journal of financial economicsamalenko/lbos.pdfjournal of financial economics 117 (2015) 607–627...

Contents lists available at ScienceDirect

Journal of Financial Economics

Journal of Financial Economics 117 (2015) 607–627

http://d0304-40

☆ We(discussHoldernKovner,Francischan, JerJessica Ythank sconfereing, theInstituteSchool,20th An

n CorrE-m

journal homepage: www.elsevier.com/locate/jfec

A theory of LBO activity based on repeateddebt-equity conflicts$

Andrey Malenko a,n, Nadya Malenko b

a MIT Sloan School of Management, Cambridge, MA 02142, USAb Boston College, Carroll School of Management, Chestnut Hill, MA 02467, USA

a r t i c l e i n f o

Article history:Received 30 January 2014Received in revised form2 September 2014Accepted 18 December 2014Available online 23 June 2015

JEL classification:G24G32G34D44

Keywords:Leveraged buyoutsPrivate equityReputationDebt-equity conflictsClub deals

x.doi.org/10.1016/j.jfineco.2015.06.0075X/& 2015 Elsevier B.V. All rights reserved.

thank the anonymous referee, Nittai Beant), David Chapman, Sergei Davydenko, Dess, Edith Hotchkiss, Oguzhan Karakas, DChristian Laux (discussant), Doron Levit,o Perez-Gonzalez, Jeffrey Pontiff, Antoinetteome Taillard, Vladimir Vladimirov, Mark Wesang, and Bilge Yilmaz (discussant) for helpfueminar participants at Boston College andnce participants at the 2014 Western Financ2014 European Finance Association meetof Private Equity Findings Symposium

the 15th Texas Finance Festival, and the Nniversary Conference for useful discussions.esponding author. Tel.: þ1 617 225 9301.ail address: [email protected] (A. Malenko)

a b s t r a c t

We develop a theory of leveraged buyout (LBO) activity based on two elements: the abilityof private equity-owned firms to borrow against their sponsors' reputation with creditorsand externalities in sponsors' reputations due to competition and club formation. Inequilibrium, the two sources of value creation in LBOs, operational improvements andfinancing, are complements. Moreover, sponsors that never add operational value cannotadd value through financing either. Club deals are beneficial ex post by allowing low-reputation bidders with high valuations to borrow reputation from high-reputationbidders with low valuations, but they can destroy value by reducing bidders' investmentin reputation. Unlike leverage of independent firms, driven only by firm-specific factors,buyout leverage is driven by economy-wide and sponsor-specific factors.

& 2015 Elsevier B.V. All rights reserved.

1. Introduction

Leveraged buyouts (LBOs) have been an importantelement of the merger and acquisition market over the

rgman, Philip Bondenis Gromb, Cliffordarren Kisgen, AnnaDmitriy Muravyev,Schoar, Philip Stra-terfield (discussant),l comments. We alsoDuke University ande Association meet-ing, the Sixth Collerat London Businessew Economic School

.

last three decades. Value creation in LBO transactions isgenerally attributed to two sources: operational improve-ments and the benefits of higher leverage involving taxshields and improved management incentives. However,private equity (PE) firms are sometimes accused of doingnothing but levering up their portfolio companies.1 Buyoutactivity has followed a boom-and-bust pattern and has

1 For example, as one former investment banker put it, “Privateequity is nothing more than incredibly brilliant financial engineering”(Gary Rivlin, Daily Beast, February 21, 2012). The academic literatureprovides evidence consistent with a positive effect of LBOs on tax benefitsfrom increased leverage (e.g., Guo, Hotchkiss, and Song, 2011; Cohn,Mills, and Towery, 2014), operating performance (e.g., Kaplan, 1989;Lichtenberg and Siegel, 1990; Boucly, Sraer, and Thesmar 2011; Acharya,Gottschalg, Hahn, and Kehoe, 2013), monitoring of the management (e.g.,Cornelli and Karakas, 2014), and innovative activity (Lerner, Sorensen,and Strömberg, 2011). See Cumming, Siegel, and Wright (2007) andKaplan and Strömberg (2009) for reviews.

www.sciencedirect.com/science/journal/0304405Xwww.elsevier.com/locate/jfechttp://dx.doi.org/10.1016/j.jfineco.2015.06.007http://dx.doi.org/10.1016/j.jfineco.2015.06.007http://dx.doi.org/10.1016/j.jfineco.2015.06.007http://crossmark.crossref.org/dialog/?doi=10.1016/j.jfineco.2015.06.007&domain=pdfhttp://crossmark.crossref.org/dialog/?doi=10.1016/j.jfineco.2015.06.007&domain=pdfhttp://crossmark.crossref.org/dialog/?doi=10.1016/j.jfineco.2015.06.007&domain=pdfmailto:[email protected]://dx.doi.org/10.1016/j.jfineco.2015.06.007

A. Malenko, N. Malenko / Journal of Financial Economics 117 (2015) 607–627608

been negatively related to aggregate credit spreads and themarket risk premium.2 Leverage in LBO transactions hasalso varied substantially over time and, unlike leverage ofpublic firms, has been strongly driven by economy-widefactors (Axelson, Jenkinson, Strömberg, and Weisbach,2013) and PE sponsor characteristics (Demiroglu andJames, 2010). The industry is characterized by a high levelof competition and frequent formation of clubs, in whichtwo or more PE firms jointly bid for the target.3

Motivated by these stylized facts, this paper develops atheory of buyout activity that analyzes the following threesets of questions. First, how is the PE sponsor's ability toadd value through financing decisions related to its abilityto add value through operational changes? Are these twosources of value creation complements or substitutes, andcan it be that PE firms add value only through financingdecisions without making any operational improvements,as is sometimes claimed? Second, what determines clubformation, and what are the positive and negative effectsof club deals on the value created in LBO transactions?Third, what factors determine aggregate LBO activity,buyout leverage, and other deal characteristics? In parti-cular, why do economy-wide factors have a much strongereffect on buyout leverage than on leverage of public firms,and how do these factors affect the composition of win-ning acquirers and PE firms' payoffs?

Our model is based on two key elements: the idea thatPE-owned firms borrow against their sponsors' reputa-tional capital and externalities between PE sponsors'reputations. Specifically, while debt is beneficial for taxand incentive reasons, it creates a conflict between share-holders and debtholders, which is especially pronouncedwhen the firm is in financial distress (e.g., Jensen andMeckling, 1976; Myers, 1977). One prominent example ofsuch conflicts of interest is dividend payouts, often accom-panied by the issuance of additional debt and leading torating downgrades.4 In 2012 alone, PE-owned firms bor-rowed a record $64 billion to fund dividends, and thisamount increased further in 2013 (Wall Street Journal,2013). PE firms have also been accused of delaying efficientliquidation of portfolio firms (a form of asset substitution)and of walking away from financially distressed firms attimes when an additional investment of capital and effort

2 See Haddad, Loualiche, and Plosser (2014), Kaplan and Strömberg(2009), and Ljungqvist, Richardson, and Wolfenzon (2008).

3 Gorbenko and Malenko (2014b) show that an average auction wonby a financial bidder involves five or more other financial bidders thatsigned confidentiality agreements. Officer, Ozbas, and Sensoy (2010) andBoone and Mulherin (2011) show that almost half of PE deals in recentyears entailed a consortium of several PE firms.

4 According to Harry Resis, a veteran portfolio manager who boughtthe bonds that were used to finance the buyout of Nalco Co. and werelater downgraded following a dividend recapitalization, “The sponsorsreward themselves with a dividend, and we got rewarded with a down-grade on our holdings” (Institutional Investor, 2004). In a more recentexample, the creditors of bankrupt Mervyn's sued Mervyn's PE ownersfor stripping Mervyn's of valuable real estate assets, while payingthemselves hundreds of millions of dollars in management fees anddividends. See “Buyout firms pay $166 mln to end suit over Mervyn'ssale” (Reuters, October 8, 2012). Other examples include lawsuits fordividend payments against the PE owners of Powermate Corporation in2008, Refco in 2007, and KB Toys in 2005.

would allow a more efficient resolution of distress (a formof debt overhang). A recent example involving multiplealleged debt-equity conflicts is the bankruptcy of EnergyFuture Holdings, an electric utility company formed in thelargest LBO in history. The creditors sued the companyalleging pervasive conflicts of interest, including exces-sively high PE firms' fees, inefficient investment to avoidbreaking up the company, delaying restructuring, and lackof diversification.5

Despite these conflicts of interest, PE firms are able tosignificantly lever up the companies they acquire. The firstkey element of our model is the idea that such highleverage is possible because, for a given leverage, debt-equity conflicts are less severe when the firm is owned bya PE sponsor than by dispersed public shareholdersbecause a PE-owned firm can use more assets to borrowagainst. If a firm is publicly owned, it can borrow onlyagainst its own assets. In contrast, a PE-owned firmborrows both against its own assets and against the PEsponsor's reputational capital with creditors. Because, inthe future, the sponsor will have other portfolio companiesthat will raise debt financing, it cares about the payoff ofcreditors in the current portfolio company, and this canpartly alleviate the debt-equity conflict of interest. PEsponsors believe that “if they get a reputation for over-leveraging portfolio companies and then leaving them totheir fate, they won't be able to execute new deals”(Institutional Investor, 2004). Consistent with this view,Moody's has been tracking PE sponsors' aggressiveness inpaying large dividends and incorporating it into its creditratings on LBO deals (Moody's, 2008).6

The second key element of our model is externalitiesbetween PE firms' reputations with creditors. Theseexternalities arise because PE firms compete for targetsand because they can form clubs. In the model, there is alarge market of long-lived PE firms and targets that areavailable every period. Each target can potentially increasein value from the buyout. This added value comes fromtwo sources: financing decisions and operational changes.Each target is matched to two potential acquirers, PE firms,that compete in an auction. We later allow PE firms toform a club and bid jointly. PE firms can differ in the skillsof their general partners (GPs): A high-skill firm is morelikely to create a higher operational value. After thebidding stage, the acquirer chooses how much to lever

5 Energy Future Holdings filed for Chapter 11 on April 29, 2014.According to the petition filed by creditors, the company “wasted nearly ayear and many hundreds of millions of dollars pursuing their doomedProject Olympus,” “continued its ‘amend and pretend’ campaign, whilecontinuing to fail to address the underlying problems in the Debtors'businesses,” and “lost opportunities to diversify its generation fleet,implement efficiency changes, and otherwise take steps to optimizeperformance and revenues.”

6 Existing empirical evidence also supports the notion that the PEsponsor's identity and reputation affect the cost of debt financing and thesponsor's post-buyout behavior (Demiroglu and James, 2010; Ivashinaand Kovner, 2011; Hotchkiss, Smith, and Strömberg, 2014; Huang, Ritter,and Zhang, 2014). For example, Hotchkiss, Smith, and Strömberg (2014)show that financially distressed firms backed by a PE firm are more likelyto receive a capital injection from their owners than firms without aPE owner, suggesting that the debt overhang problem is less pronouncedfor PE-backed firms.

A. Malenko, N. Malenko / Journal of Financial Economics 117 (2015) 607–627 609

up the target and finances the rest with the PE fund'scapital. The ability to raise debt is limited by the agencyconflict between shareholders and creditors: Shareholderscan divert value from creditors. Diversion is inefficient andresults in a deadweight loss. Ex ante, the PE firm wouldlike to commit to not diverting value because it bears thecost of diversion by paying a high interest rate on debt. ThePE firm's need to access the debt market repeatedly createsthis commitment ability endogenously. However, thiscommitment ability is limited and depends on economy-wide and sponsor-specific factors, as well as the extent ofcompetition between PE firms and their ability toform clubs.

Our results are as follows. First, the two sources ofvalue creation in LBOs are complements: PE firms' abilityto make operational improvements enhances their abilityto add value through financing, leading buyout leverage toincrease with PE sponsor skills. More generally, if a PEsponsor's operational skills are unobserved by the market,then the higher is the perceived skill of the sponsor, themore leverage can it add to the target. Moreover, theability to make operational improvements is necessary forvalue creation through financing: If PE firms never add anyoperational value, no buyouts take place.

The intuition is as follows. If the PE firm could committo not diverting value from creditors, buyout leveragewould be determined by the usual trade-offs, such as thebenefits of tax shields and management incentives versusthe costs of financial distress and inefficient investment.However, due to the agency problem, the equilibriumleverage is potentially different from this target-specificoptimal leverage and is determined by the commitmentpower of the PE firm. The PE firm takes as much debt ascreditors are willing to lend at a low interest rate, i.e., themaximum amount of debt given which it will refrain fromdiversion. A higher ability to add value through opera-tional changes increases the importance of future dealsrelative to the benefits of diversion from the currentportfolio company and thereby increases the PE sponsor'scommitment power, allowing it to take more debt. Hence,differences in PE firms' quality are amplified: Firms withmore skilled GPs gain even more advantage because oftheir higher ability to lever up their portfolio companies. Ina more general setting, in which PE firms' skills are notperfectly known to the market, the sponsor's commitmentpower is determined by its perceived skill, which isupdated over time. Interpreting the sponsor's perceivedskill as reputation for operational skill and its ability tocommit to no diversion as reputation for not expropriatingcreditors, we conclude that the two types of reputation arecomplementary to each other.

Importantly, in the extreme case, when PE firms neveradd any operational value, PE-owned firms have noborrowing advantage over non-PE-owned firms despitePE sponsors' repeated deal making. As a result, PE firmscannot add any value through financing either, and LBOactivity disappears completely. Therefore, the view thatPE firms do nothing but lever up their portfolio compa-nies is inconsistent with our results: Operationalimprovements are necessary for buyouts to take place.The reason for this result is competition among PE firms

for targets. If a PE firm can add value only by levering upits targets, it has no competitive advantage over other PEfirms, and all benefits from additional leverage accrue totarget shareholders through the buyout premium. The PEfirm thus does not earn abnormal profits in any futuredeals, which implies that it has no reputational capital toback up the debt of its current portfolio company. Ageneral point here is that a PE firm needs to have acompetitive advantage over its rivals to be able to createvalue through debt financing.

We also show that incentives of PE firms to invest inreputation for non-diversion, and therefore value createdthrough financing, are higher in common-value deals(when different PE firms obtain the same value fromacquiring the target but differ in their estimates of thisvalue) than in private-value deals (when different PE firmsobtain different values from acquiring the target). This isbecause when a PE firm diverts value, it harms its ability tocompete with other PE firms for future deals, and thedamage is considerably higher in common-value transac-tions than in private-value transactions.

Our second set of results relates to club formation.Club deals can have both a positive effect due tosynergies among club members and a negative effectdue to PE firms' reduced incentives to invest in reputa-tion for non-diversion. The positive effect of club deals isthat they allow bidders to borrow reputation for non-diversion from each other. If a low-reputation bidder canadd significant value through operational improvementsin a given target, while a high-reputation bidder cannot,then the club as a whole creates a higher value from thedeal than each of the bidders can create on its own. Thelow-reputation bidder benefits by borrowing reputationfrom the high-reputation bidder, which allows the clubto raise financing on favorable terms. The high-reputation bidder, in turn, benefits by capturing part ofthe surplus that the low-reputation bidder adds to thetarget through operations. While club deals are benefi-cial ex post, they can be detrimental ex ante because theycould negatively affect PE firms' commitment power. Abidder who trades off the benefits of diversion againstthe costs of losing its reputation with creditors couldhave lower incentives to preserve its reputation for non-diversion, realizing that it will be able to borrow reputa-tion from more reputable bidders in the future byforming a club with them. We show that this negativeex ante effect can dominate, leading to reduced buyoutactivity and a lower expected value from buyouts,despite the positive ex post effect of club formation inany given deal. This negative effect is different from thecritique that club deals transfer wealth from targetshareholders to bidders by reducing competition (e.g.,Officer, Ozbas, and Sensoy, 2010; Marquez and Singh,2013). We emphasize that club deals can have a negativeeffect on the joint surplus of bidders and the target,rather than resulting only in a wealth transfer.

Our final set of results deals with the determinants ofbuyout activity, leverage, and other deal characteristics.Consistent with the observed empirical evidence (Axelson,Jenkinson, Strömberg, and Weisbach, 2013; Demiroglu andJames, 2010), equilibrium buyout leverage is determined


not only by target-specific factors, but also by PE sponsorcharacteristics and economy-wide factors, such as dis-count rates and expectations of future deal activity. Incontrast, leverage of non-PE-owned firms in our model isdriven only by their characteristics. Intuitively, buyoutleverage is primarily determined by the PE sponsor'scommitment power. When discount rates are lower, futuredeals are more important to PE sponsors relative tocurrent deals, so the PE firm's commitment ability ishigher, increasing buyout leverage. Increased debt capa-city, in turn, implies that PE firms can add more valuethrough financing, so aggregate buyout activity increasesas well, consistent with the evidence in Haddad, Loualiche,and Plosser (2014). Interestingly, the sensitivity of buyoutactivity to discount rates depends on the informationenvironment. If uncertainty about PE sponsors' skills ishigh, then an increase in the discount rate, which destroyscommitment power of all PE firms except those with ahigh reputation for skill, causes a higher drop in buyoutactivity than if uncertainty about PE sponsors' skills is low.

Our paper builds on previous research that examinesthe role of repeated interactions in alleviating agencyconflicts due to players' reputational incentives. The ideathat repeated borrowing can help mitigate the agencyproblems of debt goes back to Jensen and Meckling (1976).Diamond (1989) models reputation acquisition in debtmarkets and studies how the incentive effect of reputationevolves over time.7 De Fontenay (2014) provides a detaileddiscussion of the importance of repeated interactions inprivate equity. Our main theoretical contribution to theliterature on reputation is the analysis of externalities inplayers' reputations that arise due to club formation andcompetition in auctions.8 Because of externalities, ourmodel is related to papers that study the effect of productmarket competition on firms' incentives to maintainreputation for producing high-quality products.9

The paper also contributes to the literature on lever-aged buyouts. One strand of this literature examines thedeterminants of aggregate buyout activity. Variation inLBO activity is often attributed to debt market mispricingor changes in the supply of credit.10 Relatedly, Martos-Vila,Rhodes-Kropf, and Harford (2014) study how debt marketovervaluation can explain the changing proportion offinancial buyers relative to strategic buyers. In contrast,our theory features efficient capital markets and nochanges in credit supply. Axelson, Strömberg, and

7 Diamond (1991) extends Diamond (1989) by distinguishingbetween bank debt and public debt. Prior to Diamond (1989), reputationin repeated borrowing was modeled by John and Nachman (1985).

8 In addition, leverage in our model is endogenous, which allows usto study the determinants of buyout leverage vis-à-vis leverage ofpublic firms.

9 See, e.g., Hörner (2002), Kranton (2003), and Bar-Isaac (2005).Relatedly, Winton and Yerramilli (2015) develop a model to study abank's incentives to monitor loans in the originate-to-distribute marketand have an extension that examines competition among banks. In thecontext of repeated interactions, Bond and Rai (2009) point to a positiveexternality between borrowers' repayment decisions, which arises if thelender's viability depends on how many borrowers repay.

10 See, e.g., Kaplan and Stein (1993), Acharya, Franks, and Servaes(2007), Kaplan and Strömberg (2009), Shivdasani and Wang (2011) andAxelson, Jenkinson, Strömberg, and Weisbach (2013).

Weisbach (2009) show how the agency conflict betweengeneral and limited partners can rationalize the financialstructure of PE funds and derive implications for dealactivity and investment performance. In Haddad,Loualiche, and Plosser (2014), buyout activity is driven bythe trade-off between higher cash flow growth underprivate ownership and underdiversification of LBO inves-tors. Burkart and Dasgupta (2015) obtain procyclicality ofhedge fund activism in a model in which hedge funds leverup target firms to signal their ability and note that theirmodel can also apply to the PE industry. Different fromthese papers, our focus is on the post-buyout conflictbetween PE firms and creditors, the role of repeated dealmaking, and externalities in PE firms' reputations due toclub formation and competition. Club formation has alsobeen examined by Marquez and Singh (2013), who focuson the trade-off between reduced competition and valuecreation due to the club aggregating individual bidders'values. In contrast, we analyze the costs and benefits ofclub deals through the lens of PE firms' reputation withcreditors.11

The remainder of the paper is organized as follows.Section 2 describes the model setup, studies the target asan independent firm, and considers the benchmark case ofa single deal. Section 3 analyzes the model in which PEfirms have identical skill, and Section 4 considers themodel in which PE firms differ in their skills. Section 5studies club formation. Section 6 analyzes the commonvalue setting. Section 7 discusses observed variation inbuyout activity and buyout leverage in the context of themodel, and Section 8 provides new empirical implicationsof the model. Finally, Section 9 concludes. The proofs of allpropositions are presented in the Appendix, whereas theproofs of all lemmas, corollaries, and supplementaryresults are relegated to the Online Appendix.

2. Model setup

Time is discrete, indexed by t ¼ 0;1;2;…, and thehorizon is infinite. There are three types of agents:creditors, PE firms, and targets. All agents are risk-neutral. The financial market is competitive, and all agentsdiscount future payoffs at the rate r. There is a continuumof measure two of infinitely lived PE firms. Each PE firm ischaracterized by skill χ of its GPs, which determines the PEfirm's ability to add value to the average target throughoperational improvements.12 In addition, in every period,there is a continuum of measure γA ½0;1� of targets.

The timeline, illustrated in Fig. 1, is as follows. At thebeginning of every period, each target is matched to twoPE firms, which are random draws from the population ofPE firms. By buying out the target, a PE firm can create

11 Relatedly, Povel and Singh (2010) study the effects of stapledfinance, which is common in LBO transactions, on competition betweenbidders and the expected price paid to the seller. Our paper abstractsfrom the incentives of sellers to arrange such financing for the bidders.

12 Empirical evidence in Kaplan and Schoar (2005), Phalippou andGottschalg (2009), and Robinson and Sensoy (2013) suggests a substan-tial variation in the perceived skills of GPs, proxied by past performanceand fund size.

13 To see this, note that, given debt D, diversion in the good state issuboptimal if β Dð Þ � g Dð Þ� D1�λ exceeds �XG . Because β

00o0 andβ0 Dn� �

o0, then β Dð ÞZmin β 0ð Þ;β Dn� �� for DA 0;Dn� �. Because β 0ð Þ ¼ 0and β Dn

� �Z�XG by Assumption 1, then β Dð ÞZmin β 0ð Þ;β Dn

� �� Z�XG .


value through two potential channels. One channel isleverage, which provides tax benefits and can restrictempire-building by the target's management. The otherchannel is operational and governance changes, such ascost-cutting, eliminating unproductive assets, improvedmonitoring, and management changes. For brevity, werefer to such changes as operational improvements. Wemodel operational improvements by assuming that thetarget can be either successful or not, and the PE sponsoraffects the probability that the target succeeds.

Formally, there are two states, sA B;Gf g, that determinewhether the target is successful (s¼G) or not (s¼B). Uponmatching with the target, PE firm i learns its probability qi ofstate G, i.e., the probability with which the target will succeedunder its management. The PE firm's skill χi determines howlikely a typical target is to succeed. In other words, the PEfirm's skill affects its ability to manage companies in a waythat minimizes chances of bad states. Specifically, if the PEfirm's skill is χi, qi is an independent draw from the distribu-tion function F �jχ i

� �with density f �jχ i

� �and full support on

½q; q�, 0rqoqr1. Section 3 considers the case in which allPE firms are identical in skill. In Section 4, PE firms differ intheir skills, and the market learns about firms' skills over time.The assumption that draws of q are PE firm specific meansthat values are private. In Section 6, we consider a model withcommon values and show that PE firms have strongerincentives to invest in reputation for non-diversion whenvalues are common.

If the target remains independent, its probability ofsuccess is given by qT A 0; qð Þ. If qT 4 q, a realization of qican be smaller than qT, corresponding to the idea that thefirm's operations are sometimes more efficient underpublic ownership. Depending on qi and the cost of finan-cing determined in equilibrium, each PE firm decideswhether to bid for the target. If no firm decides to bid,the target remains independent. If only one firm decides tobid, it makes a take-it-or-leave-it offer to the target. If bothfirms decide to bid, they compete for the target in anEnglish (open ascending-bid) auction, where the price isgradually increased until only one bidder remains. We lookat the unique equilibrium in weakly dominant strategies:Each firm bids up to its maximum willingness to pay.

The acquirer (PE firm) then decides on the capital structureof the target. It chooses amount of debt Di due at the end ofthe period and raises debt in the competitive market. The restis financed from the PE firm's own capital. Thus, the acquirerbecomes the only shareholder of the target. We assume thatPE firms are financially unconstrained, so the only reason touse debt is because debt can increase the target's value. If thetarget remains independent, its shareholders also chooseamount ~D due at the end of the period and raise debt inthe competitive market.

At the end of the period, the state sA B;Gf g (the target'ssuccess or failure) is realized and publicly observed. The valueof the target is XB in the bad state and XGþg Dð Þ in the goodstate, where ΔX � XG�XB40 and D is the debt that thetarget's shareholders took: Di if it was acquired by PE firm i and~D if it remained independent. Thus, the PE firm can add valueby operational changes (by increasing the probability of a highcash flow realization) and by using different leverage than theindependent target would use. Function g Dð Þ reflects the

benefits of debt due to tax shields and improved managementincentives and the costs of debt due to inefficient investment.We model these effects in a reduced-form way through g Dð Þand provide microfoundations for this function based on taxshields and the free cash flow argument of Jensen (1989) in thesection “Microfoundation for the benefits of debt” in theAppendix. The trade-off between these costs and benefitsdetermines the unconstrained optimal amount of debt Dn,which maximizes g Dð Þ. Specifically, we assume that g 0ð Þ ¼ 0,g0 Dð Þ40, g00 Dð Þo0, and g0 Dn� �¼ 0 for some finite Dn40.

After the state is realized, shareholders (the PE firm orcurrent owners if the target remains independent) canexpropriate creditors by diverting any amount betweenzero and the realized value. Diverting x of the cash flowsgenerates λx in value to shareholders, where λo1. Thus,diversion is inefficient. This specification encompassesdifferent ways in which the debt-equity conflict of interestcan be manifested: dividends that decrease total firmvalue, debt overhang (e.g., not injecting additional cashalthough it would allow a more efficient resolution offinancial distress), and delaying efficient liquidation.

Finally, all agents receive their payoffs. After that, thegame is repeated. Past realizations of cash flows anddiversion decisions are observable but not verifiable. Thus,we assume that diversion cannot be contracted away, forexample, by covenants. This assumption is reasonable forsome conflicts of interest (e.g., debt overhang) but could beless so for others (e.g., dividend payouts). As we discuss inthe conclusion, covenants provide an alternative, albeitcostly and imperfect, way to resolve debt-equity conflicts,and introducing them into the model would lead to anumber of additional implications.

We impose the following restriction on the parameters.

Assumption 1. 1�λ� �XBoDnr 1�λ� � XGþg Dn� �� .The first inequality implies that, given debt Dn, diver-

sion in the bad state is optimal for shareholders. This isbecause the cash flow from diversion, λXB, is greater thanthe cash flow remaining after paying out the debt,maxðXB�Dn;0Þ. Without this condition, the agency pro-blem between creditors and shareholders would not exist.The second inequality guarantees that, given debt Dn,diversion in the good state is not optimal. Moreover, giventhe properties of g Dð Þ, this assumption also implies thatdiversion is not optimal in the good state for any DrDn.13The assumption that diversion does not occur in the goodstate is not inconsistent with the evidence that PE-ownedfirms frequently pay dividends in good times as well. Thereason is that dividends paid in the good state are oftennot detrimental to debtholders and therefore should notbe considered as diversion of value. Consistent with this,Hotchkiss, Smith, and Strömberg (2014) do not findevidence that dividend recapitalizations influence theprobability of default.

(1)A target is matched to

two PE firms. EachPE firm learns thevalue it can create.

(2)PE firms bid for thetarget. The winner

chooses the leverageof the target.

(3)The state is

realized.

(4)The PE firm decideshow much value to

divert.

(5)All agents receive

cash flows.

Period Period

Beginning of the period End of the period

Fig. 1. Timeline of the model.


Finally, we introduce the following notations. Wedenote xþ �maxðx;0Þ and ½x�ba �minðmaxðx; aÞ; bÞ forbZa. In other words, xþ equals x truncated by zero frombelow, and ½x�ba equals x truncated by a from below and byb from above.

15 In unreported results, we have also analyzed the case in whichAssumption 2 is violated. All results of the main model remain the same,and the only change concerns the N-equilibrium, defined in Section 3. In

2.1. Target as an independent firm

As a prerequisite, we solve for debt policy and valuationof the target if it remains an independent firm. Beforepresenting the analysis, we discuss the important differ-ence between non-PE-owned firms and firms that areowned by a PE sponsor. The distinguishing feature of PEownership compared with public dispersed ownership isnot repeated borrowing by the PE sponsor per se. Non-PE-owned firms borrow repeatedly, too. Instead, the distin-guishing feature of a PE-owned firm is that it can back itsborrowing with more assets. A non-PE-owned firm bor-rows only against its own assets, that is, assets in place andgrowth options. In contrast, a PE-owned firm borrows bothagainst its own assets and against the reputational capitalof the PE sponsor, reflected in its payoff from future deals.Effectively, debt of a PE-owned firm is backed both by thefirm's own assets and by some fraction of the assets of thePE sponsor's future portfolio companies. To illustrate thispoint, consider a highly levered firm that suffers a negativeshock to its assets in place. This can lead to the well-known debt overhang problem (Myers, 1977): If the firm isowned by dispersed shareholders and managers maximizeshareholder value, the firm passes up positive net presentvalue (NPV) projects because shareholders bear the fullinvestment cost, while a large fraction of benefits goes todebtholders. Debt overhang takes place irrespectively ofwhether or not the firm is a repeated borrower, as thevalue of the firm's assets already incorporates all expectedfuture interactions.14 In contrast, if the firm is owned by aPE sponsor, the sponsor could invest in positive NPVprojects despite the fact that shareholders (i.e., the spon-sor) lose on them, because the sponsor gets rewardedthrough different portfolio companies in the future.

To summarize, the difference between PE-owned andnon-PE-owned firms is that PE-owned firms effectivelyback their debt with assets of other targets acquired by thePE sponsor in the future. For simplicity, we capture thisdifference by assuming that the target borrows only once ifit is non-PE-owned. But, for the above argument, the

14 The degree of debt overhang depends on the frequency ofborrowing (equivalently, average debt maturity) in a nontrivial way(Diamond and He, 2014).

reasoning would not be different if the target borrowedrepeatedly.

Consider the equilibrium debt that the target wouldtake as an independent firm. Given debt D, diversion in thebad state occurs if

XB�Dð Þþ oλXB3D4 1�λ� �

XB: ð1Þ

The target thus chooses between two options: (1)taking the unconstrained optimal debt Dn and divertingvalue from creditors in the bad state and (2) taking thehighest possible debt given which it will not divert value,D¼ 1�λ� �XB. Under the first option, the target can capturethe benefits of higher leverage, but it also has to bear the

expected deadweight loss from diversion, 1�qTð Þ 1�λð ÞXB1þ r , bypaying a high interest rate on debt. This is becausecreditors, anticipating that they will be expropriated inthe bad state, are willing to invest less given the promisedpayment Dn. We assume that the loss from diversion issufficiently high relative to the benefits of higher leverage,so that the second policy is optimal for the target. BecauseqT rq, this is ensured by the following condition.

Assumption 2. The loss from diversion is high:1�qð Þ 1�λð ÞXB

q Zg Dn

� ��g 1�λ� �XB� �.Assumption 2 is plausible because, if it were violated,

leverage of the target would sometimes decrease in abuyout, which is inconsistent with empirical evidence,except perhaps in rare cases.15 Under Assumption 2, thefollowing result holds.

Lemma 1. If the target remains independent, it takes debtD¼ 1�λ� �XB and does not divert value in the bad state. Thevalue of the target if it remains independent is

V0 �qT XGþg0

� �þ 1�qT� �XB1þr ; ð2Þ

where g0 � g 1�λ� �

XB� �

.

particular, if a PE firm is unable to commit to no diversion, then, insteadof borrowing 1�λ� �XB and not diverting value, it borrows Dn and divertsvalue if the bad state is realized. The implication of this result is thatwhen the loss from diversion is low, buyout leverage is a U-shapedfunction of drivers of deal activity and the skill of the PE sponsor.

:

16 Grim trigger strategies is a common way to model reputation inrepeated games. A player's reputation is the history of his past “goodactions.” As soon as a player deviates from a “good action,” other playersare assumed to believe that he will deviate in the future, too, so the gamemoves to a sequence of one-shot Nash equilibria. Mailath and Samuelson(2006) provide an overview of approaches to modeling reputation, one ofwhich is the grim trigger strategy approach. Another way to modelreputation is the “crazy types” approach. In the context of our model, ifsome PE firms are “honest” in the sense that they never divert value fromcreditors, then other PE firms have incentives to abstain from diversion topretend to be “honest” and thereby obtain cheap financing. In such amodel, a PE firm's reputation for non-diversion would be the outsiders'belief that it is “honest,” which would evolve over time. It is easy to seethat our results would continue to hold in such a model as well. Becauseour focus is not on how reputation evolves over time, we adopt the firstapproach for simplicity.


2.2. Single-deal setting

We next analyze debt policy under PE ownership. As abenchmark, consider the case in which each PE firm doesonly one deal.

2.2.1. Commitment caseSuppose first that each PE firm can commit to not

diverting value from creditors. The proof of Lemma 2 showsthat if PE firm i takes debt Di , its maximum willingness to

pay for the target is V0þ qi �qTð ÞΔX þqig Dið Þ�qT g01þ r , where V0 is thestand-alone value of the target, given by (2). The premiumthe bidder is willing to pay over V0 is determined by theoperational improvements it can make (if qi4qT ) and thevalue it can create by levering up the target (if g Dið Þ4g0).Therefore, if PE firm i acquires the target, it pays V0if the other bidder does not participate in the auction

(i.e., if qj�qT� �

ΔXþqjg Dj� ��qTg0o0) and pays V0þ

qj �qTð ÞΔX þqjg Djð Þ�qT g01þ r if the other bidder participates. Hence,

the payoff of PE firm i conditional on realizations qi; qj is

11þr qi�qT

� �ΔXþqig Dið Þ�qTg0� qj�qT

� �ΔXþqjg Dj

� ��qTg0� �þ� þ

ð3ÞBecause Dn maximizes g Dð Þ, the acquirer finds it opti-

mal to take debt Dn. Thus, if the PE firm is able to committo no diversion, it can always add value through financing.Lemma 2 summarizes the equilibrium in this case.

Lemma 2. Suppose PE firms can commit to no diversion. Then atarget is acquired if and only if max q1; q2

� ��qT� �ΔXþmax q1; q2

� �g Dn� ��qTg040, irrespective of r; γ. The bidder

with the highest value qi acquires the target, irrespective ofbidders' skills χ1; χ2. The acquirer takes debt D

n.

2.2.2. No commitment caseThe analysis of the no commitment case, presented in

the proof of Lemma 3, is similar to the analysis of theoptimal debt policy of a target as an independent firm.Each PE firm faces a trade-off between taking the uncon-strained optimal debt Dn but bearing the loss from diver-sion ex ante and taking a lower debt 1�λ� �XB, given whichit refrains from diversion. Under Assumption 2, the opti-mal debt level is 1�λ� �XB, which coincides with the debtof the target as an independent firm. We summarize thisanalysis in Lemma 3.

Lemma 3. Suppose PE firms cannot commit to no diversion.Then a target is acquired if and only if max q1; q2

� �4qT ,

irrespective of r; γ. The bidder with the highest value qiacquires the target, irrespective of bidders' skills χ1; χ2. Theacquirer takes debt 1�λ� �XB.

The PE firm's maximum willingness to pay for the

target, V0þ qi �qTð Þ ΔX þg0ð Þ1þ r , is strictly smaller than in thecase with commitment because it cannot add any valuethrough financing. Thus, the PE firm would like to committo not diverting value but cannot do so due to the short-term nature of its interactions with creditors. Comparedwith the case of commitment, deal activity is lower: Only

PE firms that can generate a positive value throughoperational changes, qi4qT , choose to undertake the deal.

We next analyze the general model, in which PE firmsrepeatedly find potential targets. Due to PE firms' repeateddeal making, their reputation with creditors becomes impor-tant to them. Thus, to obtain future financing on favorableterms, they might want to refrain from diverting value fromcreditors, i.e., commitment to no diversion becomes possible.This commitment ability can lead PE-owned firms to opti-mally borrow more than they would borrow as independentfirms and thereby allows PE sponsors to create value throughfinancing in addition to operational improvements.

For brevity, we use the following notations for the restof the paper: Δg � g Dn

� ��g0 andzR qð Þ �

q�qT� �

ΔXþg0� �þqΔg1R ¼ 1

1þr : ð4Þ

Here z1 qð Þ (z0 qð Þ) denotes the maximumwillingness to payfor the target above its stand-alone value V0 of a PE firmthat has (does not have) a reputation for non-diversion, i.e., that can (cannot) commit to no diversion.

3. Model with identical PE firms

In this section, we analyze the setting where PE firmsdo not differ in their skill. In each period, the PE firm thatacquires the target makes two decisions. First, it choosesthe amount of debt used to finance the buyout. Second, ifthe bad state is realized, it decides whether to divert value.By Assumption 1, diversion in the good state is suboptimaleven in the single-deal setting, and hence it is suboptimalin the repeated-deal setting as well.

We look for equilibria in symmetric pure strategies, inwhich, in every period, all PE firms take the same amountof debt and either divert all value in the bad state or do notdivert. Let D denote the equilibrium level of debt andeAf0;1g denote the equilibrium diversion decision, wheree¼0 (e¼1) stands for no diversion (diversion) in the badstate. Recall that in the single-deal setting, diversion in thebad state is optimal if and only if D4 1�λ� �XB. Hence, ifDr 1�λ� �XB, then e¼0. However, unlike in the single-deal setting, PE firms can now refrain from diversion evenif D4 1�λ� �XB due to their concerns about reputation.

We focus on equilibria in which creditors play a grimtrigger strategy.16 In particular, if a PE firm deviates fromits equilibrium strategy, either by diverting value when


e¼0 or by taking debt D0aD, creditors expect this PE firmto divert value in the future whenever it takes debtD4 1�λ� �XB.17 Note that a PE firm's deviation from theequilibrium strategy has no effect on the incentives ofother PE firms and thus does not change their actions.Therefore, considering the deviation of a single firm inisolation is sufficient.

As shown in Lemma 4 below, there always exists anequilibrium that is a repetition of the single-period gamewithout commitment characterized by Lemma 3. In thisequilibrium, to refrain from diversion, PE firms take thesame low amount of debt as the independent target,1�λ� �XB. In the repeated-deal setting, however, the needto raise financing for future deals enhances PE firms'ability to commit to no diversion, and thus more efficientequilibria can exist as well. If the ability to commit issufficiently high, an equilibrium can exist in which PEfirms take the unconstrained optimal debt Dn and yet paylow interest rates because they never divert value. Thisequilibrium is a repetition of the single-period game withcommitment characterized by Lemma 2. If the ability tocommit is in the intermediate range, PE firms will notrefrain from diversion if the amount due is Dn but willrefrain from diversion if the amount due is slightly lower.In this case, PE firms could prefer to take a lower thanoptimal debt DoDn and pay a low interest rate, ratherthan take Dn and pay a high interest rate. This argumentimplies that the set of potential equilibria is the following.

1.

widwepunnot

equ4 swhexislessexisandcon

N-equilibrium (no commitment equilibrium): PE firmstake debt 1�λ� �XB and do not divert value.

2.
C-equilibria (constrained commitment equilibria):PE firms take debt DA 1�λ� �XB;Dn� � and do notdivert value.
3.
U-equilibrium (unconstrained commitment equili-brium): PE firms take debt Dn and do not divert value.18
Lemma 4 characterizes the necessary and sufficientconditions for each of these equilibria to exist.

Lemma 4. The N-equilibrium always exists. The C-equili-brium with debt D exists if and only if

λXB� XB�Dð Þþ rγrE q1�qT� �

ΔXþq1g Dð Þ�qTg0�

� q2�qT� �

ΔXþq2g Dð Þ�qTg0� �þ iq1 g Dð Þ�g0ð Þ

0: ð5Þ

17 We assume this harsh off-equilibrium punishment to consider aer set of potential equilibria. However, the equilibria that remain afterapply the refinement in Proposition 1 can be sustained by less harshishment, where creditors punish PE firms only for diverting value, butfor taking an off-equilibrium amount of debt.18 In addition to these three types of equilibria, there can existilibria with D4Dn and no diversion. However, the proof of Lemmahows that if any such equilibrium exists, then the U-equilibrium,ich features the optimal amount of debt and hence is more efficient,ts as well. Under the efficiency refinement that we apply below, theefficient equilibrium is not selected if a more efficient equilibriumts. Hence, equilibria with no diversion and D4Dn are never selected,we do not list them in the set of potential equilibria to avoid

fusion.

The U-equilibrium exists if and only if condition (5) holds forD¼Dn.

Intuitively, PE firms refrain from diverting value only ifthe benefit from diversion today [the left-hand side ofcondition (5)] is lower than the benefit of preserving theirreputation for non-diversion and obtaining cheap finan-cing in future deals [the right-hand side of condition (5)].

Because multiple equilibria can coexist, we use theefficiency criterion to select among them. We call oneequilibrium more efficient than the other if the expectedvalue from deals, given by

γrE maxð0; q1�qT

� �ΔXþq1g Dð Þ�qTg0; q2�qT

� �ΔXþq2g Dð Þ�qTg0Þ

� �;

ð6Þ

is higher in the first equilibrium. If two equilibria coexist,we select the more efficient equilibrium. Because g Dð Þincreases in D for DrDn, (6) implies that the U-equili-brium is the most efficient among all equilibria, the N-equilibrium is the least efficient, and, among any two C-equilibria, the equilibrium with the higher debt level ismore efficient.

In the most efficient equilibrium, the PE firm takes thehighest amount of debt (up to Dn) given which it will refrainfrom diversion, i.e., the highest debt that allows financing tobe raised on favorable terms. As the PE firm's commitmentpower declines, this level of debt declines as well. Combiningthe efficiency refinement and Lemma 4 allows us to derive theequilibrium buyout leverage as a function of PE sponsors' skill,the discount rate r, and the mass of targets γ (which can beinterpreted as expectations of future buyout activity). Thereason these factors affect buyout leverage is that they affectthe PE firm's concerns about its reputation with creditors andthereby its commitment power. In particular, a more skilledPE firm expects to capture higher rents in future deals andhence cares more about its reputation with creditors, whichincreases its commitment power. Similarly, a decrease in thediscount rate or an increase in expectations about futureactivity increases the value of future deals relative to today'sbenefits of diversion and also increases the PE firm's commit-ment power. This argument implies the following result.

Proposition 1. The level of debt in the most efficient equili-brium decreases in r and increases in γ. Furthermore, supposethat q¼d q0þc, where c40 is any constant and q0 isdistributed with full support on ½q; q�. Then, the level of debtin the most efficient equilibrium increases in c.

In Section 7, we relate the implications of Proposition 1to the existing empirical evidence about the determinantsof buyout activity and leverage.

The formulation of Proposition 1 assumes that the skillof all PE firms in the economy increases simultaneously (qishifts by c for all firms). As a result, the effect of skill(parameter c) on the firm's payoff from future deals ismuted because it is accompanied by an increase incompetition. Thus, the statement will be even stronger ifwe increase the skill of one PE firm while keeping the skillof other firms fixed. In particular, consider a modifiedsetup, in which all firms but one have the same skill χ, andthe skill ~χ of the remaining firm can change independently


of χ. Lemma 5 shows that as the PE firm's skill increases inthe sense of first-order stochastic dominance (FOSD), theleverage it takes increases.

Lemma 5. Suppose that all PE firms but one have skill χ, theremaining firm has skill ~χ , and Dð ~χ Þ denotes the level of debtit takes in the most efficient equilibrium. If the distributionF qjχ� � satisfies first-order stochastic dominance, then Dð ~χ Þincreases with ~χ .

Proposition 1 and Lemma 5 show that the twosources of value creation in LBOs — operational changesand financing decisions — are complements. The higheris a PE firm's ability to add value through operations, thehigher is its ability to add value through financingbecause it can raise more debt at a low interest rate.Moreover, Proposition 2 and Corollary 1 imply that theability to add value through operations is necessary tocreate any value through financing.

Proposition 2. Suppose that the distribution of q is degenerate:Prðq¼ q0Þ ¼ 1 for some q0. Then, only the N-equilibriumexists.

Corollary 1. Suppose that PE firms never add or destroy anyoperational value: Pr q¼ qT

� �¼ 1. Then, PE firms have noborrowing advantage over independent targets and LBOsnever happen.

Corollary 1 helps evaluate the popular claim that PEfirms do not create operational improvements and onlyload up targets with debt. It shows that this claim is notconsistent with the equilibrium: Without any operationalimprovements, PE-owned firms would not be able toborrow on more favorable terms than non-PE-ownedfirms, and hence buyouts would never happen. Thus, PEfirms' ability to add value through debt financing cruciallyrelies on their ability to add operational value.

This result arises because of externalities betweencompeting PE firms. A PE firm's ability to commit to notdiverting value from creditors depends on its net payofffrom future deals. Because PE firms compete for deals,the only way for a PE firm to have a positive net payofffrom future deals is to have a competitive advantage overother PE firms in the market. If a PE firm never addsoperational value, it never has any competitive advan-tage: Even if it can take additional debt, other PE firmscan do the same, so the added value from debt isreflected in the premium and fully accrues to share-holders of the target. Thus, if a PE firm has no ability tomake operational improvements, its net payoff fromfuture deals is zero, so it cannot pledge its future payoffto credibly promise not to divert value. As a result, the PEfirm has no borrowing advantage over stand-alone targetcompanies despite doing repeated deals.

While a PE firm that never adds any operational valuecan never add value through financing, a feature ofequilibrium is that a PE firm that does not add opera-tional value in a given deal can nevertheless add valuethrough financing in this deal. In particular, if the PE firmexpects to add operational value in at least some futuredeals, then some current deals can go through even

though no operational improvements are made. Intui-tively, the ability to add operational value to some futuretargets gives the PE firm a competitive advantage andallows it to keep part of the future surplus instead ofgiving it away entirely to shareholders of the targets.Hence, the PE firm can pledge this future surplus tocredibly promise not to divert value today, which allowsit to borrow on more favorable terms than the target.

The general insight here is that, to create valuethrough debt financing, a PE firm needs a competitiveadvantage over other PE firms in the market. While inthe current model, a competitive advantage is achievedvia operational improvements that other PE firms cannotmimic, other microfoundations of a competitive advan-tage are possible, too. In Section 6, for example, we showan analog of this result in the model with commonvalues, in which the competitive advantage of a PE firmcomes through its private knowledge of the potential forvalue creation in a deal.

4. Model with heterogeneous PE firms

In this section, we extend the model of Section 3 byallowing PE firms to differ in their operational skills andfor the market to learn about PE firms' skills over time.This analysis is important because, in practice, a PE firm'sreputation for skill is arguably at least as important as itsreputation for not expropriating creditors. As we show,there is an interesting interaction between the two typesof reputation, which leads to additional implications.

Assume that each PE firm can be either of highχ ¼H� � or low χ ¼ L� � skill. If a target is managed by alow-skill PE firm, the probability of a good state, q, isdrawn from distribution f qjLð Þ. If a target is managed by ahigh-skill PE firm, the probability of a good state isdrawn from distribution f qjHð Þ, which dominates f qjLð Þin the sense of FOSD. As time goes by, new PE firmsarrive to the market and some existing PE firms leave.Every period, each existing PE firm leaves the marketwith probability φA 0;1ð Þ, and mass 2φ of new PE firmsenters the market. The total mass of PE firms in themarket thus equals two in every period. The skill of everynew PE firm is either high or low with equal probabil-ities, independent of the skill of other PE firms. The skillof a PE firm is not known by anyone when it enters themarket. However, as time goes by, participants learnabout the skill of each PE firm by observing its realiza-tions of q.

To keep the learning environment tractable, we maketwo simplifying assumptions. First, when a target ismatched to PE firms i and j, realizations qi and qj areobserved by all market participants. Thus, informationabout the skill of each PE firm is symmetric among all PEfirms and investors. Second, distributions f qjHð Þ and f qjLð Þsatisfy

f qjHð Þ ¼pf qð Þ if qA 12�d; 12þd

� �;

1�pð Þf H qð Þ if qA 12þd; q� �

;

0 otherwise;

8><>: ð7Þ


f qjLð Þ ¼pf qð Þ if qA 12�d; 12þd

� �;

1�pð Þf L qð Þ if qA q; 12�dh i

;

0 otherwise;

8>><>>: ð8Þ

where the distribution densities f �ð Þ, f H �ð Þ, and f L �ð Þ havefull support on the corresponding intervals. This distribu-tion assumption makes learning simple and tractable. IfqA 12þd; q

� �is realized, the market infers that the PE firm is

high-skill because the low-skill PE firm never gets such arealization of q. Similarly, if qA q; 12�d

h iis realized, the

market infers that the PE firm is low-skill. Finally, ifqA 12�d; 12þd

� �is realized, the market does not update its

belief about the quality of the PE firm because theseintermediate realizations of q are equally likely for bothhigh- and low-skill PE firms. It follows that at any timethere are three types of firms: firms whose skill has beenrevealed to be high, θ¼H; firms whose skill has beenrevealed to be low, θ¼ L; and firms whose skill is stillunknown, θ¼U.19

Let μH tð Þ, μL tð Þ, and μU tð Þ denote the mass of PE firms ofeach type at the beginning of period t. In what follows, wesolve for the stationary equilibria, in which the distribu-tion of types does not change over time. Lemma 6 showsthat there exists a unique stationary distribution of types,characterized by the masses μH ;μL;μU

� �of the three types

of PE firms at the beginning of each period.

Lemma 6. There exists a unique stationary distribution of PEfirms' types, characterized by μH ¼ μL ¼ 12

1�φð Þ 1�pð Þγ1� 1�φð Þ pγþ1� γð Þ

and μU ¼ φ1� 1�φð Þ pγþ1�γð Þ.Lemma 6 implies that the mass of firms of unknown

skill increases in φ, the fraction of new firms that enter themarket each period, and increases in p, the probability thata PE firm's realization of q does not reveal its skill to themarket.

To simplify the analysis, we assume that the functiong Dð Þ equals g0 on the interval ½ 1�λ

� �XB;D

nÞ and thenjumps to g Dn

� �4g0 at the point D

n. As before, g Dð Þincreases for Do 1�λ� �XB and decreases for D4Dn. Basedon the arguments in Sections 2 and 3, this assumptionimplies that PE firms take debt Dn if, given this debt level,they can commit to not diverting value and take debt1�λ� �XB otherwise.As previously, we focus on equilibria in symmetric pure

strategies, in which all firms of the same type follow thesame pure strategy in every period. The first condition ofAssumption 1 implies that diversion in the bad state isoptimal in the single-deal setting, and the second condi-tion of Assumption 1 guarantees that diversion in the goodstate never occurs. Hence, any symmetric pure strategyequilibrium is characterized by a set ϱ of types of PE firms,ϱDfH; L;Ug, such that types θAϱ take debt Dn and do not

19 A more standard assumption would be that distributions f qjHð Þand f qjLð Þ have common support and can be ranked in terms ofdominance. This assumption would complicate the learning problemsignificantly because the probability that a PE firm is high-skill could takeany value from zero to one and, thus, there would be a continuum oftypes of firms. The economic intuition behind our analysis does not relyon the fact that there are only three types in the model, so we expect ourresults to be general.

divert value (i.e., have a reputation for not expropriatingcreditors), while types θ=2ϱ take debt 1�λ� �XB (i.e., do nothave this reputation). Let R θ

� �denote whether type θ has

reputation for non-diversion: R θ� �¼ 1 if θAϱ, and

R θ� �¼ 0 otherwise.Let z denote the PE firm's maximum willingness to pay

for the target over its stand-alone value V0. Recall that zequals z0 qð Þ (z1 qð Þ) for a PE firm without (with) a reputa-tion for non-diversion, where zR qð Þ is given by Eq. (4).Denote the density of the stationary distribution of z byη zð Þ. This density depends on the stationary distribution oftypes μH, μL, and μU, calculated in Lemma 6, as well as onwhat types of PE firms are able to commit to no diversionin equilibrium, ϱ.

Let VR θ� �

denote the expected value to type θAfH; L;Ugif it has a reputation for non-diversion, and let VNR θ

� �denote the expected value to type θ if it does not have thisreputation. Then, type θ is not willing to divert value fromcreditors if and only if

λXB� XB�Dn� �þ rVR θ� ��VNR θ� �: ð9Þ

The next lemma derives VR θ� ��VNR θ� � as a function of

the stationary distribution η �ð Þ.

Lemma 7. The value from reputation for non-diversion totype θAfH; Lg is given by

VR θ� ��VNR θ� �¼ γ 1þrð Þrþφ

Z Z½z1 qð Þ�zþ �qΔg=ð1þ rÞ0 f qjθ

� �η zð Þ dq dz;

ð10Þand the value of reputation for non-diversion to type θ¼ U isgiven by

VR Uð Þ�VNR Uð Þ ¼ 12 V Hð Þ�VNR Hð ÞþV Lð Þ�VNR Lð Þð Þ

þ12

p 1þ rð Þrþφγ

þ 1�φð Þ 1�pð ÞR R ½z1 qð Þ�zþ �qΔg1R Hð Þ ¼ 0=ð1þ rÞ0�

þ½z1 qð Þ�zþ �qΔg1R Lð Þ ¼ 0=ð1þ rÞ0�f qð Þη zð Þ dq dz; ð11Þ

where V θ� �¼ VR θ� � if R θ� �¼ 1 and V θ� �¼ VNR θ� � if

R θ� �¼ 0.Similar to the model with identical firms, there exist

multiple Nash equilibria because reputation for non-diversion can be self-sustaining. To select among equili-bria, we define the following selection criterion.

Assumption 3 (Equilibrium selection). An equilibrium mustsatisfy the monotone reputation property: For any pairðθ; θ̂ÞAfH;U; Lg2, if VR θ

� ��VNR θ� �rVRðθ̂Þ�VNRðθ̂Þ forevery possible density η �ð Þ, then R θ� �rRðθ̂Þ. In case multi-ple equilibria satisfying the monotone reputation propertyexist, then the more efficient equilibrium is selected.

Intuitively, the monotone reputation property means thatif the value from reputation for non-diversion for type θ̂ isalways weakly higher than for type θ, regardless of thedistribution of bids of rival bidders, then it cannot be the casethat type θ̂ does not have commitment power, while type θdoes. While such Nash equilibria exist, to support them,expectations must be unreasonable: Types with a highervalue from reputation for non-diversion must be expected


to divert more in the future. The second part of the selectioncriterion is similar to the one used in Section 3.

Using the monotone reputation property and theexpressions for VR θ

� ��VNR θ� � derived in Lemma 7, weobtain the following characterization of equilibria.

Proposition 3. In any equilibrium, R Lð ÞrR Uð ÞrR Hð Þ.The intuition is as follows. The higher is the perceived

skill of the PE firm, the higher operational value it expectsto create in future deals. Hence, it is more important forhigher types to preserve their reputation for not expro-priating creditors to be able to finance future deals onfavorable terms (VR θ

� ��VNR θ� � increases with type). Inthat sense, Proposition 3 complements the results inSection 3 by showing that a PE firm's ability to addoperational value enhances value creation through finan-cing even in a more general setting, where PE firms'operational skills are unknown to the market. Interpretingθ as a PE firm's reputation for skill, this result suggests thatthe two types of reputation (reputation for skill andreputation for not expropriating creditors when the badstate is realized) are complementary to each other.

Proposition 3 implies that there are four possibleequilibria.

1.

N-equilibrium: All types of PE firms take debt 1�λ� �XB.

2.
H-equilibrium: Type-H firms take debt Dn and do not
divert value, while type-U and type-L firms take debt1�λ� �XB.

3.
HU-equilibrium: Type-H and type-U firms take debt Dn
and do not divert value, while type-L firms take debt1�λ� �XB.

4.

20 First, similar to the effect of φ, an increase in p increases thefraction μU of firms whose skill is unknown. In addition, a change in paffects the sustainability of different equilibria. To see this, consider theHU-equilibrium and the incentives of a type-U bidder to preserve itsreputation for non-diversion. On the one hand, an increase in p reducesthe likelihood that the bidder will soon be revealed as low-skill andbecome unable to get financing on favorable terms. On the other hand, anincrease in p decreases the reputational payoff of high-skill biddersbecause they are now less likely to add value and outbid their rivals infuture deals [the distribution f qjHð Þ decreases in terms of FOSD]. Theformer effect increases the bidder's incentives to sustain its reputation fornon-diversion, while the latter effect decreases them.

HUL-equilibrium: All types of PE firms take debt Dn anddo not divert value.

Because the expected value from deals is higher whenmore PE firms can commit to not diverting value, the HUL-equilibrium is the most efficient, followed by the HU-equilibrium, the H-equilibrium, and then the N-equili-brium. Lemma A.1 in the Online Appendix specifies thenecessary and sufficient conditions for the existence ofeach equilibrium and shows that, similar to the modelwith identical firms, a lower discount rate r improves PEfirms' ability to commit to no diversion.

We next compare the properties of the equilibria andderive implications for buyout activity and the composi-tion of acquirers.

Proposition 4. (1) The probability of a deal taking place is thehighest in the HUL-equilibrium, followed by the HU-equili-brium, the H-equilibrium, and then the N-equilibrium.(2) The fraction of targets acquired by PE firms perceived tobe high-skill is the highest in the H-equilibrium and is higherin the HU-equilibrium than in the HUL-equilibrium.

Combining the first statement of the proposition withthe efficiency refinement and the comparative statics in rfrom Lemma A.1 implies that buyout activity decreaseswith r. Intuitively, higher discount rates decrease PE firms'ability to commit to no diversion, reducing the value theycan add through financing and thus deal activity. The

second statement shows that the equilibria are character-ized by a different composition of acquirers. In the H-equilibrium, a bidder who is perceived to be high-skill canlever up the target more than bidders of unknown skill orbidders who are perceived to be low-skill. Such a biddercan therefore outbid other types of bidders even if theoperational value it creates in the current deal is the sameas, or lower than, the value created by them. As a result, PEfirms that are perceived to be high-skill acquire a dis-proportional fraction of targets in the H-equilibrium. For asimilar reason, the fraction of targets acquired by PE firmsperceived to be high-skill is higher in the HU-equilibriumthan in the HUL-equilibrium. Combining this with theefficiency refinement and the comparative statics in rimplies that the fraction of deals done by acquirersperceived to be high-skill is an inverted U-shape functionof the discount rate. Fig. 2 illustrates both implications ofProposition 1.

The equilibrium has interesting comparative statics inthe characteristics of the information environment, φ andp. Unlike the discount rate, these parameters have a non-monotonic effect on deal activity. Consider the effect of anincrease in φ, which captures how stable the PE industryis, i.e., whether there is a notable exit of existing firms andentry of new firms. Lower stability (higher φ) has twoeffects on the equilibrium. The direct effect is that itdecreases the expected lifetime of each PE firm, whichdecreases the benefit of preserving the reputation for notexpropriating creditors and makes it more difficult tocommit to no diversion. The indirect effect of a higher φis that it increases the fraction μU of firms whose skill isunknown to the market. This has a positive effect on dealactivity in the HU-equilibrium, when firms with unknownskill are able to borrow at favorable rates, and a negativeeffect on deal activity in the H-equilibrium, when onlyfirms known to be high-skill are able to borrow at favor-able rates. This argument implies that, in general, stabilityof the PE industry can have a non-monotonic effect on LBOactivity. A similar argument applies to parameter p, whichcaptures the speed with which a PE firm's skill getsrevealed in the market.20

Note also that uncertainty about PE firms' skills (whichincreases with φ and p) affects the sensitivity of buyoutactivity to large changes in discount rates. Suppose that inbust times only PE firms that are known to be high-skill areable to borrow at favorable rates (i.e., the H-equilibrium isplayed). In contrast, in boom times, PE firms of unknown skillare also able to borrow at favorable rates (i.e., the HU- or theHUL-equilibrium is played). Consider an increase in the

0.5

0.6

0.7

0.8

0.9

1.0

0.0 0.2 0.4 0.6

Mass of targets acquired

Fraction of deals by hi -gh type PE firms

Fig. 2. Buyout activity and composition of acquirers. The figure illustrates Proposition 4 by plotting how (1) deal activity (measured as the mass of targetsacquired each period) and (2) the fraction of targets acquired by private equity firms perceived to be high-skill depend on the discount rate r. Theparameters are: XB ¼ 4; XG ¼ 5; λ¼ 0:7; Dn ¼ 1:5; g Dn

� �¼ 1; g0 ¼ 0:5; qT ¼ 0:5; q ¼ 0; q ¼ 0:7; d¼ 0:1; p¼ 0:5; φ¼ 0:1; γ ¼ 0:9; and f ; f H ; f L are uniform.The discount rate r corresponds to the rate over the length of the investment.


discount rate that switches the HU- (or HUL-) equilibrium tothe H-equilibrium. The higher is the uncertainty about PEfirms' skills, the smaller is the fraction of firms known to behigh-skill. Therefore, such an increase in the discount rateleads to a greater tightening of credit and a greater drop inbuyout activity in uncertain PE markets relative to PE marketswhere sponsors' skills are known. Fig. 3 illustrates this effect:The gap between deal activity in the H -equilibrium and theHU- (or HUL-) equilibrium increases with φ. The effect ofparameter p is similar.

5. Club deals

In practice, PE firms frequently form clubs and bid forthe target as a group (Officer, Ozbas, and Sensoy, 2010;Boone and Mulherin, 2011). A common view is that PEfirms form clubs to restrict competition and therebyextract rents from shareholders of the target.21 Here, weanalyze another motive for the formation of clubs, unre-lated to competition: reputation borrowing. Specifically, ifa PE firm cannot commit to not diverting value fromcreditors but can make significant operational improve-ments in a given target, it can team up with a PE firm thatcan commit to no diversion but cannot make significantoperational improvements. Because diverting value hurtsthe reputation of all members of the club, teaming up witha high-reputation PE firm is a commitment device to notdivert value. This allows the first firm to borrow thereputation of the second firm and the second firm tocapture part of the operating value created by the firstfirm. In this section, we study the effect of club deals andreputation borrowing on buyout activity and the valuecreated in buyouts. Our main result is that even thoughclub formation is always beneficial in the context of asingle deal, it can nevertheless destroy value in the take-over market overall by lowering PE firms' ex ante incen-tives to invest in reputation for no diversion.

21 Although, in the model, we abstract from the effects of club dealson competition, our argument implies that this rent transfer can have apositive effect on the total value created in LBOs: Higher future rents canrelax the no diversion constraint of PE sponsors and thereby lower theirconflict of interest with creditors.

5.1. Timeline

We extend the model of Section 4 to allow clubformation. The timeline is as follows. In each period, thetarget is matched to two PE firms, characterized byprobabilities q1 and q2 that the target is successful undertheir ownership. These probabilities are observed by allplayers. The two bidders then decide whether to form aclub and undertake the buyout as a group. If they do so,the probability that the target is successful is max q1; q2

� �.

We assume that there are infinitesimal positive costs ofclub formation, so that the club is formed if and only if it isstrictly efficient to do so, i.e., if the bidders' joint surpluswith the club is strictly higher than in the absence of theclub. If the club is not formed, the two PE firms bid for thetarget. As in the model without club deals, bidding takesplace through the English auction, and each firm bids up toits maximum willingness to pay. If the club is formed, theclub makes a take-it-or-leave-it offer to the target. If thetarget rejects this offer, the bidders go back to competingthrough the English auction. Hence, the target's payoff ifthe club is formed is never smaller than its payoff in theabsence of the club. We deliberately abstract from theeffects of club deals on competition to focus solely on theireffects on bidders' reputation.

Finally, if the target accepts the offer and the buyouttakes place, the two bidders divide the surplus from theclub according to the Nash bargaining solution with equalsharing, in which the status quo point is the set of payoffsin the absence of the club. Thus, each club member gets itspayoff in the English auction plus half of the additionalsurplus generated due to club formation.

5.2. Analysis

According to the monotone reputation property criter-ion and the arguments in Section 4, there are again fourpotential equilibria: HUL-, HU-, H-, and N-equilibria. Wefirst find the conditions under which the club is formed.The club is not beneficial and hence is not formed if thetwo bidders have the same reputation for non-diversion.Thus, on the equilibrium path, the club can be formed onlyin the HU- and the H-equilibria and only if the two biddershave different types, only one of which can commit to nodiversion. Consider any of these two equilibria and

0.6

0.7

0.8

0.9

0.0 0.2 0.4 0.6 0.8

H- equilibrium

Mass of targets acquired

HU- equilibrium

HUL - equilibrium

N- equilibrium

Fig. 3. Effect of uncertainty about private equity firms' skills on dealactivity. The figure plots deal activity (measured as the mass of targetsacquired each period) as a function of φ in four different types ofequilibria of the model with heterogeneous private equity firms. Theparameters are: XB ¼ 4; XG ¼ 5; λ¼ 0:7; Dn ¼ 1:5; g Dn

� �¼ 1; g0 ¼ 0:5;qT ¼ 0:5; q ¼ 0; q ¼ 0:7; d¼ 0:1; p¼ 0:5; r¼ 1:15�1; γ ¼ 0:9; and f ; f H ;f L are uniform.


suppose that Bidder 1 cannot commit to no diversion andBidder 2 can. The bidders' expected joint value fromacquiring the target is V0þz1 max q1; q2

� �� with the club

and V0þmax z0 q1� �

; z1 q2� ��

without the club, where zR qð Þis given by Eq. (4). Hence, the club is formed only if q14q2,and the expected surplus from club formation is given byz1 max q1; q2

� �� ½max z0 q1� �; z1 q2� �� þ� �þ , which can berewritten as

Sclub q1; q2� �¼ ½z1 q1� ��½z1 q2� ��þ �q1Δg=ð1þ rÞ0 : ð12Þ

To analyze how club formation affects the sustainability ofequilibria with reputation for non-diversion, we formulateLemma 8.

Lemma 8. If club deals are allowed, the HUL-equilibriumexists if and only if

λXB� XB�Dn� �þ rVHULR Lð Þ�VHULNR Lð Þ� γ 1þrð Þ2 rþφ� �E Sclub χ1 ¼ L

�;�ð13Þ

the H-equilibrium exists if and only if

λXB� XB�Dn� �þ rVHR Hð Þ�VHNR Hð Þ

þ γ 1þrð Þ2 rþφ� � μLþμU

� �E½Sclubjχ1 ¼ L; χ2 ¼H�

�

�μHE Sclub χ1 ¼H; χ2 ¼H

��μU 1�pð Þ

2E Sclub χ1 ¼H;

��

q2A12þd; q

� Þ; ð14Þ

and the HU-equilibrium exists if and only if

λXB� XB�Dn� �þ rVHUR Uð Þ�VHUNR Uð Þ�ΔV ; ð15Þ

where VϱR θ� �

and VϱNR θ� �

are the values of type θ in theϱ-equilibrium without club deals without diversion and upondiversion, respectively, ϱAfHUL;HU;Hg, and ΔV 40.

First, consider how club formation affects the sustain-ability of the HUL-equilibrium. Without the last term onthe right-hand side, condition (13) is equivalent to thesustainability condition of the HUL-equilibrium in themodel without club deals. Because the last term isnegative, the ability to form clubs has a negative effecton the sustainability of the HUL-equilibrium. Intuitively,the last term reflects the reduced incentives to build a

reputation for non-diversion due to the possibility ofborrowing reputation: If a PE firm destroys its reputationby diverting value, it is able to get cheap debt financingby forming a club with a high-reputation bidder. As aconsequence, the punishment for diversion is smaller ifclub deals are possible, so the HUL-equilibrium is lesslikely to exist.

Next, consider the H-equilibrium. Different from condition(13), the right-hand side of condition (14) has three additionalterms relative to the sustainability condition of the H-equili-brium in the model without club deals, and the first term ispositive. The positive first term reflects the added incentivesto build a reputation for non-diversion due to the ability tolend reputation to a low-reputation bidder and therebyreceive part of that bidder's value from the deal. The twonegative terms are similar to the negative term in condition(13) and reflect the reduced incentives to build a reputationfor non-diversion due to reputation borrowing. The presenceof the positive term implies that the overall effect of club dealson the sustainability of the H-equilibrium is ambiguous anddepends on the stationary distribution of PE firms' types. Inparticular, as we show in the proof of Proposition 5, if φ and pare both large enough, type H finds it easier to refrain fromdiversion in the H-equilibrium if club deals are allowed.Intuitively, in this case, the probability of meeting a high-reputation bidder is very small, so reputation borrowing isunlikely.

Finally, as condition (15) demonstrates, the presence ofclub deals negatively affects the sustainability of the HU-equilibrium. Even though both the positive effect fromreputation lending and the negative effect from reputationborrowing are present in this case, the negative effectdominates. Intuitively, the effect of reputation lending issmall compared with the effect of reputation borrowingbecause relatively few high-valuation bidders need areputation for non-diversion.

While the effect of club deals on the sustainability ofthe H -equilibrium is ambiguous when there is uncertaintyabout PE firms' skill, the negative effect always dominatesif PE firms' skill is observed. (In this case, there are onlythree possible equilibria: the N-, the H-, and the HL-equilibrium.) Intuitively, in this case, the benefit of ahigh-skill PE firm from lending its reputation for non-diversion to a low-skill PE firm is relatively small becausethe low-skill firm is less likely to create a larger opera-tional value than the high-skill firm. We summarize thesetwo sets of results in Proposition 5.

Proposition 5.

1.
Allowing club deals has a negative effect on the sustainabilityof the HUL- and the HU-equilibrium and an ambiguous effecton the sustainability of the H-equilibrium.
2.
Suppose that PE firms' skill is observed, i.e., φ¼ 0. Then,allowing club deals has a negative effect on the sustain-ability of both the HL- and the H-equilibrium.
We conclude that club deals have a twofold effect onthe expected total value from LBO deals. The direct effect is


positive: Ex post, club deals increase efficiency becausethey allow synergies from reputation borrowing. However,there is also an indirect ex ante effect: Club deals affectbidders' incentives to invest in a reputation for non-diversion. In particular, when PE firms' skill is observed,which is the case in stable markets, club deals make itmore difficult for bidders to commit to no diversion.Corollary 2 describes which of the two effects dominates.

Corollary 2. Suppose that PE firms' skill is observed, i.e.,φ¼ 0, and let ρ¼ γr. There exist ρ1;ρ2;ρc1;ρc2, where ρioρci ,ρ1oρ2, ρc1oρc2, such that relative to the case in which clubdeals are not allowed, the expected value from buyouts if clubdeals are allowed is

1.

the same if ρoρ1 or ρ4ρc2;

2.

higher if ρc1oρoρ2; and

3.
lower if ρ1oρoρc1 or ρ2oρoρc2.
Intuitively, the values ρ1 and ρ1c(ρ2 and ρc2) stand for

the cutoff values of ρ above which there exists the H-equilibrium (the HL-equilibrium) without and with clubdeals, respectively. According to Part 2 of Proposition 5,ρioρci . Hence, under the efficiency refinement, when ρ isin ρ1;ρc1

� �or ρ2;ρc2

� �, the equilibrium switches from the

H- to the N-equilibrium or from the HL- to the H -equili-brium, respectively, once club deals are allowed. Thisdecreases the expected value from buyouts. In contrast,in the region ρc1;ρ2

� �, the H-equilibrium is selected both

with and without club deals. In this case, allowing clubdeals leads to additional synergies from reputation bor-rowing, which increases the expected value from buyouts.

Finally, consider how the probability of club dealschanges with drivers of buyout activity. Club deals occurwith a positive probability only in the HU- and the H-equilibrium, and hence the probability of club deals is zerowhen discount rates are very low or very high. In parti-cular, suppose that there is no uncertainty about PE firms'skills (for example, if φ¼ 0, i.e., the market is stable). Then,club deals occur only in the H-equilibrium, which isselected when ρ Aðρc1;ρc2Þ, and do not occur when ρ liesoutside this range. Hence, in this case, the probability ofclub deals follows an inverted U-shaped pattern in driversof buyout activity, such as aggregate discount rates andexpectations of future deals.

6. Common values

So far, the paper has assumed that valuations of PEfirms are private in the sense that information of onebidder about its valuation is irrelevant for the valuation ofthe other bidder. The assumption of private values isreasonable if operational gains arise due to the uniqueability of a PE firm to restructure the target that the rivalmight not possess. However, when the target is a poorlymanaged firm, whose inefficiency can be equivalentlyresolved by any PE firm, the common-value model is amore suitable one (Bulow, Huang, and Klemperer, 1999;Gorbenko and Malenko, 2014a). The two models differ in

the interpretation of where a PE firm's payoff from atransaction comes from. In the private-value setting, itcomes from the PE firm's operational skill. In thecommon-value setting, it comes from the PE firm'sinformational advantage. In this section, we show howthe model can be equivalently set up in the common-value framework.

Consider the same setup as in Section 3, but with thefollowing change. If the target is acquired by any PE firm,its probability of success is given by κZqT . The differenceκ�qT determines the inefficiency of the incumbent man-agement of the target. Parameter κ comes from twocomponents: κ ¼ κ1þκ2, where κi is an independent drawfrom distribution with cumulative distribution function (c.d.f.) Φ �ð Þ with positive support and mean E½κ�. At thebeginning of every period, each target is randomlymatched to two PE firms, and PE firm iA 1;2f g obtains animperfect signal si about κi:

si ¼κi with prob: π;~κ i with prob: 1�π;

(ð16Þ

where πA 0;1½ � and ~κ i is an independent draw fromdistribution with the same c.d.f. Φ �ð Þ. In other words, thesignal of PE firm i is fully informative about κi withprobability π and completely uninformative with prob-ability 1�π. While the PE firm observes the signal, it isunaware whether the signal is informative or not. Thissetup captures common values and can be easily extendedto allow differential informational advantage of bidders bymaking π bidder-specific (see Povel and Singh, 2006). Oncethe bidders obtain their signals, they compete in anEnglish auction. After that, the game proceeds as before.

In the proof of Proposition 6, we show that the proper-ties of the private-value setting carry over to the common-value setting. First, there is a complementarity betweenthe value that a PE sponsor creates through leverage andits skill, i.e., its ability to get private information about thevaluation. Second, if a PE firm never gets any privateinformation about potential value created (π ¼ 0), then ithas no ability to create value through leverage despiterepeated deal making.

Notwithstanding the similarities between the twomodels, they provide very different incentives for buildingreputation in the debt market. As Proposition 6 shows, aPE firm has greater incentives to invest in reputation fornon-diversion in the common-value framework.

Proposition 6. Suppose that the distribution of signals si inthe common-value model and the distribution of valuationsqi in the private-value model are such that given the samelevel of debt, the expected payoff of a PE firm from theauction is the same: E π si�s� ið Þþ

� �¼ E qi�q� i� �þh i. Let Dcvand Dpv denote the level of debt in the most efficientequilibrium in the common-value model and in the private-value model, respectively. Then, DcvZDpv.

The intuition is as follows. When deciding whether todivert value, the PE firm trades off the benefits fromdiversion with the costs of more difficult debt financingof its future portfolio companies. These costs are muchhigher in the common-value framework than in the

0.0

0.2

0.4

0.6

0.8

1.0

0 5 10 15

Debt of a non-PE-owned target

U- equilibrium

C-equilibrium

N-equilibrium

Buyout debt, D

Fig. 4. Comparative statics of buyout debt. The figure shows how buyoutdebt and debt of a non-PE-owned target change with the discount rate rand expectations of future deal activity γ in the model with identicalprivate equity firms. The x-axis corresponds to ρ¼ γr, and the y-axiscorresponds to the level of debt. The parameters are:XB ¼ 1:5; XG ¼ 4:5; λ¼ 0:7; qT ¼ 0:3; Dn ¼ 0:9; g Dð Þ ¼ 12 Dn2�ðD�DnÞ2

� �;

and qi is uniform on ½0;0:8�.


private-value framework. In common-value battles, even asmall disadvantage of a PE firm compared with its rivalscompletely destroys its ability to compete and, hence, itssurplus. Formally, as we show in the proof of Proposition

journal of financial economicsamalenko/lbos.pdfjournal of financial economics 117 (2015) 607–627...

Documents