Court of
Appeals reversed the decision as it disagreed with the lower court’s economic analysis of the
relevant market, See Am. Mem. Op. & Order, FTC v. Advocate Health Care, No. 15 C 11473
une 20, 2016); Erica Teichert, FTC Wins Appeal to Halt Penn State Hershey/PinnacleHealth
Merger, Mon. Heattucare (Sept, 27, 2016); Erica Teichert, Appeals Court Revives FTC's Bid
to Block Advocaie/NorthShore Merger, Mov. Hratricare (Oct, 31, 2016); Kristin Schorsch,
The Latest Jabs in Advocate, NorthShore Fight to Merge, Mop. Hkautncae (Dec. 16, 2016).

9 See, ¢.g., JEAN TiRoLe, THe THEORY OF INDUSTRIAL ORGANIZATION 70 (1988) (“T]he mo-
nopoly producers of complementary goods have incentive to integrate (horizontally) in order to
avoid double marginalization and an excessive demand contraction.” Id. at 175); Matteo Alvisi,
Emanuela Carbonara & Francesco Parisi, Separating Complements: The Effects of Competition
and Quality Leadership, 103 J. Econ. 107, 108 (2011) (“In fact, when complementary goods are
sold by different firms, prices are higher than those set by a monopoly selling all the complemen-
tary goods. A merger would then yield a higher consumer surplus.”); Aviv Nevo, Remarks as
Prepared for the Stanford Institute for Economic Policy Research and Comerstone Research
Conference on Antitrust in Highly Innovative Industries: Mergers that Increase Bargaining Lev-
erage 4-5 (2014) (“[I]f the [goods are complements] then bargaining separately the providers
would get... more than bargaining jointly. This might seem surprising, but it is just the counter-
part of two complements merging in a price setting framework.”)504 ANTITRUST LAW JouRNAL [Vol. 82

I. A FRAMEWORK FOR ANALYZING HOSPITAL COMPETITION

To explain how complementarities may impact hospital prices and the ef-
fect of mergers, it is first helpful to define a framework for how hospitals
compete. We present a framework in which competition takes place in three
stages."° This framework builds on models of competition used in policy set-
tings and academic studies of hospital and insurer bargaining, often with the
goal of understanding the price impacts of hospital mergers.''

1) In the first stage, hospitals negotiate with insurers over network inclu-
sion and reimbursement levels. Insure! so determine enrollee cost
sharing (in the form of copays or coinsurance) for care received at
different hospitals, which will typically be lower at in-network hospi-
tals. It is in this stage of competition that insurers and hospitals negoti-
ate prices and hence in this stage that complementarities could affect
prices. When modeling these negotiations, researchers commonly as-
sume that an insurer and a hospital will split the marginal surplus'? of
each negotiation in some fixed proportion."

2) In the second stage of competition, insurers set premiums that individ-
uals pay to enroll in their plans, and individuals select insurance plans
after observing each plan’s hospital network, cost-sharing arrange-
ments, and premiums.

3) In the third stage of competition, some enrollees require hospital treat-
ment and select a hospital. Patients choose among available hospitals
taking into account the network status of each hospital. the copay or

© This model can be described in various numbers of stages. Katherine Ho uses five stages to
describe this model, and Katherine Ho and Robin Lee use four stages, in which our first stage is
divided into two subparts. Compare Katherine Ho, Insurer-Provider Networks in the Medical
Care Market, 99 Am. Econ. Rev. 393 (2009), with Kate Ho & Robin S. Lee, Insurer Competi-
tion in Health Care Markers, 83 Economereica 379 (2017). Similarly, related models may
abstract away from the second stage listed here. See also Gowrisankaran, Nevo & Town, supra
note 4.

"Initial Decision, ProMedica Health Sys., FTC Docket No, 9346 (Dec. 12, 2011), www.fte
govsites/defaulvfiles/documents/cases/2012/0 1/12010Spromedicadecision.pdf: Capps, Dranove
& Satterthwaite, supra note 4; Gowrisankaran, Nevo & Town, supra note 4; Ho, supra note 10;
Ho & Lee, supra note 10; Robert Town & Greg Vistnes, Hospital Competition in HMO Net-
works, 20 J. HEALTH Econ. 733 (2001).

2 By “marginal surplus,” we mean the value generated by forming an agreement with the
counterparty, assuming that all other agreements have formed.

°S This formulation is a generalization of the Nash bargaining solution where the model in-
cludes interrelated bargains between multiple pairs of To our knowledge, this generaliza~
tion was first used by Henrik Horn and Asher Wolinsky. Henrik Horn & Asher Wolinsky.
Bilateral Monopolies and Incentives for Merger, 19 RAND J. Econ. 408 (1988); More recently,
Allan Collard-Wexler, Gautam Gowrisankaran, and Robin Lee showed that, under some condi-
tions, negotiations that are modeled as taking place through a series of simultaneous alternating
offers can generate these “Nash-in-Nash” payoffs. Allan Collard-Wexler, Gautam
Gowrisankaran & Robin S. Lee, “Nash-in-Nash” Bargaining: A Microfoundation for Applied
Work, 127 J. Pou. Econ. (forthcoming Feb. 2019).

nm2018] ACCOUNTING FOR COMPLEMENTARITIES IN HosprraL MERGERS 505

coinsurance amounts (in cases where there are meaningful differences
in these amounts across hospitals), a hospital’s location, and its quality
for treating particular conditions, among other potential factors.

We will refer to these three stages of competition as we discuss how hospi-
tals can be complements and how complementarities can affect negotiated
hospital prices.

Tl. HOSPITAL COMPLEMENTARITIES AND THE
IMPACT OF MERGERS

In this section, we first review the economic definition of complements. We
then discuss circumstances under which hospitals can be complements from
the perspective of an insurer negotiating over network inclusion and reim-
bursement levels at the first stage of competition. We then analyze how com-
plementarities can result in hospital mergers lowering prices.

A. Wuat Do We MEAN By ComPLEMENTS?

Formally, two products are complements when they provide more value
together than the sum of the values that each product provides on its own, An
example of two products that are complements is a left shoe and a right shoe.
Individuals usually derive more value from their left shoes because they also
have matching right shoes; one shoe on its own is not useful in most cireum-
stances. In comparison, two products are substitutes when they provide less
value together than the sum of the value that each product provides on its
own. An example of two substitute products is butter and margarine. Consum-
ers who have already chosen to buy butter will derive less value from buying
margarine, since butter can frequently take the place of margarine.

B. How Can Hosprrats Be CompLeMENTS TO INSURERS?

It is easy to envision how two hospitals can be substitutes for one another at
each stage of competition. Consider an extreme case where two hospitals, A
and B, offer the same services and amenities, the same quality of care, and are
also located right next to each other. From the perspective of a patient seeking
care at the third stage of competition, the two hospitals are interchangeable.
For an employer or an individual purchasing health insurance at the second
stage of competition, a plan with either hospital is equally valuable as a plan
with both, Finally, from the insurer’s perspective at the first stage of competi-
tion, a plan that included either hospital A or B would be more marketable
and increase profits relative to a plan without either hospital in-network, but a
plan that included both A and B would be no more valuable than a plan that
included just one. Enrollees (or employers) would not be willing to pay any
more for plans that included the second, perfectly interchangeable hospital,506 ANTITRUST LAW JouRNAL [Vol. 82

which means that the addition of the second hospital to a network that already
included one would add no value to the insurer. In this case, hospitals A and B
are perfect substitutes.

More generally, most research and policy applications assume, through
their modeling framework, that hospitals are substitutes (albeit imperfect
ones) rather than complements at the first stage of competition." This derives
from the assumption in these applications that (1) a hospital’s value to an
insurer at the first stage of competition is proportional to the average value of
the hospital to patients at the third stage of competition (where the average is
taken over all possible health conditions), that is, insurers’ preferences and
patients’ preferences are perfectly aligned; and (2) patients must select a sin-
gle hospital for treating each condition at the third stage of competition. To-
gether, these assumptions imply that hospitals are at least weak substitutes;
they rule out any possibility that having two hospitals in-network together
would add more value to the insurer than the values the insurer receives from
having either hospital in-network alone. This typical empirical model simply
does not allow for complementarities in the first stage; and this is true even if
the hospitals have differences in services, quality, and location—unlike our
extreme example above.

Despite this feature of the standard empirical model used to predict merger
effects, it is possible for real-world hospitals to be complements in the first
stage of competition, where prices are negotiated. By definition, two hospitals
are complements at this stage if the value to an insurer from including both
hospitals is higher than the sum of the values of including either hospital indi-
vidually. That is, they are complements if an insurer that includes the two
hospitals in-network will be able to add more profits than the sum of the
profits it would add from including just one or the other hospital in its net-
work. Two hospitals would have this effect on insurer profits if their joint
inclusion increases the marketability of the insurer’s health plans more than
the sum of what either one would add alone, which might allow the insurer to
attract more enrollees and/or charge higher premiums. Because the impact of
additional hospitals on plan marketability and profitability is not necessarily
proportional to the average patient value at the third stage, the value that a

4 See Kate Ho & Robin S. Lee, Insurer Competition in Health Care Markets 7-8 (Nat'l
Bureau of Econ., Working Paper 19401, June 2015), www.aeaweb.org/conference/201 6/retrieve
-php%pdfid=444 (“Relative bargaining leverage between {hospitals and insurers] depends cru-
cially on consumer demand and the extent to which other insurers and/or hospitals in the market
are good substitutes for the bargaining firms.”). Garmon discusses different methods to estimate
post-merger price effects that build upon patient choice models, including analyses of diversion
ratios, estimation of the change in willingness-to-pay, and merger simulations. Garmon, supra
note 5. Both articles implicitly assume that hospitals are substitutes through its reliance on pa-
tient choice models in which a single hospital must be chosen—such choice models are incapa-
ble of concluding that hospitals are complements.2018] ACCOUNTING FoR COMPLEMENTARITIES IN HosprraL MERGERS 507

hospital brings to an insurer from being in-network at the first stage of compe-
tition is not always perfectly aligned with the average value that it brings to
patients from being in-network at the third stage. Thus, the value generated to
plans cannot be derived by simply aggregating patients’ values for hospitals 2
the third stage of competition, as the standard model does.

To understand how complementarities might occur, and how they may cre-
ate a disconnect between value at the insurer stage of competition and the
aggregate value across patients seeking care, consider a second extreme ex-
ample where there are again only two hospitals in an area—C and D. In this
example, hospital C provides treatment for half of all diseases and hospital D
provides treatment for the other half. These two hospitals have no overlap in
services.

Considering these hospitals first from the perspective of a patient seeking
care at the third stage of competition, these two hospitals are unlikely to be
complements, even in this extreme case. Patients typically select one hospital
at which to receive treatment for a specific episode of care; finding hospitals
that cover all medical services is not necessary at that decision point. Comple-
mentarity between products simply cannot occur when consumers must make
a single discrete choice across products.'* Thus, even in this extreme example,
for the patient seeking care, hospitals C and D are neither substitutes nor com-
plements in the third stage of competition.

However, these same two hospitals C and D may very well be complements
at the second stage of competition for most employers and enrollees. Because
employers are likely to purchase health plans to meet the heterogeneous
health needs of all of their employees,'® and because different employees are
sure to need services from both hospitals C and D, a plan with only hospital C
or D may offer very limited value compared to a plan with both hospitals in-
network. This would then generate complementarities across the hospitals to
employers selecting an insurance plan to offer their employees. This is also
likely true for individuals purchasing plans, as they also do not know which
diseases they may contract and thus will want to purchase a single health plan
that covers a broad array of healthcare needs. Hence, purchasing a plan that
offers both hospitals in-network would add more value for many purchasers

°S Iris still possible that hospitals could be complements at the third stage. Imagine one hospi-
tal that has a surgery center, but no rehabilitation unit, and a rehabilitation hospital that is across
the street, A patient deciding where to get surgery may be more likely to choose this hospital for
surgery, knowing there is a rehabilitation unit nearby.

‘© John R. Moran, Michael E. Chernew & Richard A. Hirth, Preference Diversity and the
Breadth of Employee Health Insurance Options, 36 Heacrit Servs. Res. 911 (2001).508 ANTITRUST LAW JouRNAL [Vol. 82

of insurance than the sum of the value of purchasing only a plan that offers
hospital C and the value of purchasing only a plan that offers hospital D.'7

Complementarity between hospitals at the second stage of competition in
turn drives complementarity at the first stage. If employers and individuals do
not derive much value from a plan with only one of the hospitals in-network,
then an insurer could not successfully market a plan to employers without
both hospitals C and D in-network. Thus, one hospital on its own would bring
little value to the insurer since either C or D alone would not yield a marketa-
ble plan. The two together, however, would create a plan that the insurer
could profitably sell, and thus are complements from the point of view of the
insurer in the first stage of competition. Because this first stage of competition
is the stage at which insurers negotiate prices with hospitals, it is complemen-
tarities at this stage of competition (and not at the third) that matter when
determining whether a hospital merger would increase prices.

In practice, we rarely, if ever, observe cases as extreme as the above styl-
ized example. More realistically, hospitals are likely to be complements at the
first two stages of competition when each hospital offers some critical spe-
cialty services that the other does not, but also offers some overlapping ser-
vices such that they are substitutes for some patients in the third stage of
competition. Insurers could still have a hard time marketing a plan that did not
cover specialty services that employers and enrollees value. Even if the plan
met the healthcare needs of most employees, most of the time, an employer
would be reluctant to choose the plan if it omitted specialty services that a
significant share of employees were likely to need.

Suppose now that, instead of having no overlap, hospital C offered pediat-
ric care and hospital D offered oncology services. Further suppose that both
offer all other services, which we will call general acute care (GAC) services,
and are completely interchangeable for these services. At the third stage of
competition, hospitals C and D would now be substitutes overall, based on
their substitutability for patients in need of GAC services, though for individ-
ual patients seeking either pediatric or oncology care they would be neither
substitutes nor complements since only one hospital is a realistic option. Be-
cause pediatrics and oncology are both key services, which many employers
and individuals would want in-network, these hospitals would still likely be

" Purchasing two separate single-hospital health insurance plans, each at a fraction of the
premium that would be charged for a single plan with both hospitals, would be unlikely to
deliver the same value as the single unified plan, First, if some individuals choose to buy only
one of the two plans, depending on their specific health risks, health expenditures per enrollee
may increase, driving up premiums for the individual plans and raising the cost of the two plans
for those who wish to purchase insurance for both hospitals. Even if the two plans were pur-
chased together for all enrollees, the increased administrative burden on insurers, plan sponsors,
and enrollees would make a two-plan option more costly than a single plan.2018] ACCOUNTING FOR COMPLEMENTARITIES IN HosprraL MERGERS 509

complements from the perspective of insurers forming a comprehensive net-
work and enrollees selecting plans.

The key question for determining whether hospitals C and D are, on bal-
ance, complements to insurers—even though they may be substitutes for
many patient services—is whether the insurer’s profits from including both
hospitals in-network exceed the sum of the profits from including just one
hospital in-network. For this to happen, access to both service lines must be
important enough to employers and prospective enrollees that insurers can
gain significantly more enrollees and/or charge sufficiently higher premiums
when both hospitals are in-network than they can when only one or the other
is in-network.

The presence of complementarity of hospitals to insurers would result in a
number of testable implications for the pricing and market behavior of plan
sponsors, insurers, and hospitals. In particular, suppose we see a market with
two hospitals. If they are complements, employers may explicitly refuse to
purchase plans that do not include both hospitals in their network. Insurers
would also be expected to view both hospitals as “must haves” and not use
one as leverage against the other in negotiations. Deviations from this expec-
tation should be driven by circumstances that can be understood as an excep-
tion to the complementarities, such as plans marketed to specific populations
that would not expect to need the services of one or the other hospital, or
plans with access to regulated prices in the absence of a contract (e.g., Medi-
care Advantage plans).'* Hospitals would be expected to leverage their status
as complements in negotiations with insurers, and prices may be higher than
otherwise expected given predictions based on the standard model (which as-
sumes that the value of the hospital to the insurer at the first stage of competi-
tion is proportional to the average values to the patients at the third stage of
competition). Hospitals may also seek to confirm the inclusion of their com-
plement in an insurer’s network before agreeing to negotiate their own inclu-
sion in a new plan. Finally, in cases where one complementary hospital is
excluded, we would expect to see plans that have only minimal enrollment.

C. MerGers or Comp_ements May Repuce Prices

In general, the merger of firms that offer complementary products can lead
to lower rather than higher prices. In the case of two complementary hospitals,
an insurer may be able to negotiate /ower prices if two hospitals that are com-

plements at the first stage of competition merge. As we detail below, this is

'S Medicare Advantage plans have been shown to pay very similar prices to hospitals as Tradi-
tional Medicare. See Vilsa Curto et al., Healthcare Spending and Utilization in Public and Pri-
vate Medicare (Nat'l Bureau of Econ. Research, Working Paper No. 23090, 2017), www.nber
org/papers/w23090.510 ANTITRUST LAW JouRNAL [Vol. 82

because the post-merger entity has less negotiating leverage with the insurer
than the total amount of the leverage that each hospital had separately.

Independently, two hospitals that are complements at the first stage of com-
petition could each threaten to remove itself from the insurer’s network.!°
Each hospital’s leverage when negotiating with insurers is then based on the
value of completing the network and delivering the additional value from the
combination of the two hospitals. Because the two hospitals are complements,
this value, computed when the other hospital is already in the insurer’s net-
work, is higher than the value that the hospital would bring without the other
hospital in the network. In other words, each hospital adds more value when
the other hospital is in the network then it would add when the other hospital
is not in the network, and can negotiate on this basis. Hence, a hospital’s
bargaining leverage is increased by having a complementary hospital in
network.

However, if two complementary hospitals merge, the combined system
could not threaten to remove the complementary pair twice. The worst threat
that the combined system could make would be to not contract with the in-
surer at all, which would lower the value of the provider network to the in-
surer, but not by twice the amount that removing each hospital independently
would. To summarize, since the entire hospital system generates less value
than the sum of each individual hospital added under the assumption that the
other hospital is already in-network, the combined system has less leverage in
negotiations with the insurer than the combined leverage of the two hospitals
negotiating separately.

To understand the intuition of how a merger of complementary hospitals
might result in price decreases, we turn to numerical examples. We first con-
sider the case of mergers between hospitals which are substitutes to insurers
and then consider the case of mergers between hospitals which are comple-
ments to insurers.

Table 1 walks through a numerical example of the substitutes case. Suppose
now that our two substitute hospitals, A and B, are located on opposite sides
of a small city, rather than being indistinguishable from each other. Suppose
that an insurer places a value of 10 on a hospital network that includes both
hospitals A and B, based on the number of customers it could enroll and the
premiums it could charge, and a value of 0 on a network with no hospitals.
The insurer needs at least one hospital to have a marketable plan. The second
hospital is still valuable, since individuals who live near that hospital are more
likely to enroll in the plan if it is included in the network, but it is less valua-

© We focus on the first stage of competition since this is the stage that matters for the determi-
nation of prices.2018] ACCOUNTING FoR COMPLEMENTARITIES IN HosprraAL MERGERS 511

ble than the first hospital to the insurer because individuals could use the
hospital across town in the case they choose a plan with only one of the two
hospitals in-network. Consistent with these facts, suppose that the insurer
places a value of 6 on a network with just one hospital, either A or B. In this
case, the marginal value of the second hospital is 4 (= 10 — 6) and hence less
than the marginal value of the first hospital. Based on these values, hospitals
A and B are substitutes to insurers—the value of having both of them in the
network (10) is less the sum of the values of having only A or only B in the
network (6 + 6 = 12).

What prices would the parties negotiate in this situation? As noted above,
the marginal value of each hospital in this example is 4. Assuming the hospi-
tal and insurer evenly split this value, as is commonly done,” each hospital
receives 2 and the insurer pays a total of 4 to both hospitals when they are
separate entities.

In this example, the negotiated price would increase if the hospitals
merged. As mentioned, together the hospitals are worth 10 to an insurer.
Without either hospital, the insurer earns 0, so the value of reaching an agree-
ment is 10 (= 10 — 0). The insurer and the merged hospital system will split
the surplus of 10 evenly, and the hospital will receive 5. Hence, this stylized
example suggests that if hospitals A and B merged, total payments to the
hospital would rise from 4 to 5.7!

TABLE 1: PRICES WHEN HOSPITALS ARE SUBSTITUTES
Separate Hospitals __ Merged Entity

Hospitals

Hospital A Hospi AandB
1. Network Value without Hospital(s) 6 6 0
2. Network Value with Hospital(s) 10 10 10
3. Surplus from Network Inclusion (= [2] - [1]) 4 4 10
4. Hospital Price BV) 2 2 5
5. Total Paid to Hospitals, When Separate 24254 5

2 The intuition remains the same if we assume that hospitals capture a higher proportion of
the surplus than insurers. John Brooks, Avi Dor, and Herbert Wong estimate that 68% of the
surplus goes to hospitals; Kate Ho and Robin Lee estimate that about three-quarters of the sur-
plus goes to hospitals. See John M. Brooks, Avi Dor & Herbert S. Wong. Hospital-Insurer
Bargaining: An Empirical Investigation of Appendectomy Pricing, 16 J. Hrauta Econ. 417
(1997); Ho & Lee, supra note 10.

4 For simplicity, this analysis exposits price as the total revenue paid to the hospital and not
the per-person revenue. The number of patients from the insurer may also change based on
which hospitals are in-network, implying that an analysis of per-person revenues would be more
involved. Nonetheless, the same basic points, that mergers of substitutes can raise prices and
mergers of complements can lower prices, apply when considering per-person revenues as well512 ANTITRUST LAW JouRNAL [Vol. 82

Next, Table 2 walks through a numerical example of the case when hospi-
tals are complements to an insurer. Consider again the example where only
hospital C has pediatric care and only hospital D offers oncology services. As
discussed above, these hospitals are very likely to be complements for insur-
ers, though not for particular patients.

Suppose that the insurer places a value of 4 on a network with only hospital
C or D, since it would have to charge a lower premium and would enroll
fewer individuals. Based on these values, hospitals C and D are complements
for the insurer—the value of having both of them in the network (10) exceeds
the sum of the values of having only C or only D in the network (4 + 4 = 8).

As above, the insurer and hospital split this surplus evenly, so each hospital
receives a price of 3, and the insurance company pays the hospitals a total of
6. If these hospitals merged and everything else remained the same, the in-
surer would value an agreement with the system at 10 (= 10 - 0), and the
hospitals together would only be able to negotiate a price of 5. In this example
of complementary hospitals, prices would have decreased from 6 to 5 as a
result of the merger.??

TABLE 2: PRICES WHEN HOSPITALS ARE COMPLEMENTS

Separate Hospitals __ Merged Ent
Hospital C HospitalD Hospitals

(pediatric) (oncology) Cand D
1. Network Value without Hospital(s) 4 4 0
2. Network Value with Hospital(s) 10 10 10
3. Surplus from Network Inclusion (2)-D 6 6 10
4, Hospital Price 12) 3 3 3
5. Total Paid to Hospitals 343=6 5

Outside of stylized examples like the ones presented in Tables | and 2.
many factors will influence the overall effect of a merger on price, quality,
and other outcomes. Insurer concentration, interactions with hospitals not a
party to the merger, operational efficiencies, and other factors could all affect
post-merger outcomes. In some instances, these and other factors may out-
weigh a reduction in bargaining leverage due to the merger of complements,
or in other instances, may lead to lower post-merger prices separate from hos-

2 Gregory Vistnes and Yianis Sarafidis, and Craig Peters, provide similar examples to this
and also make the point that if the sum of the surpluses to the insurer from having each hospital
is less than the surplus from both hospitals, then a hospital merger will lower prices. However,
neither article explicitly considers why hospitals may be complements to insurers and, indeed,
Vistnes & Sarafidis state that this is a “perhaps unlikely situation.” Gregory S. Vistnes & Yianis
Sarafidis, Cross-Market Hospital Mergers: A Holistic Approach, 79 Antitrust LJ. 253,
272-73 (2013): Craig T. Peters, Bargaining Power and the Effects of Joint Negotiation: The
ecapture Effect” 8-9 (Econ. Analysis Grp. Discussion Paper, EAG 14-3, 2014),2018] ACCOUNTING FOR COMPLEMENTARITIES IN HosprraL MERGERS 513

pital complementarities. For example, Haas-Wilson and Garmon found that
the merger between Victory Memorial Hospital and Provena St. Therese Med-
ical Center did not raise prices.* Neither Haas-Wilson and Garmon or the
FTC—in its statement closing the investigation of this merger—mentioned
complementarities as a potential cause for the lack of a price increase.** In-
stead, the FTC statement is consistent with a better management team at the
merged hospital realizing that revenues would be increased with lower qual-
ity-adjusted prices due to increased patient volume or with the merged hospi-
tal system capturing a smaller part of the surplus than did the hospitals
individually.?>

However, the general point remains: changes in bargaining leverage will
tend to increase the prices hospitals are able to negotiate with insurers for
hospitals that are substitutes and reduce prices for hospitals that are comple-
ments to an insurer building a provider network. Even when post-merger
prices are expected to fall as a result of decreased bargaining leverage, hospi-
tals that are complements may still wish to merge for a number of reasons.
For example, there may be offsetting cost efficiencies to the merger that
would lower the cost of delivering care for most patients, including Medicare
and Medicaid, while the reduction in price would apply only to commercial
patients. Or, one of the merging parties may not be independently financially
viable, and the degradation or loss of its services may reduce not only the
value that that hospital offers to the network, but also the value that the finan-
cially strong hospital offers as a part of the complementary pair. As another
example, a merged hospital offering a wide range of services might be more
attractive to patients and expand the downstream market, increasing value
through higher volume. In addition, as payment systems move toward value-
based care models, hospitals with differentiated services may be better able to
manage the risk of such contracting or better coordinate patients’ care as a
merged entity.

Tl. WHEN DIFFERENTIATED SERVICES LEAD TO
COMPLEMENTARITIES

As discussed in Part II, hospitals with differentiated services may be com-

plements for insurers. However, differentiation is not sufficient for two hospi-
tals to be complements, and hospitals that focus on different services can still

® See Haas-Wilson & Garmon, supra note 6, at 18.
% Public Statement, Victory Mem’l Hosp./Provena St. Therese Med. Ctr., FTC Docket No
O11 0225 Wuly 1, 2004), www.fte.gov/sites/default/files/documents/closing_letters/vista-health-
acquisition-provena-st.therese-medical-center/040630ftcstatement01 10225.pdf.
25The FTC stated: [There is . . . some evidence . . . that St. Therese was pursuing a non-
(There is also evidence that St. Therese and Victory were steadily
losing market share to their rivals prior to the merger.” Hd. at 1.514 ANTITRUST LAW JouRNAL [Vol. 82

be substitutes to an insurer in forming a provider network. A key predictor of
whether two hospitals are complements is whether each hospital offers an
important service that the other does not and that other comparable in-network
hospitals—with similar quality and location—do not offer.** In this case, both
hospitals may be needed to form a complete network, particularly if the ser-
vices are important enough that enrollees or the insurer could not simply pay
out-of-network costs for these services. Under these conditions, insurers will
likely not be able to offer a marketable plan without a complete network,
which will limit the value of including only one of the hospitals, in turn gener-
ating complementarities.

Hospitals that focus on different services will not be complements for in-
surers, at least not as a result of the differentiation, if one hospital offers all
the services offered at the other. For example, a specialty hospital, such as a
cardiac hospital, will not be a complement from the point of view of insurers
to a GAC hospital if the GAC hospital also offers comprehensive cardiac care
of similar quality. This example is not just hypothetical. MedCath, a hospital
chain that specialized in cardiac treatments, had lower cost and better out-
comes than other hospitals, according to a Harvard Business School case
study.” Yet, managed care organizations did not contract with MedCath be-
cause of their long-standing relationships with GAC hospitals.’ Local hospi-
tals resisted MedCath’s inclusion in insurer networks, with the common
criticism that “MedCath was doing nothing more than ‘skimming’ hospitals’
most profitable line of business, cardiac care . . . .””°

MedCath hospitals were substitutable with the cardiac services offered at
other GAC hospitals, and insurers could and did exclude MedCath hospitals
from their networks.” For instance, one former MedCath hospital, Tucson
Heart Hospital, offered open-heart surgeries and a variety of cardiac ser-
vices.>! Other nearby hospitals in Tucson, Arizona,