Strategic Management JournalStrat. Mgmt. J., 33: 938–964 (2012)

Published online EarlyView in Wiley Online Library (wileyonlinelibrary.com) DOI: 10.1002/smj.1954

Received 17 July 2008; Final revision received 10 January 2012

A CONTEXTUAL THEORY OF ORGANIZATIONALLEARNING FROM FAILURES AND SUCCESSES: ASTUDY OF ACQUISITION COMPLETION IN THEGLOBAL NEWSPAPER INDUSTRY, 1981–2008

KATRIN MUEHLFELD,1* PADMA RAO SAHIB,2

and ARJEN VAN WITTELOOSTUIJN3

1 Utrecht University, Utrecht, the Netherlands; and University of Antwerp, Antwerp,Belgium2 University of Groningen, Groningen, the Netherlands3 University of Antwerp, Antwerp, Belgium; University of Tilburg, Tilburg, theNetherlands; and Utrecht University, Utrecht, the Netherlands

This study develops and tests theory about the context-specificity and outcome-dependenceof experiential learning in acquisition processes. First, we investigate whether learning fromexperience gained in different acquisition contexts is limited to influencing subsequent outcomesof same-context transactions. Second, we analyze whether learning patterns in response toprior successes and failures differ across acquisition contexts, depending on two propertiesof these contexts—the degree of structural variance and the level of stimulation of deliberatelearning. Learning is assessed with respect to an underexplored organizational goal variable inacquisitions: completion of a publicly announced transaction. An analysis of 4,973 acquisitionattempts in the newspaper industry in 1981–2008 largely supports our theory. Copyright 2012John Wiley & Sons, Ltd.

INTRODUCTION

In 2007, the worldwide volume of mergers andacquisitions (M&As)1 reached more than$4.74 trillion according to The Wall Street Journal(Grocer, 2007). Despite their sustained popularity,M&As remain highly controversial (Larsson and

Keywords: organizational learning; experiential learn-ing; contextualization; acquisition completion; newspaperindustry*Correspondence to: Katrin Muehlfeld, Utrecht UniversitySchool of Economics, Utrecht University, Kriekenpitplein 21-22,3584 EC Utrecht, the Netherlands. E-mail: K.Muehlfeld@uu.nl1 In line with prior work (King et al., 2004), we use the terms‘acquisition’ and ‘merger’ interchangeably to refer to interfirmactivities resulting in shared ownership in a new firm, or in oneparty acquiring stakes in another firm. In practice, the choiceof terminology is often strategically or politically motivated(Weston, Mitchell, and Mulherin, 2004).

Finkelstein, 1999). The ongoing debate about post-acquisition performance implications of M&As(King et al. 2004; Zollo and Meier, 2008) pointsto fundamental questions as to whether and underwhat conditions organizations learn from pastacquisition experience (Haleblian and Finkelstein,1999; Hayward, 2002).

Organizational learning theory holds thatexperienced-based learning promotes performancethrough its effects on knowledge creation andtransfer, and by inducing changes to organizationalpractices, strategies, and structures (Cyert andMarch 1963; Levitt and March, 1988). Empirically,positive returns to the accumulation of operatingexperience are one of the most robust findingsin organizational learning (Argote, 1999). Thewell-known learning curve—unit cost reductions

Copyright 2012 John Wiley & Sons, Ltd.

Context-Specific Learning toward M&A Completion 939

as a function of cumulative output—rests firmlyon the psychological theory of reinforcement learn-ing through repetition. However, outside of opera-tional settings, the effects of experience on learningand outcomes have proved more elusive. Com-plex, unstructured task environments—as typi-cal of corporate-level administrative and strategicactivities—impede reinforcement learning andexacerbate the identification of links between cur-rent actions and observed outcomes (Denrell, Fang,and Levinthal, 2004). Hence, empirical resultshave been mixed, as exemplified by the literatureon acquisition learning (Haleblian et al., 2009).Fowler and Schmidt (1989), for example, founda positive effect of total cumulative M&A experi-ence on subsequent acquisition performance, whileHayward (2002) and Kusewitt (1985) reporteda negative relationship. Zollo and Singh (2004)found no significant effect.

In response, scholars have advanced organiza-tional learning theory in three major directions.First, they have focused on the role of experi-ence attributes such as recency or degree of cod-ification, rather than the quantity of accumulatedexperience as such (e.g., Hayward, 2002; Zollo andSingh, 2004). Second, they have identified contin-gencies that moderate the way in which experienceaffects learning, such as the acquirer’s level ofprior M&A experience (Haleblian and Finkelstein,1999) or target firm performance (Bruton, Ovi-att, and White, 1994). Third, scholars have movedbeyond the traditional assessment of learning basedon the acquirer’s financial performance followinga specific deal. On the one hand, they have broad-ened the perspective by investigating how indi-vidual transactions fit into and contribute to theperformance of acquisition programs that consistof multiple M&As (Barkema and Schijven, 2008;Laamanen and Keil, 2008; Shi and Prescott, 2011).On the other hand, in recognition of the complexityof M&A processes and the heterogeneity of orga-nizational capabilities required during each stage,scholars have started to analyze learning in indi-vidual phases, employing stage-specific learningindicators such as the premium paid (Beckman andHaunschild, 2002; Haunschild, 1994).

Despite these significant advances, importantquestions about firms’ learning from acquisitionexperience remain unanswered. In particular, weknow little about the dependence of such learn-ing on the context in which the experience wasgained. This is an important gap because theory

and evidence suggest that learning is highly spe-cific to context, especially when it concerns com-plex, unstructured task settings (Haleblian andFinkelstein, 1999; Williams, 2007). This studyaddresses this gap by contextualizing experientiallearning in acquisition processes. Researchers inorganizational studies have approached the con-text construct with different foci, emphasizing,for instance, salience or strength of situationalfeatures (Johns, 2006). In the present study, weaccentuate context as a configuration or bun-dle of stimuli attached to certain events (acqui-sition attempts), an approach that allows us toanalyze the impact of discrete dimensions of con-text (Johns, 2006). Specifically, we consider dis-tinct contexts as delineated by industry background(intra-industry vs. diversifying), country borders(domestic vs. cross-border), and surrounding atti-tude (hostile vs. friendly). Within each of thesix contexts, we further distinguish between expe-rience with successes and failures. The way inwhich experiential learning is conditional on prioroutcomes has received little empirical attentioncompared with the literature on learning fromaggregate experience (Argote, 1999). A limitedbody of work has investigated distinct effects offailures and successes (Baum and Dahlin, 2007;Haleblian, Kim, and Rajagopalan, 2006; Hayward,2002; Madsen and Desai, 2010). Few of thesestudies have included both types of outcomes.Only one has directly compared learning fromfailure with learning from success (Madsen andDesai, 2010), and none has analyzed the context-specificity of experiential learning from successand failure, as we do here.

We investigate learning with respect to an under-explored stage of the M&A process: the publictakeover phase. It starts with the initial publicannouncement (IPA) of a transaction and ends withits consummation (Boone and Mulherin, 2007).We use the completion of an announced acquisi-tion as a simple yet comprehensive measure of thelearning that takes place during this specific stageof the M&A process (Weston et al., 2004). Thisimplies that we use the term learning to reflectorganizational processes that result in a subse-quent increase in the likelihood to complete a focalM&A. Thereby, we respond to calls to investigateorganizational behavior directed at goal variablesother than immediate firm profitability (Greve,2008). Learning and outcomes of individual phasesare important in their own right (Haspeslagh and

Copyright 2012 John Wiley & Sons, Ltd. Strat. Mgmt. J., 33: 938–964 (2012)DOI: 10.1002/smj

940 K. Muehlfeld, P. Rao Sahib, and A. Van Witteloostuijn

Jemison, 1991). First, each stage, ranging fromtarget selection to deal negotiation and integra-tion, either lays the foundations for success orfails to do so. Second, organizational resources ateach stage can be used more or less efficiently.Despite the importance of the topic, few studieshave investigated stage-specific learning. Amongthese, the predominant focus has been on the inte-gration phase (Zollo and Singh, 2004). Only twoother studies, to the best our knowledge, havesystematically explored learning during the pre-completion stage (Beckman and Haunschild, 2002;Haunschild, 1994), but both of these focused on adifferent step in the decision-making process—thepremium determination. A third study, by Dikova,Rao Sahib, and van Witteloostuijn (2010) analyzedthe public takeover phase, as we do. However,in their study, experience is considered only as amoderating variable, and the focus is on the effectof institutional differences on cross-border acqui-sitions only.

Hence our main research question is: do firmslearn from success and failure experiences—gained in different acquisition contexts—how tocomplete the final stages of acquisition prepa-ration of an announced deal? To this end, weaddress three questions. First, do firms’ experi-ences with either completing or terminating M&Asincrease their odds of completing a focal transac-tion? Second, is learning from success and failureexperiences restricted to subsequent transactionsin the same context? Third, do learning patternsdiffer across contexts depending on how muchM&As within a contextual category vary in struc-ture and depending on the extent to which thecontext stimulates deliberate learning? By analyz-ing these questions, this study makes three con-tributions to the literature. First, it adds to theorganizational learning literature by conceptualiz-ing distinct effects of successes and failures onexperiential learning and by analyzing whether theresulting learning is restricted to future M&Asin the same context. In so doing, we investigateseveral types of context that have received lit-tle attention in organizational learning studies andexamine whether they confer structural similar-ity—domestic vs. cross-border and friendly vs.hostile acquisition contexts, in addition to industrybackground (intra-industry vs. diversifying). Sec-ond, we theorize about how properties of differentcontexts give rise to specific learning patterns.Third, we contribute to acquisition learning studies

by applying our theory to a novel anchor oflearning: the intermediate goal of completing anannounced M&A.

RESEARCH SETTING: THENEWSPAPER INDUSTRY

The newspaper industry was hit by several conver-gence shocks in the past decades: in the 1980s, bythe rise of cable and satellite TV; in the 1990s, byderegulation of related information industries; and,in the 2000s, by digitization. These developmentsfundamentally changed the preconditions for infor-mation production and delivery. While the coreproducts of newspaper firms are news and adver-tising, the industry has traditionally defined itselfalso through its medium—paper. In this respect, itis an industry in decline with advertising and hardcopy subscriptions dropping rapidly (Economist,2006). In response, M&As became an importantstrategic tool aimed either at consolidation or atrelated product or geographic diversification.

Product-related diversification was largelydriven by the rise of the Internet and informa-tion and communications technology (ICT) indus-tries, enabled by regulatory changes such as the1996 U.S. Telecommunications Act. The Act abol-ished many of the cross-market barriers that pre-vented dominant players in one communicationsindustry from providing services in other sec-tors (Pritchard, 2001). It implied, for example,that newspaper firms could enter into the cableindustry and vice versa. These were importantdevelopments, as economies of scale and scopecan arise from the multiple utilization of con-tent via windowing, cross-product promotions,umbrella branding, and ‘bulk purchasing’ of inputs(van Kranenburg, 2002). Newspaper companiesresponded accordingly. For example, BloombergBusinessWeek reported on 30 May, 2003, that

‘Media General Inc., in Richmond, Va., willbe raring to go on a shopping spree. With25 daily newspapers in the Southeast, theregional powerhouse is seeking TV stationswhere it publishes its leading papers, includ-ing the Richmond Times-Dispatch. But thecompany’s ambitions don’t stop there.“Wherever we have a newspaper, we wanta television station,” says CEO J. StewartBryan III.’ (Yang and Weber, 2003)

Copyright 2012 John Wiley & Sons, Ltd. Strat. Mgmt. J., 33: 938–964 (2012)DOI: 10.1002/smj

Context-Specific Learning toward M&A Completion 941

Geographical diversification is largely drivenby divergence in industry conditions in devel-oped and emerging markets. For example, circu-lations increased by 6.72 percent in 2007 in SouthAmerica but declined by 2.14 percent in NorthAmerica and by 3.46 percent in the United King-dom (World Association of Newspapers, 2009).Further, reflecting the economic growth of coun-tries such as China and India, Asia was home to74 of the world’s 100 highest-circulation dailiesin 2007. Yet, cross-border M&As are less per-vasive than these figures might suggest. This ispartly because purchasing power is still low inemerging markets—The Times of India report-edly sold 3.5 million copies in 2007, but for justthree rupees each (about four pence) (Economist,2008)—and partly due to newspapers’ sensitivityto cultural preferences and a country’s commu-nication infrastructure (Chan-Olmsted and Chang,2003). M&As by foreign firms sometimes facerestrictions beyond usual regulatory concerns. Asnewspapers are deeply embedded in the culturaland political environment, geographic diversifica-tion typically takes the form of clustered diversi-fication—that is, firms focus on a limited set ofcountries (Chan-Olmsted and Chang, 2003).

The newspaper industry is especially well suitedfor this study. First, in declining industries, dealcompletion constitutes a particularly importantgoal variable for prospective acquirers (Aguileraand Dencker, 2008). As both consolidating anddiversifying M&As are crucial for firm survival,abandonment is even more likely to be perceivedas failure. Second, the contexts that we studyare sharply delineated in the newspaper industry,owing in part to the industry’s special politicaland cultural status (Bagdikian, 2004). Newspaperfirms have responded to consolidation and diver-sification pressures by engaging in large num-bers of M&As that are both important and dis-tinct with respect to their strategic motivation andprocedural properties. To the extent that news-papers have embraced diversifying M&As, theytypically focus on a small set of related coun-tries and a limited number of sectors, especiallyprinting and publishing, and ICT. Third, due to itsstructure with distinct local, regional, and nationalmarkets, newspaper firms of all sizes engage inM&As.

THEORY AND HYPOTHESES

Organizational learning refers to processes bywhich organizations encode inferences from expe-rience into knowledge or routines such that theirknowledge or subsequent behavior alters system-atically (Argote, 1999; Cyert and March, 1963).Experiential learning theory emphasizes the emer-gence and development of routines as the resultof experiences. Routines are collective, repetitive,and stable patterns of activities within organiza-tions (Nelson and Winter, 1982). They emerge andevolve in response to experiences organizationalmembers gain with particular activities. By extend-ing the concept of routine beyond operational andadministrative activities to include strategic deci-sion making, many types of organizational behav-ior can be understood as routinized reactions torecognized organizational events (Finkelstein andHaleblian, 2002; Winter, 1987).

Intermediate goal variables in acquisitionprocesses

Phases of the acquisition process

The process of acquiring another firm involvesmany activities that can be grouped into two broadphases: the decision-making phase and the integra-tion phase, with the official resolution as cutoff(Haspeslagh and Jemison, 1991). The decision-making phase can be subdivided into two stages(Boone and Mulherin, 2007). During the pri-vate takeover process, acquirers usually considerseveral potential targets, and vice versa. Partiestypically sign confidentiality agreements throughwhich they obtain access to private informationfor due diligence procedures and for assessingstrategic and organizational fit. They engage ininitial negotiations and sign a preliminary con-tract. Then, by officially announcing the deal, theyenter the public takeover period. The official con-summation date (informing the public about com-pletion/abandonment) marks its end. This studyfocuses on the public takeover phase. Meyer andAltenborg (2008) have illustrated its key role in theM&A process. They analyzed the merger betweenthe telecom corporations of Norway (Telenor) andSweden (Telia), announced on 20 January, 1999.Despite winning praise from analysts and the press,it was abandoned in December 1999. The analysisof the nine-month long intermediary phase showedthat organizational compatibility, complementary

Copyright 2012 John Wiley & Sons, Ltd. Strat. Mgmt. J., 33: 938–964 (2012)DOI: 10.1002/smj

942 K. Muehlfeld, P. Rao Sahib, and A. Van Witteloostuijn

resources, and synergy potential were not sufficientto ensure completion. They concluded that the ter-mination constituted ‘the most extreme case of amerger failure, where none of the potential syner-gies were realised [sic], and where the merger costswere substantial’ (Meyer and Altenborg, 2008:522).

The public takeover period involves both strate-gic and administrative activities related to compli-ance with regulations, final negotiations includingagreement on future strategy, and implementationof announcement strategies (Clougherty, 2005;Haspeslagh and Jemison, 1991; Meyer andAltenborg, 2008; Weston et al., 2004). Strategicdecisions concern, first, corporate restructuringrelated to regulatory compliance. For example,in July 1999, U.K.-based Trinity PLC disclosedthat it was seeking a buyer for its four North-ern Ireland titles (Belfast Telegraph, Sunday Life,Community Telegraph, and Farm Trader), all ofwhich were published by Belfast Telegraph Ltd(BTN), a unit of Trinity PLC. The announcementwas the result of the United Kingdom’s secre-tary of state for trade and industry’s request forTrinity to divest these four newspapers before theintended merger of Trinity with Mirror Group PLC(MG) could be realized. In response, IndependentNews & Media PLC, an Irish publishing group,acquired BTN, enabling Trinity to merge with MG(Thomson Financial Securities Data’s WorldwideMergers and Acquisitions database; henceforthThomson). In contrast, the merger between mediaconglomerates Axel Springer and ProSiebenSat.1was blocked in January 2006 by the GermanFederal Cartel Office because Axel Springer hadrefused to sell either its influential daily news-paper Bild, or one of the two main TV chan-nels of ProSiebenSat.1 (Frankfurter AllgemeineZeitung, 2006). Second, the need for strategic deci-sions arises in relation to post-acquisition inte-gration and transformation. Meyer and Altenborg(2008) found that the termination of the announcedmerger between Telenor and Telia was chieflydue to disagreement on future strategy. Admin-istrative activities include adjustments to finan-cial evaluation, handling of final negotiations, andinternal and external communication (Meyer andAltenborg, 2008; Weston et al., 2004). This entailsactivities such as providing managers with back-ground materials on the rationale for the M&A,working with human resources and communication

departments to develop key messages, and gather-ing feedback on employees’ and investors’ views(Meyer and Altenborg, 2008).

Strategic and administrative activities are oftenintertwined. During the final negotiations, the needto reassess strategic decisions may arise. Sim-ilarly, case presentations to antitrust authoritiesare closely linked to decisions on divestituresand strategy, which crucially affect the outcomeof the public takeover phase. For example, on9 January, 1989, Belfast Telegraph, a unit ofThomson, announced the acquisition of four news-papers from Century Newspapers. The deal waseventually blocked, however, by the United King-dom’s Monopolies and Mergers Commission onthe grounds that it might be against public inter-est as Thomson’s share in this regional newspapermarket would rise from 63 to 81 percent. Thom-son’s failure to close the deal was rooted in itshandling of both strategic and administrative activ-ities. The most obvious candidate would have beento obtain regulatory approval by means of divest-ing one of Thomson’s existing regional newspa-pers. Yet, a more compelling case presentationto the Commission—for example, with respect toways to ensure that public interest would continueto be served—might have resulted in less strin-gent restructuring requirements. Thomson mighthave found it easier to comply with more lenientrequirements without compromising on its regionalstrategy, thereby enabling it to complete the deal.

Acquisition completion as learning indicator

Prior studies on the pre-completion stages—private and public—suggest that prospectiveacquirers are concerned about a variety of goalvariables: the premium paid (Beckman andHaunschild, 2002; Laamanen, 2007), the durationof the phase (Dikova et al., 2010) and, most fun-damentally, whether they are able to close thedeal (Aguilera and Dencker, 2008; Dikova et al.,2010; Wong and O’Sullivan, 2001). With regardto the public takeover phase, completion repre-sents the most basic goal variable. It is compre-hensive in that it covers all key aspects relatedto an organization’s effectiveness and efficiencyin handling the final stages of M&A preparation.Prior to the IPA, target selection, due diligence,and preparations for integration, including servicesof consultants and lawyers, typically consumeformidable resources (Haspeslagh and Jemison,

Copyright 2012 John Wiley & Sons, Ltd. Strat. Mgmt. J., 33: 938–964 (2012)DOI: 10.1002/smj

Context-Specific Learning toward M&A Completion 943

1991). Opportunity costs arise as managerial atten-tion is diverted away from other investments (Hitt,Hoskisson, and Ireland, 1990). Hence, it is in thefirm’s best interest to monitor expected costs andbenefits, to terminate the process as early as possi-ble during the private stage, and to announce a dealonly after the fundamental issues have been settled(Lajoux and Elson, 2000). The IPA changes thissetting, as explained in more detail below, by pub-licizing the firm’s acquisition intention and intro-ducing a host of additional factors related to publicscrutiny, revelation of strategy, reputation dam-age, and stigmatization (Luo, 2005; Officer, 2003).Also, unilateral abandonment after this thresholdoften constitutes a breach of contract and obligesthe firms to pay heavy penalties. For example, in1984, a court ordered Getty Oil to pay more than$10 billion in damages to Pennzoil following itsbreach of their merger agreement (Luo, 2005). So,while termination after the IPA may still be prefer-able in particular circumstances, in most cases,acquirers aim to complete an acquisition once thiscritical stage has been reached.2

We assume that acquirers strive to completean M&A conditional on having made the IPAand that organizational learning during the pub-lic takeover phase is aimed at improving theirchances to achieve this goal.3 From an organiza-tional perspective, the IPA marks a fundamental

2 This aim does not imply that completion is necessarily posi-tively associated with post-acquisition performance. For exam-ple, evidence indicates that if acquisitions fail to deliver, poorintegration in the post-completion stage is the leading cause(Zollo and Singh, 2004). As a result, even a takeover that ex anteappears to make sense may end up hurting firm performance.Still, given the debate about ultimate performance implicationsof M&As, one may wonder whether abandonment may at timesnot be more in the firm’s interest than completion. However,in contrast to the vast literature on consequences of completedM&As, studies into the consequences of abandoned ones are rare(Wong and O’Sullivan, 2001), producing mixed evidence. Forexample, Bradley (1980) finds that unsuccessful bidders expe-rienced a negative share price impact after abandonment. Lehnand Zhao (2006) observe that cumulative abnormal returns are,on average, more negative for announcements of acquisitionsthat were later abandoned, compared to those that were latercompleted. Holl and Kyriazis (1996) report that acquirer abnor-mal returns were positively associated with bid success but notwith bid failure and Asquith (1983) fails to find a significantdifference between completion and abandonment, a finding thatis also supported by our own data.3 There is a fundamental difference between abandonment beforeand after the IPA. Our arguments apply solely to terminationduring the public takeover period with the ‘mistake’ being either(1) not having abandoned the transaction earlier (if it is a ‘bad’deal), or (2) not having been able to close the deal (in case of a‘good’ deal).

transformation in the relationship between acquirerand target. By publicizing the bidder’s intention toacquire the specific target, the IPA reveals privateinformation about the acquirer’s strategy (Offi-cer, 2003). Also, the IPA disrupts organizationalroutines in both firms, implying that ‘businessas usual’ ceases until the uncertainty is resolved(Jemison and Sitkin, 1986). Further, terminatingan announced deal potentially impairs the repu-tation and credibility of a firm (Luo, 2005). Thepublic’s interpretation of the IPA tends to be thatof a target the acquirer has set itself, as illustratedby Bloomberg BusinessWeek ’s comment on March14, 2006, on the acquisition of Knight Ridder byMcClatchy:

‘the deal marks a major ascension for Pruitt[McClatchy chairman and CEO], and putshim in a new and certain to be more-closely watched role. The 49-year-old CEO,long considered a major emerging executivein newspapers, now presides over a sub-stantially larger portfolio. McClatchy, whichpublishes 12 daily and 17 community news-papers, took in $1.3 billion in revenues lastyear. Including the Knight Ridder papersMcClatchy will be keeping, that figure swellsto $2.8 billion.’ (Fine, 2006)

In turn, failed efforts by rivaling Denver-basedMediaNews Group and its chief executive offi-cer (CEO) W. Dean Singleton are described asan ‘ardent desire to purchase Knight Ridder wentunconsummated’ (Fine, 2006). Abandonment,from this perspective, signals failure. If a broaderperception of incompetence, lack of vision, andweak leadership (Rhee and Valdez, 2009) catcheson, which anecdotal evidence suggests may happenfollowing a string of failed attempts, reputationdamage and stigmatization may harm businessactivities in other areas as well (Andrade andStafford, 2004). With third parties such as themedia playing a key role as reputation endorsers(Rhee and Valdez, 2009), high-profile deals andthose under public scrutiny, such as M&As thatinvolve politically sensitive sectors like newspa-pers, are particularly risky in this respect.

From the individual perspective of a manage-ment team or CEO, especially if they are newto the job or face challenges in other domains,leadership reputation is at stake (Haspeslagh and

Copyright 2012 John Wiley & Sons, Ltd. Strat. Mgmt. J., 33: 938–964 (2012)DOI: 10.1002/smj

944 K. Muehlfeld, P. Rao Sahib, and A. Van Witteloostuijn

Jemison, 1991; Jemison and Sitkin, 1986). Cara-peto, Moeller, and Faelten (2010) show that newlyhired CEOs frequently use deal making as a strate-gic tool to make a mark. While termination of anM&A is in a different category than a catastrophicorganizational crisis such as (near) bankruptcy, theemerging literature on stigmatization of corporateelites suggests that, in as far as the abandonment ofan M&A is viewed as failure by the public, man-agers associated with it may face career damage,loss of reputation, and pay decreases (Semadeniet al., 2008). The relevance of completion as agoal is apparent from evidence that managerswho would have benefited from the M&A sufferfrom motivational problems in cases of abandon-ment (Larsson and Finkelstein, 1999; Schmidt andFowler, 1990). Evidence of a positive relation-ship between compensation and firms’ acquisitivebehavior underscores the financial significance formanagers of seeing a deal through (Guest, 2009;Schmidt and Fowler, 1990).4

Learning from success and failure experiences

Theory suggests that experiential learning pro-cesses differ for success and failure experiencesbecause of limitations to organizational resour-ces—financial, cognitive, and other (Cyert andMarch, 1963; March, 1991; Sitkin, 1992). Lim-ited resources contribute to differences in learningprocesses because successes and failures stimulatedistinct search processes. Search for superior alter-native solutions poses the problem of allocatingscarce resources to exploiting existing routines vs.exploring new ones (Levinthal and March, 1993;March, 1991).

Success experience promotes ‘satisficing’ andrefinement of existing routines, which economizeson scarce resources but stifles the search for newand superior solutions (Cyert and March, 1963).In addition, each subsequent success contributesless new information, given that the firm fol-lowed existing routines without too much varia-tion (Sitkin, 1992). Further, the outcome is less

4 By pointing to disincentives for managers to pursue bad deals,recent studies suggest that the potential divergence betweenorganizational and managerial interests in pursuing M&As maybe less pronounced than previously assumed. Pay increases fromM&As that are negatively received by stock markets tend to bebalanced out by a decline in shareholding value (Guest, 2009).CEOs who engage in bad M&As are also more likely to losetheir jobs than are CEOs who have undertaken good deals (Lehnand Zhao, 2006).

surprising, compared with a failure, making theexperience less salient (Sitkin, 1992). As a result,decreasing amounts of resources are devoted toperfecting the particular activity. Such learning isself-limiting and may lead into a competency trap(Levitt and March, 1988). Together this implies,first, that firms with past success in consummatingM&As have routines in place that predispose themto complete a focal deal; second, that each suc-cess entails a strictly positive learning potential forimproving performance due to the new informationit conveys; and, third, that the contribution of eachsubsequent success to raising future completionlikelihood decreases due to the self-limiting natureof learning from success (Levitt and March, 1988;Schulz, 2002). Overall, this suggests a curvilinearrelationship in the form of positive but diminishingreturns to success experiences.

Hypothesis 1a: The relationship between prioraggregate success experience and the comple-tion likelihood of a focal M&A is positive withdiminishing returns.

Given the intention to complete an announcedM&A, abandonment signals flaws in the routinesthat the bidding firm enacts during the publictakeover stage. This failure experience induces‘problemistic search’ for superior solutions (Cyertand March, 1963). Problemistic search starts outlocally and favors incremental over radical change.If the local search fails to generate sufficientimprovements, the organization turns to non-localsearch, engages in greater risk taking, and adoptsmore radical departures from previous solutions(Cyert and March, 1963; Levitt and March, 1988).Failure has been argued to contain richer cues tocausality compared with success because it gener-ates new, unexpected types of information (Baumand Dahlin, 2007; Madsen and Desai, 2010; Sitkin,1992). Yet, the returns to exploration are sys-tematically more uncertain (March, 1991), andlearning from failure is particularly difficult incomplex tasks in which ‘repetitions’ are hetero-geneous and characterized by causal ambiguity,temporal delays, and nonlinearities (Lomi, Larsen,and Ginsberg, 1997). This implies that raisingfuture completion likelihood requires bidders, first,to correctly identify which of their routines needto be changed; second, to correctly gauge how tomodify them; and, third, to successfully implement

Copyright 2012 John Wiley & Sons, Ltd. Strat. Mgmt. J., 33: 938–964 (2012)DOI: 10.1002/smj

Context-Specific Learning toward M&A Completion 945

these changes. For example, when final negoti-ations toward closing an M&A break down, itmay be difficult to single out the key reasonsfrom among deficient negotiation tactics, unre-solved strategy issues, and personality clashes.

Given the complexity and causal ambiguity ofacquisition processes, and the local starting pointof search efforts, it is therefore unlikely thatresponses to initial failures will immediately resultin correct identification of the flawed routine(s)as well as the necessary type(s) of improvement,and their successful implementation. Hence, learn-ing from initial failures is unlikely to substantiallyraise completion likelihood. Search and changeefforts in response to these experiences have, how-ever, the potential to reduce the chances to com-plete subsequent deals. First, the adopted changesdivert attention away from other (potentially moreimportant) sources of failure (Hayward, 2002).Second, in as far as they stimulate efforts to ‘tryharder’ in routines that are (unjustly) perceived assuccessful, they will lower completion likelihood.Third, any change is an opportunity for organi-zational members to try to shape rules to theirbenefit. It increases the likelihood of conflict andthe ensuing intraorganizational bargaining processconsumes resources that could be used in moreproductive ways (Cyert and March, 1963). Lim-itations to cognitive resources further imply thatsystematic biases impact search processes, induc-ing firms to ‘oversample successes and undersam-ple failures’ (Levinthal and March, 1993: 110).Hence, initial failures are unlikely to unlock thefull potential of information embodied in the expe-riences and stimulate the appropriate changes. Tothe contrary, we expect them to further diminishthe likelihood of completion.

Accumulation of a large number of failures,however, facilitates the detection of commonpatterns as a precondition for identifying andimproving upon flawed routines. It is more likelyto stimulate sufficiently radical changes and itimplies extensive opportunity to experiment,thereby increasing the chances that these modi-fications are of the right type to increase futurecompletion likelihood. Further, by accumulatingand responding to initial failure experiences, firmsdevelop modification routines—that is, proceduresfor searching for solutions to new problems (Levittand March, 1988) and for dealing with the creationand change of routines (Nelson and Winter, 1982).This learning ‘to change by changing’ (Amburgey,

Kelly, and Barnett, 1993: 54) further improves theacquirer’s chances to finally turn the next failureinto an experience that is useful in raising com-pletion likelihood. Together this implies, first, thatfirms with past failures in consummating M&Ashave routines in place that predispose them to notcomplete a focal subsequent deal; second, that eachfailure experience entails a positive learning poten-tial due to the new information it conveys; but,third, that initial failures and the search efforts theytrigger diminish the likelihood to complete sub-sequent deals; and fourth, that focal performancewill continue to suffer until the firm engages ineffective search—that is, initiates changes to itsroutines that are sufficiently radical and of the righttype to raise future completion likelihood. Overall,this suggests a U-shaped relationship between pastfailure experiences and subsequent deal comple-tion likelihood.

Hypothesis 1b: There is a U-shaped relationshipbetween prior aggregate failure experience andthe completion likelihood of a focal M&A.

Context-specific transferability of experientiallearning

The value of experience for learning depends onthe degree to which this experience is applica-ble to a future event (March, 1991)—that is,the extent to which two events are structurallysimilar. The notion of structural similarity origi-nates from research on analogical encoding, whichrefers to processes in which learners draw compar-isons across examples (Genter, Loewenstein, andThompson, 2003). Analogical encoding facilitatesrecall and transfer of knowledge by means of thepromotion of abstract concepts. A critical condi-tion is the identification of similarity based onstructural rather than surface features. Prior workon organizational learning has transferred theseideas from the individual to the group and organi-zational level, and has demonstrated the prevalenceof mechanisms similar to analogical encoding inorganizational learning (Haleblian and Finkelstein,1999; Haunschild and Sullivan, 2002; Zollo andReuer, 2010).

This suggests that only experience gained withstructurally similar M&A attempts and not expe-rience gained with structurally dissimilar deals isapplicable to a focal transaction and should havea significant effect on its completion likelihood.

Copyright 2012 John Wiley & Sons, Ltd. Strat. Mgmt. J., 33: 938–964 (2012)DOI: 10.1002/smj

946 K. Muehlfeld, P. Rao Sahib, and A. Van Witteloostuijn

Generally, M&As can be characterized alongnumerous contextual dimensions. For example,country borders give rise to the distinction betweena domestic and a cross-border context. Yet, littleis known about which acquisition contexts conferstructural similarity—that is, within which contextcategories M&As are, on average, sufficiently sim-ilar to allow for learning. Typically, deals withinthe same industry are viewed as structurally sim-ilar, where similarity is assessed using StandardIndustrial Classification (SIC) code-based mea-sures of relatedness (Haleblian and Finkelstein,1999; Laamanen and Keil, 2008). Beyond this,conceptual arguments and empirical evidence arelimited. Hence, in identifying other types of con-texts that confer structural similarity, we draw onthe financial M&A literature as well as on proper-ties of the industry setting. For each of the fol-lowing six contexts, we propose that deals arestructurally similar based on shared core features,while M&As are structurally dissimilar across con-texts (e.g., domestic vs. cross-border).

Intra-industry acquisitions

Prior studies indicate that two M&As that sharethe same industry origin possess structural similar-ity (Haleblian and Finkelstein, 1999; Haunschild,1994; Hayward, 2002; Laamanen and Keil, 2008).Conceptually, this notion is based on Huff (1982),who argued that shared concepts are developedwithin an industry on how a firm should functionwithin that industry and that these shared con-cepts are not accessible to those outside the indus-try. Spender (1987) provided empirical supportby showing how ‘industry recipes’ are groundedin industry contexts and how they guide strate-gic thinking in these contexts. This suggests thatnewspaper firms’ experience with intra-industryM&As—but not with diversifying transactions—should significantly affect the completion likeli-hood of subsequent attempts to acquire other news-paper firms.

Diversifying acquisitions

Diversifying M&As in the newspaper industryover the past few decades were relatively moresimilar than in other sectors. Chan-Olmsted andChang (2003) argue that industry characteristicscause media firms to opt for highly related M&Asand resource alignment if they diversify. Media

firms offer dual, complementary products of con-tent and distribution. Therefore, they seek out dis-tribution products that complement their contentproducts, and vice versa, allowing, for example,for reuse of content in multiple outlets. Further,the duality of revenue sources encourages relateddiversification because a larger aggregated audi-ence enhances the appeal for advertisers. Empiri-cal studies show that diversification by newspaperfirms has predominantly taken the form of re-lated diversification, clustered in a few indus-tries: the majority of diversifying M&As since the1980s involved other printing and publishing firms,followed by communications, and data process-ing services, in particular, information-technologyrelated services (Muehlfeld, Rao Sahib, and vanWitteloostuijn, 2007).

Domestic acquisitions

In domestic M&As, common language eliminatesa potential source of friction and a confoundingfactor in identifying causes of abandonment. Fur-ther, while core elements of negotiation processesare similar across cultures, many aspects are spe-cific to national cultures (Adair and Brett, 2005).The regulatory framework provides another sourceof structural similarity (Clougherty, 2005). Experi-ence enables acquirers to learn with respect to theirhandling of regulatory requirements and communi-cation with antitrust authorities, allowing them todevelop an earlier and better view of likely strate-gic options.

Cross-border acquisitions

Firms that engage in large numbers of cross-borderM&As appear to develop generalized knowledge,strategies, and routines as to how to acquire for-eign firms (Hitt et al., 1998). Such a generalized‘dealing with a foreign country’ effect in acqui-sition preparation arises, for example, from learn-ing about sourcing from outside financial, legal,and other resources (Hitt et al., 1998) and the useof specialized M&A teams (Very and Schweiger,2001). In international M&As, firms often hireexternal local consultants (Very and Schweiger,2001). Working with them indirectly helps acquir-ers to develop generalized knowledge by extend-ing their international business network and byallowing them to refine their sets of criteria forfuture hiring in foreign countries. Indeed, Dikova

Copyright 2012 John Wiley & Sons, Ltd. Strat. Mgmt. J., 33: 938–964 (2012)DOI: 10.1002/smj

Context-Specific Learning toward M&A Completion 947

et al. (2010) report that the negative effect ofinstitutional distance on completion likelihood ofcross-border M&As was weaker if the acquirer hadaccumulated more experience with internationaldeals in general. General foreign country learningcan be expected especially if M&As are geograph-ically concentrated, as is the case in the newspaperindustry (Chan-Olmsted and Chang, 2003).

Hostile acquisitions

Hostile bids usually involve an unsolicited offerdirectly to the target’s shareholders, allowing themto sell their shares without the target’s board ofdirectors’ approval (Weston et al., 2004). They aretypically met with aggressive public rejection bythe target’s management, which implies that thekey audience that the acquirer needs to win overconsists of shareholders, not the target’s manage-ment. Further, targets often use elaborate defensivemeasures such as poison pills and white knights.The need to react has implications for acquirers’design of takeover strategy, including decisionson strategy, countermeasures, and premium. In1988, for example, Rupert Murdoch’s News Inter-national (NI) acquired the 58.3 percent of WilliamCollins that it did not already own after substan-tially raising its bid in response to Collins’s talkswith Presses de la Cite, a possible white knight.When NI matched Presses’ expected bid, Presseswithdrew as it was unable to get enough supportfrom shareholders. The deal was announced on 17November, 1988, and consummated on 19 January,1989, after a two-month takeover battle (Thom-son). By definition, issues that can be resolvedprior to the IPA in friendly M&As are addressedduring the public phase in hostile deals, hence,much later and involving greater publicity (Schw-ert, 2000).

Friendly acquisitions

Friendly acquisitions are deals that are negoti-ated with the target’s management to meet certainlegal and technical requirements (Schwert, 2000).Acquirers typically abstain from toeholds—theacquisition of target shares prior to the focalbid—because they may provoke target resistance(Betton, Eckbo, and Thorburn, 2009). During thepublic takeover phase, friendly M&As share corefeatures related to bargaining strategies and jointdevelopment of strategy (Meyer and Altenborg,

2008)—for example, concerning the relation bet-ween distribution through multiple channels (e.g.,print, online, and mobile), sourcing strategiesthrough own newsrooms vis-a-vis news agencies,coverage ranging from local to international, andthe degree of product differentiation (Economist,2006, 2009).

In summary, we have identified six acquisi-tion contexts that we expect to confer structuralsimilarity. We argue that experience—both withsuccesses and failures—gained in each of themsignificantly affects the completion likelihood ofsubsequent same-context deals, in contrast to expe-rience gained in a structurally dissimilar, alterna-tive context.5

Hypothesis 2a: Completion likelihood of a focalM&A is affected only by prior success experi-ence gained in the same context, as opposed toexperience gained in a structurally dissimilar,alternative context.

Hypothesis 2b: Completion likelihood of a focalM&A is affected only by prior failure experi-ence gained in the same context, as opposed toexperience gained in a structurally dissimilar,alternative context.

Building on the notion of context-specific trans-ferability, we further argue that inherent proper-ties of contexts map onto distinct learning pro-cesses.6 We consider two dimensions that, in par-ticular, impact the quality of analogical encodingand thereby influence organizational learning pro-cesses (Argote and Todorova, 2007): the degree towhich a context exhibits structural variance acrossM&As within this context (Haunschild and Sul-livan, 2002), and the degree to which it stimu-lates deliberate learning efforts (Zollo, 2009; Zolloand Singh, 2004). We focus on the extreme highpoints of each dimension—that is, on those twofrom among our six contexts in which structural

5 Hypotheses 2a and 2b do not predict any specific type ofrelationship, but are concerned with the significance (nonsignif-icance) of experience with respect to completing future deals inthe same (alternative) context. Hypotheses 3a/b and 4a/b formu-late expectations about context-specific types of relationship.6 The resulting hypotheses predict context-dependent learningpatterns that may differ from those predicted in Hypotheses 1aand 1b. Different context-specific relationships do not precludethat when aggregating experience, learning follows the patternssuggested in Hypotheses 1a and 1b, as those patterns result fromthe interplay of various context combinations.

Copyright 2012 John Wiley & Sons, Ltd. Strat. Mgmt. J., 33: 938–964 (2012)DOI: 10.1002/smj

948 K. Muehlfeld, P. Rao Sahib, and A. Van Witteloostuijn

variance and stimulation of deliberate learning,respectively, are particularly pronounced.

The role of structural variance within contextcategories

M&As are relatively rare, complex, and hetero-geneous compared with other corporate activities(Zollo, 2009). Even if a context as such confersstructurally similarity, the degree of intra-categorysimilarity, and hence the strength of experientiallearning, may vary substantially. Too much varia-tion hinders learning and limits the usefulness ofexperience (Hayward, 2002; Levinthal and Rerup,2006). We suggest that the six contexts vary sys-tematically in their degree of structural varianceand, thus, in the extent to which they stimulate orexacerbate learning to complete M&As.

Hostile deals are particularly high in structuralvariance, arising from three sources7: (1) targets’defense measures; (2) dispersed stakeholders; and(3) structure of hostile takeover processes. First,targets often have in place a wide range of pre-ventive defense measures or adopt reactive mea-sures after the public announcement of the hostilebid (Sudarsanam, 1991). Experience with consum-mating a deal despite the presence of a superma-jority clause is not very helpful in dealing withanother type of defense, such as a white knight.Second, compared with friendly acquisitions, therelevant stakeholder groups are much more widelydispersed. Assessment of a transaction typicallydiverges more among shareholders than within amanagement team (Harford, Jenter, and Li, 2007).Given that hostile bids are largely battles of public-ity (Schwert, 2000), communication is very impor-tant, increasing the relevance of the media as atarget audience. Third, different structure and needfor speed and secrecy (Schwert, 2000) imply thatactivities are partially shifted from the privateinto the public takeover phase (Boone and Mul-herin, 2007). For example, without direct access to

7 Another contender, the cross-border context, likely exhibits alower degree of structural variance in our setting. First, as tothe acquisition stage, prior research suggests that cross-borderM&As face more severe screening processes during early stages(Aguilera, Dencker, and Escandell, 2004). This counterbalancesthe initially greater heterogeneity between attempts in this cate-gory, once they have reached the public phase. Second, in ourindustry, structural variance between international deals is fur-ther reduced due to geographical clustering (Chan-Olmsted andChang, 2003; our data also support this observation, as the vastmajority of acquirers focus on two foreign markets at most).

sources within the target firm during pre-IPA duediligence, the need for revaluation during the pub-lic takeover phase is more likely. All this createsample scope for structural heterogeneity betweentwo hostile deals in the final stages of deal prepa-ration, beyond the moderate levels shown to bemost inductive for organizational learning (Hay-ward, 2002). As a result, acquirers will find itparticularly difficult in hostile M&As to learn fromboth past successes and failures.

Taking the failure to complete an announcedhostile bid as a starting point, improvement re-quires extensive experimentation, especially whenhigh structural variance hampers learning. If any-thing, acquirers typically have fewer such opportu-nities in hostile bids, given that they are relativelyrare, despite their prominence in public percep-tion (Schwert, 2000). Publicity likely amplifies anycognitive biases in response to failures (Hayward,2002). Due to the particular difficulties associ-ated with learning from failure in this context,we expect that the predominant effect of expe-rience with terminations is a reduced likelihoodto consummate future hostile bids. High structuralvariance also poses problems for learning fromsuccess. Recent research shows that success-basedlearning from rare events such as M&As is proneto suffer from dysfunctional mechanisms such assuperstitious learning, in which the increase inconfidence exceeds the rise in actual competencefrom experience accumulation (Zollo, 2009). Withlarger numbers of completed hostile bids, confi-dence and competence increasingly diverge. Weargue that the structural variance implied in hos-tile bids aggravates this problem because it isparticularly important for acquirers not to over-look the prevailing variance in structure. We there-fore expect that success-based learning in hostileM&As is not only limited, but follows an invertedU-shaped pattern reflecting the increasing role thatdysfunctional mechanisms play at higher levels ofexperience with hostile deals.

Hypothesis 3a: There is an inverted U-shapedrelationship between prior success experiencewith hostile acquisitions and the completionlikelihood of a focal hostile transaction.

Hypothesis 3b: There is a negative relationshipbetween prior failure experience with hostileacquisitions and the completion likelihood of afocal hostile transaction.

Copyright 2012 John Wiley & Sons, Ltd. Strat. Mgmt. J., 33: 938–964 (2012)DOI: 10.1002/smj

Context-Specific Learning toward M&A Completion 949

The role of unfamiliar institutionalenvironments in stimulating deliberatelearning efforts

Learning in infrequent and complex corporateactivities such as M&As often suffers from aber-rant phenomena such as superstitious learning, inparticular, the more it is grounded in ‘automatic’(e.g., learning-by-doing) learning processes (Zollo,2009). Research shows, though, that firms cancounterbalance these harmful effects by engagingin deliberate learning, which is often accompa-nied by more extensive knowledge codification anddocumentation (Zollo, 2009). Postmortem reports,updating of manuals, and project management soft-ware are expressions of deliberate learning inacquisitions. Deliberate, inferential, and controlledlearning processes constitute ‘mindful’ ways ofprocessing experience, involving high levels ofattention (Argote and Todorova, 2007). The degreeof mindfulness as well as the ensuing learning pro-cesses depend on properties of the organizationaland environmental context (Argote and Todorova,2007). Hence, we argue that acquisition contextsvary in the degree to which they stimulate effortsto pursue mindful learning. Following prior evi-dence (Zollo, 2009), context types that are moreconducive to mindful learning processes shouldallow firms to realize positive effects of experi-ence accumulation while circumventing its dys-functional ‘side effects.’

A property of M&A processes that captures agreat deal of bidders’ attention is an unfamiliarinstitutional environment (Barkema and Schijven,2008; Goulet and Schweiger, 2006). Perceptions ofunfamiliarity, especially those arising from culturaldifferences, appear to make acquirers sensitivetoward the need for deliberate learning (Morosiniand Singh, 1994). Especially in a sector that is asdeeply embedded in a country’s cultural, political,and social institutions as newspapers, differencein national culture is a salient feature (Bagdikian,2004). Cross-border M&As are still a sensitivetopic in many countries, generating both formaland informal obstructions. In France, for exam-ple, national laws prevent majority ownership ofnewspapers by foreign firms.

Many formal and informal institutions arenation-specific, such as judicial and political reg-ulations, culturally rooted conventions, and socialnorms (North, 1990). The association with countryborders facilitates their identification as a potential

source of frictions (Chakrabarti, Gupta-Mukherjee,and Jayaraman, 2009), compared, for example,with industry boundaries. Differences in formalinstitutions are tangible and apparent from the needto comply with multiple jurisdictional regimes,such as pre-acquisition notification and reviewrequirements (Clougherty, 2005). Differences ininformal institutions are more tacit. While detailsare difficult to decipher due to their tacit nature,their existence as such is salient through, for exam-ple, language and culture. Research on interna-tional M&As shows that firms are highly sensitiveto national culture, respond by approaching cross-border M&As more cautiously, and demonstratea strong predisposition to working toward copingwith the cultural differences (Chakrabarti et al.,2009; Goulet and Schweiger, 2006). This appearsto also entail greater sensitivity to differences inorganizational culture—often rooted in nationalculture—as compared with domestic deals (Evans,Pucik, and Barsoux, 2002).

Overall, it appears that acquirers are oftenacutely aware that they are treading on unfamil-iar ground. This caution is justified given evidencethat cross-border deals have a higher baseline riskof termination (Aguilera et al., 2008; Dikova et al.,2010; Muehlfeld et al., 2007). Yet, to the extentthat this perception stimulates deliberate learningefforts, acquirers may actually benefit comparedwith contexts that are less conducive to mindful-ness (Morosini and Singh, 1994). For example, thelower the perceived familiarity of a foreign cul-ture, the less likely organizations are to develop(unjustified) confidence in their understanding ofit (Chakrabarti et al., 2009). This suggests that, tothe extent that the international nature of a dealstimulates a more deliberate approach, learningfrom closing cross-border M&As is less subjectto myopic limits than learning in other contexts,helping acquirers avoid the competency trap bythwarting their tendency to rely on performanceimprovements via ‘semi-automatic,’ exploitation-biased learning, and by counteracting dysfunc-tional mechanisms such as superstitious learning(Zollo, 2009; Zollo and Singh, 2004). We thusexpect a predominantly positive impact of cross-border completions on the probability of closingsame-context M&As.

Key challenges regarding learning from termi-nated attempts derive from the need to identifythe causes of abandonment, from exploring alter-native approaches to strategic and administrative

Copyright 2012 John Wiley & Sons, Ltd. Strat. Mgmt. J., 33: 938–964 (2012)DOI: 10.1002/smj

950 K. Muehlfeld, P. Rao Sahib, and A. Van Witteloostuijn

activities, and from overcoming organizations’tendency to devote too little attention to fail-ures due to ‘attributional errors’ (Kelley, 1971).Greater mindfulness, heightened attention to delib-erate learning, and more extensive documentationare likely to ameliorate these problems of failure-based learning. For example, a priori defineddocumentation requirements countervail the biasto ‘undersample failures’ (Levinthal and March,1993: 110). Yet, even under favorable conditions,failure-based learning requires opportunities foridentifying causal links and for experimentation.So while firms may reach the turning point pro-posed in Hypothesis 1b earlier, we still expectthe relationship between failures in cross-borderM&As and focal completion likelihood to followa U shape.8

Hypothesis 4a: There is a positive relationshipbetween a firm’s prior success experience withcross-border acquisitions and the completionlikelihood of a focal cross-border transaction.

Hypothesis 4b: There is a U-shaped relationshipbetween a firm’s prior failure experience withcross-border acquisitions and the completionlikelihood of a focal cross-border transaction.

METHOD

Sample and data

We test our hypotheses on 4,973 M&A attemptsthat were announced during 1981–2008 in thenewspaper industry. The data are derived fromThomson (for a survey of these data, see Pryor,2001). We required at least one partner to be in thenewspaper industry (SIC code 2711). The datasetincludes all corporate transactions that involve atleast five percent of the firm’s ownership where thetransaction was valued at $1 million or more (after1992, deals of any value are covered) or where thevalue of the deal was undisclosed. Both public andprivate transactions are covered. The industry hassome large players that account for a substantial

8 Our reasoning on context-specific patterns of learning fromsuccesses and failures is based on the particular stage of theacquisition process under study. Some of those activities thatare most difficult to learn about in an international context (e.g.,target selection and post-merger integration) take place duringearlier or later phases. Hence, learning in those stages mayfollow different patterns.

volume of M&A activity. The top five acquirersin order of decreasing volume of M&A activ-ity are VNU (the Netherlands), Wolters Kluwer(the Netherlands), EMAP PLC (United Kingdom),Gannett Company Inc. (United States), and Ber-telsmann AG (Germany). Together, they accountfor 459 (13.65%) out of the 4,973 announcements.The top 10 account for 730 (21.59%) announce-ments. While there is considerable variation in thenumber of acquisitions per year, we do not observesystematic differences between the numbers ofcompleted and abandoned transactions over time.Larger firms were more likely to be active through-out the sampling period, while smaller firms wereactive during shorter intervals of time. The top10 percent of firms in terms of M&A activityhad an average of 22 years between their first andtheir last acquisition within our sampling frame(1981–2008). In contrast, if a firm made less thanfive acquisitions in total, the average number ofyears during which these transactions took placewas 3.5. Smaller firms appear on the M&A marketthroughout the sampling period.

Dependent variable

We measured the dependent variable, M&A com-pletion, by creating a dummy variable, which tookon the value of 1 if an announced M&A wascompleted, and 0 otherwise. Consistent with ourtheory, we used the date of announcement to markthe beginning of the public takeover stage, and thecompletion date as a proxy for its conclusion (bothdates as reported in Thomson). Mergers recordedas ‘intent withdrawn,’ ‘withdrawn,’ or ‘rumoredonly’ make up the abandoned category. In our data(1981–2008), we did not include deals announcedin 2009, but recorded deals that were announced inearlier years and completed in 2009. If they werenot completed by 2009, we considered them with-drawn for two reasons. First, the median number ofdays to completion is about 62 days with 94 per-cent of all deals completed within a year. Second,Thomson collects data retrospectively and updatesinformation on past deals. Hence, right-censoringneed not be a concern with our data. Left-censoringis no concern either as Thomson began collectingM&A data from 1981 onward.

Independent variables

Consistent with prior research (Haleblian andFinkelstein, 1999; Hayward, 2002; Zollo and

Copyright 2012 John Wiley & Sons, Ltd. Strat. Mgmt. J., 33: 938–964 (2012)DOI: 10.1002/smj

Context-Specific Learning toward M&A Completion 951

Singh, 2004), we measured experience as M&Aexperience prior to the focal transaction. As ourbenchmark, we used a count measure of a firm’stotal prior cumulative experience with attemptingto close publicly announced M&As up to the focaltransaction. Total experience is the total number ofM&As attempted by the larger partner in the dealprior to the focal M&A.9

We then disaggregated experience. First, we dis-tinguished between success (completed deals) andfailure (abandoned deals) experience. Second, wedisaggregated experience along the six contexts,differentiating between domestic vs. cross-border,diversifying vs. intra-industry, and friendly vs.hostile experience. Following prior studies, weused a moving window of three years prior to thefocal deal to measure a firm’s experience (Hale-blian and Finkelstein, 1999; Hayward, 2002; Laa-manen and Keil, 2008). Our choice of movingwindow is consistent with qualitative evidence forother industries, such as banking (Haleblian et al.,2006). Our sample also demonstrates substantialclustering of M&As within a short time window.Over the time period covered, 64.7 percent of firmshad all their M&As within a time span of threeyears. Only 31.2 percent of firms had more thanfive years between their least and most recent priorM&A. In robustness tests, we experimented withan alternative window of five years. Results weresimilar, but with smaller effects, suggesting thatobsolescence may play a role despite the clustering(Argote, Beckman, and Epple, 1990). The three-year moving window was used for all contexts westudied. For example, we calculated diversifyingfailure experience as the sum of all diversifyingM&As that were abandoned in the three years priorto the focal transaction.

Domestic experience and cross-border experi-ence were defined based on the firms’ headquar-ters, as stated in Thomson. Deals were classifiedas intra-industry if both firms had SIC code2711 (newspaper industry). Acquisitions of tar-gets outside of SIC code 2711 were classifiedas diversifying. Approximately 59.2 percent of allattempts were within the newspaper industry. Anadditional 16.3 percent concerned targets withinbroader printing and publishing (SIC code 27),8.1 percent involved targets in communications

9 In robustness tests, we also experimented with moving three-and five-year windows instead of the historical count of totalcumulative experience. The results were qualitatively unaffected.

(SIC code 48), and 7.7 percent related to targetsin ICT (SIC code 73). The remaining 8.7 percentwere distributed across other industries. With res-pect to the distinction between friendly and hostileM&As, we resorted to a conservative classificationbased on how a deal was presented publicly andreported in Thomson. Finally, since we focus onexperience, we included only M&As of firms thatattempted at least one deal prior to the focal oneduring the sampling period.

Control variables

For our control variables, we drew primarily onresearch in financial economics that has identifiedinfluences on pre-completion processes in M&As(Muehlfeld et al., 2007; Wong and O’Sullivan,2001), and on the literature concerned with deter-minants of post-acquisition performance (Kinget al., 2004). If a factor is positively related withpost-acquisition performance, it could affect com-pletion likelihood as well by signaling to the par-ties that this is a ‘good’ deal.

Public status acquirer and public status targetindicate whether a firm is publicly owned. Pub-lic firms are subject to more intense regulatoryscrutiny. Public ownership of the target could pos-itively affect completion likelihood as dispersedownership reduces the ability to fend off unwel-come bids. Target subsidiary is an indicator forwhether the smaller partner in the transaction wasa subsidiary (1) or not (0). Negotiations with asubsidiary differ from a conventional transactionbecause of power issues related to the parent com-pany. We also investigated if the type of parentfirm had an influence on completion probability.In the majority of cases, the industry classifica-tion of the focal newspaper firm and its parentwere identical. In robustness tests, we included acontrol variable indicating whether their industryclassifications differed. Since we did not theorizeabout effects of the parent firm, and the coefficientof this control variable was not significant, andthe magnitude and signs of the other coefficientswere virtually unchanged, we omitted it from fur-ther analyses. The method of payment (cash, stock,debt, or some combination) may impact comple-tion likelihood. Cash offers appear to create morewealth for target shareholders. In addition, cashfacilitates valuation, reducing the scope for dis-agreement during the public takeover phase. Weincluded a dummy cash payment, which is 1 if

Copyright 2012 John Wiley & Sons, Ltd. Strat. Mgmt. J., 33: 938–964 (2012)DOI: 10.1002/smj

952 K. Muehlfeld, P. Rao Sahib, and A. Van Witteloostuijn

the predominant mode of payment was cash and0 otherwise. Divestiture captures whether (1) ornot (0) the larger partner engaged in a divestitureduring the deal. It is a measure of the scaleof restructuring, which may also signal (preemp-tive) compliance with regulatory requirements.Firms learn from their own but also from theexperience of others. We accounted for this byincluding the cumulative number of recently com-pleted/abandoned M&As at the industry level,previous quarter M&A successes/failures (resultswere unchanged when we used six-month inter-vals instead of three-month intervals). This alsoallowed us to account for a potentially lower prob-ability of completion of firms that enter M&Awaves as late movers. In robustness tests, weincluded year dummies to control for changes inmacroeconomic conditions, the cyclical nature ofM&A activity, changes in legislation, and othertime-variant influences. They were not signifi-cant and when we omitted them, the magnitudeand signs of the other coefficients were virtuallyunchanged. Thus, we report results exclusive ofyear dummies.10

Estimation techniques

We estimated a binary logistic regression modelwith M&A completion as the dependent variable.While the unit of analysis is an announced M&A,the announcements were made by a sample offirms that made multiple acquisitions over the sam-pling period. Thus, our data make up an unbal-anced panel, as the number of M&As varies byfirm. If we were to treat the data as a pooledcross-section, we would ignore the within-firm cor-relation in the error term, treating our observationsas independent. Yet, our data are technically not apanel dataset—that is, we do not follow firms overtime and observe firms only when they engagein M&A activity. Nevertheless, we implementeda Hausman test to choose between a fixed andrandom effects specification. We could not rejectthe null hypothesis of no correlation between theright-hand side variables and the error term. Hence,a random effects estimation technique would be

10 In robustness analyses, we ran models with the value of thedeal included and, separately, its relative value in comparisonto other deals made by the firm. This information is availablefor a subsample of mainly completed deals. We found both ofthese measures not to be significant, and excluded them fromthe analyses reported here.

acceptable in our case. An alternative approachapplicable in this case, which we adopted, is esti-mating models using clustered standard errors thataccount for within-firm correlation. This is simi-lar to using a random effects estimation, as it alsoaddresses the issue of lack of independence due tomultiple acquisitions by the same firm. Its advan-tage over a random effects estimation is that itgives consistent estimates across a broad range ofpossible correlations (Cameron and Trivedi, 2009).The same approach is used in Pollock, Rindova,and Maggitti (2008), whose data are of a similarnature to ours, that is, not a ‘true’ panel.

RESULTS

The 4,973 M&As analyzed were undertaken by1,964 firms. An average firm attempted 2.37 acqui-sitions. Over the whole sample period, 1,307 of the4,973 announced M&As were terminated. Similarto prior evidence (Wong and O’Sullivan, 2001), thecompletion rate for the overall sample is approxi-mately 73 percent. As our interest was in studyingthe impact of closely related experience variables,which by definition were likely to be correlated,we document correlations for all right-hand sidevariables in our models in Tables A1–A5 in theAppendix. Despite the similarity in measures, thelargest correlation coefficient for two experienceconstructs included in the same model is 0.71. Werefrain from applying corrections, such as center-ing, that are sometimes used in the literature asa remedy for multicolinearity, because they canproduce meaningless and misleading colinearitydiagnostics (Belsley, 1984; Echambadi and Hess,2007). Table A6 documents correlations betweenexperience accumulated in different contexts. Theonly correlation exceeding 0.70 is between friendlysuccess and hostile failure experience (0.71).

Tables 1 and 2 report coefficients, standarderrors, the value of the likelihood function at con-vergence, and the value of the Wald chi-squaredtest for the null hypothesis that all the coeffi-cients associated with the independent variablesare simultaneously equal to zero. The null hypoth-esis could be rejected for all models. Model 1contains the control variables only. Model 2 isa benchmark model in order to assess the effectof total M&A experience (combined from suc-cesses and failures). Corroborating recent studies(Hayward, 2002; Zollo and Singh, 2004; but in

Copyright 2012 John Wiley & Sons, Ltd. Strat. Mgmt. J., 33: 938–964 (2012)DOI: 10.1002/smj

Context-Specific Learning toward M&A Completion 953

Table 1. Results of logistic regression analysis predicting M&A completion (overall sample)

Variables Model 1Controls

only

Model 2Total

experience

Model 3Ratio composite

failures tocomposite successes

Model 4Composite success

and failureexperience

Intercept 1.389∗∗∗ 1.381∗∗∗ 1.955∗∗∗ 1.345∗∗∗

(0.123) (0.126) (0.199) (0.120)Cash payment 0.532∗∗∗ 0.530∗∗∗ 0.490∗∗∗ 0.532∗∗∗

(0.096) (0.096) (0.128) (0.096)Public status acquirer −0.274∗∗∗ −0.282∗∗∗ −0.386∗∗∗ −0.235∗∗∗

(0.078) (0.086) (0.128) (0.081)Public status target −0.882∗∗∗ −0.879∗∗∗ −1.165∗∗∗ −0.869∗∗∗

(0.118) (0.119) (0.177) (0.118)Divestiture 0.008 0.007 −0.077 0.002

(0.110) (0.110) (0.161) (0.161)Target subsidiary −0.339∗∗∗ −0.337∗∗∗ −0.341∗∗ −0.329∗∗∗

(0.106) (0.106) (0.157) (0.106)Previous quarter M&A 0.022∗∗∗ 0.022∗∗∗ 0.019∗∗∗ −0.022∗∗∗

successes (0.003) (0.002) (0.004) (0.002)Previous quarter M&A −0.065∗∗∗ −0.065∗∗∗ −0.069∗∗∗ −0.064∗∗∗

failures (0.006) (0.006) (0.008) (0.006)Total experience 0.008 −0.020

(0.019) (0.021)Total experience2 −0.0004 0.0006

(0.0008) (0.001)Ratio composite failures to −0.182∗∗

composite successes (0.081)Composite success 0.092∗∗∗

experience (0.024)Composite success −0.003∗∗∗

experience2 (0.001)Composite failure −0.217∗∗∗

experience (0.047)Composite failure 0.0158∗∗∗

experience2 (0.004)Cases in the analysis 4,973 4,973 3,666 4,973Wald chi-square 205.95∗∗∗ 209.15∗∗∗ 150.00∗∗∗ 218.90∗∗∗

-Pseudo log-likelihood 2,849.10 2,848.91 1,408.68 2,836.16

∗∗∗ p<0.01, ∗∗ p<0.05, ∗ p<0.1 (standard errors in parentheses).

contrast to Haleblian and Finkelstein, 1999), thenonsignificant result indicates that this measuremay be too agglomerated.11 The results in Model2 (Table 1) can also be used to assess if firmsget better at selecting targets with experience. Onewould expect them to learn to select targets suchthat they become more likely to complete a deal.In this case, we would expect a significant pos-itive coefficient of total experience in Model 2.Note that overall experience is a historical countof all M&As attempted by the firm (not only deals

11 Experimenting with moving three- and five-year windowsinstead of the historical count of total cumulative experienceyielded nonsignificant results as well.

attempted in the three-year time window), and isthe sum of both completed and abandoned deals.The nonsignificance of total cumulative experiencecan be interpreted as evidence that more experi-enced firms are neither necessarily better at com-pleting an acquisition, nor at selecting targets suchthat chances for deal completion increase.

Model 3 adds the ratio of the sum of failuresto the sum of successes (per firm, calculated from1981 until the date of the focal M&A). This servesas a proxy of a firm’s historical failure rate. Signand significance of the term suggest that a recordof failures negatively impacts completion probabil-ity. In robustness tests, we estimated our modelsincluding also ‘one-timers’—firms that attempted

Copyright 2012 John Wiley & Sons, Ltd. Strat. Mgmt. J., 33: 938–964 (2012)DOI: 10.1002/smj

954 K. Muehlfeld, P. Rao Sahib, and A. Van Witteloostuijn

Table 2. Results of logistic regression analysis predicting M&A completion (split samples)

Variables Model 5Domestic

transactions

Model 6Cross-bordertransactions

Model 7Intra-industrytransactions

Model 8Diversifyingtransactions

Model 9Friendly

transactions

Model 10Hostile

transactions

Intercept 1.403∗∗∗ 0.881∗∗ 1.582∗∗∗ 1.328∗∗∗ 1.377∗∗∗ 1.358∗∗∗

(0.136) (0.365) (0.202) (0.220) (0.138) (0.328)Cash payment 0.251∗∗ 1.115∗∗∗ −0.037 0.696∗∗∗ 0.732∗∗∗ −0.118

(0.127) (0.236) (0.199) (0.155) (0.144) (0.227)Public status −0.209∗∗ −0.338 −0.345∗ −0.292∗∗ −0.055 −0.642∗∗∗

acquirer (0.094) (0.262) (0.179) (0.143) (0.093) (0.215)Public status target −0.905∗∗∗ −0.450∗ −1.284∗∗∗ −0.259 −0.502∗∗∗ −0.118

(0.139) (0.244) (0.209) (0.177) (0.142) (0.251)Divestiture 0.0183 −0.0527 0.018 0.267 0.032 −1.965∗∗∗

(0.130) (0.297) (0.197) (0.197) (0.124) (0.46)Target subsidiary −0.295∗∗ −0.368 −0.420∗∗ −0.476∗∗∗ −0.350∗∗∗ 0.027

(0.131) (0.305) (0.195) (0.178) (0.122) (0.284)Previous quarter 0.019∗∗∗ 0.004 0.0197∗∗ −0.002 0.023∗∗∗ −0.029∗∗∗

M&A successes (0.003) (0.006) (0.005) (0.004) (0.003) (0.009)Previous quarter −0.060∗∗∗ −0.004 −0.075∗∗∗ −0.004 −0.068∗∗∗ −0.057∗∗∗

M&A failures (0.007) (0.005) (0.011) (0.009) (0.007) (0.019)Domestic success 0.141∗∗∗ −0.0131

experience (0.039) (0.070)Domestic success −0.005∗ 0.002

experience2 (0.002) (0.003)Domestic failure −0.192∗∗ −0.270

experience (0.098) (0.251)Domestic failure 0.0135 0.076

experience2 (0.018) (0.067)Cross-border success −0.030 0.167∗∗

experience (0.093) (0.066)Cross-border success −0.005 −0.0004

experience2 (0.011) (0.002)Cross-border failure −0.126 −0.401∗∗∗

experience (0.193) (0.104)Cross-border failure 0.035 0.040∗∗∗

experience2 (0.035) (0.015)Intra-industry success 0.236∗∗ 0.066

experience (0.100) (0.111)Intra-industry success −0.019 −0.000

experience2 (0.013) (0.01)Intra-industry failure −0.498∗∗∗ 0.159

experience (0.188) (0.189)Intra-industry failure 0.138∗∗ −0.062

experience2 (0.059) (0.061)Diversifying success 0.164∗∗ 0.128∗∗∗

experience (0.082) (0.039)Diversifying success −0.018∗∗ −0.003∗∗

experience2 (0.009) (0.001)Diversifying failure −0.125 −0.440∗∗∗

experience (0.132) (0.084)Diversifying failure 0.028 0.048∗∗∗

experience2 (0.0263) (0.013)Friendly success 0.060∗∗∗ −0.044

experience (0.022) (0.033)Friendly success −0.071 0.036

experience2 (0.079) (0.085)Friendly failure −0.117∗∗ −0.110

experience (0.05) 0.101)

Copyright 2012 John Wiley & Sons, Ltd. Strat. Mgmt. J., 33: 938–964 (2012)DOI: 10.1002/smj

Context-Specific Learning toward M&A Completion 955

Table 2. continued

Variables Model 5Domestic

transactions

Model 6Cross-bordertransactions

Model 7Intra-industrytransactions

Model 8Diversifyingtransactions

Model 9Friendly

transactions

Model 10Hostile

transactions

Friendly failure 0.115 1.806experience2 (0.385) (0.883)

Hostile success −0.124 0.621∗∗∗

experience (0.097) (0.171)Hostile success 0.031 −0.084∗∗∗

experience2 (0.017) (0.027)Hostile failure −0.089 −0.335∗∗

experience (0.097) (0.143)Hostile failure 0.012 0.016

experience2 (0.015) (0.011)Cases in the analysis 4,001 972 1,680 3,293 4,330 643Wald chi-square 178.83∗∗∗ 172.20∗∗∗ 128.17∗∗∗ 105.22∗∗∗ 184.98∗∗∗ 143.15∗∗∗

-Pseudolog-likelihood

2,199.198 409.448 948.138 949.039 2,270.917 394.20

∗∗∗ p<0.01, ∗∗ p<0.05, ∗ p<0.1 (standard errors in parentheses).

only one deal during 1981–2008—and includedan indicator for firms that attempted several acqui-sitions. The experience variables were interactedwith this indicator and appeared only for thosefirms that had attempted at least one M&A inthe past. This method allowed the subgroup offirms with experience to have a different baselineprobability of completion. The indicator was notsignificant and its inclusion did not change ourresults substantially. As to the control variables,cash-financed M&As were more likely to be com-pleted, whereas deals involving public acquirersand public targets, and subsidiaries were less likelyto be completed.

Model 3 does not allow us to account for theseparate effects of positive and negative prioroutcomes. Therefore, in Model 4, experience isdecomposed into successes and failures. The re-sults indicate that composite success experiencehad a positive influence on the likelihood of com-pletion (p < 0.01) with diminishing returns, aspredicted by Hypothesis 1a. Unlike hypothesized,we found evidence of an inflection point at threepast completed M&As. Failures had a significantnegative impact up to a turning point at aroundsix failures, and a positive effect beyond this(p < 0.01). This suggests partial support for Hypo-thesis 1a and full support for Hypothesis 1b. Inrobustness tests, we investigated whether it wasone type of firm driving the successes and anothertype driving the failures. To this end, we exam-ined the breakdown of successes and failures per

firm. Of the 1,964 firms in our sample, only 310firms experienced solely failed or solely com-pleted M&As. These firms contributed in total 839attempts. The remainder of the announcements wasmade by firms that experienced a mix of failuresand successes. This is because firms that exclu-sively experienced failures (successes) typicallymade two, at most three, acquisitions over thesample period. Omitting these 839 cases did notchange the significance of the coefficient estimates,but their magnitude differed slightly.

Models 5 to 10 (Table 2; split samples) testHypotheses 2a and 2b, focusing on the context-specificity of experience. We used a sample splitto test our hypotheses in order to compare theeffects of success and failure experiences gainedin different contexts on the probability that a par-ticular type of acquisition was completed. Models5 and 6 deal with the distinction between domesticvs. cross-border deals, Models 7 and 8 with intra-industry vs. diversifying deals, and Models 9 and10 with friendly vs. hostile contexts. For five out ofsix contexts, we find that only same-context expe-rience significantly affects focal completion likeli-hood. The exception is the intra-industry context.As hypothesized, failure experience gained in thealternative context (diversifying) has no significanteffect here. However, contrary to expectations,diversifying success experience has a significantspillover effect (positive with diminishing returns).Overall, the results offer full support for Hypothe-sis 2b and partial support for Hypothesis 2a. In

Copyright 2012 John Wiley & Sons, Ltd. Strat. Mgmt. J., 33: 938–964 (2012)DOI: 10.1002/smj

956 K. Muehlfeld, P. Rao Sahib, and A. Van Witteloostuijn

robustness analyses, we explored the possibilityof learning when a firm made repeated acquisi-tions in the same foreign country. Based on thenumber of different target nations that an acquirerwas involved in, we found little evidence of multi-nationality in acquirers’ selection of targets. Forexample, 81 percent of acquirers that engaged incross-border acquisitions concentrated their activ-ity in a single foreign country. Of those engagedin M&As in several foreign countries, 90 percentlimited themselves to two foreign countries. Todifferentiate the first deal of an acquirer in a for-eign country from subsequent deals, we includeda dummy for the first acquisition, but found it notto be significant.

Further, we theorized that learning patternswould depend on a context’s degree of structuralvariance and stimulation of deliberate learning.From among our six contexts, and for our specificresearch setting, we interpreted hostile M&As andcross-border M&As as possessing the highest lev-els of structural variance and incitation of deliber-ate learning, respectively. Hypothesis 3a predictedan inverted U-shaped relationship between suc-cess with consummating hostile M&As and thecompletion likelihood of a focal hostile acquisi-tion. Based on a positive main effect (p < 0.01), anegative squared term (p < 0.01), and the inflec-tion point at three past hostile deals, Hypothesis3a receives full support (Model 10). Hypothesis3b predicted a negative effect of failed hostileattempts on focal completion likelihood in thesame context. As expected, we found a negativemain effect (p < 0.05), and no significant squaredterm (Model 10). Hypothesis 3b is therefore sup-ported. Hypothesis 4a predicted a positive effectof completed cross-border M&As on same-contextcompletion likelihood. In support of Hypothesis4a, the main effect is positive and significant(p < 0.05; Model 6), and the squared term is notsignificant. Hypothesis 4b predicted a U-shapedrelationship between terminated cross-borderM&As and focal completion likelihood in thesame context. In support of Hypothesis 4b, themain effect is negative (p < 0.01) and the squaredterm positive (p < 0.01), with the inflection pointobserved at six failures.

For the four other contexts, we expected, byand large, the same patterns as hypothesized atthe aggregate level in Hypothesis 1a for successesand Hypothesis 1b for failures, given that wedid not have strong theoretical reasons to predict

otherwise. However, only failure-based learningfrom intra-industry and diversifying M&As fullyadhered to this expectation. Success in complet-ing diversifying and domestic deals initially hada positive effect on focal completion likelihood.Beyond some threshold, the impact turned nega-tive, leading to an inverted U shape. While theseresults are contrary to our expectations, they arein line with the results for success-based learn-ing at the aggregate level. Here, we also found aninverted U shape instead of the hypothesized posi-tive but diminishing effect. As to the impact of thecontextual properties ‘intra-industry’ and ‘friendly’on success-based learning, we found no evidenceof curvilinearity, but instead a significant posi-tive effect. Similarly, evidence of curvilinearity forfailure-based learning in ‘domestic’ and ‘friendly’deals is lacking. Terminations had a negative maineffect, but the coefficients for the squared termswere not significant.

In robustness analyses, we also experimentedwith firm size. While large acquirers attemptedmore M&As, size (measured by total assets andtotal turnover) was not significant in explaining theprobability that an individual deal would be com-pleted. Also, when omitting the top five (top 10)firms in terms of M&A activity, results were virtu-ally unchanged. Finally, in supplementary analyses(available upon request), we calculated marginaleffects for the change in completion likelihoodwhen experience is increased by one M&A whileall other variables are fixed at their respectivemeans. Corroborating prior work (e.g., Madsen andDesai, 2010), we found an asymmetry in responsesto positive and negative outcomes. Across all typesof context, the magnitude of the negative effectof an additional failure was much larger than thepositive effect of an additional success.

DISCUSSION AND CONCLUSION

Despite a wealth of research on organizationallearning in M&As, we know little about how vari-ous types of context condition experiential learningfrom past successes and failures. The present studyaddresses this gap by analyzing, first, whether thetransferability of experiential learning is limitedby acquisition context. In so doing, and unlikemost prior studies (for an exception, see Mad-sen and Desai, 2010), this work explicitly distin-guishes between success and failure experiences.

Copyright 2012 John Wiley & Sons, Ltd. Strat. Mgmt. J., 33: 938–964 (2012)DOI: 10.1002/smj

Context-Specific Learning toward M&A Completion 957

Second, we investigate how particular properties ofM&A contexts—variance in structure and stimu-lation of deliberate learning—map onto distinct,context-specific learning patterns. In contrast tothe majority of work, which has assessed learningas reflected in post-acquisition performance, wefocus on learning with respect to an intermediateoutcome—the completion of an announced deal.Taking the completion of an announced acquisitionas a manifestation of organizational effectivenessand efficiency in the final stages of M&A prepa-ration, we study whether—and if so, how—thismeasure is influenced by organizational learning.

This study makes several contributions to orga-nizational learning theory in general, and to the lit-erature on acquisition learning in particular. First,it highlights an important boundary condition ofexperiential learning—context. We extend priortheory on the limits of experience transfer (Finkel-stein and Haleblian, 2002; Haleblian and Finkel-stein, 1999; Hayward, 2002) by defining structuralsimilarity not only based on industry context butalso in relation to other contextual dimensions(country borders and surrounding attitude). Empir-ically, we obtain strong support for the notion thattransferability is restricted within these contexts(Williams, 2007)—that is, that the six analyzedM&A contexts confer structural similarity. Domes-tic experience, for example, affects the completionlikelihood of subsequent domestic but not of cross-border deals.

Second, this study answers significant questionsconcerning distinct learning processes in responseto success and failure, with context as a power-ful moderating influence. Our results suggest thatlearning from success can enhance future perfor-mance but in more complex ways than suggestedby the myopic, self-limiting process tradition-ally conceptualized in experiential learning theory(Schulz, 2002). We reveal that prior successes hada positive and non-diminishing influence in somecontexts, whereas in other contexts, and at theaggregate level, the impact was positive for smallnumbers of completed transactions but turned neg-ative for larger numbers. These results may helpreconcile ambiguities discussed in the literature.On the one hand, success experience appears tofoster subsequent performance by facilitating therefinement of successful routines. On the otherhand, recent work has cautioned scholars aboutorganizations’ learning from success, arguing thatperceptions of success may hamper learning by

stimulating dysfunctional reactions such as super-stition, especially in infrequent activities (Levittand March, 1988; Zollo, 2009). Our results suggestthat which of the effects dominates may depend onthe specific context and its properties, such as thedegree of structural variance and extent of stimu-lation of deliberate learning. The precise nature oflearning from failure similarly depends on context.In most contexts, the predicted U-shaped patternprevails, indicating that after a string of failures,firms reach a turning point beyond which they areable to successfully encode lessons from furtherterminated attempts into their routines and ulti-mately improve the chances of completing futuredeals. Yet, in domestic and hostile M&A contexts,the adverse effects of failures dominate, suggest-ing that certain contextual characteristics, such as ahigh degree of structural variance within a context,obstruct learning. The observation that firms needto accumulate a relatively large number of failuresbefore they are able to reap the benefits of learn-ing from failure, together with the finding that theyare unable to extract any valuable lessons fromcertain types of failure, corroborates prior workthat has documented difficulties of firms’ learningfrom failure in complex corporate activities (e.g.,Hayward, 2002; Madsen and Desai, 2010) and con-trasts with more positive accounts of learning fromfailure in settings such as railroad accidents (Baumand Dahlin, 2007) and airline accidents and inci-dents (Haunschild and Sullivan, 2002).

Third, this work presents a first attempt to cap-ture how specific contexts map onto distinct learn-ing patterns, depending on certain generic dimen-sions. We argue that intra-context structural vari-ance and the extent to which a context stimulatesdeliberate learning are two important determinantsof firms’ learning toward the completion of pub-licly announced M&As, and that contexts vary sys-tematically along these dimensions. We identifiedhostile acquisitions as particularly high in struc-tural variance and cross-border deals as most con-ducive to mindful learning from among the ana-lyzed six contexts. As predicted, firms are largelyunable to extract valuable lessons from terminatedhostile takeovers, and additional completed hostileM&As reduced the chances for closing future dealsafter a few successes. These dampening effectsof structural variance within a context relate tostudies that point to the dangers of inappropri-ate generalization from experience (Haleblian andFinkelstein, 1999; Nadolska and Barkema, 2007).

Copyright 2012 John Wiley & Sons, Ltd. Strat. Mgmt. J., 33: 938–964 (2012)DOI: 10.1002/smj

958 K. Muehlfeld, P. Rao Sahib, and A. Van Witteloostuijn

The general argument is that experience accumu-lation harms future performance if it induces firmsto inappropriately generalize from past experiencethat shares only surface similarities with the focalevent.

Our results add to this literature by highlight-ing the ambiguous role of context. On the onehand, organizations indeed appear to categorizeM&As based on the distinction between hostileand friendly context, as only same-context experi-ence had a significant impact on subsequent com-pletion likelihood. On the other hand, firms’ learn-ing form hostile acquisition experience seems tosuffer from a high degree of heterogeneity betweendeals within this category, and possibly a tendencyto inappropriately generalize from one hostile dealto the other, given that they belong in the samecategory. In contrast, we find that firms’ learn-ing in the final stages of cross-border transactionspositively deviates from aggregate learning pat-terns. We attribute this result to greater opennessbrought about by unfamiliar institutional settingsand heightened awareness of the need to engagein deliberate learning. In that sense, our find-ings underscore the importance of deliberate learn-ing, especially in complex domains that are nothighly amenable to semi-automatic, exploitation-biased learning (Madsen and Desai, 2010; Zolloand Singh, 2004).

Overall, the results suggest that properties of thecontext in which experience is gained influence thenature of learning processes—both from successesand failures—and create substantial variation inthese processes across contexts. We reveal that inaddition to the hypothesized context-specific learn-ing patterns for cross-border and hostile acqui-sitions, the relationship between experience andcompletion likelihood also diverges from theaggregate pattern in some of the other contexts.First, firms managed to capitalize more success-fully than expected on past completions of bothintra-industry and friendly deals. Second, learningfrom abandonment in a domestic and friendly con-text appears challenging. We speculate that proper-ties not captured in our framework—for example,the degree of industry consensus about the needand direction of consolidation, which arguably isunusually high in the newspaper industry—mightplay a role in shaping context-specific learningprocesses in domestic, intra-industry, and friendlyacquisitions. In sum, our results indicate the needfor future research into the moderating role of

context on experiential learning related to othercorporate activities, goal variables, and contextualproperties.

Managerial implications

For managers, our results have three broad impli-cations. First, they draw attention to an impor-tant boundary condition of learning from expe-rience: its context. Second, the results highlightthat learning how to close M&As from whatwent wrong in past deals is possible but chal-lenging. Developing superior strategies for copingwith failure should allow organizations to reapsubstantial gains. Extracting lessons from termi-nated M&As, such as learning to spot disadvan-tageous deals early on and abandon them priorto the IPA or improving negotiation strategies,should allow firms to use their resources moreefficiently. Prior research has uncovered substan-tial variation in firms’ dealing with failures (Mad-sen and Desai, 2010). Generally, organizationalmembers frequently refuse to acknowledge failure,for example, due to associated stigmata (March,Sproull, and Tamuz, 1991). Organizations that areunable to institutionalize practices that counter-act these tendencies deprive themselves of valu-able learning opportunities (Madsen and Desai,2010). In contrast, firms that manage to imple-ment suitable communication structures and createa climate characterized by tolerance for disagree-ment and openness are likely to stimulate discus-sion about mistakes and increase rates of errordetection (Edmondson, 2004). Third, the observedcontext-specific learning patterns point to facil-itators of learning. They underscore the crucialrole of deliberate learning efforts (Madsen andDesai, 2010; Zollo and Singh, 2004), suggestingthat institutional structures that improve learningconditions such as documentation requirementsand discussion forums that raise awareness of thedistinct properties of acquisition experience maycounterbalance the dampening effects of selec-tive managerial perception and organizational iner-tia. These measures may be particularly in orderwhen high intra-category structural variance ham-pers learning, as with hostile M&As: learning doesnot happen automatically with experience accu-mulation. Instead, managers need to recognize thesupportive role of deliberate and focused learning.

Copyright 2012 John Wiley & Sons, Ltd. Strat. Mgmt. J., 33: 938–964 (2012)DOI: 10.1002/smj

Context-Specific Learning toward M&A Completion 959

Limitations and suggestions for future research

Our study has several limitations that could beaddressed in future work. First, we focused ona specific setting, in the form of a single indus-try—newspapers—during a particular stage of theindustry life cycle—its decline in major regionalmarkets. Single-industry samples have strengthsand weaknesses (Baum and Dahlin, 2007; Hale-blian et al., 2006). While they allow researchersto control for exogenous industry effects and areoften justified on the grounds of suitability ofan industry, they limit generalizability. This isa particular concern when the importance of thedependent variable may vary with industry con-ditions, as in the case of acquisition completion(Aguilera and Dencker, 2008). Generalizability toother sectors and stages of the industry life cycleshould, thus, be probed in future research. Anothersampling-related limitation pertains to our crite-rion for including an acquisition attempt, namelythat the intention to complete it has been publiclyannounced. Ideally, one would like to account fordifferences in announcement policies. As it stands,we cannot empirically exclude the possibility thatfirms that came across as superior learners werethose that delayed announcements, leading us tooverestimate learning effects since our sample onlycovers attempts after they are announced. Stricter,more comprehensive disclosure requirements easesuch concerns for publicly listed firms by spec-ifying a point beyond which the IPA may not bedelayed. Given, thus, that the majority of acquirersin our sample are publicly owned, we are confi-dent about the basic validity of our learning-relatedinterpretation.

Our outcome measure—completion of an an-nounced acquisition—allowed us to focus onlearning aimed at a novel organizational goalvariable during an underexplored phase of theacquisition process (Aguilera and Dencker, 2008;Dikova et al., 2010). However, precisely becauseso little is known about learning processes inacquisition preparation, future studies should in-vestigate and compare alternative indicators, suchas the time needed to complete a deal (Dikovaet al., 2010) or the premium paid (Haunschild,1994; Laamanen, 2007). Is there a trade-offbetween different measures? Does a higher proba-bility of closing a deal require sacrifices in terms oflonger duration or a higher premium paid? Alter-natively, learning during the final stages could

result in both a higher likelihood of completionand lower premiums, pointing, for example, tosuperior negotiating skills. Related, our study isconfined to the public takeover phase. Learningduring the earlier stages relates to different activ-ities such as screening, target selection, and duediligence (Jemison and Sitkin, 1986). The mech-anisms that drive learning during these phasesmay differ and warrant an explicit study in theirown right. Future research should also addressthe relationship between goal variables at the pre-and post-completion stages. Prior work has arguedthat acquirers frequently overpay and that posi-tive effects of M&As primarily accrue to targetshareholders. Do firms that are more proficientin completing deals suffer with regard to post-acquisition performance? Or, conversely, are pre-dictors of completion positively associated withpost-acquisition performance? If so, what is driv-ing this relation: are firms that are superior learnersduring early stages also better at capitalizing onlearning opportunities during integration? Theseresearch issues directly emanate from our studyand will hopefully stimulate future conceptual andempirical advances in the analysis of organiza-tional learning in M&As.

ACKNOWLEDGEMENTS

We are grateful to Editor Edward Zajac andtwo anonymous reviewers for their valuable com-ments that helped us greatly in improving thispaper. We thank seminar audiences at the Uni-versity of Antwerp, the University of Gronin-gen, Utrecht University, and at the AOM 2008and EGOS 2008 conferences for useful commentson earlier drafts. Katrin Muehlfeld and Arjenvan Witteloostuijn gratefully acknowledge fund-ing from the Antwerp Centre of EvolutionaryDemography (ACED), which is financed throughthe Odysseus program of the Flemish ScienceFoundation (FWO). All remaining errors are ourresponsibility.

REFERENCES

Adair WL, Brett JM. 2005. The negotiation dance: timeculture, and behavioral sequences in negotiation.Organization Science 16(1): 33–51.

Aguilera RV, Dencker JC. 2008. Determinants of acqui-sition completion: a relational perspective. Working

Copyright 2012 John Wiley & Sons, Ltd. Strat. Mgmt. J., 33: 938–964 (2012)DOI: 10.1002/smj

960 K. Muehlfeld, P. Rao Sahib, and A. Van Witteloostuijn

paper, College of Business, University of Illinois atUrbana-Champaign, IL.

Aguilera RV, Dencker JC, Escandell X. 2004. Left atthe altar: an institutional analysis of global mergerand acquisition announcements in the 1990s. Workingpaper 04–0117, College of Business, University ofIllinois at Urbana-Champaign, IL.

Amburgey TL, Kelly D, Barnett WP. 1993. Resettingthe clock: the dynamics of organizational changeand failure. Administrative Science Quarterly 38(1):51–73.

Andrade G, Stafford E. 2004. Investigating the economicrole of mergers. Journal of Corporate Finance 10(1):1–36.

Argote L. 1999. Organizational Learning: Creating,Retaining and Transferring Knowledge. KluwerAcademic: Boston, MA.

Argote L, Beckman SL, Epple D. 1990. The persistenceand transfer of learning in industrial settings.Management Science 36(2): 140–154.

Argote L, Todorova G. 2007. Organizational learning. InInternational Review of Industrial and OrganizationalPsychology , Hodgkinson GP, Ford JK (eds). Wiley:Chichester, UK; 193–234.

Asquith P. 1983. Merger bids, uncertainty, and share-holder returns. Journal of Financial Economics11(1–4): 51–83.

Bagdikian BH. 2004. The New Media Monopoly . BeaconPress: Boston, MA.

Barkema HG, Schijven M. 2008. Toward unlocking thefull potential of acquisitions: the role of organizationalrestructuring. Academy of Management Journal 51(4):696–722.

Baum JAC, Dahlin KB. 2007. Aspiration performanceand railroads’ patterns of learning from train wrecksand crashes. Organization Science 18(3): 368–385.

Beckman CM, Haunschild PR. 2002. Network learning:the effects of partners’ heterogeneity of experienceon corporate acquisitions. Administrative ScienceQuarterly 47(1): 92–124.

Belsley DA. 1984. Demeaning conditioning diagnosticsthrough centering. American Statistician 38(2):73–77.

Betton S, Eckbo BE, Thorburn KS. 2009. Mergernegotiations and the toehold puzzle. Journal ofFinancial Economics 91(2): 158–178.

Boone AL, Mulherin JH. 2007. Do termination provi-sions truncate the takeover bidding process? Reviewof Financial Studies 20(2): 461–489.

Bradley M. 1980. Interfirm tender offers and the marketfor corporate control. Journal of Business 53(4):345–376.

Bruton GD, Oviatt BM, White MA. 1994. Performanceof acquisitions of distressed firms. Academy ofManagement Journal 37(4): 972–989.

Cameron AC, Trivedi PK. 2009. MicroeconometricsUsing Stata . Stata Press: College Station, TX.

Carapeto M, Moeller S, Faelten A. 2010. What shouldI do next? CEO succession and subsequent M&Astrategy. Working paper, Cass Business School, CityUniversity of London: London, UK.

Chakrabarti R, Gupta-Mukherjee S, Jayaraman N. 2009.Mars-Venus marriages: culture and cross-borderM&As. Journal of International Business Studies40(2): 216–236.

Chan-Olmsted SM, Chang BH. 2003. Diversificationstrategy of global media conglomerates: examiningits patterns and determinants. Journal of MediaEconomics 16(4): 213–233.

Clougherty JA. 2005. Antitrust holdup source, cross-national institutional variation, and corporate politicalstrategy implications for domestic mergers in aglobal context. Strategic Management Journal 26(8):769–790.

Cyert RM, March JG. 1963. A Behavioral Theory of theFirm . Prentice Hall: Englewood Cliffs, NJ.

Denrell J, Fang C, Levinthal DA. 2004. From T-mazesto labyrinths: learning from model-based feedback.Management Science 50(10): 1366–1378.

Dikova D, Rao Sahib P, van Witteloostuijn A. 2010.Cross-border acquisition abandonment and comple-tion: the effect of institutional differences and organi-zational learning in the international business serviceindustry, 1981–2001. Journal of International Busi-ness Studies 41(2): 223–245.

Echambadi R, Hess JD. 2007. Mean-centering doesnot alleviate collinearity problems in moderatedmultiple regression models. Marketing Science 26(3):438–445.

Economist. 2006. More media, less news: newspapers aremaking progress with the Internet, but most are stilltoo timid, defensive or high-minded. The NewspaperIndustry, Special Report. 26 August, 50–52.

Economist. 2008. Calling the shots: India’s newspapersembrace a profitable but questionable new sideline.India Media, Business. 3 May, 70.

Economist. 2009. High wires: with newspapers in crisis,newswires may learn to live without them. NewsAgencies, Business. 14 February: 68.

Edmondson AC. 2004. Learning from mistakes is easiersaid than done: group and organizational influences onthe detection and correction of human error. Journalof Applied Behavioral Science 40(1): 66–90.

Evans P, Pucik V, Barsoux JL. 2002. The GlobalChallenge: Frameworks for International HumanResource Management . McGraw-Hill: Boston, MA.

Fine J. 2006. McClatchy’s big bet on old media.Bloomberg BusinessWeek. 14 March. http://www.businessweek.com/investor/content/mar2006/pi20060314 402352.htm (21 March 2011).

Finkelstein S, Haleblian J. 2002. Understanding acquisi-tion performance: the role of transfer effects. Organi-zation Science 13(1): 36–47.

Fowler KL, Schmidt DR. 1989. Determinants of tenderoffer post-acquisition financial performance. StrategicManagement Journal 10(4): 339–350.

Frankfurter Allgemeine Zeitung . 2006. SpringersRuckzug. 2 February, 1.

Genter D, Loewenstein J, Thompson L. 2003. Learningand transfer: a general role for analogical encoding.Journal of Educational Psychology 95(2): 393–408.

Goulet PK, Schweiger DM. 2006. Managing cultureand human resources in mergers and acquisitions.

Copyright 2012 John Wiley & Sons, Ltd. Strat. Mgmt. J., 33: 938–964 (2012)DOI: 10.1002/smj

Context-Specific Learning toward M&A Completion 961

In Handbook of Research in International HumanResource Management , Stahl GK, Bjorkman I (eds).Edward Elgar: Cheltenham, UK; 405–429.

Greve HR. 2008. A behavioral theory of firm growth:sequential attention to size and performance goals.Academy of Management Journal 51(3): 476–494.

Grocer S. 2007. A holiday toast to biggest year inM&A history. The Wall Street Journal, 21 Decem-ber. http://blogs.wsj.com/deals/2007/12/21/a-holiday-toast-to-biggest-year-in-ma-history/(21 March 2011).

Guest P. 2009. The impact of mergers and acquisitionson executive pay in the United Kingdom. Economica76(301): 149–175.

Haleblian J, Devers CE, McNamara G, Carpenter MA,Davison RB. 2009. Taking stock of what we knowabout mergers and acquisitions: a review and researchagenda. Journal of Management 35(3): 469–502.

Haleblian J, Finkelstein S. 1999. The influence oforganizational acquisition experience on acquisitionperformance: a behavioral learning perspective.Administrative Science Quarterly 44(1): 29–56.

Haleblian J, Kim J, Rajagopalan N. 2006. The influenceof acquisition experience and performance on acqui-sition behavior: evidence from the U.S. commercialbanking industry. Academy of Management Journal49(2): 357–370.

Harford J, Jenter D, Li K. 2007. Conflicts of interestsamong shareholders: the case of corporate acquisi-tions. Working paper no. 4653-07. MIT Sloan Schoolof Management: Cambridge, MA.

Haspeslagh P, Jemison D. 1991. Managing Acquisitions .Free Press: New York.

Haunschild PR. 1994. How much is that companyworth? Interorganizational relationships, uncertainty,and acquisition premiums. Administrative ScienceQuarterly 39(3): 391–411.

Haunschild PR, Sullivan BN. 2002. Learning fromcomplexity: effects of prior accidents and incidentson airlines’ learning. Administrative Science Quarterly47(4): 609–643.

Hayward ML. 2002. When do firms learn from theiracquisition experience? Evidence from 1990–1995.Strategic Management Journal 23(1): 21–29.

Hitt MA, Harrison JS, Ireland RD, Best A. 1998. Attri-butes of successful and unsuccessful acquisitions ofU.S. firms. British Journal of Management 9(2):91–114.

Hitt MA, Hoskisson RE, Ireland RD. 1990. Mergers andacquisitions and managerial commitment to innovationin M-form firms. Strategic Management Journal ,Summer Special Issue 11: 29–47.

Holl P, Kyriazis D. 1996. The determinants of outcomein U.K. takeover bids. International Journal of theEconomics of Business 3(2): 165–184.

Huff A. 1982. Industry influences on strategy reformula-tion. Strategic Management Journal 3(2): 119–131.

Jemison DB, Sitkin SB. 1986. Corporate acquisitions: aprocess perspective. Academy of Management Review11(1): 145–163.

Johns G. 2006. The essential impact of context onorganizational behavior. Academy of ManagementReview 31(2): 386–408.

Kelley HH. 1971. Attribution in Social Interactions .General Learning Press: New York.

King DR, Dalton DR, Daily CM, Covin JG. 2004. Meta-analyses of post-acquisition performance: indicatorsof unidentified moderators. Strategic ManagementJournal 25(2): 187–200.

Kusewitt JB. 1985. An exploratory study of strategicacquisition factors relating to performance. StrategicManagement Journal 6(2): 151–169.

Laamanen T. 2007. On the role of acquisition premiumin acquisition research. Strategic Management Journal28(13): 1359–1369.

Laamanen T, Keil T. 2008. Performance of serialacquirers: toward an acquisition program perspective.Strategic Management Journal 29(6): 663–672.

Lajoux AR, Elson CM. 2000. The Art of M&A DueDiligence. McGraw-Hill: New York.

Larsson R, Finkelstein S. 1999. Integrating strategic,organizational, and human resource perspectives onmergers and acquisitions: a case survey realization.Organization Science 10(1): 1–26.

Lehn KM, Zhao M. 2006. CEO turnover after acquisi-tions: are bad bidders fired? Journal of Finance 61(4):1759–1811.

Levinthal DA, March JG. 1993. The myopia of learning.Strategic Management Journal , Winter Special Issue14: 95–112.

Levinthal DA, Rerup C. 2006. Crossing an apparentchasm: bridging mindful and less-mindful perspectiveson organizational learning. Organization Science17(4): 502–513.

Levitt B, March JG. 1988. Organizational learning.Annual Review of Sociology 14: 319–340.

Lomi A, Larsen ER, Ginsberg A. 1997. Adaptive learn-ing in organizations: a system dynamics-based explo-ration. Journal of Management 23(4): 561–582.

Luo Y. 2005. Do insiders learn from outsiders? Evidencefrom mergers and acquisitions. Journal of Finance60(4): 1951–1982.

Madsen PM, Desai V. 2010. Failing to learn? The effectsof failure and success on organizational learning inthe global orbital launch vehicle industry. Academy ofManagement Journal 53(3): 451–476.

March JG. 1991. Exploration and exploitation inorganizational learning. Organization Science 2(1):71–87.

March JG, Sproull LS, Tamuz M. 1991. Learning fromsamples of one or fewer. Organization Science 2(1):1–13.

Meyer CB, Altenborg E. 2008. Incompatible strategies ininternational mergers: the failed merger between Teliaand Telenor. Journal of International Business Studies39(3): 508–525.

Morosini P, Singh H. 1994. Post-cross border acqui-sitions: Implementing ‘national culture-compatible’strategies to improve performance. European Manage-ment Journal 12(4): 390–400.

Muehlfeld K, Rao Sahib P, van Witteloostuijn A. 2007.Completion or abandonment of mergers and acqui-sitions: evidence from the newspaper industry1981–2001. Journal of Media Economics 20(2):107–137.

Copyright 2012 John Wiley & Sons, Ltd. Strat. Mgmt. J., 33: 938–964 (2012)DOI: 10.1002/smj

962 K. Muehlfeld, P. Rao Sahib, and A. Van Witteloostuijn

Nadolska A, Barkema HG. 2007. Learning to interna-tionalise: the pace and success of foreign acquisi-tions. Journal of International Business Studies 38(7):1170–1186.

Nelson RR, Winter SG. 1982. An Evolutionary Theory ofEconomic Change. Belknap: Cambridge, MA.

North D. 1990. Institutions, Institutional Change andEconomic Performance. Cambridge University Press:Cambridge, UK.

Officer MS. 2003. Termination fees in mergers andacquisitions. Journal of Financial Economics 69(3):431–467.

Pollock TG, Rindova VP, Maggitti PG. 2008. Marketwatch: information and availability cascades amongthe media and investors in the U.S. IPO market.Academy of Management Journal 51(2): 335–358.

Pritchard D. 2001. A tale of three cities: ‘diverse andantagonistic’ information in situations of local news-paper/broadcast cross-ownership. Federal Communi-cations Law Journal 54(1): 31–52.

Pryor FL. 2001. Dimensions of the worldwide mergerboom. Journal of Economic Issues 35(4): 825–840.

Rhee M, Valdez ME. 2009. Contextual factors surround-ing reputation damage with potential implications forreputation repair. Academy of Management Review34(1): 146–168.

Schmidt DR, Fowler KL. 1990. Post-acquisition financialperformance and executive compensation. StrategicManagement Journal 11(7): 559–569.

Schulz M. 2002. Organizational learning. In Companionto Organizations , Baum JAC, Rowley TJ (eds).Blackwell: Malden, MA; 415–441.

Schwert GW. 2000. Hostility in takeovers: in the eyes ofthe beholder? Journal of Finance 55(6): 2599–2640.

Semadeni M, Cannella AA Jr, Fraser DR, Lee DS. 2008.Fight or flight: managing stigma in executive careers.Strategic Management Journal 29(5): 557–567.

Shi W, Prescott JE. 2011. Sequence patterns of firms’acquisition and alliance behaviour and their perfor-mance implications. Journal of Management Studies48(5): 1044–1070.

Sitkin S. 1992. Learning through failure: the strategy ofsmall losses. In Research in Organizational Behavior(Volume 14), Cummings LL, Staw BM (eds). JAIPress: Greenwich, CT; 231–266.

Spender JC. 1987. Industry Recipes: An Enquiry inthe Nature and Sources of Managerial Judgment .Blackwell: Oxford, UK.

Sudarsanam PS. 1991. Defensive strategies of targetcompanies in UK contested takeovers. ManagerialFinance 17(6): 47–56.

van Kranenburg H. 2002. Mobility and market structurein the Dutch daily newspaper market segments.Journal of Media Economics 15(2): 107–123.

Very P, Schweiger DM. 2001. The acquisition processas learning process: evidence from a study of criticalproblems and solutions in domestic and cross-borderdeals. Journal of World Business 36(1): 11–31.

Weston JF, Mitchell ML, Mulherin JH. 2004. Takeovers,Restructuring, and Corporate Governance (4th edn).Prentice Hall: Upper Saddle River, NJ.

Williams C. 2007. Transfer in context: replicationand adaptation in knowledge transfer relationships.Strategic Management Journal 28(9): 867–889.

Winter S. 1987. Knowledge and competence as strategicassets. In The Competitive Challenge: Strategies forIndustrial Innovation and Renewal, Teece DJ (ed).Ballinger: Cambridge, MA; 159–184.

Wong P, O’Sullivan N. 2001. The determinants andconsequences of abandoned takeovers. Journal ofEconomic Surveys 15(2): 145–186.

World Association of Newspapers. 2009. World PressTrends 2009 . Zenith Media: Paris, France.

Yang C, Weber J. 2003. Media merger mania: the firstwave. Bloomberg BusinessWeek,30 May. http://www.businessweek.com/bwdaily/dnflash/may2003/nf20030530 9857 db038.htm (21 March 2011).

Zollo M. 2009. Superstitious learning with rare strategicdecisions: theory and evidence from corporateacquisitions. Organization Science 20(5): 894–908.

Zollo M, Meier D. 2008. What is M&A performance?Academy of Management Perspectives 22(3): 55–77.

Zollo M, Reuer JJ. 2010. Experience spillovers acrosscorporate development activities. Organization Sci-ence 21(6): 1195–1212.

Zollo M, Singh H. 2004. Deliberate learning in corporateacquisitions: post-acquisition strategies and integrationcapability in U.S. bank mergers. Strategic Manage-ment Journal 25(13): 1233–1256.

Copyright 2012 John Wiley & Sons, Ltd. Strat. Mgmt. J., 33: 938–964 (2012)DOI: 10.1002/smj

Context-Specific Learning toward M&A Completion 963

AP

PE

ND

IX

Tabl

eA

1.Ta

ble

ofco

rrel

atio

nsfo

rva

riab

les

inM

odel

s1

–3

inTa

ble

1

Mea

nS.

D.

(1)

(2)

(3)

(4)

(5)

(6)

(7)

(8)

(1)

Cas

hpa

ymen

t0.

240.

48(2

)Pu

blic

stat

usac

quir

er0.

460.

490.

26∗

(3)

Publ

icst

atus

targ

et0.

150.

360.

31∗

0.23

(4)

Div

estit

ure

0.30

0.46

0.00

−0.0

7∗−0

.26

(5)

Targ

etsu

bsid

iary

0.37

0.48

−0.0

3∗−0

.03∗

−0.3

4∗0.

72(6

)Pr

evio

usqu

arte

rM

&A

succ

esse

s42

.38

15.1

5−0

.04∗

−0.0

3∗−0

.04∗

0.01

−0.0

2(7

)Pr

evio

usqu

arte

rM

&A

failu

res

15.5

97.

15−0

.04∗

−0.0

0−0

.02

0.00

−0.0

10.

06∗

(8)

Tota

lex

peri

ence

3.51

5.52

0.04

∗0.

36∗

0.09

∗−0

.05∗

−0.0

2∗0.

06∗

0.06

∗

(9)

Rat

ioco

mpo

site

failu

res

toco

mpo

site

succ

esse

s0.

420.

710.

05∗

0.19

∗0.

06∗

0.0∗

−0.0

20.

040.

04∗

0.07

∗

∗C

orre

latio

nis

sign

ifica

ntat

the

0.05

leve

l(t

wo-

taile

d).

Tabl

eA

2.Ta

ble

ofco

rrel

atio

nsfo

rva

riab

les

inM

odel

4in

Tabl

e1

Mea

nS.

D.

(1)

(2)

(3)

(4)

(5)

(6)

(7)

(10)

(10)

Com

posi

tesu

cces

sex

peri

ence

2.51

3.98

0.02

∗0.

29∗

0.07

∗−0

.05∗

−0.0

2∗0.

04∗

0.05

∗

(11)

Com

posi

tefa

ilure

expe

rien

ce1.

182.

270.

06∗

0.39

∗0.

10∗

−0.0

4∗−0

.01

−0.0

6∗0.

10∗

0.58

∗

∗C

orre

latio

nis

sign

ifica

ntat

the

0.05

leve

l(t

wo-

taile

d);

vari

able

num

beri

ngfo

r(1

)–(7

)fo

llow

sTa

ble

A1

abov

e.

Tabl

eA

3.Ta

ble

ofco

rrel

atio

nsfo

rva

riab

les

inM

odel

s5

–6

inTa

ble

2

Mea

nS.

D.

(1)

(2)

(3)

(4)

(5)

(6)

(7)

(10)

(11)

(12)

(10)

Dom

estic

succ

ess

expe

rien

ce1.

472.

690.

09∗

0.32

∗0.

04∗

−0.0

3−0

.03

0.07

0.06

(11)

Dom

estic

failu

reex

peri

ence

0.49

1.04

0.10

∗0.

35∗

0.09

∗−0

.05

−0.0

2−0

.03∗

0.05

∗0.

50∗

(12)

Cro

ss-b

orde

rsu

cces

sex

peri

ence

0.96

3.16

−0.0

8∗0.

26∗

0.00

−0.0

2−0

.01

0.14

∗0.

10∗

0.21

∗0.

11(1

3)C

ross

-bor

der

failu

reex

peri

ence

0.38

1.02

0.00

0.29

∗0.

05−0

.03

−0.0

20.

120.

09∗

0.25

∗0.

21∗

0.68

∗

∗C

orre

latio

nis

sign

ifica

ntat

the

0.05

leve

l(t

wo-

taile

d);

vari

able

num

beri

ngfo

r(1

)–(7

)fo

llow

sTa

ble

A1

abov

e.

Copyright 2012 John Wiley & Sons, Ltd. Strat. Mgmt. J., 33: 938–964 (2012)DOI: 10.1002/smj

964 K. Muehlfeld, P. Rao Sahib, and A. Van Witteloostuijn

Tabl

eA

4.Ta

ble

ofco

rrel

atio

nsfo

rva

riab

les

inM

odel

s7

and

8in

Tabl

e2

Mea

nS.

D.

(1)

(2)

(3)

(4)

(5)

(6)

(7)

(10)

(11)

(12)

(10)

Intr

a-in

dust

rysu

cces

sex

peri

ence

0.63

1.37

0.07

∗0.

25∗

0.07

∗−0

.02

−0.0

1∗0.

02∗

0.02

∗

(11)

Intr

a-in

dust

ryfa

ilure

expe

rien

ce0.

290.

720.

12∗

0.33

∗0.

14∗

−0.0

3−0

.02

0.03

∗0.

05∗

0.43

∗

(12)

Div

ersi

fyin

gsu

cces

sex

peri

ence

1.81

4.07

0.03

∗0.

33∗

−0.0

1−0

.03

−0.0

30.

15∗

0.11

∗0.

21∗

0.22

∗

(13)

Div

ersi

fyin

gfa

ilure

expe

rien

ce0.

531.

29−0

.02∗

0.33

∗0.

040.

04−0

.02

0.09

0.09

0.16

∗0.

22∗

0.63

∗

∗C

orre

latio

nis

sign

ifica

ntat

the

0.05

leve

l(t

wo-

taile

d);

vari

able

num

beri

ngfo

r(1

)-(7

)fo

llow

sTa

ble

A1

abov

e.

Tabl

eA

5.Ta

ble

ofco

rrel

atio

nsfo

rva

riab

les

inM

odel

s9

and

10in

Tabl

e2

Mea

nS.

D.

(1)

(2)

(3)

(4)

(5)

(6)

(7)

(10)

(11)

(12)

(10)

Frie

ndly

succ

ess

expe

rien

ce4.

197.

490.

09∗

0.36

∗0.

09∗∗

−0.0

4∗0.

00−0

.03

−0.0

5∗

(11)

Frie

ndly

failu

reex

peri

ence

1.23

2.47

0.09

∗0.

38∗

0.09

∗−0

.04∗

0.00

−0.0

3∗−0

.04∗

0.70

(12)

Hos

tile

succ

ess

expe

rien

ce0.

541.

310.

11∗

0.32

∗0.

14∗

−0.0

7∗−0

.09∗

−0.0

4∗−0

.05∗

0.63

∗0.

54∗

(13)

Hos

tile

failu

reex

peri

ence

0.51

1.68

0.38

∗0.

38∗

0.09

∗−0

.04

0.00

−0.0

3∗−0

.04∗

0.71

∗0.

66∗

0.54

∗

∗C

orre

latio

nis

sign

ifica

ntat

the

0.05

leve

l(t

wo-

taile

d);

vari

able

num

beri

ngfo

r(1

)–(7

)fo

llow

sTa

ble

A1

abov

e.

Tabl

eA

6.Ta

ble

ofco

rrel

atio

nsfo

rdi

ffer

ent

expe

rien

ceco

ntex

ts(n

ote

that

not

all

expe

rien

ceva

riab

les

appe

arin

the

sam

em

odel

)

(1)

(2)

(3)

(4)

(5)

(6)

(7)

(8)

(9)

(10)

(11)

(1)

Dom

estic

succ

ess

expe

rien

ce(2

)D

omes

ticfa

ilure

expe

rien

ce0.

54∗

(3)

Cro

ss-b

orde

rsu

cces

sex

peri

ence

0.21

∗0.

20(4

)C

ross

-bor

der

failu

reex

peri

ence

0.25

∗0.

27∗

0.68

∗

(5)

Intr

a-in

dust

rysu

cces

sex

peri

ence

0.56

∗0.

30∗

0.35

∗0.

21∗

(6)

Intr

a-in

dust

ryfa

ilure

expe

rien

ce0.

38∗

0.68

∗0.

25∗

0.30

∗0.

43∗

(7)

Div

ersi

fyin

gsu

cces

sex

peri

ence

0.64

∗0.

43∗

0.65

∗0.

47∗

0.27

∗0.

16∗

(8)

Div

ersi

fyin

gfa

ilure

expe

rien

ce0.

32∗

0.61

∗0.

57∗

0.70

∗0.

16∗

0.15

∗0.

27∗

(9)

Frie

ndly

succ

ess

expe

rien

ce0.

54∗

0.44

∗0.

42∗

0.35

∗0.

37∗

0.25

∗0.

56∗

0.43

∗

(10)

Frie

ndly

failu

reex

peri

ence

0.36

∗0.

45∗

0.47

∗0.

47∗

0.31

∗0.

61∗

0.46

∗0.

50∗

0.54

∗

(11)

Hos

tile

succ

ess

expe

rien

ce0.

35∗

0.30

∗0.

34∗

0.30

∗0.

27∗

0.26

∗0.

39∗

0.30

∗0.

63∗

0.54

∗

(12)

Hos

tile

failu

reex

peri

ence

0.36

∗0.

45∗

0.47

∗0.

47∗

0.31

∗0.

48∗

0.46

∗0.

510.

71∗

0.65

∗0.

54∗

∗C

orre

latio

nis

sign

ifica

ntat

the

0.05

leve

l(t

wo-

taile

d).

Copyright 2012 John Wiley & Sons, Ltd. Strat. Mgmt. J., 33: 938–964 (2012)DOI: 10.1002/smj