4m ago

2 Views

0 Downloads

619.66 KB

56 Pages

Transcription

Merger Waves: Theory and EvidenceJinghua Yan*First draft: February, 2006This version: December 17, 2006ABSTRACTThis paper presents a model that incorporates product market competition into the standardneoclassical framework. The model explains why value-maximizing firms conduct mergersthat appear to lower shareholder value. In a Cournot setting, the model demonstrates aprisoners’ dilemma for merging firms in a merger wave. Consistent with the model’simplications, the paper empirically documents that horizontal mergers are followed bysubstantially worse performance when they occur during waves. Moreover, furtherempirical tests show that the empirical relation between performance and merger waves isindependent of the method of payment and increasing in the acquirer’s managerialownership. These findings are difficult to reconcile with alternative interpretations fromexisting theories.*The Wharton School at the University of Pennsylvania. Email: [email protected] thank Philip Bond, Gary Gorton, Wei Jiang, Jozsef Molnar, David Musto, Vinay Nair, Michael Roberts,Pavel Savor, Geoffrey Tate, Ding Wu, Julie Wulf, Jianfeng Yu, Paul Zurek, and seminar participants atWharton, the CRSP Forum, and Lehman Brothers for helpful discussions. I am especially grateful to RobStambaugh and my committee members, Franklin Allen (chair), Chris Géczy, Itay Goldstein, João Gomes,Robert Holthausen, and Andrew Metrick for their comments, guidance, and encouragement.1

1IntroductionOne of the most enduring puzzles in modern corporate nance is why many mergers appear to lower shareholder value.1 The existing neoclassical theory, which assumes valuemaximization and market e ciency, fails to provide an explanation. By deviating fromthe standard neoclassical assumptions, two strands of literature have provided solutions tothis puzzle. Agency theory attributes the negative post-merger stock performance to aprincipal-agent problem. Market timing theory attributes it to an overdue correction ofmispricing. In the absence of agency costs and market ine ciencies, this paper proposesan explanation by incorporating the role of product market competition into the standardneoclassical framework. In a neoclassical setting where mergers facilitate technology transfer between rms, mergers that take place outside merger waves (hereafter, o -the-wavemergers) increase shareholder value due to the value maximization principle. However, ifsuch horizontal mergers take place in a wave that is driven by technology shocks,2 the improved technology of merging rms and an increasingly concentrated market structure altersthe competitive landscape for non-merging rival rms. When merging rms’improvementin production e ciency is su ciently high, stand-alone rivals in an on-going merger wavemay face a declining pro t margin and a shrinking market share. The merger wave thusresembles a game of prisoners’dilemma: each individual pair chooses to merge despite thefact that their combined value is less than that prior to the merger wave. Therefore, mergers that take place in a merger wave (hereafter, on-the-wave mergers) may appear to lowershareholder value. The poor performance following on-the-wave mergers can neverthelessbe consistent with value maximization.Guided by the model developed in the paper, I discover that horizontal mergers arefollowed by substantially worse performance when they occur during waves. Waves areidenti ed here using the concentration or “clusteredness”of contemporaneous same-industryM&A activity. Among all horizontal mergers announced during the period from 1979 to2004, acquirer stocks in the most clustered quintile of mergers underperform those in the leastclustered quintile by 15% over one year and by 40% over two years. This relation is robustto a number of performance measures, industry classi cations, and empirical approaches.Moreover, as is predicted by the theory model, the underperformance of on-the-wave mergers1See Jensen and Ruback (1983) and Andrade, Mitchell, and Sta ord (2001) for surveys of this literature.Several papers have found that merger waves are triggered by industry-level technology or deregulationshocks, e.g., Mitchell and Mulherin (1996), Rhodes-Kropf, Robinson, Viswanathan (2005), and Harford(2005).22

is more pronounced in less concentrated or non-durable goods industries. Finally, industryrivals’performance following on-the-wave mergers is also worse than that following o -thewave mergers.The empirical relation between performance and clusteredness documented here couldpotentially be consistent with market-timing and agency theories as well. The markettiming theory, exempli ed by Shleifer and Vishny (2003) and Rhodes-Kropf and Viswanathan(2004), suggests that the acquirer uses its relatively overvalued stock as currency to purchasethe target company’s stock. Such stock market driven mergers have poor long-run stockperformance due to the correction of misvaluation. A central prediction of the market timingtheory is that stock deal acquirers underperform cash deal acquirers in the long run.3 Toexamine the possibility of the market timing theory as an alternative explanation, I showthat in the data the relation between performance and clusteredness is independent of themethod of payment. Moreover, this relation is weakened when the net sales of insider sharesare positive, indicating that market overvaluation perceived by company insiders does notdrive down stock price following merger waves. These ndings are di cult to reconcile withthe misvaluation explanation provided by the market timing theory.The agency theory of mergers, rst proposed by Jensen (1986), suggests that valuedestroying mergers are driven by the manager’s incentive to grow the rm beyond its optimalsize. More recently, Gorton, Kahl, and Rosen (2005) show that when managers have privatebene ts of control, fundamental shocks may trigger defensive merger waves. One of the keypredictions of agency theory is that low managerial ownership in the acquirer rm leadsto poor post-merger performance.4 In the data, I show that the negative relation betweenstock performance and clusteredness strengthens as acquirer managerial ownership increases,which fails to support the agency theory as an alternative explanation.The following table compares the current model with existing theories and shows the keydi erences in assumptions and empirical implications. The empirical results in this paperdistinguish among these alternatives only in the context of on-the-wave versus o -the-wavemergers.34See Loughran and Vijh (1997) and Rau and Vermaelen (1998) for evidence supporting this prediction.See Lewellen, Loderer, and Rosenfeld (1985) for evidence supporting this prediction.3

Theory (literature)Neoclassical(JR 2002)Agency(Jensen 1986,GKR 2005)Market Timing(SV 2003, RKV 2004)Neoclassical withimperfect competition(this paper)AssumptionsValueMarketmaximization e ciencyYesYesNoYesYesNoYesYesEmpirical ImplicationsCharacterizationPrediction onof merger wavesperformanceFundamentalAlways non-negativeshocksPreemptiveMixedwaves(managerial ownership:low high)MisvaluationMixedwaves(method of payment:stock cash)FundamentalMixedshocks(merger waves:on- o -the-wave)The rest of this paper is organized as follows: Section 2 presents a model with imperfectproduct market competition, which demonstrates the prisoners’dilemma faced by rms ina merger wave. Section 3 tests the empirical implications of the model and addresses anumber of alternative interpretations of the results from competing theories and hypotheses.Section 4 relates the model and ndings of this paper to the existing literature on mergers.Section 5 concludes.2The ModelThe objective of this model is to demonstrate the theoretical possibility of a merger waveequilibrium where the value-maximization principle is upheld and merging rms’ value islowered. To do this, I choose a simple framework: a static Cournot equilibrium with oneperiod of production. The current model is not the rst to investigate the role of Cournotcompetition on horizontal mergers. Prior literature has focused on product pricing andwelfare (Salant, Switzer, and Reynolds (1983) and Farrell and Shapiro (1990)), strategiccoordination on merging decisions (Fauli-Oller (2000)), and endogenous dynamics of mergers(Gowrisankaran (1999)). While these papers examine rms’incentive to merge, they do notconsider explicitly the main focus of this model regarding the implications for rm value,i.e., comparing rm value across two equilibria, the status quo and the merger wave.5 In a5Molnar (2006) is an exception in this regard: it examines mergers’ implication on rm value in asequential auction framework with two acquirers competing for one target. In contrast, this paper featuresa more general framework with randomly paired rms making simultaneous merger decisions.4

Cournot setting, the improved technology of merging rms alters the competitive landscapefor non-merging rival rms. In a merger wave equilibrium, each individual merger enhancesthe value of merging rms. Nonetheless, the value of a merged rm under the merger waveequilibrium need not be higher than the total value of two stand-alone rms under the statusquo. Thus, merging rms may face a prisoners’dilemma: each individually value-maximizingpair of rms conduct mergers that appear to lower shareholder value.Davidson and Deneckere (1985) argue that quantity setting games, such as Cournot,understate rms’ incentive to merge. They advocate the use of Bertrand competitionwith product di erentiation instead. The choice of Cournot framework in this paper isfor expositional simplicity. It can be shown that in a Bertrand framework with productdi erentiation as in Davidson and Deneckere (1985) the main conclusions of this paper remainunchanged.62.1A Simple Model of MergersConsider an existing economy with N rms, where N is even. One half of the rms ( N2 )produce using high technology and the other half produce using low technology. Let C h(C l ) denote the cost of production for high (low) technology rms (C h C l ). Each rm isendowed with K0 units of capital stock. All rms produce identical goods and strategicallyset the quantity of production in a Cournot setting. Stand-alone rms of type i produceiunits of goods, where i h; l.qi KCiAt t 0, high and low technology rms form N2 identical pairs, and each pair decideswhether or not to merge. If two rms remain stand-alone, their costs of production willremain unchanged. If two rms decide to merge, the combined rm’s technology, denoted asC hl , satis es C1hl C1h C1l , where0.is a measure of complementarity between thetwo merging rms. For example, Adidas has higher overall productivity than Reebok as isre‡ected by their market share, but Reebok has stronger distribution channels in Europeanand Asian markets than Adidas does. If the merged rm integrates the strength of each rm,its productivity will be higher than both stand-alone rms. Thus, gains from the mergerare higher if the degree of complementarity is higher. The notion of complementarity inthis model is similar to that in Rhodes-Kropf and Robinson (2006).7 If 0, then we havethe special case that C hl C h , as in Jovanovic and Rousseau (2002). When the degree of6The main conclusions also hold in a quantity-setting game with tangible assets, as in Perry and Porter(1985).7The supermodularity condition in Rhodes-Kropf and Robinson (2006) would require 1. In thisregard, the complementarity assumption in this paper, i.e.,0, is less restrictive.5

complementarity is zero, the low type rm’s technology will be completely replaced by thehigh type rm’s technology.I rule out "mergers of likes", i.e., mergers of two rms of the same type, for tractabilityreasons. One can relax this assumption and the key results remain.8 In the special caseC l C h , i.e., all rms are initially identical at t 0, this assumption becomes irrelevant. Inaddition, I rule out one rm matching with multiple rms. It takes a considerable amountof time for the acquirer to conduct a thorough due diligence on the target.9Let Ai denote the decision of pair i, where i 2 f1; 2; :::; N2 g. Ai 1 if two rms mergeand Ai 0 otherwise. Let x denote the number of pairs that decide to merge at t 0,P N2Ai . It follows that among the remaining N x rms in the economy, x rmsi.e., x i 1produce under C hl , N2x rms produce under C h , and N2x rms produce under C l . Irule out entry into the industry.At t 1, each merged or stand-alone rm adjusts its capital stock level K i such that themarginal revenue of each additional unit of capital stock equals the external cost of capitaldRiWithout loss of generality, I normalize k to 1. Thus,stock (denoted as k), i.e., dKi k. rm i’s pro t is given byKi(K i K0 );iCwhere the market clearing price for rms’output is assumed to satisfy an inverse demandfunction of a constant elasticity form,i(K i ; P ) PP X Q ;where Q denotes the aggregate industry output, i.e., Q NXq i , and1is the price elasticityi 1of demand. For expositional simplicity, X, which describes the condition of the aggregateeconomy, is normalized to 1. Since the external capital stock market is frictionless in thismodel, i.e., there is no capital adjustment cost, the initial capital stock level K0 becomesirrelevant to rms’incentive to merge. Without loss of generality, I assume that K0 0.As in the existing literature on mergers and imperfect product market competition,10 this8The empirical evidence on "mergers of likes" is mixed. Servaes (1991) nds that mergers of high M/Band low M/B have higher total returns. In contrast, Rhodes-Kropf and Robinson (2006) nd that mergerstypically pair together rms with similar M/B ratios.9Empirically, it is uncommon for one rm to make multiple sizable acquisitions within the same year.10See, for example, McCardle and Viswanathan (1994), Salant, Switzer, and Reynolds (1983), Farrell andShapiro (1990), and Gowrisankaran (1999).6

paper uses backward induction to solve the model. Production and pro ts are determined bya Cournot production game among the remaining rms in the industry. Thus, I characterizethe optimal strategy for each pair of high-low technology rms as follows: there are x mergersin the existing economy, Ai (x) 1 if and only ifhl(x 1)hI (x) l(1)(x);where I denotes the integration costs (or xed cost savings, if I 0).11A merger wave equilibrium is a pure-strategy equilibrium when the optimal response forthe N2 th pair conditional on all other ( N2 1) pairs merging is to merge:De nition 1 Merger wave equilibrium: x 1) 1, where Ai (x) is given by (1).N2is a pure-strategy equilibrium, i.e., A( N2Conventional event studies12 draw conclusions on change in rm value by comparingthe post-merger value, i.e., hl ( N2 ) I, with the status quo value, h (0) l (0), hence thefollowing de nition on the types of merger waves:De nition 2 A merger wave is value-creating (value-destroying) ifl(0) ( hl ( N2 ) I h (0) l (0)).hl N(2)I h(0) Given the number of mergers in the existing economy (x), the solution for rm pro t ina Cournot setting at t 1 is standard: Firm’s revenue depends on its capital stock directly,because it uses the capital stock to produce the revenue generating good, and indirectly,because the price of the good depends, partly, on the rm’s production:d iPK i dP dK iCiC i dK iDi erentiating the inverse demand function P QdP dQdP dK idQ dK i1 0:with respect to K i gives,P;C iQand substituting into the previous equation together with q i qi P) ;Q Ci1 (1KiCiyields,(2)11Integration costs are often dependent on rm size and synergies. The simpli ed assumption here is notcrucial to the conclusion.12See, for example, Andrade, Mitchell, and Sta ord (2001).7

which indicates that the rm internalizes the price impact in proportion to its market share,qi. Since all rms’marginal valuations of capital equate to the cost of the capital stock, (2)Qqi 1qj 1holds for all i. Therefore, (1) (1) , for any j 2 f1; 2; :::; N xg. SummingQ CiQ Cjover rms yields,C(x) (1 N x )C iqi;(3) QC(x)subject to the constraints8x 2 [0;whereC(x) N21];( N21N xCi(4) 1;C(x)x)C h ( N2 x)C l xC hl:N xC denotes the equally weighted industry average capital requirement per unit of production.1I rewrite (2) as k ( CN x )P . Hence,P (x) C(x)1(5);N xand the inverse of the demand function givesQ(x) P (x)1 (C(x)1N xC(x)(1)1:The market share of each rm becomesiq (x) C(x)(1N x)C iC(x)Q(x) N x)C iC(x)(C(x)1)1:N xFinally, the pro t function of each rm is derived as follows,1i1(x) C(x) 11{zNxaggregate economic condition111} (1N xC(x){zmarket share)C i!} 1(1N xC(x){zpro t margin)C i!;}(6)where i h; l; and hl.A rm’s pro t function is the product of three terms: the condition of the aggregateeconomy, the rm’s market share, and the rm’s pro t margin. The latter two terms8

both depend on the rm’s cost of production, C i ; relative to the average cost of productionin the economy, C, and the number of remaining rms in the economy, N x. Whenthe number of mergers increases in the economy, there are fewer rms remaining and theeconomy moves closer to monopoly, thus increasing the rival rms’market share and pro tmargin. However, when the degree of complementarity between merging rms is su cientlyhigh, the decrease in the number of competitors does not fully compensate for the increasein average competitiveness. In this case, a merger will lower the stand-alone value of rival rms.2.2Existence of Value-Destroying Merger Wave EquilibriumDepending on the values of a few key parameters, a merger wave equilibrium may enhanceor lower shareholders’value under the status quo. This section starts with the special caseof 0 as in Jovanovic and Rousseau (2002) to examine conditions for value-creating andvalue-destroying merger waves.Proposition 1 A merger wave equilibrium x N2 always payo dominates the statusquo (x 0), i.e., hl ( N2 ) I h (0) l (0), if (i) 0 and (ii) regularity conditions (4)hold.The Appendix provides a detailed proof of this proposition. Proposition 1 shows thatunder constant returns to scale and the standard neoclassical assumptions on technologicaladvancement in mergers, a merger can only have positive externalities on its rivals, i.e., thestand-alone values of rivals improve as the number of mergers in the industry increases.For rival rms, the gain from having fewer competitors (lower N x) always outweighs theloss from facing (on average) more technologically advanced competitors (lower C). If themerger wave is an equilibrium, then each pair of rms must be better o than standingstill conditional on all other pairs merging. By transitivity, merger wave equilibrium mustpayo dominate the status quo. This statement holds for all values of l. The followingproposition demonstrates that a non-zero degree of complementarity is not only necessarybut also su cient for a merger wave equilibrium to be payo dominated by the status quo.Proposition 2 There exists a value-destroying merger wave equilibrium x hl N( 2 ) I h (0) l (0); if(i) , where , the complementarity threshold, is given by l(ii) I 2hl N(2)h(0)l(0);hl N(2)h N(291)l N(2N( N2 )(1 l)1) , andN21 ;, i.e.,

(iii) regularity conditions (4) hold.The Appendix provides a detailed proof of this proposition. When the degree of complementarity between two rms is su ciently high, i.e., , mergers bring about negativeexternalities. For rival rms, the gains from having fewer competitors (lower N x) canbe outweighed by facing tougher competition in the product market (lower C). Therefore,in a merger-wave equilibrium, every pair of rms may be worse o than under the statusquo, even though each individual pair’s strategy is value maximizing. Such a merger wavewould be labeled as "value-destroying" by conventional event studies. It can neverthelessbe consistent with value maximization.Proposition 2 derives the set of conditions for value-destroying merger waves centeredupon two key parameters: and I. Since integration costs are highly idiosyncratic andunobservable, I will focus on the degree of complementarity ( ) to derive the main empiricalimplications of the model. It is a well established fact that merger waves are driven bytechnology or deregulation shocks. Such a shock would translate into a sudden increase incomplementarity in this model. For example, a technology breakthrough in online paymentprocessing leads to high complementarity between a conventional bookstore and an internetretailer. Or, a deregulation that allows telecommunication companies to operate across different states leads to an increase in complementarity between two phone companies initiallyoperating in di erent states. An increase in has two important e ects: rst, it increases rms’ incentive to merge, thus moving the economy from its status quo to a merger-waveequilibrium. This e ect of technology shocks on mergers is similar to that in Jovanovic andRousseau (2002). Second, Proposition 2 shows that when is su ciently high, the mergerwave may lower shareholder value. Thus, the model establishes an inherent link betweentwo of the most well-known empirical facts about mergers, namely merger waves and poorpost-merger performance.2.3Comparative StaticsI have shown that a merger wave may increase or lower shareholder value depending on thevalue of . This section examines the comparative statics on , the minimum threshold fora merger wave to be value-destroying, with regard to two relevant industry characteristics.Corollary 3 The complementarity threshold ( ) is decreasing in the initial number of rmsin the economy (N ). Moreover, as N ! 1, ! 0.10

Corollary 3 states that the minimum degree of complementarity required for a mergerwave to destroy private value ( ) is decreasing in the number of rms operating in theindustry (N ). The intuition is as follows: the externalities of a horizontal merger are twofold. On the one hand, a merger improves rivals’value due to higher oligopoly rents. Onthe other hand, it lowers rivals’value due to tougher competition. The increase in oligopolyrents is decreasing in the initial number of companies, e.g., a merger wave that reduces thenumber of rms from 10 to 5 yields less oligopoly rents for remaining rms than one thatreduces the number of rms from 2 to 1. Therefore, the threshold of technological synergy( ) for the competition e ect to dominate must also be decreasing in the initial number ofcompanies (N ). Hence, the model predicts that in a concentrated industry (low N ), islikely to be high and merger waves are less likely to destroy value.Corollary 4 The complementarity threshold ( ) is decreasing in the price elasticity of demand ( 1 ).Corollary 4 states that , the minimum degree of complementarity required for a mergerwave to destroy private value, is decreasing in the price elasticity of demand of the industry( 1 ). The intuition is that as price elasticity of demand increases, product market competition toughens and the competition e ect will be more pronounced everything else equal.Therefore, the minimum degree of complementarity for a merger to have destructive impacton rival rms’values is lower for highly competitive industries (high 1 ). Hence, the modelpredicts that in industries with low price elasticity of demand, merger waves are less likelyto destroy value.2.4On-the-wave and O -the-wave MergersThe baseline model demonstrates a prisoners’dilemma problem for merging rms in a mergerwave, thus establishing a link between merger wave and post-merger performance. Due toits simpli ed assumptions, the baseline model cannot accommodate two stylized empiricalfacts: (a) some mergers take place outside a merger wave, and (b) some rms remain standalone during merger waves.13 I can overcome this limitation by making the integration costsidiosyncratic, i.e.,13In the baseline model, when integration costs are extremely high, an increase in may not triggera merger wave in a sense that only the rst few pairs of rms will merge. This result of the model isless consistent with the established fact that technology shocks trigger merger waves, probably because thethreshold on integration costs is unrealistically high.11

I i I "iIwhere "iI denotes the idiosyncratic variation in integration costs for the ith pair of rms. Theassumption is motivated by the idiosyncratic nature of both physical integration (such ascomputer systems) and cultural integration. Under this simple characterization, the modeldescribes the merger activities in an industry as follows:Under normal economic conditions, the degree of complementarity is low and the incentive0)to merge is low. Only pairs with extremely low realizations of integration costs ("iI14merge outside a merger wave. The proposed merger between O ceDepot and Staples wassuch an example.15 Participants of these o -the-wave mergers are always better o thanthe sum of their stand-alone value under the status quo due to value maximization. Thisprediction has been veri ed by existing literature that indicates that horizontal mergers onaverage create value for shareholders (see, for example, Mitchell and Mulherin (1996)).Technology innovations or deregulation shocks increase the degree of complementarityand trigger a merger wave. Proposition 2 shows that when is su ciently high, a mergerwave may leave each pair of rms worse o than under the status quo. Therefore, the centralprediction of the model is that on-the-wave mergers may lower shareholder value. Moreover,in a value-destroying merger wave, some matched pairs have high positive idiosyncratic integration costs ("iI ), thus remaining stand-alone. These stand-alone rms also absorb negativeexternalities brought about by value-destroying merger waves of their rivals. Therefore, themodel also predicts that rival rms’value may fall following horizontal merger waves.Finally, the model also has two cross-industry predictions: Corollary 3 predicts thatindustries with higher concentration (lower N ) are less likely to have value-destroying mergerwaves; Corollary 4 predicts that industries with low price elasticity of demand (higher ) areless likely to have value-destroying merger waves.2.5Product Market Prices and MergersAlthough this is not the focus of this paper, the model has empirical implications on therelation between mergers and product market prices. A horizontal merger generates twocountervailing e ects on product market prices, a market power e ect and a productivee ciency e ect. When productive synergies are low, the market power e ect dominates and14The merger was later rejected by an anti-trust review by the Federal Trade Commission.In the data used in this paper, there are no other horizontal mergers in the same 4-digit SIC code overa 7-month period centered on the event month of this merger.1512

price rises. When productive synergies are high, the productive e ciency e ect prevails andproduct market price falls. The empirical evidence on the impact of horizontal mergers onproduct market pricing is mixed yet consistent with my model. Using a sample of airlinemergers from 1985 to 1988, Kim and Singal (1993) showed that merging rms raised airlineticket prices by 9% relative to the routes una ected by the mergers.16 In the sample usedin this paper, the M&A activities in the transportation industry, i.e., Fama-French industry40, are low during Kim and Singal’s sample period relative to other periods, e.g., only 38 outof the total 323 mergers were announced during this 4-year window. Therefore, the increasein product price is consistent with this paper’s theoretical prediction that low-synergy, o the-wave mergers lead to higher product market price. More recently, Focarelli and Panetta(2003) argue that it takes time to realize productive synergies. Using a unique dataset ofdeposit rates of Italian banks, they show that deposit rates fall in the short run but rise inthe long run after mergers, i.e., the market power e ect dominates in the short run, but theproductive e ciency e ect dominates in the long run. Thus, the competition e ect identi edin this paper is not limited to the sample of domestic mergers studied in this paper.33.1Empirical Methods and ResultsData and MethodsFrom Thomson Financial’s Securities Data Corporation (SDC), I obtain all domestic completed mergers or tender-o er bids from 1979 to 2004. I exclude all repurchases and leveragedbuyouts.17 I assign each acquirer and target to one of the Fama-French 48 industry groupsbased on their SIC codes recorded by SDC at the time of the announcement. If the acquirerand the target are in the same industry group, then the merger is identi ed as horizontal.For each horizontal merger, I use the number of contemporaneous horizontal mergers inthe same industry, as a measure of "clusteredness," or concentration of merger announcements. That is, I sum the number of horizontal mergers in the same industry announcedduring the announcement month, the previous 3 months, and the following 3 months. I thennormalize this number by the total number of mergers announced in that industry over the16In another related paper, Prager and Hannan (1998) nd that large horizontal mergers of banks substantially reduce deposit rates, but nancial companies are excluded in the sample due to heavy regulation.17I do not exclude deals for which less than 100% of the target shares are acquired. To avoid doublecountingof multiple announcements of the same merger, I keep one observation per calendar year for each unique pairof acquirer and target. In the nal sample, deals in which 100% of the target share was acquired accountfor over 90% of the observations.13

entire sample period. This adjusted number of contemporaneous horizontal mergers will bethe measure of clusteredness.Harford (2005) calculat

For example, Adidas has higher overall productivity than Reebok as is re ected by their market share, but Reebok has stronger distribution channels in European and Asian markets than Adidas does. If the merged –rm integrates the strength of each –rm, its productivity will be higher than both stand-alone –rms. Thus, gains from the merger