Strategic Management JournalStrat. Mgmt. J. (2015)
Published online EarlyView in Wiley Online Library (wileyonlinelibrary.com) DOI: 10.1002/smj.2389Received 17 December 2013; Final revision received 12 March 2015
A CAPABILITIES-BASED PERSPECTIVE ON TARGETSELECTION IN ACQUISITIONS
ASEEM KAUL1* and BRIAN WU2
1 Strategic Management & Entrepreneurship Department, Carlson School ofManagement, University of Minnesota, Minneapolis, Minnesota, U.S.A.2 Strategy Department, Stephen M. Ross School of Business, University of Michigan,Ann Arbor, Michigan, U.S.A.
The study of mergers and acquisitions (M&A) is atopic of central interest to the corporate strategy lit-erature. While early work on acquisitions focusedon their role in enhancing scale economies (Singhand Montgomery, 1987), and increasing marketpower (Chatterjee, 1986; Kim and Singal, 1993),a growing body of strategy literature has empha-sized a capabilities-based perspective on acquisi-tions, viewing acquisitions as a means for firmsto access and deploy capabilities and resources,1
especially those whose services cannot be directly
Keywords: acquisitions; organizational capabilities;acquisition capabilities; geographic diversification; targetchoice*Correspondence to: Aseem Kaul, Carlson School of Manage-ment, University of Minnesota, 321 19th Avenue S, 3-412 CSOM,Minneapolis MN 55455, U.S.A. E-mail: [email protected] contributed equally and are listed in alphabetical order.1We distinguish conceptually between resources, which aredefined as stocks of available factors, and capabilities, which arethe firm’s capacity to deploy these resources (Amit and Schoe-maker, 1993; Capron and Mitchell, 2009). Our focus in this study
transacted through the factor market, and that there-fore require the firm to take ownership of the assetin order to make use of it (Capron, Dussauge, andMitchell, 1998; Capron and Mitchell, 2009; Villa-longa and McGahan, 2005). More specifically, therecent literature suggests two distinct sources ofvalue from acquisitions: on the one hand, acqui-sitions may be a means for firms to deploy theirexisting resources and capabilities (Capron, 1999;Capron et al., 1998; Kaul, 2012) creating value byimproving the performance of the acquired firm(Berchicci, Dowell, and King, 2012; Jovanovic andRousseau, 2002). On the other hand, acquisitionsmay be a means for firms to acquire new resourcesand capabilities (Ahuja and Katila, 2001; Graeb-ner, 2004; Karim and Mitchell, 2000; Ranft andLord, 2002; Puranam, Singh, and Chaudhuri, 2009),
is on acquisitions as a means of deploying or acquiring capabili-ties, though to the extent that this will often require the deploymentor acquisition of the associated resources (Capron and Mitchell,2009; Karim and Mitchell, 2000), we also build on prior workthat has examined the acquisition and deployment of resourcesthrough acquisition.
A. Kaul and B. Wu
allowing them to bridge capability gaps (Capronand Mitchell, 2009) and enter new markets (Helfatand Lieberman, 2002; Lee and Lieberman, 2010).
The implications of these different sourcesof acquisition value on the acquirer’s choice oftarget remain to be fully explored, however. Priorresearch has emphasized the importance of strate-gic fit between acquirer and target, arguing andshowing that acquisition value comes from com-bining resources and capabilities that are distinctbut related and therefore complementary (Kim andFinkelstein, 2009; Larsson and Finkelstein, 1999;Makri, Hitt, and Lane, 2010; Shelton, 1988). Incontrast, studies examining acquirer target selectionhave generally found a preference for similar orless distant targets (Baum, Li, and Usher, 2000;Berchicci et al., 2012; Chakrabarti and Mitchell,2013; Schildt and Laamanen, 2006). The questionof what acquirers look for when assessing targetsthus remains open.
In this paper we study the antecedents of acquirertarget selection from a capabilities-based perspec-tive. We contend that when assessing target fitwe need to distinguish between the level of targetcapabilities and their context (Capron and Mitchell,2009), while considering both these dimensionssimultaneously. Drawing on this distinction, weargue that acquirers seeking to create value bydeploying their existing capabilities will prefertargets with weak capabilities in existing contexts,while those seeking to benefit from the acquisitionof new capabilities will prefer targets with strongcapabilities in new (though related) contexts.Between the two, we expect capability deploymentto be more strongly preferred than capability acqui-sition because of the higher costs of acquisitions innew contexts, so that the firm’s overall preferencewill be for targets in existing contexts and withweak capabilities, with the preference for weaktargets being stronger in existing contexts than innew contexts. Moreover, we expect that firms withweak acquisition capabilities will limit themselvesto acquiring inferior targets in existing markets,and only those with strong acquisition capabilitieswill pursue targets with superior capabilities and innew markets.
We formalize these arguments using a simpleanalytical model that allows us to consider thevarious benefits and costs associated with anacquisition as a function of the level and contextof target capability in an integrated, coherent, andrigorous way. The model is used to develop a set
of testable hypotheses regarding target choice forone specific type of capability in one specific typeof context—the choice of targets with high or lowmanufacturing productivity in existing or new geo-graphic markets. These hypotheses are then testedby examining horizontal acquisitions in the Chinesebrewing industry from 1998 to 2007. Using detailedproductivity data for the entire population of firmsin this industry, we show that acquirers generallyprefer targets with low levels of productivity intheir existing market, consistent with our argumentsfor capability deployment. When acquirers do buytargets in new markets, however, they are relativelymore willing to pursue superior targets, in line withcapability acquisition. These preferences are mod-erated by the acquisition capabilities of the acquirer,with weak acquisition capability firms limitingthemselves to inferior targets in existing markets,while geographically diversified or more experi-enced acquirers pursue a wider range of targets.
Our study thus contributes to a capabilities-basedunderstanding of acquisitions, highlighting the the-oretical distinction between capability deployingand capability acquiring benefits and mappingthese two distinct sources of value to the differenttypes of targets associated with them. Doing so notonly extends our understanding of what constitutesstrategic fit, it also addresses a long-standingdebate about the benefits of similarity vs. differ-ence in acquisition (Harrison et al., 1991; Kim andFinkelstein, 2009) by adopting a multidimensionalperspective (Tanriverdi and Venkatraman, 2005). Inaddition, our study provides a substantially richerunderstanding of the antecedents of acquirer targetchoice, a topic that remains relatively unexplored(Chakrabarti and Mitchell, 2013; Schildt andLaamanen, 2006). We provide both a more rigorousformal account of this key decision and a strongempirical test using detailed panel data on theentire population of potential targets.
THEORY AND HYPOTHESES
Capability level, capability context, and typesof strategic fit
As mentioned above, a substantial body of priorwork has compared the resources and capabilitiesof acquirers and targets in trying to explain eithertarget selection or acquisition performance, withsome studies arguing for the need for comple-mentarity between acquirer and target (Kim and
A Capabilities-Based Perspective on Target Selection
Finkelstein, 2009; King, Slotegraaf, and Kesner,2008; Krishnan, Miller, and Judge, 1997; Larssonand Finkelstein, 1999; Makri et al., 2010; Shelton,1988), while others highlight the importance ofsimilarity (Datta, 1991; Ramaswamy, 1997) andproximity (Baum et al., 2000; Chakrabarti andMitchell, 2013; Schildt and Laamanen, 2006). Oneway to disentangle these seemingly contradictoryarguments is to recognize that the capabilities of thetarget may be assessed along multiple dimensions.Specifically, a given functional capability of thetarget, compared with the same type of capabilityof the acquirer, could be high or low in terms oflevel and similar or different in terms of context(Capron and Mitchell, 2009), where context couldmean either product market, geographic market,or technology field. The level of the target’scapability captures how much stronger or weakerit is compared to the acquirer, while differencesin capability context determine how relevant thecapabilities of one firm are to the other (Capron andMitchell, 2009; Lee, 2008). The two dimensionsare independent of each other, moreover, so that atarget could have strong or weak capabilities in thesame or different context as the acquirer. In orderto understand strategic fit fully, then, we need toadopt a multidimensional perspective (Tanriverdiand Venkatraman, 2005; Zaheer, Castaner, andSouder, 2013) and consider simultaneously boththe level and context of a target’s capabilities.
In order to apply such a multidimensional per-spective to the acquirer’s choice of target, weconsider the sources of value creation and cap-ture from an acquisition. Prior literature suggeststwo potential sources of acquisition value from acapabilities-based perspective.2 On the one hand,acquirers can realize value by deploying their exist-ing capabilities in order to improve target per-formance (Berchicci et al., 2012; Jovanovic andRousseau, 2002). On the other hand, acquirerscan benefit by acquiring new capabilities fromthe target (Capron et al., 1998; Graebner, 2004;Kim and Finkelstein, 2009; Ranft and Lord, 2002;Rhodes-Kropf and Robinson, 2008), combiningthese with their existing capabilities in order to plugcapability gaps (Capron and Mitchell, 2009) and to
2Prior literature also suggests several sources of acquisition valuesuch as market power (Chatterjee, 1986; Kim and Singal, 1993)and economies of scale (Singh and Montgomery, 1987) that areunrelated to capabilities. We limit ourselves to capabilities-basedarguments in this study.
deepen or extend their existing capabilities (Karimand Mitchell, 2000).3
First, consider capability deployment. Since thevalue from capability deployment comes from rais-ing the target’s capability level, an acquirer wouldprefer a target with weak capabilities, since theweaker a target’s capabilities, the greater the poten-tial for improvement.4 Acquirers seeking to deploytheir existing capabilities will also prefer targetsoperating in the same or similar contexts. Deploy-ing acquirer’s capabilities to the target will only bevaluable if the two capabilities operate in identicalor overlapping contexts, since only in such a casewill the acquirer’s capabilities be relevant to the tar-get (Kim and Miner, 2007; Lee, 2008). Attempts todeploy an acquirer’s capabilities to distant contextsare unlikely to be of value and may even be harm-ful (Haleblian and Finkelstein, 1999; Kim, Kim,and Miner, 2009; Kim and Miner, 2007; Levitt andMarch, 1988; Zollo, 2009; Zollo and Reuer, 2010).We thus expect that acquirers pursuing capabilitydeployment will prefer targets with weak capabili-ties in existing contexts (Berchicci et al., 2012; Bru-ton, Oviatt, and White, 1994).
Next, consider the acquisition of new capabil-ities. Clearly, acquirers looking to acquire newcapabilities from the target will prefer targets withstrong capabilities so as to maximize value creation(King et al., 2008). And since the purpose here isnot to replace the weaker firm’s capabilities withthose of the stronger firm (as in capability deployingacquisitions) but to combine the capabilities of thetwo firms to create joint value, acquirers should beable to capture some part of the joint value createdthrough acquiring and recombining target capabil-ities, so long as they possess strong and distinctivecapabilities of their own (Capron and Pistre, 2002).This need for distinctive capabilities also meansthat acquirers seeking to acquire new capabilitieswill prefer targets in new and nonoverlapping
3Conceptually, a third source of capabilities-based value fromacquisitions could result from the complementarity between dif-ferent types of functional capabilities; for instance, by combiningthe marketing capabilities of the acquirer with the technologicalcapabilities of the target. While we do not deny the potential forsuch complementarities, our focus in this paper is limited to devel-oping theory about differences in the level and context betweencapabilities of the same (functional) type.4While acquirers could create value by using target’s capabilitiesto replace their own, they are unlikely to capture much of this valuesince it results from the target’s superior capabilities and is thuslikely to be captured by the target as a result of competitive biddingin the market (Barney, 1988; Capron and Pistre, 2002).
contexts, so as to acquire capabilities that they donot already possess (Capron and Mitchell, 2009;Karim and Mitchell, 2000; Ranft and Lord, 2002).Targets in highly overlapping contexts will havecapabilities that are largely redundant for the targetand are therefore less likely to be valuable (Kimand Finkelstein, 2009; Makri et al., 2010). Ofcourse, the two contexts will need to be related insome way (Kim and Finkelstein, 2009; Makri et al.,2010; Singh and Montgomery, 1987; Uhlenbruck,Hitt, and Semadeni, 2006) to ensure that the target’scapabilities are relevant to the acquirer (Kim andMiner, 2007; Lee, 2008). Firms seeking to acquirenew capabilities will thus prefer targets in new, butrelated, contexts.
Differences in level and context also have impli-cations for the costs of acquisition. Acquirers faceproblems of information asymmetry in identifyingand evaluating targets (Capron and Shen, 2007;Reuer, Shenkar, and Ragozzino, 2004), and theseproblems are likely to be more severe as they pur-sue targets in new and less familiar contexts (Baumet al., 2000; Chakrabarti and Mitchell, 2013; Schildtand Laamanen, 2006; Villalonga and McGahan,2005; Yang, Lin, and Lin, 2010). Differences incontext will also be associated with ex post inte-gration challenges (Haspeslagh and Jemison, 1991;Jemison and Sitkin, 1986), given differences in cul-ture (Ranft and Lord, 2002; Stahl and Voigt, 2008)and greater internal resistance (Larsson and Finkel-stein, 1999), though these challenges may be off-set by a lower need for integration (Mitchell andShaver, 2003; Zaheer et al., 2013). Integration chal-lenges are also likely to be higher when acquir-ing more capable targets, given the need to pro-tect and maintain their existing capabilities andresources from the disruptive effects of acquisition(Paruchuri, Nerkar, and Hambrick, 2006; Puranam,Singh, and Zollo, 2006b; Puranam and Srikanth,2007), though acquirers may also find it difficultto integrate extremely weak targets that may lackthe absorptive capacity to benefit from capabilitydeployment.
Together, these arguments suggest that acquir-ers will pursue two distinct types of targets, eachassociated with a distinct source of value. On theone hand, they will target firms with low levelsof capability in existing or close contexts, seekingto realize value by deploying their existing capa-bilities to these targets in order to improve theirperformance. On the other hand, they will targetfirms with high levels of capability in new (though
related) contexts, seeking to benefit from the acqui-sition of capabilities that they can recombine withtheir own. Of the two, capability deploying acquisi-tions will face both lower ex ante information costsand lower ex post integration costs than capabilityacquiring acquisitions, and thus are more likely tobe pursued.
A simple model of capabilities-based targetselection
In order to lay out these arguments more rig-orously, we develop a simple formal model ofcapabilities-based target selection. The modelis helpful because it allows us to consider thecombined effect of the various benefits and costsassociated with the acquisition in an integrated andcoherent way, to clarify our conceptual argumentin unambiguous terms, and to develop severalfine-grained and nonintuitive predictions regardingcapabilities-based target selection.
Consider two firms, A and B. The stand-alonevalue of each firm (i.e. the net present value ofits expected future cash flows,5 V), is determinedby the combination of its focal capability 𝜃, anda vector of other complementary capabilities andresources 𝜂, so that VA = 𝜃A𝜂A and VB = 𝜃B𝜂Bwhere 𝜃A > 0, 𝜂A > 0, 𝜃B > 0, 𝜂B > 0. Since weare interested in the effect of the focal capability𝜃 on target choice, we make the parsimoniousassumption that 𝜂A = 𝜂B = 𝜂.
The two firms operate in distinct but overlap-ping contexts, with the extent of overlap betweenthem captured by parameter r where higher val-ues of r mean greater overlap between contexts and1 ≥ r ≥ 0. A value of r equal to 1 means the two con-texts are identical, while a value of 0 means they areentirely unrelated.
Next, consider, without loss of generality, thecase where firm A acquires firm B. As discussedabove, a capabilities-based perspective suggeststwo sources of potential value from such an acqui-sition. First, to the extent that the capabilities ofthe two firms overlap (captured by r), the strongerfirm can deploy its capabilities to the weakerfirm, raising the weaker firm’s capabilities to itsown level; i.e., to max(𝜃A, 𝜃B). Thus, the value
5For simplicity, we assume that each firm is accurately valued bythe market (i.e. that the market value of the firm reflects the bestestimate of future cash flows and there are thus no opportunitiesfor purely speculative gains).
A Capabilities-Based Perspective on Target Selection
created through capability deployment is equal to𝜂r(2 max(𝜃A, 𝜃B)− 𝜃A − 𝜃B).
Second, to the extent that the capabilities arenonoverlapping (captured by 1 − r), each firmcould benefit from acquiring the nonoverlappingportion of the other’s capabilities. The extent of thisbenefit will be limited by the extent to which thetwo capabilities are related, however. We assumethat the extent to which each firm benefits from theother’s nonoverlapping capabilities is proportionalto the extent of overlap (r) between them. Thus,firm A will acquire additional capabilities equalto r(1− r)𝜃B, and firm B will acquire additionalcapabilities equal to r(1− r)𝜃A.6 The benefit fromcapability acquisition is thus strongest when thetarget and acquirer have moderately overlappingcapabilities (r = 0.5), consistent with prior liter-ature (Kim and Finkelstein, 2009; Makri et al.,2010)—firms have little to gain from acquiringcapabilities that are either extremely similar (andtherefore redundant) or completely unrelated (andtherefore irrelevant).
In addition to these sources of potential value,the acquisition will also be associated with severalcosts. First, the acquiring firm will face an ex anteinformation cost associated with the difficulty ofidentifying and evaluating a potential target. Asprior literature has shown, acquirers are likely tooverpay for targets, due to factors such as poor duediligence, escalation of commitment, and manage-rial hubris (Haunschild, Davis-Blake, and Fichman,1994; Hayward and Hambrick, 1997; Puranam,Powell, and Singh, 2006a). Such overpaymentrepresents a cost to the acquirer and is likely to behigher the more difficult it is for firms to accuratelyassess the value of potential synergies (Laamanen,2007). Specifically, we assume that this cost willincrease with the distance between the contextsof the two firms (Chakrabarti and Mitchell, 2013;Schildt and Laamanen, 2006) and reduce withthe buyer’s acquisition capabilities (Eisenhardtand Martin, 2000; Zollo and Singh, 2004), sincemore capable acquirers will be able to identify andevaluate targets better (Kim, Haleblian, and Finkel-stein, 2011; Laamanen and Keil, 2008). We thusmodel the information cost incurred by the acquirer
6These additional capabilities may be thought of as eitherenhancing the firm’s existing capabilities or being sepa-rately combined with the firm’s complementary resources.The two are equivalent in terms of the model since𝜂(𝜃A + r(1− r)𝜃B)= 𝜂𝜃A + 𝜂r(1− r)𝜃B.
as equal to (1 − r) I𝛼
where 𝛼 > 0 is a parameterreflecting firm A’s acquisition capabilities andI ≥ 0 is a parameter reflecting information costsspecific to the target. I may depend upon a numberof factors, such as the nature of the target, theinformation context (Capron and Shen, 2007), andprior ties between acquirer and target (Vanhaver-beke, Duysters, and Noorderhaven, 2002; Zaheer,Hernandez, and Banerjee, 2010).
Second, the acquiring firm will face an ex postcost resulting from the challenges associated withintegrating the operations of two distinct firms.We expect these costs first to increase and then todecrease with the level of target capability. On theone hand, targets with very weak capabilities willlack the absorptive capacity (Cohen and Levinthal,1990) necessary to benefit from the deployment ofacquirer capabilities, and acquirers may thus facesevere challenges in integrating significantly infe-rior targets. On the other hand, targets with capabil-ities substantially superior to those of the acquirerwill need to be handled carefully in order to pro-tect and maintain their capabilities and thus alsopose a significant integration challenge for acquir-ers (Paruchuri et al., 2006; Puranam and Srikanth,2007; Puranam et al., 2006b). We therefore expectintegration costs to first decrease and then increasewith target capability, being lowest when targetcapabilities are at the same level as those of theacquirer. Integration will also become more difficultas the distance between contexts increases. Evenas distance increases the difficulty of integration,however, it will also reduce the need for integra-tion (Mitchell and Shaver, 2003), so that integrationcosts will be high for moderately related targets butlow for both closely related targets (that are rela-tively easy to integrate) and unrelated targets (thatdo not require integration).
In line with these arguments, we model theintegration cost as r
[(1 − r) +
(||𝜃B − 𝜃A||) 𝛿𝜂] C
𝛼,
where C is a parameter reflecting integration costs(C ≥ 0) and 𝛿 is a parameter reflecting the difficultyof integrating a target with different capability levelrelative to the difficulty of integrating a target ina different context (𝛿 ≥ 0). As with informationcosts, we expect the cost of integration to be drivenin part by the buyer’s acquisition capabilities (𝛼)and in part by other contextual variables, such asdifferences in organizational culture and ownershipor the prior relationship between the two firms,reflected here in the parameter C.
Given these assumptions, the value of the mergedfirm7 is given by
Vmerged = 𝜂A
(rmax
(𝜃A, 𝜃B
)+(1−r)
(𝜃A+ r𝜃B
))+𝜂B
(r max
(𝜃A, 𝜃B
)+ (1 − r)
(𝜃B + r𝜃A
))−1𝛼
[r((1 − r) +
(||𝜃B − 𝜃A||) 𝛿𝜂)C
]= VA + VB +
[r𝜂(2max
(𝜃A, 𝜃B
)−(𝜃A + 𝜃B
))]+[𝜂r(1 − r)
(𝜃B + 𝜃A
) ]−1𝛼
[r((1 − r) +
(||𝜃B − 𝜃A||) 𝛿𝜂)C
](1)
setting 𝜂A = 𝜂B = 𝜂 and rearranging terms.Expression 1 shows that the difference between
the stand-alone value of A and B and their mergedvalue is driven by three factors: the value createdthrough the deployment of the superior firm’s capa-bilities to the inferior firm (the first term in squarebrackets), the value created through the combina-tion of the two firms’ (unrelated but relevant) capa-bilities (the second term in square brackets), and theintegration costs of the acquisition (the third term insquare brackets).
Since our focus is on A’s decision to acquireB, however, we need to consider the share of thisvalue that will be captured by A. To determinethis, we make three assumptions. First, we assumethat value created from capability deploymentwill be captured by the stronger firm (i.e., by thefirm whose capabilities are being deployed tocreate this value).8 Second, we assume that thevalue created from the combination of nonover-lapping capabilities is split equally between thetwo firms.9 Third, we assume that the information
7Note that the value of the merged firm is unaffected by theinformation cost of acquisition which is simply a transfer of valuefrom acquirer to target.8Strictly speaking, our predictions require only that the weakerfirm capture a substantially smaller share of the value fromcapability deployment than the stronger firm and that this sharenot increase with the capabilities of the stronger firm. Thus,even if acquirers were to capture some small benefit frombuying superior targets and using their capabilities to substitutefor the acquirer’s own, the predictions from our model wouldstill hold.9We assume an equal split of value for simplicity. The sharethat each party captures of the value they jointly create will bethe outcome of a complex bargaining process, modeling whichwould require making additional assumptions about the relativebargaining position of the two parties (their opportunity costs,utility functions, risk preferences, etc.) and is beyond our currentscope.
and integration costs are borne exclusively by theacquirer. Given these assumptions, the price paidby acquirer A for target B (PB
A) is
PBA = VB + max
(0, r𝜂
(𝜃B − 𝜃A
))+
r (1 − r) 𝜂(𝜃A + 𝜃B
)2
+ (1 − r) I𝛼
(2)
That is, the acquirer pays the stand-alone valueof the target, plus the target’s share of potentialsynergies, plus some excess amount resulting frominformation challenges.10 Note that VB ≤ PB
A, sothat the target always benefits from the acquisition.Using Expressions 1 and 2, we can derive the valuecaptured by the acquirer (𝜋A) as
𝜋A = Vmerged − VA − PBA
= r𝜂
[(𝜃A − 𝜃B
)+
(1 − r)(𝜃A + 𝜃B
)2
]−
1𝛼
[(1 − r) I + r
((1 − r) +
(𝜃A − 𝜃B
)𝛿𝜂
)C]
if 𝜃A ≥ 𝜃B;
r (1 − r) 𝜂(𝜃A + 𝜃B
)2
− 1𝛼
[(1 − r) I + r ((1 − r)
+(𝜃B − 𝜃A
)𝛿𝜂
)C]
if 𝜃A < 𝜃B (3)
Expression 3 highlights a key asymmetrybetween acquiring inferior and superior targets,with the benefit from acquiring a target with supe-rior capability coming entirely from the acquisitionof new capabilities, while that from a target withinferior capability comes from a combination ofcapability deployment and capability acquisition.More generally, Expression 3 shows that the valuecaptured by the acquirer is impacted by the leveland context of target capabilities. Increases intarget capability cause the benefit from capabilitydeployment to decline and those from capabilityacquisition to increase, while integration costsfirst decrease and then increase. Increases inthe distance between capability contexts causethe benefits of capability deployment to declineand the information costs to increase, while both
10The premium paid by the acquirer is given by PBA − VB and
includes both acquirer overpayment and the target’s share ofsynergies.
A Capabilities-Based Perspective on Target Selection
capability acquisition benefits and integration costsfirst increase and then decrease.
Using Expression 3, we can examine the relation-ship between target capability and the value cap-tured by the acquirer. Taking the partial derivativeof Expression 3 with respect to 𝜃B, we get
𝜕𝜋A
𝜕𝜃B= r𝜂
(𝛿C𝛼
− (1 + r)2
)if 𝜃B ≤ 𝜃A ;
r𝜂(1 − r
2− 𝛿C
𝛼
)if 𝜃B > 𝜃A (4)
Expression 4 shows that the effect of targetcapability on acquirer value capture is decreasingfor inferior targets, so long as (1+r)
2>
𝛿C𝛼
, whichis likely to be the case for potentially valuabletargets.11 Thus, acquirers pursuing inferior targetswill always prefer targets with low levels of capa-bility, so as to maximize the benefits of capabilitydeployment. Moreover, this effect is decreasingwith relatedness,12 meaning that the firm’s pref-erence for inferior targets is stronger in existingor close contexts than in new or distant contexts.For superior targets, however, the effect of targetcapability depends upon the relatedness in context.Specifically, we can define r∗ = 1 − 2𝛿C
𝛼such that
the acquirer value is increasing in target capabilitylevel for r < r*, but decreasing for r > r*. Fortargets with superior productivity, our model thuspredicts that target value will continue to fall withtarget capability level in existing contexts but willstart to rise with target capability in new contexts.Overall, the model predicts that the acquirer willalways prefer weak targets in existing or close con-texts but that this effect will grow weaker as it startsto consider targets in newer, more distant contexts.
These relationships are shown graphically inFigures 1–3. Figure 1 shows a three-dimensionalplot of the value captured by the acquirer (𝜋A)as a function of target capability level (𝜃B) andthe relatedness of target capability context (r). Itshows that the highest peak in acquirer value cap-ture occurs where targets have inferior capability
11Note that if 𝛿C𝛼
≥12, then the acquirer stands to lose half or more
of the target’s value as integration costs—a case in which theacquirer is unlikely to realize value in any case. We therefore limitour subsequent discussion to the case where 𝛿C
𝛼<
12.
12Formally, 𝜕2𝜋A
𝜕𝜃B𝜕r=−𝜂
[12+ r− 𝛿C
𝛼
]if 𝜃B ≤ 𝜃A; 𝜂
(12− r− 𝛿C
𝛼
)if 𝜃B > 𝜃A
Con
text
Rel
ated
ness
(r)
Val
ue C
aptu
re (
π A)
Target Capability (θB)
Figure 1. Effect of level and context of target capabilityon acquirer’s value capture
Val
ue C
aptu
re (
π A)
Target Capability (θB)
θA
New
Existing
Context
Figure 2. Effect of target capability level on acquirer’svalue capture
Val
ue C
aptu
re (
π A)
Target Capability (θB)
High α, Existing
High α, New
Low α, Existing
Low α, New
θA
Figure 3. Moderating effect of acquisition capability onacquirer’s value capture
and high relatedness. When pursuing targets withsimilar or superior levels of capability, however,moderate values of relatedness are better for theacquirer than high values, with Figure 1 showing asecond, though lower, peak for targets with supe-rior capability and moderate relatedness. These twopeaks correspond to the two types of strategic fitin the theory section above—a capability deploy-ing peak for targets with low levels of capabil-ity in existing contexts and a capability acquiringpeak for targets with high levels of capability in
new, though related, contexts. Moreover, the rel-ative height of the two peaks confirms our intu-ition that firms will, on average, prefer acquisitionsthat are primarily capability deploying. Figure 1also shows that acquirers do not capture value fromtargets in very distant contexts (r close to 0),13
suggesting that acquirers are very unlikely to buyextremely distant targets irrespective of the target’scapability level.
Figure 2 then shows a simplified two-dimensional version of these relationships,plotting the relationship between acquirer valuecapture and target capability level for existingcontexts and, separately, for new (moderatelyrelated) contexts. It shows that value captured bythe acquirer declines with target capability forinferior targets, with the decline being steeper fortargets in existing contexts. For superior targets,the value captured by acquirer increases with targetcapability in new contexts, while continuing todecline with target capability in existing contexts,albeit at a slower rate. Figure 2 thus predicts thatacquirers will generally prefer targets with weakercapability levels in existing contexts, with thispreference being stronger for inferior targets; whilein new contexts, the effect of target capability willbe negative for inferior targets and positive forsuperior targets. It also predicts that acquirers willgenerally prefer targets in existing contexts to thosein new contexts.
Finally, Figure 3 shows how the lines in Figure 2shift with changes in the buyer’s acquisitioncapability. As expected, the figure shows thatfirms with strong acquisition capabilities capturemore value from acquisition than those with weakacquisition capabilities, with the advantage beinggreater in new contexts than in existing contexts.In particular, Figure 3 suggests that low acquisi-tion capability firms may find it unprofitable topursue targets in new contexts, on account of thehigh information and integration costs associatedwith such targets, and may therefore focus theirattention on firms with inferior capabilities inexisting contexts. In contrast, firms with strongacquisition capabilities may generally be morewilling to pursue superior targets, and especiallylikely to do so in new contexts, on account of their
13While 𝜋A could be negative in principle, we assumethat the acquirer will never buy a target with negativeexpected value capture, so that the minimum value of 𝜋Aplotted is 0.
superior ability to keep information and integrationcosts low.
Hypotheses development
Having proposed a general theory of capabilities-based target selection and formalized it in a simplemodel, we now turn to define hypotheses based onthe theory, so as to put it to empirical test. Whilewe believe that our theory applies across a rangeof different types of capabilities and contexts, ourempirical tests in this study focus on a single typeof capability—manufacturing productivity—and asingle definition of context—geographic markets.Our purpose, then, is to define testable hypothesesabout the selection of targets by acquirers based onthe level of the target’s manufacturing productivityand its location in existing or new geographicmarkets.
Our decision to focus on geographic markets asthe salient context builds on prior work that hasargued that firms face significant challenges whenacquiring in new or distant geographic markets(Yang et al., 2010), including ex ante informationchallenges in identifying and evaluating targets(Baum et al., 2000; Chakrabarti and Mitchell,2013) and ex post integration challenges result-ing from cultural differences across geography(Bjorkman, Stahl, and Vaara, 2007; Weber andCamerer, 2003; Weber, Shenkar, and Raveh, 1996)as well as the ongoing challenges of managingacross geographical distance (Berry, Guillen, andZhou, 2010). Consistent with this work, as wellas our theoretical argument that acquirers will,on average, prefer targets in existing contexts, wepropose this baseline hypothesis:
Hypothesis 1: The probability of a potentialtarget being acquired will be lower for targets inmarkets that are new to the acquirer than in theacquirer’s existing markets.
Next, consider manufacturing productivity asour focal capability. As predicted by our model,we expect the effect of target manufacturingproductivity to vary based on whether the targetis in an existing context or a new context. Inexisting contexts the primary source of value foracquirers is capability deployment, the potentialfor which declines with target productivity. In newcontexts acquirers have less to gain from capability
A Capabilities-Based Perspective on Target Selection
deployment relative to existing contexts but havethe potential to capture value through capabilityacquisition, the benefit from which increases withtarget productivity. Thus, as our model predicts andFigure 2 shows, target productivity will have a con-sistently negative effect on acquisition likelihood inexisting markets, but in new markets this negativeeffect will be weaker for inferior targets and willbecome positive for superior targets.
In terms of hypotheses, these arguments suggesttwo things. First, they suggest that, on average,the manufacturing productivity of the target willhave a negative effect on the likelihood of it beingacquired. This follows from the fact that acquisitionlikelihood only increases with target productivityfor superior targets in new markets and falls in allother cases. Second, they suggest that this generalpreference for less productive targets will be weakerin new markets than in existing markets.14 Wetherefore predict
Hypothesis 2: The probability of a potentialtarget being acquired will be greater, the lowerthe manufacturing productivity of the target.
Hypothesis 3: The negative relation betweenthe productivity of a potential target and itsacquisition likelihood will be weaker for targetsin markets that are new to the acquirer than inthe acquirer’s existing markets.
While the hypotheses above make no distinc-tion between acquirers, our theoretical discussionsuggests that acquirers’ target preferences willvary with their acquisition capabilities. Specifically,as predicted by our formal model and shown inFigure 3, we expect that firms with weak acqui-sition capabilities will generally limit themselvesto pursuing weaker targets in existing regions, onaccount of the high information and integrationcosts of pursuing other types of targets. In con-trast, firms with strong acquisition capabilities willbe better able to overcome these information andintegration costs and are therefore likely to bothundertake more acquisitions and pursue a wider setof targets (Laamanen and Keil, 2008; Mitchell and
14In principle, our theory suggests that we should test the effectof target productivity separately for targets inferior to the acquirerand those superior to the acquirer. While we do so empiricallyusing a two-slope model (described in detail below), we do notdefine separate hypotheses for the two cases for the sake of brevity.
Shaver, 2003; Zollo and Singh, 2004). Specifically,we expect such firms to be more willing to pursuesuperior targets and those in new regions. We there-fore hypothesize
Hypothesis 4a: The negative relation between theproductivity of a potential target and its acquisi-tion likelihood will be weaker, the stronger thebuyer’s acquisition capabilities.
Hypothesis 4b: The positive moderating effectof new regions on the relation between targetproductivity and acquisition likelihood will bestronger, the stronger the buyer’s acquisitioncapabilities.
DATA AND METHODS
Chinese brewing industry
We test these predictions by examining acquisi-tions in the Chinese brewing industry from 1998to 2007, a period of substantial growth and con-solidation for China’s brewing industry. During thisperiod, industry output increased from 38.7 billionRMB to 83.1 billion RMB (all numbers are in 1998RMB), making China the largest beer market inthe world. This rapid growth was accompanied byincreasing consolidation achieved through aggres-sive acquisition activity, with the eight-firm con-centration ratio increasing from 28.7 to 67.5 percentduring the same period, turning a fragmented indus-try with more than 400 small, local brewers intoa consolidated industry with large national play-ers – a consolidation not dissimilar to the one thatoccurred in the U.S. brewing industry in the 1950s(McGahan, 1991).
Underlying this rapid growth and consolidationwas a cross-industry change in Chinese governmentpolicy. Prior to the mid-1990s, Chinese industryhad been largely regional, with high trade barriersbetween administrative regions within the coun-try. There are 31 such administrative regions inmainland China, each with different subcultures,dialects, income levels, and levels of marketdevelopment and competition (Chang and Wu,2014). Prior to deregulation, each region operatedas a self-contained market with the objective ofeach regional administration being to maximizelocal economic growth.
With the progress of economic reform and liber-alization, there was a growing impetus away from
self-contained regional markets toward greaterexploitation of scale and scope economies on anational scale (Gilley, 2001), enabled by the low-ering of inter-regional barriers by the government.The Chinese brewing industry was no exception tothis policy change, with the government loweringor removing restrictions on the sale of beer acrossregions and explicitly favoring the growth of largenational brewers.
These policy changes provide a unique opportu-nity to study target selection because they representan arguably exogenous change that makes salient alarge number of acquirer–target combinations thatwere previously untenable. A key challenge withstudying acquirer target selection empirically is thedifficulty of accounting for the endogeneity of theacquirer and target’s pre-acquisition positions. Inthe case of the Chinese brewing industry, however,there is a strong policy rationale for why acquirersdid not consider either acquiring in other regionsor consolidating within their existing region priorto our study period. As a result we have the uniqueopportunity to observe firms choosing betweena set of potential targets that they may not haveconsidered before for reasons exogenous to theacquirers themselves.
There are several other factors that make the Chi-nese brewing industry a good setting to test ourtheory empirically. First, the high levels of acqui-sition activity in a short period of time mean thatwe have sufficient variance to test our predictionsin a single industry in a single country. Second,the regional and fragmented nature of the indus-try before 1998, as well as the size and geographicdiversity of China, enables us to treat the differentregions of the country as distinct markets. Third, thestudy context provides access to detailed, compre-hensive and statutory data on the entire populationof firms in the industry, allowing us to consider thecomplete pool of potential targets.
Data
This study uses the Annual Industrial Sur-vey Database (1998–2007) from the ChineseNational Bureau of Statistics (NBS).15 The NBScollects financial information on all industrialestablishments whose sales are more than 5MRMB (roughly US$ 685,000 using the 2007
15We end our study in 2007 because it is the last year for whichdata from the NBS is available to us.
exchange rate) with each plant being treated as aseparate establishment, as it is a tax-paying legalentity.16 By law, all qualified plants in China arerequired to cooperate with the survey and submitthe requested financial information. From this fullNBS database, we extract data for the brewingindustry based on the Chinese four-digit stan-dard industry classification code (SIC 1513 prior to2003 and 1522 thereafter). Based on this plant-leveldataset, we manually identify parent firms for eachplant for each year. To do so, we first search eachfirm’s annual reports (if they are publicly listed) andwebsite. We then search newspaper and magazinearticles and analyst reports, both in Chinese andin English, through the China National KnowledgeInfrastructure (CNKI), Baidu.com, Google.com,Business Monitor Online, IBISWorld, ABI/InformGlobal, and Business Source Complete, and usethese other sources to verify and cross-check ourmatching. We consider an acquisition to haveoccurred when a change in parent firm occurs.Using this method, we identify 184 total acquisi-tions during our study time period (1998–2007).17
Finally, we construct firm-level variables byaggregating plant-level measures where necessary.
Measures
Dependent variable
Since we wish to understand the decision ofan acquiring firm to acquire a potential targetplant, we create all possible combinations ofacquirer-target-year, which is our unit of analysis.The dependent variable Acquisition is a binaryvariable, taking the value of 1 if a given firmacquires a given target plant in a given year andtaking the value of 0 if not. All plants existing attime t are considered potential acquisition targetsby a given acquiring firm. An acquiring firm or atarget plant is dropped from the sample if it dis-solves. Note that this approach allows us to accountfor the complete population of potential targetsand eliminates any concern of sample selection
16Because plants are independent legal entities in the Chinesecontext, they are called “firms” in the dataset. For the sake ofconsistent presentation, however, we call the reporting entity aplant and the ultimate owner a firm throughout the paper.17Non-beer industry firms acquired breweries in 34 cases. Becausethese non-beer industry parents became valid participants in thebrewing industry only after these acquisitions, we exclude these34 cases. These corporate parents do however enter our analysisas potential acquirers and targets after these events.
A Capabilities-Based Perspective on Target Selection
bias (Berchicci et al., 2012). Overall, there are1,229,057 acquirer-target-year combinations,including the 184 actual acquisitions.
Main predictors
We measure a firm’s manufacturing productivityusing the productivity index developed by Caves,Christensen, and Diewert (1982) and later mod-ified by Aw, Chung, and Roberts (2003).18 Thisproductivity index has several advantages overconventional parametric measures, such as theresiduals from the Cobb-Douglas production func-tion and its variants (Van Biesebroeck, 2007). Theindex is straightforward in computation, flexiblein allowing heterogeneous production technology,and allows for consistent comparison of plant levelproductivity across years. Our main independentvariable Target Productivity is the value of thisproductivity index for the potential target plant inthe previous year. Note that we include a control forAcquirer Productivity (measured in the same way)in all our models, so that the coefficient of TargetProductivity captures the effect of target capabilitycontrolling for that of the acquiring firm. In order tocalculate Acquirer Productivity for acquirers withmore than one plant, we aggregate their productivityto its weighted average, using plant sales as weights.
We operationalize markets as geographic regions,with each geographic region being treated as adistinct market. Our main predictor is then a binary
18The productivity index is defined as follows:
Productivityit =(
lnYit − lnYt
)+
t∑𝜏=2
(lnY𝜏 − lnY𝜏−1
)−[
m∑j=1
12
(Sijt + Sjt
)(lnXijt − lnXjt
)+
t∑𝜏=2
m∑j=1
12
(Sj𝜏 + Sj𝜏−1
)(lnXj𝜏 − lnXj𝜏−1
)]
where i denotes firm, t year, and j type of input (j= 1,… ,m).Yit denotes output, and Xijt denotes inputs including labor input,material input, and capital stock. Sijt denotes input shares, definedas the ratio of labor costs to output for labor input, the ratio ofmaterial costs to output for material input, and one minus laborshare and material share for capital input. The first term in thisequation captures the deviation of a firm’s output in year t from theindustry average output in that year. The second term reflects thechange in industry average output across all years. The third andfourth terms repeat the same for each input j, which are summedusing input share for each firm (Sijt) and the average input share for
each 3-digit industry (Sjt in the third term and Sj𝜏−1 in the fourthterm) in each year as weights. The productivity index measuresthe proportional difference between the productivity of firm i inyear t relative to the hypothetical firm in the base year.
variable New Region which takes the value 1 if thetarget operates in a region where the acquirer has noexisting presence and 0 otherwise.
We consider two alternate measures of acqui-sition capability. First, we consider the extent ofa firm’s geographic diversification, on the basisthat geographically diversified firms are likely tohave both greater experience, coordinating andorganizing across multiple markets and moregeneralized capabilities (Goerzen and Beamish,2003; Levinthal and Wu, 2010; Montgomery andWernerfelt, 1988; Villalonga and McGahan, 2005),and that this will enable them to better evaluate andintegrate new targets (Barkema and Vermeulen,1998; Chakrabarti and Mitchell, 2013; Zollo andWinter, 2002). Acquirer Geographic Diversifica-tion is calculated as one minus the Herfindahl indexof its sales distribution across regions.
Second, we consider the firm’s prior AcquisitionExperience (measured as the count of acquisitionsthe firm has undertaken in the past) as a measure ofacquisition capability, consistent with prior work(Haleblian and Finkelstein, 1999; Haleblian, Kim,and Rajagopalan, 2006; Kim et al., 2011; Puranamand Srikanth, 2007; Villalonga and McGahan,2005). While the two measures represent distincttheoretical constructs, they are highly related inour empirical context, since most expansion intonew areas is undertaken through acquisition. Wetherefore use both geographic diversification andacquisition experience as measures of generalacquisition capabilities.
Control variables
In addition to these main independent variables,we include a number of controls. First, to accountfor the role of market power and economies ofscale in driving acquisitions (Chandler, 1990),we include controls for (logged) values of bothAcquirer Sales and Target Sales.19 Second, wecontrol for the acquirer’s financial constraintby including a measure of Acquirer Debt Levelcalculated as the ratio of the acquiring firm’s totaldebt to its total equity. Third, we control for the
19To account for the possibility that firms are buying inferiorplants with the intent of closing them to eliminate competition, wealso look at whether targets were closed shortly after acquisition.We find no plants that were closed within four years of beingacquired, and only four acquired plants that were ever closed,suggesting that this was not a major driver of acquisitions in ourcontext, perhaps due to rapid industry growth.
nature of acquirer ownership by including dummyvariables for whether the acquirer is MajorityState Owned or Majority Foreign Owned. Wealso include a control for Ownership Difference,which takes the value 1 if the majority ownerof the acquirer is of a different type (using afive-part classification of ownership types as state,foreign, private, collective, or incorporated) fromthe majority owner of the target. Fourth, we controlfor Business Group Affiliation, an institutionalfactor in emerging markets that can influence afirm’s acquisition propensity by affecting access tointernal capital market, agency behavior, and risksharing (Chang, 2003; La Porta, Lopez-de-Silanes,and Shleifer, 1999; Ma, Yao, and Xi, 2006). Fifth,to account for differences in the richness of theinformation context (Capron and Shen, 2007),we include a Region Information Index, whichmeasures the availability of information intermedi-aries (specifically lawyers and accountants) in thetarget’s region, and is a subindex of a marketizationindex of Chinese provinces created by the NationalEconomic Research Institute (Chang and Wu,2014). Finally, to control for the extent of rivalry inthe target market we include a measure of RegionConcentration, measured as the Herfindahl indexof industry sales in the target region.
Summary statistics and correlations of these vari-ables are provided in Table 1.
Model
The dependent variable for our study is a dichoto-mous variable that captures the acquisition decisionfor all possible combinations of acquirers and tar-gets; we use a logit regression to estimate the model.A conventional logit model estimates the acquisi-tion probability with the following functional form:
lnPijt+1(
1 − Pijt+1
) =𝛽0 + 𝛽1Xit + 𝛽2Yjt + 𝛽3Zijt
+ year + region + 𝜀ijt
where Pijt+1 is the probability that the acquisitionevent occurs ( i.e. that the ith acquirer will acquirethe jth target plant in year t+1). The log odds of theprobability are estimated to be linearly affected20
20In supplementary analysis (available upon request) we alsoinclude the square of Target Productivity in both our one-slopeand two-slope models in order to test for curvilinear effects. Wefind no evidence for a significant effect of these square terms.
by a vector of the acquiring firm’s characteris-tics (Xit), target plant characteristics (Yjt) includ-ing Target Productivity and characteristics of thetarget’s region, and the New Region and Owner-ship Difference measures (Zijt). We lag all the inde-pendent variables by one year. All models containyear and region dummies. Since a given acquirercould potentially acquire multiple target plants overmultiple years, robust standard errors are used toaccount for intra-firm nonindependence of observa-tions (Rogers, 1993; White, 1980).
Since our analytical model predicts that theeffect of target productivity will vary dependingupon whether the target’s productivity is superioror inferior to that of the acquirer, we also use atwo-slope model (Baum et al., 2005; Greene, 1993),splitting our main Target Productivity measureinto a measure of Superior Productivity=TargetProductivity if Target Productivity>AcquirerProductivity and 0 otherwise, and a measure ofInferior Productivity=Target Productivity if TargetProductivity≤Acquirer Productivity and 0 other-wise, and including these two new measures andtheir interaction with our New Region dummy inour regression. For completeness, we also include adummy variable for Superior Target, which equals1 if Target Productivity>Acquirer Productivityand 0 otherwise.21
One concern with our analysis is that our sam-ple is overwhelmingly dominated by nonevents(we have 184 events out of a total of 1,229,057observations), so that a traditional logit model mayunderestimate the probability of rare events, in turnbiasing its estimation of coefficients (King andZeng, 2001). To address this issue, we used therare event logit model developed by King and Zeng(2001) and used by other researchers (Henisz andDelios, 2001; Jensen, 2003; Sorenson and Stuart,2001; Zhou, 2011).
RESULTS
The results of our analysis are shown in Table 2.Model I is the baseline model with all controls.Model II then includes our main predictors. Con-sistent with Hypothesis 1, New Region enters theregression with a negative and significant sign,
2152.4 percent of potential targets in our final sample have produc-tivity greater than the acquirer, though only 30.4 percent of targetsthat are acquired have superior productivity.
Significance *< 0.1; **< 0.05, ***< 0.01Rare-event logit models. Figures in parentheses are robust standard errors. All models include region and year fixed effects.
implying that firms are less likely to pursue targetsin new markets. Model II also shows support forHypothesis 2, with the coefficient of Target Pro-ductivity being negative and significant, implyingthat firms are less likely to buy a potential tar-get, the greater its productivity. Finally, the interac-tion between New Region and Target Productivityenters the regression with a positive and significant
coefficient. This implies that acquirers’ preferencefor less productive targets is significantly weaker innew markets than in existing markets, supportingHypothesis 3.
Model III is a two-slope model, testing our the-oretical prediction that the effect of target produc-tivity on acquisition likelihood varies with whetherthe target is superior or inferior to the acquirer. For
A Capabilities-Based Perspective on Target Selection
0
0.00002
0.00004
0.00006
0.00008
0.0001
0.00012
0.00014
0.00016
-2.82 -1.41 0.00 1.36 2.77
Acq
uisi
tion
Lik
elih
ood
Target Productivity
Existing
New
AcquirerProductivity(Mean)
-0.0004
-0.00035
-0.0003
-0.00025
-0.0002
-0.00015
-0.0001
-0.00005
0
0.00005
-2.82 -1.41 0.00 1.36 2.77
Acq
isit
ion
Lik
elih
ood
Dif
fere
nce
(Exi
stin
g -N
ew)
Target Productivity
AcquirerProductivity(Mean)
(a) (b)
Figure 4. (a) Predicted likelihood of acquisition. (b) Difference in acquisition likelihood between existing and newregions
Inferior Productivity, our model predicts a nega-tive effect for both existing and new regions withthis effect being weaker in new regions. Model IIIshows that Inferior Productivity does in fact have asignificant negative effect on acquisition likelihood,consistent with our prediction and further confirm-ing Hypothesis 2, but the interaction between Infe-rior Productivity and New Region, while positive(as predicted), is not significant. Turning to Supe-rior Productivity, our analytical model predicts thatits effect will be negative for existing regions, butpositive for new regions. Model III shows sup-port for this prediction, with the main effect ofSuperior Productivity being significant and nega-tive, while its interaction with New Region is sig-nificant and positive. We thus see a reversal ofslope in the effect of Superior Productivity betweenexisting and new regions, with a significant differ-ence between them, which is consistent with ouranalytical model and provides strong support forHypothesis 3.
Given the nonlinear nature of our model, wecannot directly interpret the interaction effectsin Table 2 (Hoetker, 2007). To understand theseinteractions better, we graph them out using thesimulation-based approach suggested by Zelner(2009). Figure 4(a) shows the predicted likelihoodof acquisition as a function of target productiv-ity for targets in new and existing regions sepa-rately (based on Model III).22 All other indepen-dent variables are set to either their sample mean
22Because we consider the entire population of potential targets,the baseline probability of acquisition is very low (184 eventsout of 1,229,057 observations), reflecting the fact that the chanceof one firm in the industry acquiring another in a given year isgenerally negligible.
(for nonbinary variables) or their sample mode (forbinary variables) (Zelner, 2009). Figure 4(b) thenplots the difference between the new and existingregion lines shown in Figure 4(a), along with a95 percent confidence interval around the predicteddifference. These plots show that acquisition likeli-hood declines with target productivity for inferiortargets in both new and existing regions, thoughthe likelihood of acquisition is significantly higherfor targets in existing regions than in new regions.For superior targets, acquisition likelihood contin-ues to decline with target productivity for targets inexisting regions but starts to rise for targets in newregions, with the difference in the two slopes beingsignificant. As the figure shows, acquisition likeli-hood is significantly lower in new regions than inexisting regions for targets with productivity similarto the acquirer but becomes higher in new regionsthan in existing regions (though not significantlyso) for targets with substantially higher productiv-ity than the acquirer, consistent with Hypothesis 3.In general, Figure 4(a) is strongly consistent withFigure 2, providing support for our theory.
These results are economically significant. Hold-ing all other variables at their average level, a targetin an existing region with manufacturing productiv-ity one standard deviation below that of the acquireris 51 percent more likely to be acquired than a targetwith productivity equal to the acquirer, while a tar-get with productivity one standard deviation abovethe acquirer is 42 percent less likely to be acquired.For targets in new regions, a target with produc-tivity one standard deviation below the acquirer is29 percent more likely to be acquired, and a targetwith productivity one standard deviation above theacquirer is 61 percent more likely to be acquired,
compared to a target with productivity equal to theacquirer.
To test Hypotheses 4a and 4b we turn to asplit sample analysis, which is shown in ModelsIV to VII. Models IV and V show the resultsof our two-slope regression in the subsamplesof geographically diversified and focused (i.e.,single region) firms, respectively, while ModelsVI and VII show them for subsamples of firmswith and without prior acquisition experience.Both sets of results show a similar pattern: first,consistent with Hypothesis 4a, the coefficient ofInferior Productivity is more negative for firmswith weak acquisition capabilities than for thosewith strong acquisition capabilities, with thedifference in coefficients being significant forfocused vs. diversified acquirers (z= 3.80***),though not for experienced vs. inexperiencedacquirers (z= 1.33). Hypothesis 4a is thus partiallysupported. Second, consistent with Hypothesis4b, we see a positive and significant interactionbetween Inferior Productivity and New Region inless capable acquirers but an insignificant (andnegative) coefficient for high capability acquirers,with the difference between them being significant(z= 3.10*** for the difference between diversifiedand focused acquirers, and z= 2.33** for thedifference between experienced and inexperiencedacquirers). Hypothesis 4b is thus supported. Third,we also see that the effects of Superior Productivityare only significant in the case of strong acquisitioncapability firms, i.e., those that are geographicallydiversified or have prior acquisition experience. Inparticular, the coefficient of the interaction betweenSuperior Productivity and New Region is positiveand highly significant for strong acquisition capa-bility firms but insignificantly different from zerofor weak acquisition capability firms. Though thedifference in these coefficients between the twotypes of acquirers is not statistically significant ineither case (largely due to the high standard error ofthe coefficients in the weak acquisition capabilitycase), these results are directionally consistent withour predictions in Hypotheses 4a and 4b.
To interpret these results more clearly, we graphthe results for Models IV and V in Figure 5(a,b),respectively, using the same approach as that usedin Figure 4(a)23 (graphs for Models VI and VII, notshown, are similar). Consistent with our theoretical
23All other variables are set to the subsample mean or modewhen drawing these figures. Note that the figures use different
predictions, these graphs show that firms withweak acquisition capabilities generally restrictthemselves to acquiring inferior targets in existingregions. They almost never acquire targets in newregions and are also very unlikely to buy superiortargets. In contrast, firms with strong acquisitioncapabilities are not only more likely to undertakeacquisitions in general, but they are open to abroader range of targets, including targets withsuperior productivity and those in new regions.Interestingly, these figures show that geographi-cally diversified acquirers have a higher propensityto acquire targets with similar capability levels inexisting areas than predicted by our model. It maybe that firms with strong acquisition capabilities areable to capture some value from “cream-skimming”acquisitions or to realize complementaritiesbetween different types of functional capabilities.That one difference aside, the consistency betweenour theoretical predictions in Figure 3 and theobserved empirical relationships in Figure 5(a,b)strongly confirm our theoretical arguments.
CONCLUSION
These empirical findings provide strong support forour theory. As predicted by our analytical model,we find that acquirers pursue weak targets in exist-ing markets, consistent with capability deployment,but are relatively more willing to acquire superiortargets when entering new markets, consistent withcapability acquisition. These preferences are mod-erated by the buyer’s acquisition capabilities, withfirms that have weak acquisition capabilities gener-ally limiting themselves to buying inferior targetsin existing markets, while those with strong acqui-sition capabilities pursue a broader range of targetsand are relatively more willing to enter new marketsand acquire superior targets.
By predicting and successfully testing theseresults, our study contributes to the M&A lit-erature in a number of ways. To begin with, itcomplements and extends recent work that offersa capabilities-based perspective on acquisitions(Capron and Mitchell, 2009; Karim and Mitchell,2000), applying this perspective to the question ofacquirer target selection. We highlight two sourcesof value from acquisition (Rhodes-Kropf and
scales, reflecting the greater baseline probability of acquisition bydiversified firms.
Robinson, 2008)—the value from deploying theacquirer’s existing capabilities to improve targetperformance (Berchicci et al., 2012; Jovanovic andRousseau, 2002) and the value of acquiring newcapabilities from the target to combine with thoseof the acquirer (Ahuja and Katila, 2001; Capronand Mitchell, 2009; Karim and Mitchell, 2000;Ranft and Lord, 2002)—and map these two distinctsources of value to two distinct types of targets, thusexpanding our conception of strategic fit (Kim andFinkelstein, 2009; Larsson and Finkelstein, 1999).In doing so, we shed new light on the long-standingdebate between the need for similarity or differ-ence in acquisitions (Harrison et al., 1991; Kim andFinkelstein, 2009), and emphasize the need to adopta multidimensional perspective when comparingcapabilities (Tanriverdi and Venkatraman, 2005;Zaheer et al., 2013).
Our study also advances our understandingof the relatively underexplored question of theantecedents of acquirer target selection. Whileprior work on target selection has mainly focusedon the information challenges associated withidentifying and valuing targets (Baum et al., 2000;Chakrabarti and Mitchell, 2013; Schildt andLaamanen, 2006), we bring a capabilities-basedperspective to bear on the antecedents of targetchoice, highlighting the different benefits froman acquisition and their implications for targetselection. We do so, moreover, by developinga rigorous formal model, one that allows us toconsider the multiple benefits and costs associatedwith acquisition in an integrated and coherent way.We are then able to validate the predictions of thismodel in a longitudinal empirical setting whileaccounting for the complete set of potential targets.
Finally, our study also contributes to the literatureon dynamic capabilities (Eisenhardt and Martin,
2000; Helfat et al., 2007; Teece, Pisano, and Shuen,1997), especially work on acquisition capabilities(Laamanen and Keil, 2008; Zollo and Singh, 2004).We show that such acquisition capabilities havea significant effect on acquirer target selection,enabling more capable acquirers to pursue sourcesof acquisition value that may be unavailable to otherfirms (Mitchell and Shaver, 2003).
As with any study, our work has limitations,which provide the opportunity for future researchand improvement. First, while our theory andresults are consistent with firms pursuing targetsthat are likely to maximize acquisition value, we donot directly test the performance of the acquisitionswe study. We therefore cannot be sure that theseacquisitions are, in fact, resulting in benefit to theacquirers; nor can we empirically confirm that thesebenefits are the result of capability deployment oracquisition, since we do not observe the deploymentor recombination of capabilities post-acquisition.Second, while we believe that our theory appliesbroadly to many different types of capabilitiesacross many different contexts, we are only ableto test our predictions for one type of capability(manufacturing productivity) across one type ofcontext (geographic markets). Future work couldbuild on our study by extending it to other empiricalcontexts, using our theory and formal model todevelop specific predictions for these contexts.Future work could also use our model to developadditional predictions, such as predictions aboutother factors that impact information or integrationcosts or predictions about differences in the levelof complementary resources. Finally, our empiricalanalysis is limited to a single industry (brewing) ina single country (China), during a period of rapidgrowth. Future work could test the generalizabilityof our findings across other industries and countries.
We thank Nick Argyres, Sea-Jin Chang, SaminaKarim and Aks Zaheer, as well as participants atthe Atlanta Competitive Advantage conference, theAcademy of Management conference, the Academyof International Business conference, and theStrategic Management Society conference for theircomments and suggestions on earlier versions ofthis paper. We are also grateful to Associate EditorAnita McGahan and two anonymous reviewers fortheir support and feedback. Brian Wu acknowledgesfinancial support from the Lieberthal-Rogel Centerfor Chinese Studies at the University of Michigan.
REFERENCES
Ahuja G, Katila R. 2001. Technological acquisitions andthe innovation performance of acquiring firms: a lon-gitudinal study. Strategic Management Journal 22(3):197–220.
Amit R, Schoemaker PJ. 1993. Strategic assets and orga-nizational rent. Strategic Management Journal 14:33–46.
Aw BY, Chung S, Roberts MJ. 2003. Productivity, output,and failure: a comparison of Taiwanese and Koreanmanufacturers. Economic Journal 113: 485–510.
Barkema HG, Vermeulen F. 1998. International expansionthrough start-up or acquisition: a learning perspective.Academy of Management Journal 41(1): 7–26.
Barney JB. 1988. Returns to bidding firms in mergers andacquisitions: reconsidering the relatedness hypothesis.Strategic Management Journal 9(S1): 71–78.
Baum JAC, Li SX, Usher JM. 2000. Making the nextmove: how experiential and vicarious learning shapethe location of chains’ acquisitions. AdministrativeScience Quarterly 45(4): 766–801.
Baum JAC, Rowley TJ, Shipilov AV, Chuang Y. 2005.Dancing with strangers: aspiration performance and thesearch for underwriting syndication partners. Adminis-trative Science Quarterly 50(4): 536–575.
Berchicci L, Dowell G, King AA. 2012. Environmentalcapabilities and corporate strategy: exploring acquisi-tions among US manufacturing firms. Strategic Man-agement Journal 33(9): 1053–1071.
Berry H, Guillen MF, Zhou N. 2010. An institutionalapproach to cross-national distance. Journal of Inter-national Business Studies 41(9): 1460–1480.
Bjorkman I, Stahl GK, Vaara E. 2007. Cultural differencesand capability transfer in cross-border acquisitions: themediating role of capability complementarity, absorp-tive capacity and social integration. Journal of Interna-tional Business Studies 38(4): 658–672.
Bruton GD, Oviatt BM, White MA. 1994. Performanceof acquisition of distressed firms. Academy of Manage-ment Journal 37(4): 972–989.
Capron L. 1999. The long-term performance of hori-zontal acquisitions. Strategic Management Journal 20:987–1018.
Capron L, Dussauge P, Mitchell W. 1998. Resourceredeployment following horizontal mergers and acqui-sitions in Europe and North America, 1988–1992.Strategic Management Journal 19(7): 631–661.
Capron L, Mitchell W. 2009. Selection capability: howcapability gaps and internal social frictions affectinternal and external strategic renewal. OrganizationScience 20(2): 294–312.
Capron L, Pistre N. 2002. When do acquirers earn abnor-mal returns? Strategic Management Journal 23(9):781–794.
Capron L, Shen J. 2007. Acquisitions of private vs. pub-lic firms: private information, target selection, andacquirer returns. Strategic Management Journal 28(9):891–911.
Caves DW, Christensen LR, Diewert WE. 1982. The eco-nomic theory of index numbers and the measurementof input, output, and productivity. Econometrica 50(6):1393–1414.
Chakrabarti A, Mitchell W. 2013. The persistent effectof geographic distance in acquisition target selection.Organization Science 24(6): 1805–1826.
Chandler ADJ. 1990. Scale and Scope: The Dynamicsof Industrial Capitalism. Belknap/Harvard UniversityPress: Cambridge, MA.
Chang SJ. 2003. Ownership structure, expropriationand performance of group affiliated companies inKorea. Academy of Management Journal 46(2):238–253.
Chang SJ, Wu B. 2014. Institutional barriers and indus-try dynamics. Strategic Management Journal 35(8):1103–1123.
Chatterjee S. 1986. Types of synergy and economic value:the impact of acquisitions on merging and rival firms.Strategic Management Journal 7(2): 119–139.
Cohen WM, Levinthal DA. 1990. Absorptive capacity: anew perspective on learning and innovation. Adminis-trative Science Quarterly 35(1): 128–152.
Datta DK. 1991. Organizational fit and acquisition perfor-mance: effects of post-acquisition integration. StrategicManagement Journal 12(4): 281–297.
Eisenhardt KM, Martin JA. 2000. Dynamic capabilities:what are they? Strategic Management Journal 21:1105–1121.
Gilley B. 2001. Breaking barriers. Far Eastern EconomicReview 164(27): 14–19.
Goerzen A, Beamish PW. 2003. Geographic scope andmultinational enterprise performance. Strategic Man-agement Journal 24(13): 1289–1306.
Graebner ME. 2004. Momentum and serendipity: howacquired leaders create value in the integration oftechnology firms. Strategic Management Journal 25:751–777.
Greene WH. 1993. Econometric Analysis (2nd edn).Macmillan: New York.
Haleblian J, Finkelstein S. 1999. The influence oforganizational acquisition experience on acquisitionperformance: a behavioral learning perspective.Administrative Science Quarterly 44(1): 29–56.
A Capabilities-Based Perspective on Target Selection
Haleblian JJ, Kim JYJ, Rajagopalan N. 2006. The influ-ence of acquisition experience and performance onacquisition behavior: evidence from the US commer-cial banking industry. Academy of Management Journal49(2): 357–370.
Harrison JS, Hitt MA, Hoskisson RE, Ireland RD.1991. Synergies and post-acquisition performance:differences versus similarities in resource allocations.Journal of Management 17(1): 173–190.
Haspeslagh PC, Jemison DB. 1991. Managing Acquisi-tions: Creating Value through Corporate Renewal. FreePress: New York.
Haunschild PR, Davis-Blake A, Fichman M. 1994. Man-agerial overcommitment in corporate acquisition pro-cesses. Organization Science 5(4): 528–540.
Hayward MLA, Hambrick DC. 1997. Explaining thepremiums paid for large acquisitions: evidence ofCEO hubris. Administrative Science Quarterly 42(1):103–127.
Helfat CE, Lieberman MB. 2002. The birth of capabilities:market entry and the importance of pre-history. Indus-trial and Corporate Change 11(4): 725–760.
Henisz WJ, Delios A. 2001. Uncertainty, imitation, andplant location: Japanese multinational corporations,1990–1996. Administrative Science Quarterly 46(3):443–475.
Hoetker G. 2007. The use of logit and probit models instrategic management research: critical issues. StrategicManagement Journal 28(4): 331–343.
Jemison DB, Sitkin SB. 1986. Corporate acquisitions: aprocess perspective. Academy of Management Review11(1): 145–163.
Jensen M. 2003. The role of network resources in marketentry: commercial bank’s entry into investment banking1991–1997. Administrative Science Quarterly 48(3):466–497.
Jovanovic B, Rousseau PL. 2002. The Q-theory of merg-ers. American Economic Review 92(2): 198–204.
Karim S, Mitchell W. 2000. Path-dependent andpath-breaking change: reconfiguring businessresources following acquisitions in the US medicalsector, 1978–1995. Strategic Management Journal 21:1061–1081.
Kaul A. 2012. Technology and corporate scope: firm andrival innovation as antecedents of corporate transac-tions. Strategic Management Journal 33(4): 347–367.
Kim JY, Finkelstein S. 2009. The effects of strategic andmarket complementarity on acquisition performance:evidence from the US commercial banking indus-try, 1989–2001. Strategic Management Journal 30(6):617–646.
Kim JY, Haleblian J, Finkelstein S. 2011. When firms aredesperate to grow via acquisition: the effect of growthpatterns and acquisition experience on acquisitionpremiums. Administrative Science Quarterly 56(1):26–60.
Kim JY, Kim JY, Miner AS. 2009. Organizational learningfrom extreme performance experience: the impact of
success and recovery experience. Organization Science20(6): 958–978.
Kim JY, Miner AS. 2007. Vicarious learning from thefailures and near-failures of others: evidence fromthe U.S. commercial banking industry. Academy ofManagement Journal 50(2): 687–714.
Kim EH, Singal V. 1993. Mergers and market power:evidence from the airline industry. American EconomicReview 83(3): 549–569.
King DR, Slotegraaf RJ, Kesner I. 2008. Performanceimplications of firm resource interactions in the acqui-sition of R&D-intensive firms. Organization Science19(2): 327–340.
King G, Zeng L. 2001. Logistic regression in rare eventsdata. Political Analysis 9(2): 137–163.
Krishnan HA, Miller A, Judge WQ. 1997. Diversificationand top management team complementarity: is per-formance improved by merging similar or dissimilarteams? Strategic Management Journal 18(5): 361–374.
Laamanen T. 2007. On the role of acquisition premiumin acquisition research. Strategic Management Journal28: 1359–1369.
Laamanen T, Keil T. 2008. Performance of serial acquir-ers: toward an acquisition program perspective. Strate-gic Management Journal 29(6): 663–672.
La Porta R, Lopez-de-Silanes F, Shleifer A. 1999. Corpo-rate ownership around the world. Journal of Finance54(2): 471–517.
Larsson R, Finkelstein S. 1999. Integrating strategic, orga-nizational, and human resource perspectives on mergersand acquisitions: a case survey of synergy realization.Organization Science 10(1): 1–26.
Lee GK. 2008. Relevance of organizational capabilitiesand its dynamics: what to learn about entrants’ prod-uct portfolios about the determinants of entry timing.Strategic Management Journal 29(11): 1257–1280.
Lee GK, Lieberman MB. 2010. Acquisition vs. internaldevelopment as modes of market entry. Strategic Man-agement Journal 31(2): 140–158.
Levinthal DA, Wu B. 2010. Opportunity costs andnon-scale free capabilities: profit maximization, corpo-rate scope and profit margins. Strategic ManagementJournal 31(7): 780–801.
Levitt B, March JG. 1988. Organizational learning. AnnualReview of Sociology 14: 319–340.
Ma X, Yao X, Xi Y. 2006. Business group affiliation andfirm performance in a transition economy: a focus onownership voids. Asia Pacific Journal of Management23(4): 467–483.
Makri M, Hitt MA, Lane PJ. 2010. Complementary tech-nologies, knowledge relatedness and invention out-comes in high technology mergers and acquisitions.Strategic Management Journal 31(6): 602–628.
McGahan AM. 1991. The emergence of the nationalbrewing oligopoly: competition in the American market1933–1958. Business History Review 65(2): 229–284.
Mitchell W, Shaver JM. 2003. Who buys what? Howintegration capability affects acquisition incidence andtarget choice. Strategic Organization 1(2): 171–201.
Montgomery CA, Wernerfelt B. 1988. Diversification,Ricardian rents, and Tobin’s Q. Rand Journal of Eco-nomics 19(4): 623–632.
Paruchuri S, Nerkar A, Hambrick DC. 2006. Acquisitionintegration and productivity losses in the technicalcore: disruption of inventors in acquired companies.Organization Science 17(5): 545–562.
Puranam P, Powell BC, Singh H. 2006a. Due diligencefailure as a signal detection problem. Strategic Orga-nization 4(4): 319–348.
Puranam P, Singh H, Zollo M. 2006b. Organizing for inno-vation: managing the coordination-autonomy dilemmain technology acquisitions. Academy of ManagementJournal 49(2): 263–280.
Puranam P, Singh H, Chaudhuri S. 2009. Integratingacquired capabilities: when structural uncertaintyis (un)necessary. Organization Science 20(2):313–328.
Puranam P, Srikanth K. 2007. What they know vs.what they do: how acquirers leverage technologyacquisitions. Strategic Management Journal 28(8):805–825.
Ramaswamy K. 1997. The performance impact of strategicsimilarity in horizontal mergers: evidence from the USbanking industry. Academy of Management Journal40(3): 697–715.
Ranft AL, Lord MD. 2002. Acquiring new technolo-gies and capabilities: a grounded model of acqui-sition implementation. Organization Science 13(4):420–441.
Reuer JJ, Shenkar O, Ragozzino R. 2004. Mitigating riskin international mergers and acquisitions: the role ofcontingent payouts. Journal of International BusinessStudies 35: 19–32.
Rhodes-Kropf M, Robinson DT. 2008. The market formergers and the boundaries of the firm. Journal ofFinance 63(3): 1169–1211.
Rogers WH. 1993. Regression standard errors in clusteredsamples. Stata Technical Bulletin 13: 19–23.
Schildt HA, Laamanen T. 2006. Who buys whom:information environments and organizational boundaryspanning through acquisitions. Strategic Organization4(2): 111–113.
Shelton LM. 1988. Strategic business fits and corporateacquisition: empirical evidence. Strategic ManagementJournal 9(3): 279–287.
Singh H, Montgomery CA. 1987. Corporate acquisitionstrategies and economic performance. Strategic Man-agement Journal 8(4): 377–386.
Sorenson O, Stuart TE. 2001. Syndication networksand the spatial distribution of venture capital invest-ments. American Journal of Sociology 106(6):1546–1588.
Stahl GK, Voigt A. 2008. Do cultural differencesmatter in mergers and acquisitions? A tentativemodel and examination. Organization Science 19(1):160–176.
Tanriverdi H, Venkatraman N. 2005. Knowledge relat-edness and the performance of multibusiness firms.Strategic Management Journal 26(2): 97–119.
Teece DJ, Pisano G, Shuen A. 1997. Dynamic capabili-ties and strategic management. Strategic ManagementJournal 18(7): 509–533.
Uhlenbruck K, Hitt MA, Semadeni M. 2006. Marketvalue effects of acquisitions involving internet firms:a resource-based analysis. Strategic Management Jour-nal 27(10): 899–913.
Van Biesebroeck J. 2007. Robustness of productivityestimates. Journal of Industrial Economics 55(3):529–569.
Vanhaverbeke W, Duysters G, Noorderhaven N. 2002.External technology sourcing through alliances oracquisitions: an analysis of the application-specificintegrated circuits industry. Organization Science13(6): 714–733.
Villalonga B, McGahan AM. 2005. The choice amongacquisitions, alliances, and divestitures. Strategic Man-agement Journal 26(13): 1183–1208.
Weber RA, Camerer CF. 2003. Cultural conflict andmerger failure: an experimental approach. ManagementScience 49(4): 400–415.
Weber Y, Shenkar O, Raveh A. 1996. National and corpo-rate culture fit in mergers/acquisitions: an exploratorystudy. Management Science 42(8): 1215–1227.
White H. 1980. A heteroskedasticity-consistent covariancematrix estimator and a direct test for heteroskedasticity.Econometrica 48(4): 817–838.
Yang H, Lin Z, Lin Y. 2010. A multilevel frameworkof firm boundaries: firm characteristics, dyadic differ-ences, and network attributes. Strategic ManagementJournal 31(3): 237–261.
Zaheer A, Castaner X, Souder D. 2013. Synergy sources,target autonomy, and integration in acquisitions. Jour-nal of Management 39(3): 604–632.
Zaheer A, Hernandez E, Banerjee S. 2010. Prioralliances with targets and acquisition performance inknowledge-intensive industries. Organization Science21(5): 1072–1091.
Zelner BA. 2009. Using simulation to interpret results fromlogit, probit, and other non-linear models. StrategicManagement Journal 30(12): 1335–1348.
Zhou YM. 2011. Synergy, coordination costs, and diver-sification choices. Strategic Management Journal 32:624–639.
Zollo M. 2009. Superstitious learning with rare strategicdecisions: theory and evidence from corporate acquisi-tions. Organization Science 20(5): 894–908.
Zollo M, Reuer JJ. 2010. Experience spillovers acrosscorporate development activities. Organization Science21(6): 1195–1212.
Zollo M, Singh H. 2004. Deliberate learning in corporateacquisitions: post-acquisition strategies and integrationcapability in US bank mergers. Strategic ManagementJournal 25(13): 1233–1256.
Zollo M, Winter SG. 2002. Deliberate learning and the evo-lution of dynamic capabilities. Organization Science13(3): 339–351.
