whose experience matters in the boardroom? the effects of … · 2019. 12. 12. · argote and...

Strategic Management JournalStrat. Mgmt. J. (2013)

Published online EarlyView in Wiley Online Library (wileyonlinelibrary.com) DOI: 10.1002/smj.2100

Received 16 January 2012 ; Final revision received 10 December 2012

WHOSE EXPERIENCE MATTERS IN THEBOARDROOM? THE EFFECTS OF EXPERIENTIALAND VICARIOUS LEARNING ON EMERGING MARKETENTRYANJA TUSCHKE,1* WM. GERARD SANDERS,2 and EXEQUIEL HERNANDEZ31 Munich School of Management, University of Munich, München, Germany2 Jones Graduate School of Business, Rice University, Houston, Texas, U.S.A.3 Olin Business School, Washington University in St. Louis, St. Louis, Missouri,U.S.A.

Using an organizational learning perspective, we develop arguments about vicarious learningthrough board interlocks and its relation to experiential learning. Although it is well establishedthat firms learn from board interlocks, little attention has focused on which types of interlocksare most consequential and why. We distinguish between the relative advantages of various tieattributes such as experience, authority, and credibility and argue that these distinctions leadto measureable differences in learning outcomes. We further demonstrate that whether vicariouslearning substitutes or complements focal firm experiential learning depends upon the type ofinterlock involved. After accounting for the endogeneity of ties, we find support for our frameworkin a longitudinal analysis of foreign investments by German firms in emerging economies between1990 and 2003. Copyright 2013 John Wiley & Sons, Ltd.

INTRODUCTION

Board interlocks are established when an indi-vidual affiliated with one firm serves on theboard of another firm, and function as one ofthe most important sources of learning regard-ing strategic issues. Interlocks have been shownto influence key decisions such as acquisitions(Beckman and Haunschild, 2002), organizationalstructure (Palmer, Jennings, and Zhou, 1993), busi-ness group affiliation (Khanna and Rivkin, 2006),and strategic persistence (Geletkanycz and Ham-brick, 1997). Most studies implicitly assume thatall interlocks matter relatively equally regardless

Keywords: vicarious learning; experiential learning;board interlocks; FDI; emerging markets*Correspondence to: Anja Tuschke, Munich School of Manage-ment, University of Munich, Ludwigstr. 28 RG, 80539 München,Germany. E-mail: [email protected]

Copyright 2013 John Wiley & Sons, Ltd.

of the characteristics of the individual establishingthe tie (Shropshire, 2010). Yet individuals creat-ing board interlocks vary along important dimen-sions likely to affect learning such as the natureof their experience (e.g., first- or second-hand),their credibility as transmitters of knowledge, andthe influence they have over focal firm decisions.As a result, we expect that some ties mattermore than others in helping firms learn vicariouslyabout strategic opportunities and thus have varyingdegrees of impact on the strategic actions of firms.

In addition, we have a limited understandingabout how experiential learning (the firm’s inter-nal knowledge acquisition) relates to the vicar-ious learning that can be obtained via boardinterlocks—which represents a key limitation inthe organizational learning literature more gen-erally (Argote and Miron-Spektor, 2011). Yetthere is good reason to believe that the outcomesfrom vicarious learning, including which types

A. Tuschke, Wm. G. Sanders, and E. Hernandez

of interlocks are most influential, are dependenton how much experience the focal firm has withthe issues on which interlocks provide knowledge(Levitt and March, 1988).

We argue that differences in learning outcomesarise from two important points of distinctionacross individuals forming board interlocks. First,directors differ in the type of experience theypossess regarding specific domains of knowl-edge (Carpenter and Westphal, 2001). Somedirectors have developed knowledge about par-ticular strategies through first-hand action (Kroll,Walters, and Wright, 2008), while others havemore coarse-grained knowledge acquired fromsecond-hand observations of other firms’ strategies(Geletkanycz and Hambrick, 1997; Haunschild,1993). Those with first-hand experience are likelyto be better transmitters of knowledge due to theprimacy of their experience, the vivid caselikedescriptions they can provide, and the credibil-ity they have with other board members. Second,some directors who create board ties are CEOswith ultimate decisional authority in the focal firmor in their home organization, while others are non-CEO top managers lacking this ultimate authority.This difference in hierarchical authority may affectthe credibility of the director and the potential heor she has to influence the focal firm’s adoptionof novel strategies (Kroll et al ., 2008). We buildupon these dimensions to present a theoreticallyderived model of interlock influence that helpsus to study which ties matter most. We furtherdemonstrate that experiential learning substitutesthe vicarious learning from some types of inter-locks but complements the learning from others,depending on the nature of knowledge brought bythe individual forming the interlock. By address-ing this issue, we make an important theoreticalcontribution by demonstrating when vicarious andexperiential sources of learning function as substi-tutes or complements.

Another important contribution comes fromempirically accounting for the potential endogene-ity of interlocks. Extant literature in this domainhas not addressed the possibility that interlocksmay influence firms’ actions because of selectionrather than learning. That is, it could be that cer-tain individuals are simply easier to attract to afocal firm’s board or have more attractive charac-teristics to begin with—making interlocks influ-ential not because of learning but due to differ-ential selection. We find support for our learning

propositions while directly accounting for the pos-sibility of selection in our analysis.

Emerging market entry as the learning context

Our empirical context provides a unique opportu-nity to test our hypotheses. We examine Germanfirms’ foreign direct investments into 21 formerWarsaw Pact countries between 1990 and 2003.After the sudden fall of communism in 1990,an entire new set of markets was made avail-able to foreign investors. However, uncertaintyregarding the viability of such investments wasparticularly high because Western firms werevirtually nonexistent in countries behind the IronCurtain. It was rare to find individuals or entitieswith rich, first-hand experience operating foreignbusinesses in those countries. This setting makesthe need for external information sourcing fromboard interlocks particularly salient, allowingus to assess the differential impact of differentkinds of board ties and their relation to first-handexperience. Moreover, we can observe the gradualbuild-up of firms’ own investment experience inthe region under study without the concerns of leftcensoring and thus credibly assess the interactionof vicarious and experiential learning.

One of the central questions in the FDI literaturepertains to where firms choose to invest and whatinfluences location choice (Dunning, 1998). Exist-ing research emphasizes an experiential learningtheory of foreign expansion, in which firms learnfrom their experiences with prior investments inrelated countries or markets, which reduces theuncertainty that would otherwise preclude or sig-nificantly reduce the likelihood of entry (Deliosand Henisz, 2003; Johanson and Vahlne, 1977).However, when firms possess little or no experi-ence in a particular region, prior investment expe-rience provides little guidance. This seems to bethe case when it comes to emerging or transi-tion countries, which present firms with uniquechallenges—including unusually rapid growth,political turmoil, institutional voids, liberalizationand privatization, and lack of developed markets(Hoskisson et al ., 2000). These challenges makeprior experience in nonemerging markets less use-ful when evaluating opportunities in such highlyrisky locations. Consequently, vicarious learningfrom connections to experienced interlock partnersshould be especially valuable in helping firms inthe process of entering emerging foreign countries.

Copyright 2013 John Wiley & Sons, Ltd. Strat. Mgmt. J. (2013)DOI: 10.1002/smj

Board Ties, Learning, and Emerging Market Entry

THEORY AND HYPOTHESES

Organizational learning is the systematic change inknowledge or behavior deriving from prior expe-rience (Argote, 1999). Argote and Miron-Spektor(2011) argued that the most fundamental dimen-sion of experience is whether it is acquired directlyby the focal organization or indirectly from oth-ers. Learning can arise from first-hand experience,generally referred to as experiential learning (e.g.Levitt and March, 1988). For instance, firms canlearn to perform better at acquisitions by engag-ing in multiple acquisitions, especially of a similarnature (Baum, Li, and Usher, 2000). In addition,organizations learn vicariously by observing thebehaviors and experiences of other organizations(e.g. Kim and Miner, 2007).

However, organizations vary in their abilityto learn from other organizations (Cohen andLevinthal, 1990). Heterogeneity in organizationallearning is a function of firm experience andthe nature of ties to other organizations in theenvironment (Ahuja, 2000). Research suggests thatin some contexts firms seem unable to benefitfrom the knowledge and experiences of other firmsin their environment, which might be caused bytheir own lack of experience (Baum et al ., 2000).Consequently, the interplay of direct experienceand vicarious experience has drawn increasedattention (e.g. Argote and Miron-Spektor, 2011).Experiential learning and vicarious learning maybe substitutes or complements, an issue aboutwhich research is equivocal (Bresman, 2010; Haasand Hansen, 2005). In fact, Argote and Miron-Spektor (2011) suggested that understanding whenthese two sources of learning are complements orsubstitutes is one of the key frontiers in the fieldof organizational learning.

One important context in which organizations’own experiences intersect with external experi-ences is in the boardroom (Westphal, 1999). Thecorporate governance milieu brings together mem-bers of the corporate elite from a variety ofother organizations into a formal setting wherethey interact with senior managers of a focalfirm. These formal ties between organizationscan function as important sources of knowledgetransfer and vicarious learning regarding strate-gic choices (Kalnins, Swaminathan, and Mitchell,2006). The information flowing through boardinterlocks is relatively inexpensive, reliable, andfocuses on high-level strategic issues (Haunschild

Focal Firm

Board of Directors

Top Management

Experienced Firm A

Board of Directors

Top Management

Experienced Firm B Experienced Firm C

Other Party

Indirectties

Outcom

ing

ties

Incomingties

Board of Directors

Top Management

Board of Directors

Top Management

Figure 1. Outgoing, incoming, and indirect interlocks

and Beckman, 1998)—exposing the focal firm tothe actions, processes, and reasoning behind otherfirms’ choices. Such connections are particularlyuseful when firms are facing novel and uncertaindecisions, as demonstrated by a volume of empir-ical work (e.g. Beckman and Haunschild, 2002;Palmer et al ., 1993). We thus take as a startingpoint the proposition that the more interlocks toother firms with experience in emerging marketsthe more likely the focal firm is to learn vicari-ously and enter the same markets as its interlockpartners.

Heterogeneity of interlock influence

Our key point of departure from prior researchis in relaxing the implicit assumption that allboard interlocks provide equal opportunities forlearning. Extant work has recognized differenttypes of board interlocks (e.g. Haunschild, 1993;Palmer et al ., 1993) but has generally treatedthese as equally influential both theoreticallyand empirically. Building on this literature, wedistinguish between three ways in which focalfirm managers and outsiders create board con-nections to companies with relevant knowledgeabout specific emerging markets (as depicted inFigure 1). Outgoing ties are created by managersof the focal firm who serve on the board of othercompanies that have already entered particularemerging markets. Incoming ties are establishedby managers of firms with market entry experiencewho serve on the focal firm’s board. Indirectties are formed by individuals who serve on theboard of the focal firm as well as on the board ofanother experienced firm but are managers neitherof the experienced firm nor of the focal firm. Priorliterature has used various labels for these differ-ent types of interlocks (e.g. ‘sent’ for outgoing,‘received’ for incoming, ‘neutral’ for indirect).



The individuals forming outgoing, incoming,and indirect ties are distinct in two importantways: in the nature of the experience (first- orsecond-hand) they bring to the focal firm, andin whether they serve as CEOs or play otherroles (e.g. top manager) in the focal or anotherfirm. We argue that these two characteristics leadto differences in the relative strength of learningfrom directors forming outgoing, incoming, andindirect ties—which become manifested in thestrategic decision to invest in a foreign country.Thus, our first set of hypotheses focuses on howmuch each type of tie matters for learning. Afterour hypotheses regarding this heterogeneity ofinfluence, we shift our focus to when each type oftie matters most by exploring the moderating effectof the focal firm’s own prior experience. Table 1provides a brief outline of the key argumentsbehind each hypothesis.

Discussions in the boardroom revolve aroundapplying board members’ relevant knowledge in astrategic domain—in our case, entry into formerlycommunist countries—to the strategic needs of thefocal firm. Research on advice seeking from net-work partners suggests that when the focal firmlacks its own expertise in a domain of knowl-edge, the most important criteria in learning fromanother source is the expertise and credibility ofthe informant (Nebus, 2006). The nature of expe-rience possessed by an incoming tie is fundamen-tally different from that possessed by an outgo-ing or indirect tie. An incoming tie is created bya focal firm director with first-hand experiencewith evaluating, directing, and overseeing emerg-ing market investments. Consequently, she bringsher experiences to the focal firm directly, withoutan intermediary. This first-hand experience givessuch a director fine-grained knowledge, whichincreases the quality and credibility of the informa-tion transferred relative to the third-party learningprovided by directors forming indirect ties. Forexample, incoming directors can bring rich storiesand accounts of personal experiences formulatingand implementing an emerging market entry strat-egy. The credibility of these directors’ opinions tothe focal firm should be especially high becausevivid case-type information is of high fidelity andmore influential than abstract recounting of some-one else’s experience (Nisbett and Ross, 1980).

Alternatively, the information or knowledgebrought to the focal firm by individuals formingoutgoing or indirect ties is second-hand in nature.

A manager creating an outgoing tie brings backsecond-hand information obtained while supervis-ing another firm with on-the-ground experience.Similarly, an indirect tie is formed by a directorwho brings second-hand knowledge obtained in asupervisory role as a director of an experiencedcompany. In this capacity, individuals creatingoutgoing or indirect ties are able to learn vicar-iously and absorb relevant information. Neverthe-less, such second-hand knowledge is not as rich asthat of a manager who orchestrates a strategy andoversees its daily operations. As a result, the first-hand information possessed by incoming experi-enced directors is likely to have a greater influenceon the focal firm’s decisions than the second-hand information provided by directors formingoutgoing or indirect ties—especially for such anuncertain and unprecedented decision as enteringan emerging country. This logic leads to the fol-lowing hypothesis:

Hypothesis 1a: Incoming interlocks formed byboard members with first-hand experience enter-ing a particular foreign market have a strongereffect on the focal firm’s entry into the sameforeign market than outgoing or indirect inter-locks formed by board members with second-hand experience obtained supervising anotherfirm’s entry into the same market .

As just reviewed, incoming ties possess first-hand knowledge, which has significant learningadvantages in boardroom discussions of the focalfirm. Alternatively, both outgoing and indirect tiesbring second-hand knowledge to boardroom dis-cussions. While outgoing and indirect ties are onsimilar footing in that regard, the individuals form-ing these ties differ in one important respect thatmay affect their relative influence on the focal firm:whether they are insiders or outsiders. Outgoingdirectors are insiders in the focal firm, thus moreintimately aware of its operations, capabilities, andneeds because they oversee its ongoing operations.As such, they have superior understanding of therelevance to the focal firm of second-hand knowl-edge they acquire in their experiences as direc-tors of other firms (Haunschild, 1993). Individualsforming indirect ties lack this insider perspectivecompared to those forming outgoing ties. As out-siders without direct experience, they may be lessin tune with the potential application of the expe-riences of another firm to the strategic needs of



Table 1. Summary of learning mechanisms for outgoing, incoming, and indirect interlocks

Type of board interlock

Incoming Outgoing Indirect

Type ofvicarious experience

High level of expertise andcredibility due to first-handinformation (H1a,b)

Second-hand informationbut insider familiarity(H1a,b)

Second-hand informationwith outsider perspective(H1a,b)

CEO vs. TMT Outsider CEO has greaterstatus and credibility inboard discussions thanoutsider non-CEOs (H2b)

Focal firm CEO hasgreater decisionalauthority than internalTMT members (H2a)

Not applicable

Focal firm experience External knowledge from outgoing and incoming tiessubstitutes for lack of focal firm knowledge. Oncefocal firm gains experience, external knowledgebecomes redundant, less influential (H3b,c)

External knowledge fromIndirect ties is weak but lesslikely to overlap with directlearning by focal firm,becomes complementary asfirm gains first-handexperience (H3a)

the focal firm. Consequently, the distinguishingfeature of outgoing and indirect ties—which bothlack the first-hand knowledge possessed by incom-ing ties—is that the individual forming outgoingties is an insider in the focal firm. Thus,

Hypothesis 1b. Outgoing interlocks formed byboard members with second-hand experienceobtained supervising another firm’s entry into aparticular foreign market have a stronger effecton the focal firm’s entry into the same foreignmarket than indirect interlocks formed by boardmembers with second-hand experience obtainedsupervising another firm’s entry into the samemarket .

Formal authority: CEOs vs. non-CEOs

The differences in interlock influence establishedso far consider the nature of directors’ experienceregarding emerging markets. However, organiza-tional learning from board ties could also be afunction of the formal authority possessed by man-agers of the focal or of the experienced firm—whoultimately serve as the carriers of vicarious knowl-edge to the focal firm and make the case forimplementing a given set of actions. Those form-ing indirect ties do not possess formal authority ateither firm, so the arguments in this section applyonly to outgoing and incoming interlocks.

While all outgoing and incoming interlocks areestablished by directors who serve as top managersat either the focal or partner firm, they differ intheir potential to utilize that knowledge in ways toshape the focal firm’s strategy. Some executivesare likely to have significantly greater discretionto put their knowledge to use and to shape thefocal firm’s actions proactively (Hambrick andFinkelstein, 1987). While multiple forces affectexecutive discretion, we focus on the formalauthority vested in CEOs relative to non-CEOexecutives. CEOs are more likely to be persuasivewhen bringing their knowledge to bear on thefirm’s existing strategy and to affect the courseof the firm in ways that other executives cannot(Kroll et al ., 2008). The mechanisms are slightlydifferent for CEOs of the focal firm creatingoutgoing interlocks and CEOs of other firmscreating incoming ties. We discuss each in turn.

Managerial characteristics vary between mem-bers of the top management team who serve exter-nally on other firms’ boards. While these managersare well acquainted with the strategy and decisionmaking process of the focal firm, formal authoritygives some executives the ability to establish poli-cies, procedures, and long-term objectives directly.Within a top management team, a CEO clearly hasformal authority and power to shape consequen-tial decisions such as entry into emerging markets.Non-CEO managers creating outgoing ties havesimilar opportunities as the CEO to learn from



other firms, but they have a disadvantage withrespect to authority and discretion to use theirlearning to redirect the focal firm’s strategy. Con-sequently, we expect that CEOs creating outgoinginterlocks will be in the best position to utilizethe knowledge gained about other firms’ emergingmarket entry decisions and integrate such knowl-edge into the focal firm’s strategic planning in thefuture, suggesting

H2a: Outgoing interlocks formed by CEOs whosupervise other firms with experience enteringa particular emerging market have a strongereffect on the focal firm’s entry into the sameforeign market than outgoing interlocks formedby non-CEOs .

Focal firm managers seek for help on strategicmatters from other firms’ top managers. Accessto executives of other firms is valuable in partbecause they can provide an outsider’s moreindependent and objective perspective on issuesfacing the firm (Geletkanycz and Hambrick,1997). While research identifies external directorsas one key source of outside advice, directorexperience is often treated without regard for theidentification of the director who possesses thevaluable experience. However, the acceptance ofadvice is largely based on who is providing it(Berlo et al ., 1969). Advice givers, even advisorson boards of directors, are not uniformly viewedas credible on all issues.

Source credibility is the most potent means ofpersuasion and is a function of expertise, veracity,and benevolence (McCroskey and Young, 1981).In as much as firms are unlikely to appoint boardmembers who are untruthful or malevolent, wefocus our discussion on expertise. Credibility isthe assessment of believability of an informationsource and whether that source is a reliable guideto belief and behavior (O’Keefe, 2002). Thecredibility of a board member’s expertise withemerging markets may vary based on the degreeto which the director is perceived as the architectof the market entry strategy. Because CEOs areultimately responsible for the formulation, imple-mentation, and performance of their company’sstrategies, they are likely to be perceived as havinga higher level of expertise than other incomingboard members who are not CEOs. This logicsuggests that incoming ties established by CEOs

of other firms are likely to be more influential thanincoming ties established by non-CEOs. Thus:

H2b: Incoming interlocks formed by CEOs offirms with experience entering a particularemerging market have a stronger effect on thefocal firm’s entry into the same foreign marketthan incoming interlocks formed by non-CEOs .

When each type of tie matters most: the roleof focal firm experience

Having argued for varying degrees of interlockinfluence, we now turn to the important issueof how experiential learning may modify whichsources of vicarious learning are most relevantand thus change the impact of different interlocks.Haunschild and Beckman (1998) reported thatalternative external sources of information—forexample, industry associations—are substitutes forthe vicarious learning gained from board inter-locks. We build upon this finding but focus insteadon a more direct and accessible source of knowl-edge residing inside the firm: its own experientiallearning. As firms learn from their own experi-ences with previously unexplored emerging mar-kets, their informational needs are likely to change.This shift in strategic needs should affect the wayin which different types of interlocks influence afirm’s decisions. We argue that prior experienceenhances the impact of some interlocks on firms’actions while diminishing the importance of others.

What remains unclear is whether experientialand vicarious information are substitutes or com-plements. On the one hand, board interlocks maybe most valuable for strategic decision makingwhen they transfer information not available else-where (Haunschild and Beckman, 1998). Accord-ing to this logic, interlocks become less criticaland influential if equivalent information is avail-able inside the firm due to knowledge redundancy,leading firm experience to substitute for learningfrom interlocks. On the other hand, the conceptof absorptive capacity suggests that internal andexternal sources of knowledge may be comple-mentary (Cohen and Levinthal, 1990). A firm withsome experience in a particular domain gains atleast a rudimentary level of expertise, putting man-agers in a better position to judge the relevanceof an interlock partner’s experience. The key tosolving this apparent paradox may lie in consider-ing who creates the interlock.



Related experience as a complement to inter-locks . Research has shown that, compared to moredirect relationships, firms connected through indi-rect ties obtain relatively weaker, less specific,but nonredundant information from each otherabout any single issue (Hansen, 1999; Lin, 2001).This observation resonates with existing fieldworkregarding the value of indirect interlocks, whichsuggests that they serve as conduits of broaderknowledge about ‘what’s out there’ regarding thelatest actions being taken by firms in the market(Useem, 1984). Information obtained from indirectties seems thus most valuable when combined withthe firm’s own experience.

Because of its nonredundancy and second-handnature, knowledge provided by indirect ties mostlikely comes to the firm more obliquely thanthat stemming from direct requests for informa-tion regarding specific issues on which the firmseeks advice from experienced individuals. Con-sequently, knowledge from indirect ties becomespotentially complementary to a firm’s internalexperiences. Importantly, however, a focal firmlacking absorptive capacity developed throughfirst-hand experience is not likely to benefitmuch from this broad but less-specific informationbecause it is harder to relate to existing knowl-edge and to apply to the firm’s current concerns.However, once the firm begins gaining its ownexperience in the general domain on which thosecreating indirect ties are advising it, its abilityto translate that novel information into actionablestrategies increases. This suggests that indirect tieswill complement firm experience.

Hypothesis 3a. As the focal firm gains experi-ence with similar emerging markets, the impactof directors forming indirect ties to experiencedpartners on emerging market entry becomesstronger .

Related experience as a substitute to interlocks .In contrast, scholars have found that direct ties,such as outgoing and incoming interlocks, aremore conducive to the ‘thick’ transfer of tacitknowledge relative to indirect ties (Hansen, 1999).Specifically, the information is rich, first-hand, andspecific because of the directness of the connectionbetween informant and receiver. Because theinformation from incoming and outgoing ties ismore self-contained and comes through a richer

medium, it is more likely to be applicable in casesin which the focal firm lacks direct experience(Daft and Lengel, 1986). Such pointed informationwill be especially valuable for firms lacking first-hand experience.

However, as the focal firm gains related first-hand experience, this specific information becomessomewhat redundant and internal information mayoften be sufficient to meet the needs of thefirm’s local search when engaging in additionalforeign entries. Consequently, an inexperiencedfirm may be influenced more by vicarious learningopportunities provided by incoming and outgoingties than an experienced firm. This is consistentwith observations of scholars who report that afterthe transfer of highly specific knowledge, directties are less appropriate for firms seeking moregeneral and nonredundant information (Hansen,1999; Lin, 2001). This logic suggests that outgoingand incoming ties function as substitutes for focalfirm experience. Thus, we posit:

Hypothesis 3b. As the focal firm gains experi-ence with similar emerging markets, the impactof directors forming incoming ties to expe-rienced partners on emerging market entrybecomes weaker .

Hypothesis 3c. As the focal firm gains experi-ence with similar emerging markets, the impactof directors forming outgoing ties to expe-rienced partners on emerging market entrybecomes weaker .

DATA AND METHODS

We study the effect of domestic board interlocks onGerman firms’ foreign direct investments into 21former Warsaw Pact countries between 1990 and2003. We first describe in more detail the contextof German corporate governance as it pertains toboard structure and ownership. We then detail theparticulars of the data, analysis, and findings.

Institutional setting

Compared to the U.S., the institutional contextin Germany shows some important similaritiesand differences. Of relevance for our analysisare the two-tier board system with workerco-determination, the prevalence of cross-shareholdings among firms, and the prominent



role of large domestic banks and insurancecompanies. We describe each of these in turn.

The German corporate governance code man-dates a two-tier board system consisting of amanagement board and a supervisory board. Akinto the top management team in U.S. firms, themanagement board is responsible for strategic andoperational decisions. It reports to the supervisoryboard, which is similar to the board of directorsin the U.S. According to the German CorporateGovernance Code (2002: 1), ‘the supervisoryboard appoints, supervises and advises the mem-bers of the management board and is directlyinvolved in decisions of fundamental importanceto the enterprise’.

In line with this mandate, members of bothboards are expected to work together, exchangeinformation, and coordinate the firm’s strategy.The German Corporate Governance Code alsostates that the chairman of the supervisory board isespecially supposed to keep close contact with themanagement board and to consult with the chair-man of the management board on ‘strategy, busi-ness development, and risk management’ (2002:9). An important difference between the one-tierboard structure of Anglo-American firms and thetwo-tier board structure in Germany is that mem-bers of the management board, including the CEO,are not allowed to serve on their own firm’s super-visory board.

To ensure that internationalization strategies areindeed discussed in the boardroom, we conductedinterviews with several directors of German firms.These interviews revealed that entries into thenewly developing markets in the former WarsawPact area were strategic issues of importance. Forinstance, the chairman of the supervisory boardof one firm stated:

‘Overall, firms’ regional strategies are oftendiscussed in the boardroom. Regional strate-gies are very important, they are tangible,and board members can provide input. Often,board members have experience with runningoperations in specific regions—and they canprovide helpful input on opportunities andrisks. Product strategies are discussed lessoften. [ . . . ] It is much easier to provide inputon regional strategies.’

A peculiarity of the German two-tier boardstructure is mandatory worker co-determination.

Rooted in the Co-determination Law(Mitbestimmungsgesetz ) and tracing back tothe Cooperative Management Law (Montan-mitbestimmungsgesetz ) of 1951, the supervisoryboard consists not only of shareholder representa-tives but also of employee representatives. Thesemay be employees or trade union representatives.The number of seats assigned to shareholder andemployee representatives depends on the size ofthe firm. In large stock corporations such as theones in our sample, employee representativesare ensured half of the seats in the supervisoryboard. However, the chairman of the supervisoryboard is always a shareholder representative andis allotted two votes in case of a tie in the voting.

A widely discussed characteristic of Germancorporate governance is the prevalence of cross-shareholdings between firms compared to coun-tries with more dispersed ownership such asthe U.S. or the U.K. La Porta, Lopez deSilanes, and Shleifer (1999) describe the case ofAllianz—a large insurance company—that hadcross-shareholdings with most of its large corpo-rate shareholders. Large sample studies have foundthat it is not uncommon for German companiesto own substantial minority stakes in other Ger-man firms, with majority stakes also being preva-lent, though not nearly as common (Franks andMayer, 2001; Windolf and Beyer, 1996). Some-times referred to as ‘Germany Incorporated,’ thissystem of cross-shareholdings arose partly to offerprotection against hostile takeovers by relyingmore on internal mechanisms of control ratherthan on external capital market interventions.With increasingly global capital markets, cross-shareholdings started to diminish in the mid-1990s.

The largest corporate owners in Germany havetraditionally been large domestic banks and insur-ance companies (e.g. La Porta et al ., 1999).These companies were among the most importantfinanciers of industrialization in the late 19th cen-tury and helped to rebuild the economy after WorldWar II. Besides providing equity, the relationshipbetween banks and other stock corporations wasstrengthened through long-term credit agreementsthat are typical of the German governance sys-tem (Shleifer and Vishny, 1997). Reflecting thisstrong influence in financial matters, managers ofthese influential financial companies often servedas directors on other firms’ supervisory boards.

In sum, the institutional characteristics of theGerman governance system present important



similarities and differences with the U.S.setting—which has been the context of mostprior work studying learning via interlocks. Fromour review of prior literature and our interviewsof German directors, we conclude that learningabout foreign market opportunities seems to occurin this setting. At the same time, the foregoingdescription suggests a need to account empiricallyfor the influence of several unique features ofGerman corporate governance. We explain ourefforts to do so next.

Data and variables

The sampling frame was comprised of firms listedin the index of the 100 largest stock corporationsin Germany (the DAX 100). This index repre-sents a broad spectrum of the German economy,with the largest firm equivalent to a company inthe top five of the Fortune 500, and the smallestfirm falling in the range of mid-caps in the U.S.,allowing us to observe a great deal of variance infirm size within the sample. Similar samples of thelargest German companies have been used in priororganizational studies (e.g. Sanders and Tuschke,2007). We removed seven firms that were sub-sidiaries of other organizations. For the remainingfirms, we obtained panel data on their entries into21 former Warsaw Pact countries for the 14-yearperiod between 1990 and 2003. The target coun-tries in our sample include Armenia, Azerbaijan,Belarus, Bulgaria, Czech Republic, Estonia, Geor-gia, Hungary, Kazakhstan, Kyrgyzstan, Latvia,Lithuania, Moldova, Romania, Poland, Russia,Slovakia, Tadzhikistan, Turkmenistan, Ukraine,and Uzbekistan. The period of observation helpedus avoid left censoring, as direct investments ofWestern firms in countries behind the Iron Curtainwere virtually nonexistent before 1990.

Information on foreign investments in formerWarsaw Pact countries is difficult to obtain forGerman firms. While some firms provide a com-plete list of foreign properties in their annualreports, large firms tend to restrict their disclosuresto investments that exceed a certain percentage ofequity. Thus, to obtain complete information wefollowed three procedures. First, we referred to the‘list of share properties’ in the firms’ annual report.Second, we contacted each firm’s IR departmentand asked them to provide detailed information onentries into the countries under study or to send therespective ‘list of share properties’ for the period

1989–2003. If we could not obtain an answer froma firm’s IR department, we searched LexisNexisfor press reports that contained the name of thecompany combined with terms related to foreignentry. Furthermore, we referred to information onthe firm’s foreign investment in the Handbook ofGerman Listed Companies . We then contacted thefirms again and asked them to confirm or cor-rect our information. Third, we contacted the reg-istration offices of the respective district courtsand looked at the firms’ original filings to cor-rect inconsistencies and reduce missing data. Ourexhaustive effort resulted in complete data for 82firms. There were several reasons that data wereunavailable for the remaining DAX 100 firms,but approximately 80 percent of the missing casesoccurred because the firm was delisted during thestudy’s time frame (e.g., acquired by another firm),making it impossible to obtain reliable records ofinvestment in the countries under study. In a fewother cases, firms became newly listed in the DAXtoward the end of our time frame and did not havecomplete historical records of market entry.

Ownership data were collected from the annualHoppenstedt directories, the most authoritativelisting of equity ownership in Germany. Dataon the economic, political, and demographiccharacteristics of the 21 countries were collectedfrom statistics published by the United Nations, theInternational Labor Organization, and the WorldBank. Data on the business climate in the countriesunder study were obtained from BERI S.A. Firm-level variables were collected from annual reports.

Variables

Dependent variable

Our dependent variable, entry jit, is an indicatorcoded as 1 if firm j entered host country i in yeart . Following the definition of direct investmentof the Organization for Economic Co-operationand Development (OECD) and of the InternationalMonetary Fund (IMF), we ascertained marketentry if a firm established a wholly owned facilityor acquired at least 10 percent of the ordinaryshares of a host country firm. The same measurehas been used in other studies on internationalmarket entry decisions (e.g. Gimeno et al ., 2005).

Independent variables

All independent and control variables are lagged(year t − 1 ) with respect to the dependent variable



(year t). Our theory focuses on how ties to experi-enced partners affect firms’ market entry choices.Accordingly, we measure interlocks by countingties to other firms that have previously entered for-mer Warsaw Pact countries. Importantly, we mea-sure interlock experience on a market-by-marketbasis. Thus, the number of ties depends on part-ners’ prior market entries into specific host coun-tries because a partner may, for example, haveexperience in Poland but not in Bulgaria. Weupdate the interlock variables annually for eachfocal firm-target market combination.

The variable aggregated ties captures the countof all ties to experienced partners, regardlessof the type of tie. This variable is used in abaseline model for the purpose of establishing ageneral effect for interlocks. In models testingour hypotheses, we disaggregated board networkties into counts of outgoing ties created by thefocal firm’s managers who serve on the board ofother experienced firms, incoming ties formed bymanagers of experienced firms who serve on thefocal firm’s board, and indirect ties representingfocal firm’s board members who are directorsof another firm with experience in a particularemerging market but play no managerial role ineither firm. To test H2a and H2b, we furtherdistinguished between ties that are created bya CEO (outgoing CEO and incoming CEO)and by other members of the top managementteam of the focal firm (outgoing non-CEO) oranother experienced firm (incoming non-CEO),respectively. We again note that indirect ties arenot distinguished as CEO or non-CEO because theindividuals creating them do not have a managerialresponsibility in the focal or the experienced firm.

To capture a firm’s related experience, wemeasured each firm’s general regional experiencewith running operations in former Warsaw Pactcountries as the number of years since a firm’sfirst market entry in any of the 21 countries (e.g.Henisz and Delios, 2001). In robustness checks, weused an alternative measure capturing the numberof countries a firm had entered as of year t−1 , andour results remained consistent.

Control variables

We included control variables at the country,industry, year, and firm levels. Empirical studieshave revealed a positive correlation between themarket size of a host country and a firm’s

propensity for market entry (e.g. Coughlin, Terza,and Arromdee, 1991). Firms prefer to invest incountries with larger market size in order tocompensate for the risks and resource requirementsassociated with foreign market entry. To capturethe size of the host country’s market we usedGDP . Growth in GDP per capita serves as proxyfor market growth and customer purchasing power(Ford and Strange, 1999). These two controlsreflect the attractiveness of a host country formarket seeking foreign direct investment. Theavailability of labor is also seen as influencingthe attractiveness of a specific host country,especially for factor seeking FDI. In countries witha comparatively high rate of unemployment, theworkforce is expected to have a higher level ofjob appreciation and thus a willingness to acceptlower wages and longer work hours (Billington,1999). To capture this notion, we included theyearly rate of unemployment for each country.We also included a control for the geographicdistance to each emerging market, measured asthe distance between the firm’s headquarters inGermany and the capital of the host country (logof kilometers). As one moves deeper into EasternEurope, differences in the languages, cultures,and historical relationships between Germany andthe former Warsaw Pact countries increase, anddistance imposes higher logistical costs, whichshould make entry less likely.

To account for the risk and uncertainty asso-ciated with a firm’s entry in an emerging market,we included information on the business climate ineach target country. We used an annually updatedindex of market risk that measures the favorabil-ity of the operating climate for foreign businessin each country. We obtained this measure fromBERI S.A., which surveys a panel of 105 expertswho evaluate the overall quality of the businessclimate (including bureaucratic barriers, stabilityof government policies, and the degree to whichnational firms are given preferential treatment). Asprovided by BERI, market risk scores range from 0to 100, with risk decreasing as scores increase. Forease of interpretation, we reverse scored this mea-sure by subtracting the raw value from 100 so thathigher values correspond to higher levels of risk.

At the firm level, we controlled for firm sizeby including the log of employees . Firm sizeis related to the ability to enter foreign marketsbecause larger firms tend to have greater financialand social resources. Additionally, we controlled



for profitability by including the firm’s ROA.Profitable firms are deemed to be more capableof absorbing the costs and risks involved withentering a foreign emerging market. In addition,firms with strong general international orientationmay be more likely to enter newly emergingmarkets in former Warsaw Pact countries, whichwe captured by controlling for its foreign salesratio (Sullivan, 1994).

Informal interactions and trust created by boardinterlocks may also have an important impact onfirms’ actions (McDonald, Khanna, and Westphal,2008). One way to capture mutual trust is toaccount for reciprocated ties . We do this bysumming the number of cases in which the focalfirm simultaneously had an outgoing and incomingtie with another firm. Such multiplexity of con-nections should increase the chances of informalencounters that may produce opportunities tolearn and transfer tacit knowledge in settings notbound by some of the legal restrictions of formalmeetings and interactions. We also controlledfor learning from interlock partners’ experi-ence entering other countries in the region. Forinstance, a firm contemplating entering the CzechRepublic might be influenced by an interlockpartner’s experience in the surrounding region.We include three control variables to account forthis. Regional outgoing ties, regional incomingties , and regional indirect ties were measured asthe number of board ties to companies with priorexperience entering any country in the regionexcept the host country in question.

Our hypotheses focused on learning throughboard interlocks. Given the unique institutionalattributes of the German governance system, analternative hypothesis might be that interlockssimply are reflective of coordination and controlamong firms bound by ownership links or domi-nated by influential stakeholders. Thus, we includeseveral controls to account for the influence ofparticularly powerful actors. To get at the strongrole played by financial organizations in Germany,we control for two types of ties. Ties to influentialfinancial firms was measured as the numberof outgoing board ties to banks and insurancecompanies. Ties from influential financial firmswas measured as the count of incoming boardlinkages from such firms. We directly address theissue of cross-holdings by including a measureof equity owned by experienced firms , capturedas the total percentage of the focal firm’s equity

owned directly by other firms in the sample thathave experience in a particular emerging market.Finally, Windolf and Beyer (1996) suggestedthat intra-industry board ties can be mechanismsof coordination in Germany. Consequently, wealso included two controls for ties within broadindustry sectors: consumer, machinery, automo-tive, chemical and pharmaceutical, building andconstruction, banking, health care, steel, energy,and insurance. Within sector outgoing ties wasmeasured as the count of direct board ties to otherfirms within the same industry sector. Likewise,within sector incoming ties was measured as thenumber of direct board ties from other firmswithin the same industry sector.

To control for unobserved industry effects, weincluded a dummy variable for nine of the tenbroad industry categories just described. Thesecategories came from an adapted version of theclassification of the Deutsche Boerse Group, anequivalent of the SEC in the U.S. To account forunobserved, idiosyncratic factors making entrymore or less common in some years, we alsoincluded year dummy variables. For clarity ofpresentation, the effects of the industry and yeardummies are not reported in our tables, but theyare included in all models.

Analysis

We used a discrete-time logit specification ofevent history with each spell corresponding to ayear and obtained the results through maximumlikelihood estimation (Allison, 1984). This methodallows us to estimate the propensity of foreignentry for the same organization at multiple inter-vals and accounts for right-censored observationsfor firms that never engaged in foreign entryduring the period of observation. Because eachspell corresponds to a year, we have a total of 14spells between 1990 and 2003. The model has thefollowing form:

logPjit

1 − Pjit = at + b1X1ji + b2X2ji,t−1

where logPjit

1−Pjit represents the logarithmic oddsthat firm j will enter foreign country i at any pointduring time t ; a represents the baseline hazardrate of entry occurring at time t and allows thehazard rate to be different in each of the 14 years



under study when all other time-variant and time-invariant variables are held constant; b1 representsthe change in the log-odds for each one-unitincrease in a time-invariant covariate X 1ji ; and b2represents the change in the log-odds for each one-unit increase in a time-varying covariate X2ji,t-1 .

Each year (beginning in 1990) represents a spellin which firms may potentially engage in foreigninvestment. If a firm did not enter a particularmarket throughout the observation period, thespell was right censored by the end of 2003.Spells were updated at the end of each year toaccommodate the annual time-varying covariates.Once a firm entered a specific country in anygiven year, the next year’s risk set was diminishedby the firm-country spells for which a marketentry had already occurred. This yielded a total of19,882 firm-country-years spells (had no entriesoccurred by any firms in any market in any year,the total firm-country-year spells would have been24,108). We accounted for the correlation betweeninvestment decisions by the same firm in the samecountry across different years by using a robustvariance estimator clustered by firm-country com-binations. In robustness tests, described later, weused several alternative event history estimatorsand our results remained unchanged.

RESULTS

Table 2 presents descriptive statistics and correla-tions. The mean of entry (0.02 or 2%) reportedin the table was calculated by dividing the 404investments in the sample by the 19,882 firm-country-year spells created for the event historyanalysis. Thus, investment in any given countryin any given year was relatively infrequent, but404 investments throughout the period is appre-ciable given the number of firms in the sampleand the risk associated with transition economies.The average values for the interlock measures werealso small (e.g. 0.79 for aggregated ties). Suchsmall averages are driven partly by the rarity ofsuch experience and also by the fact that for manycountry-year combinations there were no entriesby any firm. We note that the interlock variablesshow significant variance across firms and coun-tries, which is partly masked by the crude summarystatistics shown in Table 2.

Table 3 contains the results of our main anal-yses. The coefficients represent the effect of each

variable on the log-odds of foreign entry. Model 1reports the findings for only the control variables.In Model 2 we included the measure of aggregatedties . In line with prior research on board interlocks,we found that interlocks to experienced firmsincreased the likelihood that the focal firm wouldenter that same market (p < 0.001). Hypothesis1a suggested that incoming interlocks formed byboard members with first-hand experience wouldhave stronger effects than outgoing or indirectinterlocks, while Hypothesis 1b suggested that out-going interlocks would have stronger effects thanindirect interlocks. In Model 3 we disaggregatethe interlocks into outgoing, incoming, and indi-rect. Incoming ties had the strongest influence onthe focal firm, followed by outgoing ties, andindirect ties. To verify that the effects were sig-nificantly different in magnitude, we conducted aWald test based on the marginal effects for incom-ing, outgoing, and indirect ties. The three-way testrejects the null hypothesis that the magnitudes ofall three types of ties were equal (p < 0.01). Two-way tests revealed that incoming ties had a signifi-cantly stronger effect than outgoing ties (p < 0.05)and indirect ties (p < 0.001), and the differencebetween outgoing ties and indirect ties was alsomarginally significant (p < 0.1). Thus, the findingsare consistent with H1a and H1b.

Hypotheses 2a and 2b predicted that ties createdby CEOs would have stronger effects than tiescreated by managers who were not CEOs. Asrevealed in Model 4, we found support for bothhypotheses. Whereas outgoing CEO ties (H2a)had a positive influence on the focal firm’smarket entry decision (p < 0.01), the coefficientfor outgoing non-CEO ties was not significant. Inaddition, the effect of incoming CEO ties (H2b)was greater than the effect of incoming non-CEOties per a comparison of their marginal effects(p < 0.05).

Hypotheses 3a–3c predicted that experientiallearning would moderate the effect of vicari-ous learning in different ways depending on thenature of the tie. Hypothesis 3a argued that focalfirm experiential learning would complement indi-rect interlocks, while H3b and H3c conjecturedthat experiential learning would be a substitutefor incoming and outgoing ties. We tested thesehypotheses by assessing the interaction of out-going, incoming, and indirect ties with the focalfirm’s prior experience in Models 5 and 6. Schol-ars have recently shown that in nonlinear models,



Tabl

e2.

Des

crip

tive

stat

istic

san

dco

rrel

atio

ns

Var

iabl

eM

ean

SDM

inM

ax1

23

45

67

89

1011

1E

ntry

0.02

0.14

01

2A

ggre

gate

dtie

s0.

792.

180

230.

193

Inco

min

gtie

s0.

180.

570

70.

150.

734

Out

goin

gtie

s0.

090.

570

170.

120.

510.

245

Indi

rect

ties

0.57

1.62

018

0.17

0.94

0.56

0.43

6O

utgo

ing

CE

Otie

s0.

050.

300

70.

120.

460.

240.

760.

417

Out

goin

gno

n-C

EO

ties

0.04

0.39

014

0.08

0.40

0.17

0.88

0.32

0.36

8In

com

ing

CE

Otie

s0.

070.

310

60.

130.

570.

690.

220.

440.

270.

129

Inco

min

gno

n-C

EO

ties

0.11

0.42

06

0.11

0.58

0.85

0.17

0.44

0.14

0.14

0.21

10R

elat

edex

peri

ence

3.70

3.83

014

0.00

0.11

0.11

0.07

0.11

0.09

0.03

0.06

0.11

11G

DP

(in

mill

ions

)1.

671.

440.

19.

00.

150.

390.

300.

160.

360.

150.

130.

220.

25−0

.01

12G

row

thin

GD

P−1

.33

10.1

0−4

4.9

17.8

0.01

0.09

0.08

0.04

0.09

0.04

0.02

0.06

0.06

0.34

0.18

13U

nem

ploy

men

t5.

244.

820.

0820

0.09

0.30

0.24

0.13

0.27

0.11

0.10

0.17

0.21

0.19

0.31

14M

arke

tri

sk62

.72

4.43

5174

−0.1

1−0

.28

−0.2

2−0

.11

−0.2

5−0

.11

−0.0

8−0

.16

−0.1

8−0

.26

−0.6

615

Em

ploy

ees

(log

)9.

541.

484.

813

.10.

050.

220.

140.

160.

240.

160.

110.

140.

090.

17−0

.06

16G

eogr

aphi

cdi

stan

ce(l

og)

24.3

714

.63

1.8

51.1

−0.1

3−0

.29

−0.2

3−0

.12

−0.2

7−0

.12

−0.0

9−0

.17

−0.1

90.

11−0

.64

17R

OA

0.04

0.05

−0.3

0.4

0.00

−0.0

6−0

.06

−0.0

6−0

.06

−0.0

6−0

.04

−0.0

6−0

.04

−0.0

10.

0018

Fore

ign

sale

sra

tio0.

460.

260

0.9

0.02

0.05

0.06

−0.0

10.

050.

02−0

.02

0.05

0.04

0.38

−0.0

319

Rec

ipro

cate

dtie

s0.

040.

210

20.

010.

040.

010.

100.

040.

150.

040.

04−0

.01

0.04

0.00

20R

egio

nal

inco

min

g1.

523.

030

330.

000.

190.

300.

030.

150.

030.

020.

190.

270.

12−0

.02

21R

egio

nal

outg

oing

0.75

2.91

045

0.01

0.13

0.04

0.32

0.12

0.28

0.26

0.05

0.02

0.07

−0.0

322

Reg

iona

lin

dire

ct5.

5010

.17

010

30.

010.

270.

170.

100.

290.

110.

070.

160.

110.

10−0

.02

23W

ithin

sect

orin

com

ing

0.10

0.31

02

0.01

0.01

0.05

−0.0

20.

01−0

.02

−0.0

10.

090.

01−0

.08

0.00

24W

ithin

sect

orou

tgoi

ng0.

090.

350

30.

010.

020.

000.

050.

020.

050.

030.

02−0

.01

0.02

−0.0

325

Tie

sfr

omfin

anci

alfir

ms

0.79

0.94

04

0.04

0.19

0.20

0.04

0.17

0.05

0.02

0.15

0.16

−0.0

3−0

.05

26T

ies

tofin

anci

alfir

ms

0.25

0.65

05

0.02

0.09

0.01

0.19

0.10

0.23

0.10

0.07

−0.0

30.

09−0

.05

27E

quity

owne

dby

expe

rien

ced

firm

s0.

010.

060

10.

050.

280.

320.

060.

230.

060.

050.

190.

310.

040.

17

Var

iabl

e12

1314

1516

1718

1920

2122

2324

2526

13U

nem

ploy

men

t0.

1414

Mar

ket

risk

−0.3

6−0

.24

15E

mpl

oyee

s(l

og)

0.00

−0.0

30.

0216

Geo

grap

hic

dist

ance

(log

)0.

02−0

.29

0.29

0.05

17R

OA

0.04

0.02

−0.0

3−0

.17

0.01

18Fo

reig

nsa

les

ratio

0.11

0.04

−0.0

70.

270.

040.

1019

Rec

ipro

cate

dtie

s0.

03−0

.02

0.00

0.17

0.01

−0.0

30.

0020

Reg

iona

lin

com

ing

0.04

0.06

−0.0

60.

210.

02−0

.11

0.07

0.02

21R

egio

nal

outg

oing

−0.0

10.

02−0

.01

0.25

0.02

−0.1

0−0

.02

0.15

0.18

22R

egio

nal

indi

rect

0.02

0.06

−0.0

60.

340.

02−0

.10

0.06

0.06

0.64

0.39

23W

ithin

sect

orin

com

ing

−0.0

3−0

.02

0.03

0.01

−0.0

1−0

.01

0.05

−0.0

20.

07−0

.02

0.01

24W

ithin

sect

orou

tgoi

ng−0

.03

−0.0

40.

040.

230.

02−0

.02

0.09

0.07

−0.0

10.

070.

01−0

.06

25T

ies

from

finan

cial

firm

s−0

.07

−0.0

60.

070.

480.

01−0

.09

0.14

0.11

0.31

0.06

0.30

0.03

0.19

26T

ies

tofin

anci

alfir

ms

−0.0

1−0

.05

0.04

0.48

0.03

−0.0

80.

050.

370.

000.

290.

12−0

.04

0.18

0.27

27E

quity

owne

dby

expe

rien

ced

firm

s0.

040.

13−0

.10

0.00

−0.1

2−0

.04

−0.0

4−0

.01

0.06

−0.0

10.

030.

03−0

.02

0.06

−0.0

2

N=

19,8

82;

corr

elat

ions

grea

ter

than

—0.

02—

are

sign

ifica

ntat

p<

0.05

.



Table 3. Discrete time event history analysis

Model 1Controls

only

Model 2Aggregated

ties

Model 3Disaggregated

ties

Model 4Hierarchical

position

Model 5Experience

low

Model 6Experience

high

Model 7Experience

low

Model 8Experience

high

Constant −3.060† −2.831† −2.955† −3.114† −2.876 −7.054** −2.924 −7.123**(1.71) (1.71) (1.71) (1.72) (2.78) (2.71) (2.79) (2.72)

GDP (in millions) 0.122† 0.056 0.067 0.072 0.317* 0.063 0.316* 0.068(0.07) (0.07) (0.07) (0.07) (0.13) (0.09) (0.13) (0.09)

Growth in GDP 0.020* 0.020* 0.020* 0.020* 0.014 0.022 0.013 0.022(0.01) (0.01) (0.01) (0.01) (0.01) (0.02) (0.01) (0.02)

Unemployment 0.072*** 0.058*** 0.060*** 0.060*** 0.072*** 0.061*** 0.072*** 0.061***

(0.01) (0.01) (0.01) (0.01) (0.02) (0.02) (0.02) (0.02)Geographic distance (log) −0.076*** −0.074*** −0.073*** −0.073*** −0.064*** −0.070*** −0.064*** −0.068***

(0.01) (0.01) (0.01) (0.01) (0.01) (0.02) (0.01) (0.02)Market risk −0.089*** −0.077*** −0.077*** −0.077*** −0.084** −0.014 −0.084** −0.014

(0.02) (0.02) (0.02) (0.02) (0.03) (0.03) (0.03) (0.03)Employees (log) 0.300*** 0.232*** 0.244*** 0.245*** 0.198** 0.261* 0.203** 0.255*

(0.05) (0.05) (0.05) (0.05) (0.07) (0.11) (0.07) (0.11)ROA 1.880† 1.901† 2.054† 2.089† 1.148 2.756† 1.219 2.728†

(1.09) (1.12) (1.11) (1.12) (1.69) (1.49) (1.69) (1.48)Foreign sales ratio 1.002** 0.958** 0.922** 0.899** 0.870* −0.003 0.865† −0.022

(0.32) (0.32) (0.32) (0.32) (0.44) (0.51) (0.44) (0.51)Reciprocated ties 0.090 0.121 0.099 0.021 −0.559 1.015** −0.611 0.976**

(0.27) (0.28) (0.28) (0.30) (0.44) (0.35) (0.46) (0.37)Regional incoming ties −0.026 −0.020 −0.060* −0.053† −0.090† −0.054 −0.094† −0.045

(0.03) (0.02) (0.03) (0.03) (0.05) (0.04) (0.05) (0.04)Regional outgoing ties −0.020 −0.021 −0.040† −0.044* −0.075† −0.067† −0.072† −0.071*

(0.02) (0.02) (0.02) (0.02) (0.04) (0.04) (0.04) (0.04)Regional indirect ties 0.000 −0.009 0.004 0.003 0.008 0.016 0.009 0.013

(0.01) (0.01) (0.01) (0.01) (0.01) (0.01) (0.01) (0.01)Within sector 0.280† 0.296† 0.247 0.244 −0.060 0.534† −0.054 0.514†

incoming ties (0.16) (0.16) (0.17) (0.17) (0.23) (0.28) (0.23) (0.30)Within sector 0.171 0.208 0.208 0.209 0.066 0.430† 0.080 0.433†

outgoing ties (0.14) (0.14) (0.14) (0.14) (0.18) (0.24) (0.18) (0.24)Ties from influential 0.147* 0.069 0.047 0.062 0.054 −0.037 0.051 −0.012

financial firms (0.06) (0.07) (0.07) (0.07) (0.10) (0.11) (0.10) (0.11)Ties to influential −0.144 −0.114 −0.119 −0.142 0.068 −0.345† 0.057 −0.365*

financial firms (0.10) (0.10) (0.10) (0.10) (0.15) (0.18) (0.15) (0.18)Equity owned by 0.860† 0.252 0.146 0.267 0.296 −0.565 0.364 −0.518

experienced firms (0.51) (0.60) (0.61) (0.62) (0.84) (1.01) (0.86) (1.04)Year dummies Y Y Y Y Y Y Y YIndustry dummies Y Y Y Y Y Y Y YRelated experience 0.062* 0.063* 0.060† 0.061* 0.249*** 0.093 0.243*** 0.098

(0.03) (0.03) (0.03) (0.03) (0.07) (0.06) (0.07) (0.06)Aggregated ties 0.101***

(0.02)Incoming ties 0.311*** 0.371*** 0.296**

(0.08) (0.12) (0.10)Outgoing ties 0.127* 0.251*** −0.039

(0.07) (0.08) (0.09)Indirect ties 0.024 0.023 0.014 0.078* 0.010 0.083*

(0.03) (0.03) (0.04) (0.04) (0.04) (0.05)Outgoing CEO ties 0.309** 0.367* 0.101

(0.12) (0.19) (0.17)Outgoing non-CEO ties 0.045 0.191* −0.135

(0.08) (0.12) (0.13)Incoming CEO ties 0.408*** 0.417** 0.406***

(0.10) (0.18) (0.13)Incoming non-CEO ties 0.208* 0.339* 0.190†

(0.10) (0.17) (0.13)N 19,882 19,882 19,882 19,882 11,133 8,749 11,133 8,749Model Chi2 594.22*** 644.43*** 648.26*** 686.88*** 551.30*** 325.51*** 555.09*** 343.08***

Dependent variable: log-odds of foreign investment; coefficients reported with cluster-robust standard errors in parentheses.† < 0.10; *< 0.05; **< 0.01; ***< 0.001 (one-tailed hypothesis tests).



like those reported here, conventional interactionsmay lead to inappropriate conclusions because thecoefficient of an interaction term does not alwaysrepresent the correct sign or magnitude (Ai andNorton, 2003). Consequently, we tested the inter-action hypotheses using the procedure suggestedby Shaver (2007). Specifically, we split the sam-ple into firms with experience levels below (Model5) and above the mean (Model 6) and comparedthe effect of each kind of tie across subsamplesthrough a test of significance. Comparing acrossmodels, we found that indirect ties had a strongereffect among experienced firms than among firmswith little experience (H3a). In contrast, we foundthe opposite with respect to incoming and outgoingties on market entry; these effects were higher forfirms with low levels of experience than for thosewith high levels of experience (H3b, H3c). Weconducted t-tests comparing the marginal effectsof each kind of interlock across subsamples withhigh and low levels of prior experience. These testsconfirmed that indirect ties had a stronger positiveimpact on the likelihood of foreign entry for firmswith high levels of experience (p < 0.001). Incom-ing and outgoing interlocks, on the other hand, hada stronger positive effect on firms with low lev-els of experience (p < 0.001). The findings providesupport for hypotheses H3a–H3c.

We examined the effect of CEO and non-CEOinterlocks across low and high levels of focalfirm experience (Models 7 and 8) as an additionalway to assess the moderating effect of experientiallearning, though we did not have formal hypothe-ses in this regard. Under low levels of experi-ence, the coefficients for outgoing and incomingCEO ties were positive and significant. In addi-tion, outgoing non-CEO ties and incoming non-CEO ties had a significant influence. For firmswith high levels of experience, incoming CEO ties(p < 0.001) and incoming non-CEO ties (p < 0.1)remained significant. Comparing the respectivemarginal effects across subsamples we find thatthe effect of outgoing and incoming CEO ties andnon-CEO ties was significantly stronger in firmswith low levels of related experience (p < 0.001).This suggests that incoming and outgoing inter-locks lose influence as firm experience increasesregardless of the formal position of the managerestablishing the interlock, though incoming tiesremain significant. We note that indirect ties con-tinue to increase in impact for firms with higherexperience.

Additional tests

Endogeneity of incoming ties

Our theoretical explanation of the results centerson organizational learning. An alternative expla-nation for the hierarchy of interlock influence wefind may be that the strong effect of incoming tiesis driven by selection rather than learning. Thismay be plausible if firms have greater latitude inestablishing incoming ties to experienced individ-uals than in forming outgoing ties, particularlybecause a focal firm has greater control over whomit invites to its own board than over procuringinvitations for its managers to serve on the boardsof other firms. In addition, firms may be likely toprefer direct incoming ties with experienced firmsover indirect ties that do not bring an experiencedindividual directly into the boardroom. If thiswere the case, the stronger impact of incomingties relative to outgoing and indirect ones could bedriven by a higher propensity to choose incomingties to experienced partners rather than by superiorlearning from incoming interlocks. While ourarguments are not based on an assumption of suchinstrumentality in the selection of different typesof interlocks1, we believe it important to dismissthis possibility empirically.

The solution to this concern is to either controlfor factors that may allow firms to attract incomingexperienced directors or to use an estimator thataccounts for the endogenous selection of incomingties. We have followed both approaches. Wenote that the models summarized in Table 3include variables that get at attributes of firms thatmake them more attractive to incoming directors(firm size, profitability, links to other influentialfirms) or that suggest a need for external adviceon foreign market strategies from experienceddirectors (foreign sales, prior experience in formerWarsaw Pact countries). Nevertheless, to addressthe selection concern more carefully and directly,we implemented the two-step treatment regressionmodel derived by Maddala (1983). In the firststep, we estimated the hazard of establishing

1 In our interviews with German managers and board members,we found no evidence that interlocks of any type wereestablished with the ex ante goal to learn about a very specificdomain, with the exception of an occasional need for financialexpertise. Rather, directors are appointed based on their generalmanagerial experience, and learning about specific issues onwhich directors have expertise seems to arise after ties areformed.



an incoming interlock, and in the second stageestimated our main results controlling for thehazard.

To identify the hazard of forming incomingties properly in the first stage, we included fiveinstruments that affected the likelihood of havingincoming ties but not the probability of entryin the second stage. We reasoned that certainattributes of firms’ headquarters (HQ) city wouldpartially explain the ability of firms to attractincoming directors but be unrelated to the choiceto invest in Eastern Europe. The five instrumentswere the population of the HQ city, the numberof cities within 50 kilometers of HQ, whetherthe HQ city had a first division (Bundesliga)soccer team, the yearly precipitation in the HQregion, and whether the HQ were located in astate that provided economic assistance to otherGerman states. Since treatment regression requiresthe endogenous variable to be dichotomous, wecreated an indicator coded as 1 if the firm had oneor more incoming ties and 0 otherwise. To be ableto compare the effect of incoming ties to othertypes of ties, we also dichotomized outgoing andindirect ties2. The results are shown in Table 4.

In Models 9–11, we treat incoming ties (regard-less of whether they are formed by CEOs or not)as potentially endogenous. Model 9 demonstratesthat, after accounting for selection, the pattern oftie influence remains as predicted by H1a andH1b. Models 10 and 11 compare effects acrossfirms with high and low experience in other East-ern European countries. As in H3a–H3c, indirectties increase in influence whereas incoming andoutgoing ties decrease in influence as firms gaintheir own experience (p < 0.05). In Model 12 wetreat incoming CEO ties as potentially endoge-nous. Our reasoning is that firms would be espe-cially desirous to invite CEOs of other firms totheir boards to benefit from the status and expe-rience of these individuals. We continue to findsupport for H2b (p < 0.01). Our hypotheses con-tinue to be supported after accounting for theselection of incoming ties—providing further evi-dence that learning is the mechanism explainingthese results.

2 We also attempted to estimate our models treating incominginterlocks as a continuous outcome in the first stage but foundthat identification was extremely weak or impossible becauseof the high prevalence of firms with no incoming interlocks todirectors with experience in particular countries.

Other tests

The high correlation between incoming and indi-rect interlocks (see Table 2) raises a concern thatmulticollinearity may affect our results. We tooktwo steps to address this concern. First, we esti-mated VIF scores for each variable and for theoverall model (Allison, 1984). None of the scoresapproached problematic levels. Second, we ran ouranalysis excluding indirect interlocks, and our find-ings regarding incoming and outgoing interlockswere unaltered. Thus, the correlation betweenincoming and indirect interlocks is not biasing theresults.

We also assessed the robustness of our resultsto alternative event history estimation proceduresincluding the rare event procedure suggested byKing and Zeng (2001), the Cox (1972) propor-tional hazard model, and piecewise exponentialregression (Lee, 2007). In all cases, our findingsremained robust.

DISCUSSION

Our paper has implications for research on therole of boards of directors on strategy formula-tion, on experiential and vicarious learning morebroadly, and for work on FDI in emerging mar-kets. We address each of these contributions inturn.

Prior work on the influence of board interlockson strategy formulation has largely been agnosticwith respect to attributes of the individuals creatingthe ties. The findings of this study clearly revealthat there are differences across types of ties thataffect how influential their information might befor the focal firm. In line with our theoreticalassumptions, outgoing, incoming, and indirect tiesaffect the focal firm’s market entry strategy—butunder different conditions and with differentialimpact. As expected, incoming ties exhibit thestrongest effect, particularly when the focal firmlacks its own first-hand experience, followed byoutgoing and indirect ties. In novel situations suchas emerging market entry, the focal firm seemsto be most likely to act upon rich and fine-grained information provided by top managers ofother firms that bring first-hand experience. Thistype of rich knowledge appears to outweigh theknowledge gained by managers forming outgoingties. In line with our hypotheses, indirect ties have



Table 4. Treatment regression results

Model 9Disaggregated

ties

Model 10Experience

low

Model 11Experience

high

Model 12Incoming CEO vs.incoming non-CEO

Second stage estimation (DV = probability of foreign entry)Incoming ties 0.061** 0.097** 0.042**

(0.01) (0.02) (0.02)Outgoing ties 0.040** 0.067** 0.016*

(0.01) (0.01) (0.01)Indirect ties 0.009* 0.006 0.013* 0.012**

(0.00) (0.01) (0.01) (0.00)Related experience 0.001* 0.004** 0.002† 0.001*

(0.00) (0.00) (0.00) (0.00)Outgoing CEO ties 0.040**

(0.01)Outgoing non-CEO ties 0.026**

(0.01)Incoming CEO ties 0.057**

(0.01)Incoming non-CEO ties 0.026**

(0.00)Inverse mill’s ratio −0.019*** −0.032*** −0.015* −0.015*

(0.01) (0.01) (0.01) (0.01)First stage estimation (DV = probability of having incoming ties)

HQ in a donor state −0.226** −0.197** −0.150* −0.378**(0.05) (0.07) (0.08) (0.07)

Yearly precipitation 0.000 0.000 0.000 0.001**

(0.00) (0.00) (0.00) (0.00)Soccer team in HQ city 0.010 −0.300** 0.256** 0.300**

(0.04) (0.07) (0.06) (0.05)Cities near HQ (50 km) 0.006† 0.025** −0.011† 0.017**

(0.01) (0.01) (0.01) (0.01)HQ city size (millions) −0.068* 0.013 −0.113* −0.431**

(0.03) (0.06) (0.05) (0.05)N 19,615 10,848 8,767 19,615Model Chi2 4,747.56*** 2,565.82*** 2,213.70*** 3,024.76***

†< 0.10; *< 0.05; **< 0.01; ***< 0.001 (one-tailed hypothesis tests).For presentational simplicity, only the independent variables of interest and the excluded first stage variables are shown. Not shown,but included in these models, are all the control variables detailed in Tables 2 and 3.Incoming ties is treated as endogenous in Models 9–11, and incoming CEO ties is treated as endogenous in Model 12.Treatment regression requires the endogenous variable to be dichotomous. We thus created an indicator coded as 1 if the firm hadone or more incoming ties and 0 otherwise. To be able to compare coefficients, we also dichotomized outgoing and indirect ties.

the least influence on the focal firm when focalfirm experience is low, but the more subtle andabstract experiences of indirect ties become moreinfluential once the focal firm acquires a baselineof experience that allows it to activate the insightsbrought by indirect ties.

Our understanding of which interlocks mattermost becomes more nuanced when considering theformal authority of managers creating outgoingor incoming ties. Our results show that tiescreated by CEOs are more influential than thosecreated by non-CEOs. Incoming CEO ties are

more influential than outgoing CEO ties—againsuggesting that the first-hand experience of theincoming CEO is more impactful than the second-hand knowledge brought back to the firm by theoutgoing CEO. The difference between incomingand outgoing non-CEO ties seems noteworthy.Whereas non-CEOs of other experienced firmsinfluence the focal firm’s market entry decisions,non-CEOs of the focal firm who serve on theboards of experienced firms have no significantinfluence. This suggests that only the CEO of thefocal firm seems to have enough decision-making



authority to influence the firm’s strategy basedon what s/he vicariously learns in the processof serving as a director. Beyond the applicationto board interlocks in particular, this suggeststhat organizational learning is a function of notonly the characteristics of the knowledge obtainedvicariously—as many studies emphasize—butalso of the authority possessed by boundaryspanners who carry knowledge into the firm.These findings help to make sense of inconsistentfindings in prior research. Some studies (while notcomparing the effects of all types of interlocks)have reported significant effects for outgoing ties(e.g. Haunschild, 1993), while other studies havereported weak or no effect (e.g. Geletkanycz andHambrick, 1997). Our results suggest that withrespect to outgoing ties, the influence is driven bywhether the outgoing manager is the focal firmCEO.

Prior research has suggested—but not empir-ically shown—that indirect interlocks have theweakest influence (Haunschild and Beckman,1998). In fact, some recent studies go as far asexcluding indirect (often called ‘neutral’) inter-locks from analysis altogether (e.g. Beckman,Haunschild, and Phillips, 2004). Our findings sug-gest that indirect ties may be more influential thanpreviously thought by highlighting when such tiesare most influential. It appears that to fully exploitthe coarse-grained information available from indi-rect ties, firms need a baseline of experience inorder to have the absorptive capacity (Cohen andLevinthal, 1990). This insight echoes research oninnovation that finds that firms investing in inter-national R&D extract learning benefits only whenthey already possess a knowledge base in areasthat are technologically related to the new ven-ture (Penner-Hahn and Shaver, 2005). Interest-ingly, direct ties, which provide the most ‘frontal’information to the firm, diminish in importanceonce the firm gains a baseline of first-hand experi-ence. These findings challenge scholars to specifymore carefully boundary conditions that clarifywhen they expect various interlocks to influencefirms.

These results contribute to the field of orga-nizational learning, which distinguishes betweenexperiential and vicarious learning. An importanttension in this literature pertains to whether thetwo types of knowledge are substitutes or com-plements, and reasonable arguments exist to sup-port both effects. We propose that substitution

or complementarity between the two sources oflearning is contingent upon the type of vicariousknowledge brought by external ties such as inter-locks. Prior focal firm experience and vicariouslearning through direct ties to other experiencedfirms appear to function as substitutes because theyboth bring specific, pointed knowledge. In contrast,indirect ties affect firms more strongly only oncea firm gains sufficient levels of prior experience.This finding may have applicability beyond boardties in particular, but future work must verify thisimplication.

The results of our study also have implicationsfor the FDI literature, particularly as it appliesto emerging market entry, which are particularlyfraught with risk and uncertainty. Existing FDItheory suggests that firms learn from their expe-riences with prior investments in related markets,which reduces the uncertainty that might other-wise preclude or significantly reduce the likeli-hood of entry (Delios and Henisz, 2003; Johansonand Vahlne, 1977). The theoretical challenge hasbeen that learning from first-hand experience maybe impossible when the firm seeks to enter anemerging market so different or novel that priorforeign investments provide little guidance. Wehave built on a small but important body of lit-erature that has begun to explore how vicariouslearning from firms with relevant foreign marketexperience affects entry choices (Guillen, 2002;Martin, Swaminathan, and Mitchell, 1998). Ourresearch suggests that characteristics of the indi-viduals creating the tie, which have not beenstudied in this context, have a significant effecton firms’ decisions to enter high-risk emergingmarkets.

An important contribution was to accountempirically for two alternative explanations tolearning: coordination or control, and endogenousselection. Besides fostering learning, board inter-locks are mechanisms of coordination and control,and we included covariates related to the institu-tional context to account for this alternative mech-anism of interlock influence. Recently scholarshave recognized endogeneity of networks as one ofthe most critical issues in interorganizational stud-ies (Ahuja, Soda, and Zaheer, 2012). To our knowl-edge, this is the first study to control for selectioneffects in interlock research. The need to addressthis issue arises because the effects of interlocks onfirm learning may not be independent of the under-lying processes the lead firms to appoint directors



to their boards in the first place. Addressing thisissue becomes fundamental for scholars to distin-guish between learning effects and those drivenby omitted variables. While all interlocks may tosome extent be subject to this concern, when com-paring the influences of different types of ties theconcern is particularly salient in the case of incom-ing interlocks (as we explained already). Notwith-standing this important empirical advance, we real-ize that ultimately we do not observe directly whatis discussed in the boardroom—a limitation com-mon to interlock research. Our interviews withGerman directors and executives go some waytoward mitigating this concern, but future workcould more systematically employ field work toshed light on the knowledge exchange dynamicsin boards.

This research adds to an important and emergingemphasis on differences in learning from differ-ent types of sources of vicarious and experientiallearning by focusing on the information communi-cated by individuals who differ with respect to thenature of experience they bring to the boardroomand their hierarchical position. We find that outgo-ing, incoming, and indirect ties transmit informa-tion with different vicarious learning impact. Thedifferences are further accentuated when we takethe hierarchical position (CEO vs. non-CEO) of theperson creating the tie into account. In addition, weconsider the interaction of these external sourcesof information with internal knowledge availablefrom past experience. This allows us to explainnot only the relative learning effects of differentexternal sources, but also when each type willsubstitute or complement the firm’s experientiallearning.

ACKNOWLEDGMENTS

All authors contributed equally. We thank coeditorWilliam Mitchell and two anonymous reviewersfor their insightful comments on prior versionsof the paper. We also thank Daniel Levin, ArikLifschitz, Anne Miner, David Souder, Aks Zaheer,and the participants at brown bag discussions atArizona State University, Brigham Young Uni-versity, FU Berlin, HKUST, Pennsylvania StateUniversity, Purdue University, Rice University,Singapore Management University, University ofIllinois at Urbana-Champaign, University of St.Gallen, and the University of Wisconsin for their

helpful comments. We thank Susanne Krenn forher help with collecting the data for this study andMyles Shaver for valuable suggestions regardingdata analysis.

REFERENCES

Ahuja G. 2000. Collaboration networks, structural holes,and innovation: a longitudinal study. AdministrativeScience Quarterly 45(3): 425–455.

Ahuja G, Soda G, Zaheer A. 2012. Introduction tothe special issue: the genesis and dynamics oforganizati

whose experience matters in the boardroom? the effects of … · 2019. 12. 12. · argote and...

Documents