MNES AND CORRUPTION: THE IMPACT OF NATIONAL … · 2017-09-09 · MNEs and Host Country Corruption...

Strategic Management JournalStrat. Mgmt. J., 32: 280–300 (2011)

Published online EarlyView in Wiley Online Library (wileyonlinelibrary.com) DOI: 10.1002/smj.874

Received 23 February 2007; Final revision received 6 June 2010

MNES AND CORRUPTION: THE IMPACTOF NATIONAL INSTITUTIONS AND SUBSIDIARYSTRATEGY

JENNIFER SPENCER1* and CAROLINA GOMEZ2,3

1 Department of International Business, George Washington University School ofBusiness, Washington, DC 20052, U.S.A.2 Department of Management and International Business, College of BusinessAdministration, Florida International University, Miami, Florida, U.S.A.3 Department of Management, Love School of Business, Elon University, Elon, NorthCarolina, U.S.A.

We argue that the pressure MNE subsidiaries face to engage in corrupt practices in their hostcountry varies positively with the institutionalization of corrupt practices in both host and homecountry environments. We further argue that the relationship between an MNE’s home countryenvironment and the pressure it faces in the host country is moderated by its localization strategy.Results suggest a positive relationship between the host country corruption environment and thepressure subsidiaries face to engage in bribery locally. Mixed results emerged concerning MNEsfrom home countries participating in the OECD Convention for Combating Bribery. Resultsconcerning the impact of the home country corruption environment are best viewed in light ofsignificant moderating effects. When MNEs did not have local partners, firms from less corrupthome countries reported less pressure to engage in corrupt practices locally; however, thepresence of local partners eliminated this relationship. Results will help managers understandthe pressures their firm is likely to face when operating in corrupt host country environments, andalso offer guidance concerning how the firm might reduce its exposure to those local institutionalpressures. Copyright 2010 John Wiley & Sons, Ltd.

INTRODUCTION

Despite the efforts of governments, non-govern-mental organizations, and multilateral institutionsto reduce corruption levels worldwide, subsidiariesof multinational enterprises (MNEs) operating inemerging and developing countries regularly en-counter pressure to engage in corrupt practices

Keywords: corruption; multinational enterprises; bribery;institutional theory; alliances; developing countries*Correspondence to: Jennifer Spencer, Department of Interna-tional Business, George Washington University School of Busi-ness, Washington, DC 20052, U.S.A. E-mail: [email protected]

such as bribery (Beets, 2005). Indeed, since cor-ruption remains the norm rather than the exceptionaround the world today (Beets, 2005), managersoften must make a critical strategic choice. Impor-tantly, a decision on whether to violate or conformto the corruption norms that have been institu-tionalized locally will affect the MNE’s opera-tions, reputation, and performance—both locallyand worldwide.

The strategic implications of a subsidiary’s deci-sions regarding corrupt practices are particularlyacute for MNEs from home countries in whichanticorruption norms have been institutionalized.

Copyright 2010 John Wiley & Sons, Ltd.

MNEs and Host Country Corruption 281

Recent scholarship has suggested a lesser tendencyfor such MNEs to invest in more corrupt envi-ronments (Cuervo-Cazurra, 2006), possibly due tothe challenges inherent in conforming to conflict-ing home and host institutional environments. Butwhen MNEs do invest, what factors influence thedegree of local pressure they face to engage incorrupt practices? And do strategic decisions madeby the MNE have an impact on the level of thatpressure?

Transparency International (2008) defines cor-ruption, for purposes of their Corruption Percep-tion Index (CPI), as the ‘abuse of public power forprivate gain’ and their measure includes question-able business practices such as bribery, kickbacks,and embezzlement of funds. In this paper, we linkthe CPI of an MNE’s home and host countries to aspecific manifestation of corruption—the propen-sity of local government officials to impose pres-sure on firms to make ‘unofficial payments’ (i.e.,bribes) to get business done. We begin our theorydevelopment with the notion that characteristicsof corruption can become so institutionalized thatthey become fundamental components of a coun-try’s institutional environment (Collins and Uhlen-bruck, 2004). We further draw on the recognitionthat MNE subsidiaries face unique circumstancesby virtue of their embeddedness in ‘multiple, frag-mented, nested, or often conflicting institutionalenvironments’ (Kostova, Roth, and Dacin, 2008:998). Building on both perspectives, we arguethat the pressure an MNE subsidiary faces variesaccording to the institutionalization of corruptionin both its host and home countries. Further, wesuggest that certain strategies that MNEs pursue torespond to local conditions will weaken the influ-ence of the home country institutional environ-ment, thereby increasing pressure on the subsidiaryto conform to host country corruption norms.

To test our arguments, we use two indepen-dent sets of data. The first is an archival databaseof firm-level survey responses across a range ofEastern European countries. The second draws onprimary survey data collected among MNE sub-sidiary managers in Ghana. We operationalize ourdependent variable, local pressure to engage incorrupt practices, in two complementary ways inour two datasets. In the Eastern Europe dataset, thisvariable reflects the degree to which an MNE sub-sidiary needs to bribe local officials to get businessdone, as perceived by the subsidiary’s manager. Inthe Ghana dataset, this variable reflects the degree

to which local corruption pressures—includinggovernment officials’ demands for bribes, a ten-dency for the subsidiary’s competitors to offerbribes, and a general perception of corruptionamong local officials—pose an obstacle to thefirm’s performance, as perceived by the sub-sidiary’s manager. By drawing on complementarydata sources, we demonstrate the robustness ofresults beyond the methodology and specific word-ing of items in a single survey, which provides astronger test of our arguments.

In the next sections, we describe institutionaltheory in the context of the MNE, and develophypotheses concerning the pressure subsidiariesface to engage in corrupt practices in their hostcountries and the influence of particular localiza-tion strategies. We then summarize the methodol-ogy, which employs hierarchical linear modeling,and present the results of our analysis. We end bydiscussing theoretical and practical implications.

THE ROLE OF NATIONALINSTITUTIONAL ENVIRONMENTS

Institutional theory and the MNE

Firms face institutional pressure to conform tosocietal conventions and expectations by way ofat least three processes (DiMaggio and Pow-ell, 1983). Coercive processes reflect pressuresimposed by an authority; normative pressuresreflect established paradigms in the society; andmimetic processes reflect pressures for firms toimitate successful enterprises in their organiza-tional field.1 Such conformity can legitimate a firm(Parsons, 1960) by contributing to its acceptanceand endorsement by relevant actors in the institu-tional environment, and the resulting legitimacycan be critical to the firm’s survival and per-formance (Deephouse, 1999; Meyer and Rowan,1977; Miller and Chen, 1995).

Because an MNE’s activities to achieve legiti-macy in one country can affect its legitimacy else-where (Kostova and Zaheer, 1999; Westney, 1993),headquarters’ managers often impose pressure forsubsidiaries to adopt practices used in home coun-try (i.e., headquarters) operations (Kostova and

1 An organizational field comprises the set of organizations that,taken as a group, ‘constitute a recognized area of institutionallife’ (DiMaggio and Powell, 1983: 148), exhibiting interactionand mutual recognition among participants (Phillips and Tracey,2009).

Copyright 2010 John Wiley & Sons, Ltd. Strat. Mgmt. J., 32: 280–300 (2011)DOI: 10.1002/smj

282 J. Spencer and C. Gomez

Zaheer, 1999). Indeed, Kostova et al. (2008) pro-posed that internal constituencies can exert strongerpressure on MNE subsidiaries than external hostcountry constituencies. At the same time, sub-sidiary managers may resist such internal pressurewhen they perceive that the practices promotedby headquarters run counter to local expectations(Kostova and Roth, 2002). In the context of cor-ruption, this means that a subsidiary may encountersignificant host country pressure to engage inbribery, while facing a mandate from headquar-ters to conform to home country expectations byavoiding such corrupt practices.

Recent scholarship has noted the challenges aris-ing for organizations striving to achieve legit-imacy across diverse institutional environments(Kraatz and Block, 2008). Given conflicting pres-sures, Kostova et al. (2008) proposed that MNEs’receipt of legitimacy can be viewed as a nego-tiated process between subsidiary managers andrelevant local actors. Similarly, Kraatz and Block(2008) noted that organizations may deal with suchconflicting pressures through one of four strate-gies. First, they may attempt to eliminate the con-flict by denying the validity of one constituency’sdemands or co-opting that constituency to controlits expectations. Second, they can compartmental-ize their identities in order to relate to each con-stituency independently. Third, they may attemptto balance demands, play constituencies againstone another, or find cooperative solutions to ten-sions. And finally, firms may forge a strong enoughidentity to emerge as institutions in their own right,capable of legitimating their own actions (Kraatzand Block, 2008). Given the variety of ways inwhich firms may deal with institutional pluralism,it is important to understand the diverse pressuresan MNE may face regarding corruption.

Impact of the host country institutionalenvironment on local corruption pressure

When managers perceive that a corrupt practicesuch as bribery has become institutionalized ina society, they are more likely to conform tothose societal expectations to obtain legitimacy.For instance, Collins and Uhlenbruck (2004) foundempirical evidence that managers in India who per-ceived corruption to reflect ‘the way things aredone’ locally were more likely to engage in cor-rupt practices, even when they personally viewed

them negatively. In addition, as corrupt activi-ties become more taken for granted in a country,government officials may impose more coercivepressure on MNE subsidiaries to participate inactivities such as bribery. Moreover, the more thathigh performing local organizations engage in cor-ruption, the greater will be the mimetic pressure onan MNE subsidiary to offer bribes in the course ofits business operations. Therefore, as a baselineproposition, we suggest that,

Proposition 1: The more that corruption normsare institutionalized in a host country, thegreater the pressure the MNE subsidiary willface to engage in corrupt practices such asbribery in the local environment.

We assume that the institutionalization of cor-ruption corresponds with the prevalence ofcorruption in a given country, as measured byTransparency International’s CPI, and we testthis relationship using the Eastern Europe dataset.Thus,

Hypothesis 1: The corruption level of an MNE’shost country will relate positively to its sub-sidiary’s need to engage in bribery in that hostcountry.

Impact of the home country institutionalenvironment on local corruption pressure

Stinchcombe (1965) described how firms are ‘im-printed’ by the conditions present at their found-ing, and argued that organizational characteristicsacquired in response to these initial environmentalpressures tend to be retained even as the envi-ronment changes. Institutional pressures are par-ticularly salient at the time an organization iscreated, and the organizational practices thatfounders devise generally reflect this institutionalenvironment (Dacin, 1997). Because these orga-nizational practices become taken for granted as‘the way things are done’ (Berger and Luckmann,1966), they tend to persist in the organization overtime, even in the face of environmental changes(Scott, 1991).

Applying this concept to MNEs, we expect thatorganizational practices devised at founding willpersist as the MNE expands abroad into new insti-tutional environments (Kogut, 1993). In this way,an MNE’s response to corruption pressure in its



foreign subsidiaries will partially reflect practicesdevised to conform to its home country institu-tional environment. Additionally, researchers havedocumented MNEs’ efforts to foster common orga-nizational cultures (Jaeger, 1983) and implementcommon organizational practices in their world-wide subsidiary networks (Kostova and Roth,2002). Since the corruption domain is subjectto intense worldwide scrutiny, often couched inmoral, ethical, and legal arguments, a violationof home country expectations can create substan-tial negative publicity for an MNE—what Kos-tova and Zaheer (1999: 64) called negative ‘legit-imacy spillovers.’ Given that MNEs from lesscorrupt institutional environments are more ex-posed to potential legitimacy spillovers by virtue ofthe greater disdain for corruption present in theirhome environments, they will be more likely toconsciously invest in an internal culture and orga-nizational policies and practices that discouragecorruption worldwide. These effects should haveimportant implications at the country level, withsystematic differences in MNEs’ behavior acrosshome countries translating into reputation effectsthat can influence the pressure all subsidiaries fromthat country face to engage in corrupt practices ina host country.

Increasingly, government officials face uncer-tainty when they demand bribes due to expandedenforcement efforts (Gyimah-Boadi, 2004; Wid-jajabrata and Zacchea, 2004). Given that somefirms acquiesce quietly to bribe requests and oth-ers are more likely to report the demands orappeal to other officials or agencies for protec-tion (Uhlenbruck et al, 2006; Rodriguez, Uhlen-bruck, and Eden, 2005), we believe that localofficials will adjust their tactics and level of pres-sure based on their assessments of the targetedfirm’s likely reaction. Supporting this idea, anec-dotal evidence suggests that individual firms canbuild reputations for uncorrupt behavior, and thatsuch reputations have allowed some MNEs toavoid engaging in bribery, even within environ-ments in which corruption is strongly institution-alized. For example, Habib and Zurawicki (2002)suggested that McDonalds’ worldwide image help-ed it take a stand against corruption in Russia, andGratchev (2001) reported a similar effect among3M’s subsidiaries overseas. Likewise, a managerwith extensive experience doing business in someof the most corrupt countries in the world notedthat firms that make it absolutely clear that they

will not engage in corrupt practices tend to receivefar fewer bribe requests than firms that appearmore willing to pay (Pingle, 2009).

Many times, however, firm-specific policies arenot visible externally; in these cases, govern-ment officials must resort to heuristics such asstereotypes (Kostova and Zaheer, 1999) to assessa subsidiary’s likely behavior. Hence, when thesubsidiary’s home country has a reputation foranticorruption norms, officials may rely on thisstereotype. In this way, we argue that officialswill be more aggressive when targeting firms fromrelatively corrupt home countries, and less aggres-sive in pressuring MNEs from countries in whichanticorruption norms have been institutionalized.In addition, when subsidiary managers from lesscorrupt countries refuse an initial overture out ofconformity with MNE policy, they may be sub-jected to less subsequent pressure, making firmsfrom these countries better able to resist demandsto pay a bribe. For these reasons, we propose thatwhen operating in corrupt environments,

Proposition 2: The less that corruption normsare institutionalized in the MNE’s home country,the weaker the pressure the MNE subsidiarywill face to engage in corrupt practices such asbribery in its host country.

More specifically, we expect that within theEastern Europe sample, fewer bribe requests andless aggressive tactics by host country officials willmean that subsidiaries of MNEs from less corrupthome countries will face a lesser need to engagein bribery. Thus,

Hypothesis 2a: The corruption level of anMNE’s home country will relate positively to itssubsidiary’s need to engage in bribery in its hostcountry.

Likewise, this same restraint by governmentofficials suggests that in the Ghana sample, cor-ruption will pose a smaller obstacle to businessperformance among subsidiaries of MNEs fromless corrupt home countries. Thus,

Hypothesis 2b: The corruption level of theMNE’s home country will relate positively tothe degree to which local corruption pressureposes an obstacle to its host country subsidiary’sperformance.



A number of national governments have cho-sen to participate in supranational efforts to reducecorruption levels worldwide, such as the Organi-sation for Economic Co-operation and Develop-ment (OECD) Convention on Combating Briberyof Foreign Public Officials in International Busi-ness Transactions. This convention calls for homecountries to prosecute bribery of foreign officialsas crimes. Empirical evidence has suggested thatthese laws deterred foreign direct investment sothat MNEs from signatory countries engage inless investment in corrupt host countries (Cuervo-Cazurra, 2006). In addition, countries that haveentered into the force of the OECD conventioneffectively send a signal that their MNEs will resistbribery, which may insulate subsidiaries from localcorruption pressure. Hence, we expect that withinthe Eastern Europe sample, subsidiaries of MNEsfrom home countries that have entered the OECDconvention will face fewer bribe requests and lessaggressive tactics by host country officials, andthey will thus encounter less need to engage inlocal bribery. Thus,

Hypothesis 2c: Home country participation inthe OECD convention will relate negatively tothe MNE subsidiary’s need to engage in briberyin its host country.

This same restraint by local officials shouldmean that, in the Ghana sample, corruption pres-sures will pose a smaller obstacle to business per-formance for subsidiaries of MNEs from homecountries that have entered the OECD convention.

Hypothesis 2d: Home country participation inthe OECD convention will relate negatively tothe degree to which local corruption pressureposes an obstacle to an MNE subsidiary’s per-formance in its host country.

THE ROLE OF LOCALIZATIONSTRATEGIES

Firms engage in localization strategies in order toadapt to host country conditions. Such strategiesinclude taking on local partners (e.g., local firmsor investors with partial stakes in the host countrysubsidiary; strategic alliance partners), and localiz-ing control (e.g., decentralizing decision making to

the host country subsidiary; using locals as man-aging directors). We suggest that when MNEs hailfrom countries in which anticorruption norms havebeen institutionalized, such localization strategieswill reduce the insulation afforded by the firm’shome country institutional environment.

Local partners

Just as local government officials may assume thatan MNE subsidiary will conform to the expec-tations of its home country environment, theymay also conclude that local partners will exhibitbehavior more consistent with host country con-ventions and expectations. Thus, when an MNEcomes from a home country in which anticorrup-tion norms have been institutionalized, governmentofficials may find it safer to target their coerciontoward the local partner, which is less likely toresist, and less likely to appeal or report the impro-priety. Indeed, in interviews in Ghana, multiplemanagers noted that when MNEs from less cor-rupt home countries work with local partners, it isthe partner that generally undertakes the corrupttransaction. In this way, the presence of a localpartner provides an avenue for host country offi-cials to pressure the subsidiary to engage in corruptpractices. The presence of a local partner shouldhave a smaller impact when the MNE comes froma country in which corruption is more institutional-ized, since officials would see little risk in directlypressuring the MNE subsidiary itself.

Consequently, we propose that the relationshipbetween an MNE’s home country environmentand local pressure to engage in corruption will beweaker when a local partner is present. Thus,

Proposition 3: The presence of a local partnerwill moderate the relationship between the insti-tutionalization of corruption in the MNE’s homecountry and the pressure its subsidiary faces toengage in corrupt practices such as bribery inthe host country.

More specifically, we expect that in the EasternEurope sample, there will be a weaker relationshipbetween the home country corruption environmentand the degree to which the subsidiary needs toengage in bribery among MNEs that have partiallocal ownership than among those that are fullyforeign owned. Hence,



Hypothesis 3a: Partial local ownership willreduce the effect of an MNE’s home countrycorruption level on the degree to which its sub-sidiary needs to engage in bribery in the hostcountry.

Likewise, we propose that in the Ghana sample,a weaker relationship will emerge between anMNE’s home country corruption environment andthe degree to which corruption pressure poses anobstacle to performance when the firm has analliance partner. Thus,

Hypothesis 3b: The presence of a local alliancepartner will reduce the effect of an MNE’s homecountry corruption level on the degree to whichlocal corruption pressure poses an obstacle toperformance in the host country.

Localizing control of operations

MNEs also make strategic decisions regarding thelevel of control and coordination that they imposefrom headquarters (Martinez and Jarillo, 1991)based on their needs for global integration ver-sus local adaptation (Bartlett and Ghoshal, 1989).These strategies correspond to different levels andtypes of interdependence among the MNE’s sub-sidiaries (Kostova and Roth, 2002). For exam-ple, an MNE striving to achieve global integrationwill have stronger subsidiary interdependence, andthus may strengthen headquarters’ control throughformal mechanisms such as centralized decisionmaking.

Centralized decision making should help MNEsfrom less corrupt countries in their efforts toenforce policies reflecting the anticorruption expec-tations of the home institutional environment. Inturn, MNEs that impose strong anticorruption poli-cies via headquarters’ control are more likely todisplay their credible commitment toward uncor-rupt activities, and thus develop firm-specific repu-tations for avoiding corrupt activities. This shouldcomplement the MNE’s home country reputa-tion to deter government officials from aggres-sively demanding payments. Even when officialsare not aware of an MNE’s internal control mech-anisms, subsidiary managers in these firms willlikely display greater resistance to initial requests.Such resistance can confirm officials’ expecta-tions regarding the risks or payoffs from applying

pressure on firms from less corrupt home countries,and thus strengthen the deterrent effect.

In contrast, when major decisions are localizedto host country managers, a home country’s anti-corruption norms will likely be less internalized bylocal decision makers, and organizational policiesprohibiting bribes are less likely to be monitoredand enforced by the global organization. Gov-ernment officials may anticipate the presence ofrelatively weaker anticorruption norms when theyknow that local managers hold greater levels ofautonomy; or, these norms may become evidentwhen subsidiary managers display weaker resis-tance to initial requests or even offer payments ontheir own initiative. Thus,

Proposition 4: Localized control will moderatethe relationship between the institutionalizationof corruption in the MNE’s home country andthe pressure its subsidiary faces to engage incorrupt practices in the host country.

Given data constraints, we are only able to testthis argument with the Ghana dataset. Therefore,we propose that in the Ghana sample, a weakerrelationship will emerge between the level of cor-ruption in an MNE’s home environment and thedegree to which local corruption pressure poses anobstacle to performance when the firm decentral-izes decision making rather than imposing strongcontrol from headquarters.

Hypothesis 4a: Decentralized decision makingwill reduce the effect of an MNE’s home coun-try corruption level on the degree to which localcorruption pressure poses an obstacle to perfor-mance in its host country.

Another strategy to localize control is reflectedin the use of host country managers rather thanhome country nationals to head the subsidiary.Home country nationals are more likely to havebeen exposed to the MNE’s organizational norms,and to have internalized the norms of the homecountry institutional environment; they are thusmore likely to implement the anticorruption direc-tives imposed by headquarters. In contrast, hostcountry nationals will have had less exposure tothe MNE’s home country norms by virtue of fewerties to the home country and professional net-works there, and are therefore less likely to haveinternalized anticorruption norms into their own



philosophies. We argue that managers who haveinternalized home country norms will mountstronger resistance to bribe requests (reinforcingthe home country reputation effect), whereas man-agers who have not internalized the MNE’s orga-nizational norms will be more likely to acquiesceor volunteer bribes (compromising the home coun-try reputation effect). Therefore, we propose thatin the Ghana sample, a weaker relationship willemerge between an MNE’s home country corrup-tion environment and the degree to which cor-ruption pressure poses an obstacle to performancewhen a local rather than expatriate manager headsthe subsidiary.

Hypothesis 4b: The use of a local national tohead the subsidiary will reduce the effect of anMNE’s home country corruption level on thedegree to which corruption pressure poses anobstacle to performance in the host country.

METHODOLOGY

As noted above, we drew from two distinct data-sets. Hypotheses concerning whether managersperceived a need to engage in bribery were testedusing the Business Environment and EnterprisePerformance Survey (BEEPS) database. TheBEEPS database, developed jointly by the WorldBank and the European Bank for Reconstruc-tion and Development, contains survey responsesfrom firms in Eastern European countries from1999–2000. Because of lack of availability ofcontemporaneous data for some countries, thisstudy was conducted using data on MNEs thathad operations in: Armenia, Azerbaijan, Belarus,Bulgaria, Croatia, Czech Republic, Estonia, Hun-gary, Kazakhstan, Latvia, Lithuania, Moldova,Poland, Romania, Russia, Slovak Republic, Slove-nia, Turkey, Ukraine, and Uzbekistan. Each coun-try displays a relatively high level of corrup-tion, with ratings on Transparency International’sCPI ranging from 1.5 to 5.7 (with 10 reflectingthe least corrupt environment). The database ispublicly available on the World Bank Web siteand contains 3,565 foreign- and locally ownedfirms in the above countries (after excluding pub-lic agencies, charities, non-governmental organiza-tions, and farms). In this analysis, we consideredonly firms with foreign ownership, leading to an

effective N, after deleting observations with miss-ing values, of 151 firms.

Hypotheses regarding the degree to which cor-ruption poses an obstacle to business performancewere tested using data from firms operating inGhana. Interviews of foreign and domestic firmswere undertaken by one of the authors in Accraand Tema, Ghana in order to pretest a surveyinstrument and gain deeper insight into the phe-nomenon. Data were obtained from surveys thatwere administered to foreign- and locally ownedfirms operating in Ghana at two different times:the first in 2004–2005 and the second in 2007.Participant firms were randomly identified throughdata provided by the Ghana Investment Promo-tion Center, the Ghanaian Chamber of Commerce,and the Ghanaian Yellow Pages. Graduate studentsat the University of Ghana and Ashesi Univer-sity in Ghana, employed as research assistants,hand delivered surveys to managing directors at thefirms, and either waited while they were completedor scheduled a date for an unlabeled, sealed surveyenvelope to be picked up.2 All respondents weremanaging directors, vice presidents, or accountantsin the Ghanaian organization. In 2004–2005, 400firms were sampled and 174 usable responses werereceived. In 2007, 365 firms were sampled and141 usable responses were received. This yieldeda response rate of 41 percent overall. This studyrelates only to foreign-owned firms that listed theirheadquarters country to be outside of Ghana. Fifty-five responses met this criterion, and listwise dele-tion of responses with missing values yielded aneffective N of 46 firms.

Although the response rate was high, we ex-plored for nonresponse bias by testing for differ-ences among those completing the survey in fulland those refraining from participation on the cor-ruption variables, the most sensitive questions inthe survey. We explored differences in ownership,home country CPI score, participation in a strate-gic alliance, and origin of managing director, aswell as the more general variables of age andsize. Means between respondents and nonrespon-dents were not significantly different at the 0.10level.

2 This data collection mechanism was used due to the poorreliability of the postal system. In three cases, the manager chosea second option of mailing the survey back to the author in theUnited States. However, all other respondents relied on manualpick up from research assistants.



Dependent variables

The level of local pressure a firm faces to engage incorruption in its host country was operationalizedin two complementary ways. In the study of East-ern European firms, the dependent variable (a sub-sidiary’s need to engage in bribery) took the formof six items in which respondents were asked theirperceptions with respect to the question, ‘Howoften do firms like yours nowadays need to makeextra, unofficial payments to public officials. . .’across a range of activities (see Appendix 1 for allitems). As with other research on corruption, theitems were phrased indirectly and, thus, the respon-dents were not implicating themselves of wrongdo-ing (Svensson, 2003). Such phrasing should reduceconcerns about social desirability. In its origi-nal form, responses ranged from 1, ‘always,’ to6, ‘never.’ For ease of interpretation, we reversecoded this variable so that higher numbers indi-cated a greater need to offer bribes. A factor anal-ysis using varimax rotation of survey items withcommon anchors confirmed the presence of a cor-ruption factor; Cronbach’s alpha was 0.89.

In the study of Ghanaian firms, subsidiary man-agers were asked to report their perceptions con-cerning whether corruption pressures posed anobstacle to their firm’s business performance. Thewording of this scale not only provides a comple-mentary perspective to the other dataset, but alsoframes the question in a manner that is less sen-sitive to socially desirable responses. Respondentswere not asked to divulge whether they had actu-ally participated in corrupt transactions, but theextent to which they viewed the country’s corrup-tion environment as an obstacle to their businessactivities. Because these questions do not implythat the MNE did or did not engage in bribery,managers should be less concerned about indirectlyrevealing sensitive information. (See Appendix 1for the scale.)

Unlike the earlier measure, this measure wasframed as a criticism of the country’s businessenvironment. Sometimes managers with bad feel-ings toward a local environment may display a per-ception bias, termed a propensity to “‘kvetch’. . .theYiddish expression for habitual complainer” (Kauf-man and Wei, 1999: 13), which can create system-atic biases. In this dataset, managers of foreignenterprises perceived greater obstacles to businessperformance across a wide range of issues than didlocal, Ghanaian, firms (p < 0.01). This difference

may stem from variations in tendencies to expresscriticism, additional challenges reflecting MNEs’‘liability of foreignness,’ or differences in framesof reference. To account for these potential system-atic biases, the corruption scale was standardizedto account for respondents’ perceptions regardingobstacles created by 17 different issues (such asthe uncertainty about property rights, the impor-tance of family relationships in the business envi-ronment, difficulty understanding local customers’needs, government bureaucracy, and difficulty hir-ing skilled workers). Thus, this construct reflectsthe degree to which corruption creates more or lessof an obstacle to business performance than otherfactors in the Ghanaian environment; Cronbach’salpha was 0.86.

Independent variables

Country-level (level 2) variables

Host country corruption level. The host coun-try corruption environment was measured usingTransparency International’s CPI, which draws on18 separate surveys administered by twelve inde-pendent organizations pertaining to 146 countries,and reflects the perceptions of resident and non-resident managers and country analysts. The CPIasks respondents questions related to the misuseof public office for personal benefit and includesitems related to bribery, kickbacks, embezzlement,and the strength and effectiveness of anticorrup-tion efforts. As it is published, high scores indi-cate a less corrupt national environment; for easeof interpreting our results, we reverse coded andstandardized this variable (using z-scores). Thus,in tables and figures in the next section, highernumbers reflect more corrupt institutional environ-ments. Countries’ scores from 2003 (TransparencyInternational, 2003) were used because severalEastern European countries made their first appear-ance in Transparency International’s CPI in 2003,affording the ability to include a greater numberof countries and observations.3

Home country corruption level. Similarly, anMNE’s home country corruption level was oper-ationalized by the CPI score of its home coun-try. As above, CPI scores were reversed and

3 The correlation between 2000 and 2003 CPI scores (Trans-parency International, 2000, 2003) was 0.93 for the countries inthis dataset.



standardized so that higher numbers reflect morecorrupt environments. Since the BEEPS surveywas administered in 1999–2000, and no additionalobservations were able to be included by usinglater data, the home country CPI variable was oper-ationalized using the CPI scores for 2000. Foreignfirms in the Eastern Europe sample came from:Austria, Belgium, Canada, Denmark, France, Ger-many, Greece, Japan, Russia, Turkey, the UnitedKingdom, and the United States.

Data for the Ghana sample were collected in2004–2007; thus Transparency International’s2004 CPI scores were used (Transparency Inter-national, 2004). This analysis included data onfirms from: Belgium, Botswana, Brazil, Canada,China, Cyprus, Denmark, Finland, France, Ger-many, Guinea, Hong Kong, India, Israel, Italy,Ivory Coast, Japan, Kenya, Korea, Lebanon, Mau-ritania, the Netherlands, Nigeria, Russia, Senegal,Singapore, South Africa, Spain, Sweden, Switzer-land, Syria, Thailand, the United Kingdom, and theUnited States.

Home country entry into force of the OECD con-vention. For the Eastern Europe sample, MNEs’home countries were coded as ‘1’ if they hadentered into the force of the OECD Convention onCombating Bribery of Foreign Public Officials inInternational Business Transactions on or before 1January, 1999, and as 0 otherwise. By 1999, eightof the 12 home countries represented in our samplehad entered into the force of the convention. Forthe Ghana sample, home countries were coded as‘1’ if they had entered into the force of the OECDConvention in or before June, 2007, and as ‘0’otherwise. Sixteen of the 34 home countries repre-sented in our sample had entered into the force ofthe convention by this time, and no country enteredbetween 2004 and August, 2007.

Firm-level (level 1) variables

Partial local ownership. For the Eastern Euro-pean firms, partial local ownership was measuredas a dichotomous variable; firms with any localownership were coded as ‘1’ and those that werefully foreign owned were coded as ‘0.’ MNEs thatreport partial local ownership could fall into twocategories. On the one hand, the firm could havea single local equity partner that plays a role inthe firm’s strategy and operations. On the otherhand, the subsidiary could have more dispersed

local ownership, with controlling interest held bythe MNE, but a noncontrolling portion of sharesheld by local stockholders. Due to data constraints,we were unable to identify the composition of thislocal ownership. We therefore assume that firmsreporting partial local ownership in the EasternEurope sample fall into the first category. To theextent that some firms in the sample have moredispersed local ownership, less significant resultsshould arise in the empirical analysis.

Local alliances. In the Ghana dataset, managerswere asked to report the number of strategicalliances and joint ventures in which their firmhad participated with Ghanaian-owned companies.Since few firms reported more than one localalliance, this variable was dummy coded as ‘1’when an alliance was present and as ‘0’ when noalliances were reported.

Local decision making. In the Ghana dataset,decentralized decision making was operationalizedbased on a three-item scale (see Appendix 1) inwhich subsidiary managers were asked if strategicdecisions were decentralized to the host countrysubsidiary. A factor analysis using varimax rota-tion of survey items with common anchors con-firmed the presence of a decision making factor;Cronbach’s alpha for this scale was 0.70.

Nationality of managing director. In the Ghanadataset, this variable was coded as ‘1’ whenthe managing director was Ghanaian and as ‘0’otherwise.

Control variables

In the Eastern Europe sample, we controlled forindustry, MNE’s worldwide employment (morethan 500 employees, 100–499 employees, 50–99employees) and engagement in international trad-ing activities. The analysis of Ghanaian firmsincluded similar controls including industrydummy variables and subsidiary size (number ofemployees in Ghana).

Although country-level independent variableswere drawn from distinct sources, the depen-dent variables in our analyses came from thesame sources as the firm-level explanatory vari-ables; thus, we assessed the potential for commonmethod bias. First, we noted that the form of thequestions measuring the dependent and moderator



variables was quite different, with the questionsoperationalizing the moderator variables asking forobjective information such as the percentage ofcapital owned by various categories of owners, thenumber of alliances with local firms, and whetherthe subsidiary’s managing director was a home-,host-, or third-country national. In both surveysthere were multiple pages separating these objec-tive questions from the perception-based depen-dent variables. In the Ghana sample, since oneof the moderator variables (local decision mak-ing) was measured on the same Likert scale as thedependent variable, we performed Harman’s sin-gle factor test. We entered all scaled variables andno general factor was apparent; thus the possibil-ity that common method bias drives the empiricalresults is not of great concern.

Analysis

We used hierarchical linear modeling (HLM) forour analysis. HLM is used to predict values ona dependent variable when independent variablesoccur at more than one level so that some indepen-dent variables are embedded within higher levelsof analysis—in our case, firms embedded withincountry contexts. In the Eastern Europe dataset,we have a hierarchical crossed-and-nested designin which each subsidiary is embedded within twoindependent contexts—that of its home countryand that of its host country. Consequently, weemployed a cross-classified random-effect model(HCM2) and thus estimated models using fullmaximum likelihood (FML) estimation of HCM2using the software program HLM 6.02a. Given thatgroup mean centering is not viable in HCM2, alllevel-1 variables entered the equations uncentered;level-2 variables were grand-mean centered. Whenemploying grand mean centering, the country-level coefficients show the relationship betweencountry-level predictors and bribery, absent theinfluence of firm-level predictors. HCM2 mod-els can employ either fixed or random effects.Fixed effects would assume that the relationshipbetween firms’ characteristics and perceived needto bribe is invariant across national contexts. Ran-dom effects do not impose this assumption onthe model. We tested each of our level-2 vari-ables against this assumption and found fixedeffects to be appropriate for the host country CPIand home country OECD convention variables,but not for the home country CPI. Therefore,

we use the random-effects specification for homecountry CPI.

For the Ghana sample, we had only one coun-try context—that of the MNE’s home country.Thus, we estimated models using an HLM2 model.We performed a test to assess whether the modelrequired a full maximum likelihood (FML), ratherthan restricted maximum likelihood (REML), esti-mation given that we had a moderate number oflevel-2 observations, and FML produced a bet-ter fit (p < 0.000). All level-1 variables enteredthe equations group-mean centered, as is preferredin HLM2 models with interaction effects (Hof-mann and Gavin, 1998). All level-2 variables weregrand-mean centered.

We examined our hypotheses related to sub-sidiaries’ embeddedness in home and host countrycontexts using an incremental approach. For eachdataset, we first specified a null model (Model1), that included only the intercept—no firm- orcountry-level variables. Model 2 then added firm-level variables but no country-level variables. Thenext models added country effects, including hostcountry context (for the Eastern Europe sample)and home country context. The last column ofTable 3 and the last two columns of Table 4 pro-vide results for each of the interaction terms.Appendix 2 contains equations for both the maineffects and fully specified models.

RESULTS

Tables 1 and 2 provide descriptive statistics andcorrelation matrices for both samples. We firstassessed normality, heteroskedasticity, the pres-ence of systematic variation across groups, andmulticolinearity. No issues were found in eithersample. In the Ghana sample, the bivariate corre-lation between the two home country variables—home country CPI and home country participationin the OECD Convention—was high at 0.667 (p <

0.05). For this reason, we entered these variablesinto the analysis separately. The home countryvariables did not exhibit a significant correlation(0.371; n.s.) in the Eastern Europe dataset; thus,variables were not separated in the analysis pre-sented here, though robustness checks showed nosubstantial differences when the variables enteredthe analysis separately.

Results for tests of Hypotheses 1, 2a, 2c, and3a using Eastern European firms are presented in



Table 1. Descriptive statistics and correlations for Eastern Europe sample

Mean St. dev. (1) (2) (3) (4) (5) (6) (7) (8)

(1) More than 500 employees 0.03 0.17 1.00(2) 100–499 employees 0.10 0.30 −0.06 1.00(3) 50–99 employees 0.05 0.23 −0.04 −0.08 1.00(4) Partial local ownership 0.34 0.48 0.22 0.35 −0.04 1.00(5) Manufacturing industry 0.06 0.24 0.01 −0.05 0.09 −0.18 1.00(6) Construction industry 0.02 0.15 −0.04 −0.01 0.15 −0.11 −0.04 1.00(7) Mining industry 0.72 0.45 0.09 0.11 0.02 0.22 −0.05 −0.07 1.00(8) Import/export orientation 0.70 0.46 0.05 0.03 0.03 0.17 0.11 0.02 0.02 1.00

Correlations above 0.18 are significant (p < 0.05). Bivariate correlations at country level between home country CPI and homecountry OECD participation are 0.37 (nonsignificant).

Table 2. Descriptive statistics and correlations for Ghana sample

Mean St. dev. (1) (2) (3) (4) (5) (6) (7)

(1) Construction industry 0.03 0.17 1.00(2) Manufacturing industry 0.31 0.46 −0.11 1.00(3) Service industry 0.38 0.49 −0.13 −0.53 1.00(4) Employees in Ghana 118.19 235.84 0.06 0.17 −0.19 1.00(5) Local alliance 0.33 0.47 0.00 −0.01 0.06 −0.01 1.00(6) Localized decision making (LDM) 3.07 0.91 −0.09 −0.17 0.48 −0.18 0.12 1.00(7) Ghanaian managing director (GMD) 0.45 0.50 0.16 −0.10 0.18 0.02 0.10 0.01 1.00

Correlations above 0.48 are significant (p < 0.05). Bivariate correlations at country level between home country CPI and homecountry OECD participation are 0.67 (p < 0.05).

Table 3. Results for tests of Hypotheses 2b, 2d, 3b,4a, and 4b using the Ghana data, are presented inTable 4.

Firm-level main effects

Table 3, Model 2 shows that among the firm-levelcontrol variables in the Eastern Europe sample,only two variables—importing/exporting and oneof the size variables—were marginally related tothe dependent variable (p < 0.10). In the Ghanasample, Table 4, Model 2 shows that only thevariable for Ghanaian managing director emergedas significant (p < 0.05). The negative coefficienthere suggests that subsidiaries with Ghanaian man-aging directors found corruption to be less of anobstacle than other subsidiaries.

Propositions 1 and 2: country-leveldeterminants

Table 3, Model 3 presents results testing hypothe-ses in the Eastern Europe sample. Host countryCPI emerged as a significant main effect (p <

0.001), indicating that subsidiaries operating in

more corrupt host countries faced a larger needto offer bribes. This result supports Hypothesis1. However, neither home country CPI (Hypoth-esis 2a) nor home country participation in theOECD Convention (Hypothesis 2c) emerged assignificant direct effects. Results from similarmodels in the Ghana sample are presented inTable 4, Models 3 and 4. Home country CPIemerged as positive and significant (p < 0.05),indicating that subsidiaries of MNEs from morecorrupt home countries faced greater obstaclesfrom corruption pressures in Ghana and, like-wise, subsidiaries from home countries in whichcorruption was less institutionalized faced lesserobstacles from local corruption pressures; thus,Hypothesis 2b was supported. The OECD con-vention variable emerged as negative and signif-icant (p < 0.01), indicating that subsidiaries ofMNEs from home countries that had entered theforce of the OECD convention faced lesser obsta-cles from corruption pressures in Ghana, andproviding support for Hypothesis 2d. Thus, thegeneral proposition that subsidiaries of MNEsfrom countries in which anticorruption norms



Table 3. Results of HCM2 estimation for need for MNE bribery in Eastern Europe sample

Variable (coefficient) Model 1Intercept

only

Model 2Firm-level

main effects

Model 3Country-levelmain effects

Model 4Local ownership

by home CPIinteraction

Intercept 2.13 (0.16)∗∗∗ 1.87 (0.24)∗∗∗ 1.86 (0.22)∗∗∗ 1.98 (0.22)∗∗∗

Firm-level variablesMore than 500 employees −0.23 (0.31) −0.19 (0.30) −0.18 (0.30)100–499 employees −0.16 (0.21) −0.06 (0.20) −0.06 (0.20)50–99 employees 0.41 (0.24)+ 0.48 (0.23)∗ 0.48 (0.23)∗

Manufacturing industry −0.02 (0.20) 0.01 (0.19) 0.02 (0.19)Construction industry 0.33 (0.32) 0.24 (0.31) 0.25 (0.31)Mining industry 0.69 (0.51) 0.56 (0.50) 0.58 (0.49)Import/export 0.31 (0.17)+ 0.35 (0.17)∗ 0.38 (0.17)∗

Partial local ownership (PLO) 0.00 (0.17) −0.06 (0.17) −0.21(0.19)Country-level variablesHost country CPI 0.47 (0.11)∗∗∗ 0.45 (0.11)∗∗∗

Home OECD participation 0.01 (0.26) 0.07 (0.27)Home country CPI 0.09 (0.16) 0.60 (0.34)∗

Interaction termsPLO X home country CPI −0.53 (0.31)∗

Model fitDeviance (−2 log likelihood) 516.58 503.20 484.49 481.661 Deviance 13.38173+ 18.70652∗∗ 2.83+

+ p < 0.10; ∗ p < 0.05; ∗∗ p < 0.01; ∗∗∗ p < 0.001.

have been institutionalized are insulated from hostcountry corruption pressures was supported in theGhana sample, but not in the Eastern Europesample.

Propositions 3 and 4: localization interactiondeterminants

Hypothesis 3a proposed that partial local own-ership would moderate the relationship betweenan MNE’s home country corruption environmentand the perceived need to engage in briberyin the host country. Table 3, Model 4 presentsresults from the Eastern Europe sample. The par-tial local ownership by home country CPI inter-action term emerged as negative and significant(p < 0.05); thus, Hypothesis 3a was supported.Hypotheses 3b, 4a, and 4b predicted moderatingeffects of alliances, decentralized decision mak-ing, and the use of a local managing director inGhana. Results are presented in Table 4, Mod-els 5, 6, and 7. The alliance by home countryCPI interaction term was a negative and signif-icant predictor (p < 0.05), indicating support forHypothesis 3b. Neither the localized decision mak-ing by home country CPI interaction (Hypoth-esis 4a) nor the managing director by home

country CPI interaction (Hypothesis 4) emergedas significant.4

Figures 1 and 2 illustrate the significant inter-actions for partial local ownership in the East-ern Europe sample and alliances in the Ghanasample. Only the solid black line in each figureemerged as significantly different from zero (p <

0.05) (Preacher, Curran, and Bauer, 2006).

DISCUSSION

Building from recent discussions on the implica-tions of MNEs’ embeddedness in diverse insti-tutional environments (Kraatz and Block, 2008;Phillips and Tracey, 2009), our results support theargument that MNE subsidiaries can and do faceconflicting institutional pressures in their variousnational environments. Indeed, our results suggestthat home and host country institutions interact in

4 Because the BEEPs sample contains a relatively small numberof level-2 (i.e., country-level) observations, we verified theresults using a Score Test I (Berkhof and Snijders, 2001), whichprovides a conservative test of variance components at smallerlevel-2 sample sizes. Results of the Score Test I were consistentwith the findings presented above for both row and columncoefficients (i.e., home and host country effects).



Tabl

e4.

Res

ults

ofH

LM

2es

timat

ion

for

perc

eptio

nth

atco

rrup

tion

isan

obst

acle

inG

hana

sam

ple

Var

iabl

e(c

oeffi

cien

t)M

odel

1In

terc

ept

only

Mod

el2

Firm

-lev

elm

ain

effe

cts

Mod

el3

Hom

eco

untr

yC

PI

Mod

el4

OE

CD

part

icip

atio

n

Mod

el5

Hom

eC

PIX

allia

nce

Mod

el6

Hom

eC

PIX

deci

sion

mak

ing

Mod

el7

Hom

eC

PIX

man

agin

gdi

rect

or

Inte

rcep

t1.

05(0

.04)

∗∗∗

1.08

(0.0

6)∗∗

∗1.

09(0

.05)

∗∗∗

1.10

(0.0

5)∗∗

∗1.

09(0

.05)

∗∗∗

1.09

(0.0

5)∗∗

∗1.

09(0

.05)

∗∗∗

Fir

m-l

evel

vari

able

sC

onst

ruct

ion

ind.

−0.0

8(0

.34)

−0.0

8(0

.35)

−0.0

8(0.

34)

0.09

(0.3

3)−0

.08

(0.3

5)−0

.10

(0.3

6)M

anuf

actu

ring

ind.

−0.0

7(0

.14)

−0.0

7(0

.14)

−0.0

7(0

.14)

−0.0

4(0

.13)

−0.0

7(0

.14)

−0.0

7(0

.14)

Serv

ice

ind.

−0.1

5(0

.15)

−0.1

5(0

.15)

−0.1

5(0

.15)

−0.1

5(0

.14)

−0.1

5(0

.15)

−0.1

5(0

.15)

Em

ploy

ees

inG

hana

0.00

(0.0

0)0.

00(0

.00)

0.00

(0.0

0)0.

00(0

.00)

0.00

(0.0

0)0.

00(0

.00)

Loc

alal

lianc

e0.

06(0

.11)

0.06

(0.1

1)0.

06(0

.11)

−0.0

2(0

.11)

0.06

(0.1

1)0.

05(0

.12)

Loc

aliz

edde

cisi

onm

akin

g(L

DM

)0.

09(0

.07)

0.09

(0.0

7)0.

09(0

.07)

0.08

(0.0

7)0.

09(0

.08)

0.09

(0.0

8)G

hana

ian

man

agin

gdi

rect

or(G

MD

)−0

.27

(0.1

2)∗

−0.2

7(0

.12)

∗−0

.27

(0.1

2)∗

−0.2

4(0

.11)

∗−0

.27

(0.1

2)∗

−0.2

3(0

.21)

Cou

ntry

-lev

elva

riab

les

Hom

eco

untr

yC

PI0.

13(0

.06)

∗0.

13(0

.06)

∗0.

13(0

.06)

∗0.

13(0

.06)

∗

Hom

eO

EC

Dpa

rtic

ipat

ion

−0.3

1(0

.10)

∗∗

Inte

ract

ion

term

sA

llian

ceX

hom

eC

PI−0

.22

(0.1

2)∗

LD

MX

hom

eC

PI0.

00(0

.08)

GM

DX

hom

eC

PI0.

05(0

.24)

Mod

elfit

Dev

ianc

e(-

2lo

glik

elih

ood)

54.4

79.

215.

131.

631.

755.

135.

091

Dev

ianc

eA,B

45.2

6∗∗∗

4.08

∗7.

59∗∗

3.38

+0.

000.

04R

21/R

22,C

0.11

/0.0

00.

26/0

.17

0.33

/0.2

50.

28/0

.16

0.26

/0.1

70.

27/0

.17

+p

<0.

10;

∗p

<0.

05;

∗∗p

<0.

01;

∗∗∗

p<

0.00

1.A1

Dev

ianc

efo

rM

odel

4of

fers

com

pari

son

with

Mod

el2.1

Dev

ianc

efo

rM

odel

s5,

6,an

d7

offe

rco

mpa

riso

nsw

ithM

odel

3.B

The

diff

eren

cebe

twee

nM

odel

s3

and

4is

nons

igni

fican

t.C

Inan

HL

M2

mod

el,

R2

1is

inte

rpre

ted

asth

epr

opor

tiona

lre

duct

ion

iner

ror

for

pred

ictin

gan

indi

vidu

al-l

evel

(lev

el-1

)ou

tcom

e,an

dR

22

isin

terp

rete

das

the

prop

ortio

nal

redu

ctio

nof

erro

rfo

rpr

edic

ting

agr

oup

(lev

el-2

)m

ean

(Luk

e,20

04).



Figure 1. Partial local ownership by home country corruption level

sophisticated ways depending on the MNE’s local-ization strategy.

Recently, Kostova et al. (2008) suggested thatby virtue of their foreignness and distinctive con-tributions, MNEs may be shielded from pressureto conform to host country institutions. However,in this study we found support for our hypothesisthat, at least in the corruption domain, MNE sub-sidiaries indeed encounter pressure to conform tohost country conventions and expectations. Theseempirical findings may stem from our focus on asingle host country constituency (local governmentofficials), rather than a broader set of institutionalactors. They may also arise from our considera-tion of the specific domain of corruption, whichmay be subject to different dynamics than domainsin which the focal activity is not inherently ille-gal. Nevertheless, these results point to the needfor further empirical inquiry into the pressures thatMNE subsidiaries face to conform to their variedinstitutional environments.

Several other interesting results emerged fromthe empirical tests. First, in terms of the directeffect of the home country’s corruption environ-ment on pressures faced by subsidiaries in their

host country, results were mixed—the direct effectwas statistically significant in the Ghana sample,but not the Eastern Europe sample. However, theseresults must be considered in light of interactioneffects. In both samples, when MNEs had no localpartner or local ownership, the home country cor-ruption environment was a significant predictor ofthe pressure the subsidiary faced to engage in cor-rupt practices locally, and in both samples the pres-ence of a local partner eliminated that relationship.

Additionally, in the Ghana sample but not theEastern Europe sample, results suggested thatMNEs from countries participating in the OECDConvention for Combating Bribery faced less localpressure to engage in corruption than MNEs fromnonparticipating countries. The presence of signif-icant results for the OECD convention variablein the Ghana sample but not the Eastern Europesample may reflect a timing issue. The survey ofEastern European firms took place in the middleof the primary ratification time frame for OECDmember countries. Thus, if some countries simplycompleted their legislative procedures and enteredinto the force of the convention more quickly thanothers, then the cut-off date used here may be



Figure 2. Alliance by home country corruption level

problematic. In contrast, data collection for theGhana sample took place several years after coun-tries had entered into the force of the conven-tion; thus, a country’s participation may have heldgreater signaling power to local officials. Together,these home country institutional effects offer somesupport for our arguments that when countries haveinstitutionalized anticorruption norms at home orimplemented legal constraints on their MNEs, theymay establish reputations and send signals thattheir MNEs will resist bribery, even when oper-ating in corrupt host countries.

Kostova et al. (2008: 998) argued that, espe-cially in areas such as social responsibility, MNEsmay comprise their own ‘metainstitutional field.’This suggests that, rather than belonging to orga-nizational fields in each of their home and hostcountries, MNEs should be seen as a ‘recognized’category in themselves with expectations and, thus,demands regarding their behavior, established asa class. If key constituencies recognize MNEsas comprising their own metainstitutional field,we would expect those actors to impose similar

expectations upon all MNEs, regardless of a par-ticular firm’s attributes. The empirical findings inthis paper do not provide strong support for thisnotion in the corruption domain, however. Theconstituency considered here—host governmentofficials—does not appear to expect all MNEsto operate according to shared rules or norms.Instead, we found empirical support for our argu-ments that government officials impose differentlevels of pressure on subsidiaries based on theMNE’s home country institutional environment.

Although our results cast doubt on the notionthat this particular constituency recognizes MNEsas belonging to a common metainstitutional field,it is feasible that such a field forms around a nar-rower set of MNEs with particular characteristics.For example, the fact that an empirical distinc-tion arose in the Ghana sample between MNEscoming from home countries that had implementedthe OECD convention and those that had not sug-gests that a metainstitutional field may form, notaround all MNEs, but around MNEs that arebound together by certain institutional constraints.



In addition, other key constituencies such as con-sumers may view MNEs as belonging to a commonfield.

Our empirical results also suggest that firms donot have equal capabilities to navigate pluralis-tic institutional environments. We found that firmswith local partners and with partial local owner-ship faced greater pressure to engage in corruptpractices than their counterparts without such localrelationships. This suggests that when they expectthat an MNE subsidiary will react negatively to abribe request, local government officials may iden-tify alternative access points into the subsidiary. Alocal alliance partner or local individual or firmthat has an equity stake in the subsidiary wouldserve as an ideal access point for the corrupt offi-cial, since that entity is likely more dependent onthe local institutional environment.

In other words, an MNE’s localization strategymay compromise the firm’s ability to implementsome of the strategies that Kraatz and Block (2008)proposed for firms encountering conflicting insti-tutional pressures. For example, by virtue of itsforeignness, an MNE may be able to effectivelydeny the validity, or escape the jurisdiction, oflocal officials’ demands. Subsidiary managers mayalso be able to convince local officials that theMNE had made such a strong worldwide com-mitment to anticorruption policies that the firmwould withdraw from the host country or incura large expense rather than violate internal poli-cies. Either strategy may allow MNE subsidiariesto stand up to the pressure imposed by host coun-try officials. However, once the MNE has takenon a local partner, the official’s authority becomesmore difficult to deny—particularly when pres-sure is placed on the local partner directly. TheMNE, then, may instead need to pursue strate-gies of compartmentalizing its national identitiesor balancing demands imposed by its various insti-tutional environments.

An alternative interpretation of the empiricalresults is that when MNEs find themselves in situa-tions in which they believe corruption will be aparticular obstacle, they tend to take on a localpartner to either smooth the way or outsourcethe corrupt activities. In this alternative explana-tion, taking a local partner would not cause moredemands for bribes, but would instead serve asa mechanism for the MNE to compartmentalizeits identities in its home and host country envi-ronments. Although in violation of many MNEs’

home country laws and norms, such practiceswould be less transparent to stakeholders, offersubsidiary managers greater deniability, and serveas a form of ceremonial adoption of the MNE’spractices.

The differences in these two explanations shouldbe reflected in which firms choose to take on localowners and alliance partners. If MNEs tend to takeon local partners to deliberately outsource corruptactivities, one would expect that MNEs from lesscorrupt countries, in an effort to maintain a cleanpublic image, would use local partners more oftenthan MNEs from more corrupt countries. How-ever, in the Ghana sample, no differences emergedacross these groups, and in the Eastern Europesample, the opposite was true. Firms with no localownership came from less corrupt home countriesthan firms with partial local ownership (p < 0.05).Although based on a simple analysis, this findingis compatible with Smarzynska and Wei’s (2002)observation that U.S. firms appear averse to jointventures in corrupt countries.

Contrary to our expectations, there was no sig-nificant interaction effect between an MNE’s homecountry corruption environment and either local-ized decision making or use of host country nation-als as managing director. The lack of supportfor these hypotheses may result from the factthat MNEs are increasingly using controls suchas organizational culture (Martinez and Jarillo,1989) to guide the behavior and strategies pur-sued by their managers. Thus, if firms in ourGhana sample use a preponderance of behavioraland cultural controls, then local managers, regard-less of their nationality or formal decision mak-ing authority, should internalize the organization’score values.

The findings in this study have important impli-cations for managers of MNEs. Managers havea strong interest in understanding what kinds ofpressure their firms will likely face to engage incorruption in a given host country environmentand what they might do to reduce that pressure.Although some have argued that the presence ofcorruption may increase firms’ efficiency by allow-ing them to avoid red tape, empirical evidencesuggests that firms that encounter prevalent briberysee greater unpredictability and higher costs inbusiness operations (Kaufmann and Wei, 1999;Wei, 1997). Indeed, in a study across 73 coun-tries, Kaufmann and Wei (1999) found that a firm’s



payment of bribes corresponded with more man-agement time wasted with government bureau-crats and a higher cost of capital. Thus, insteadof facilitating business operations, firms operatingin environments in which corruption is prevalentmay well see lower economic efficiency. In addi-tion, the costs likely go beyond these immediateinefficiencies to include the negative effect suchactions can have on the firm’s image and reputa-tion worldwide.

In particular, MNE managers should be awareof the implications of their localization strate-gies for the image building process in which theyare engaged. Although MNEs may aim to takeon local partners to gain legitimacy and buildsocial capital (Rodriguez et al., 2005), this strategymay also unintentionally increase their exposure tohost country corruption pressure. Therefore, MNEswith strong corporate social responsibility goals ora major concern about legitimacy spillovers mayconsider foregoing other strategic benefits fromtaking on local partners in more corrupt host coun-tries. Indeed, if one subsidiary’s decision to engagein bribery—even indirectly by way of the localpartner—becomes widely known, it could com-promise the ability of the MNE as a whole to builda reputation for standing up against corruption inits other markets. The fact that our post hoc analy-sis indicates that firms from less corrupt countriesseem to engage in fewer alliances in very corrupthost countries indicates that some managers mayrecognize this issue.

For policy makers, the implications are twofold.First, a host government policy that has historicallybeen fairly common in developing countries—thatof encouraging foreign firms to work with localpartners—may actually run counter to an objectiveof reducing corruption in the host country envi-ronment. Second, these empirical results may givepolicy makers a roadmap of where to spend thetight resources that they have devoted to address-ing corruption by helping them identify where cor-ruption is most likely to take place.

As with any study there are certain limitationsthat should be acknowledged. First, it is importantto note that our dependent variables are perceptualin nature. Given the sensitive nature of the depen-dent variable, particularly in the BEEPs dataset, itis conceivable that managers of MNEs from lesscorrupt home countries misrepresent their honestperceptions concerning the need for bribery in thehost country, perhaps out of a fear of indirectly

implicating their firm in wrongdoing. The factthat the partial local ownership variable moderatesthe relationship runs counter to this explanation,however, since there is little reason to believethat managers’ misrepresentations would dependupon their MNE’s strategy to take on a partiallocal owner. In addition, the variable measuredin the Ghana sample was less likely to promptsuch misrepresentations. Managers of MNEs oper-ating in Ghana were asked to report the degree towhich local corruption pressure poses an obsta-cle to their firm’s attempts to increase profits.Because these survey items do not indicate howthe MNE navigates such obstacles (i.e., whetherit actually engages in bribery), managers shouldview these questions as less sensitive. It should benoted that while the indirect wording of the depen-dent variable prevented managers from implicat-ing themselves, it may have led to a differentresponse than if managers were noting how oftentheir firm actually pays bribes. Thus, the data pre-sented here should not be interpreted as an indi-cation of how much a given firm actually engagedin bribery, but rather the pressures they faced todo so.

This paper begins to shed light on how anMNE’s host and home country institutional envi-ronments influence the pressure its subsidiariesface to engage in corruption. It suggests that insome cases, conflicting institutions may accom-modate one another, so that the intensity of insti-tutional pressure imposed by actors in the localenvironment varies in response to the firm’s insti-tutional pressure in other environments. It alsoreveals that the relationship is a complex one,influenced by the localization strategy that thefirm adopts in the host country. Thus, once again,MNEs are caught in the delicate balancing act ofmanaging their diverse institutional environments.

ACKNOWLEDGEMENTS

The authors are grateful to Associate Editor SteveTallman and anonymous SMJ reviewers, whosedetailed comments greatly strengthened the manu-script. The project was supported by grants fromthe National Science Foundation Award # 0551121and the George Washington University Centerfor International Business Education and Research(GW-CIBER).



REFERENCES

Bartlett C, Ghoshal S. 1989. Managing Across Borders:The Transnational Solution . Harvard Business SchoolPress: Boston MA.

Beets SD. 2005. Understanding the demand-side: issuesof international corruption. Journal of Business Ethics57: 65–81.

Berger PL, Luckmann T. 1966. The Social Constructionof Reality . Doubleday: New York.

Berkhof J, Snijders TAB. 2001. Variance componenttesting in multilevel models. Journal of Educationaland Behavioral Statistics 26: 133–152.

Collins J, Uhlenbruck K. 2004. How firms respond togovernment corruption: insights from India. Academyof Management Best Paper Proceedings A1–A6.

Cuervo-Cazurra A. 2006. Who cares about corrup-tion? Journal of International Business Studies 37:807–822.

Dacin MT. 1997. Isomorphism in context: the powerand prescription of institutional norms. Academy ofManagement Journal 40: 76–81.

Deephouse DL. 1999. To be different, or to be thesame? It’s a question (and theory) of strategic balance.Strategic Management Journal 20(2): 147–166.

DiMaggio PJ, Powell WW. 1983. The iron cage revisited:institutional isomorphism and collective rationality inorganizational fields. American Sociological Review48: 147–160.

Gratchev MV. 2001. Forethought: making the most ofcultural differences. Harvard Business Review 79(9):28–29.

Gyimah-Boadi E. 2004. Ghana: a case study. Paperpresented at New Anticorruption Governments: TheChallenge of Delivery Conference co-organized by theGovernment of Kenya, TI-Kenya, and TransparencyInternational. Nairobi, Kenya.

Habib M, Zurawicki L. 2002. Corruption and foreigndirect investment. Journal of International BusinessStudies 33: 291–308.

Hofmann DA, Gavin MB. 1998. Centering decisions inhierarchical linear models: implications for research inorganizations. Journal of Management 24: 623–641.

Jaeger AM. 1983. The transfer of organizational cultureoverseas: an approach to control in the multinationalcorporation. Journal of International Business Studies14: 91–114.

Kaufmann D, Wei SJ. 1999. Does ‘grease money’ speedup the wheels of commerce? NBER Working paper# 7093. National Bureau of Economic Research,Cambridge, MA.

Kogut B. 1993. Learning, or the importance of beinginert: country imprinting and international competi-tion. In Organization Theory and the MultinationalCorporation , Ghoshal S, Westney DE (eds). St. Mar-tin’s Press: New York; 136–154.

Kostova T, Roth K. 2002. Adoption of an organizationalpractice by subsidiaries of multinational corporations:institutional and relational effects. Academy ofManagement Journal 45: 215–233.

Kostova T, Roth K, Dacin MT. 2008. Institutional theoryin the study of multinational corporations: a critique

and new directions. Academy of Management Review33: 994–1006.

Kostova T, Zaheer S. 1999. Organizational legitimacyunder conditions of complexity: the case of themultinational enterprise. Academy of ManagementReview 24: 64–81.

Kraatz M, Block ES. 2008. Organizational implicationsof institutional pluralism. In The Handbook of Orga-nizational Institutionalism , Greenwood R, Oliver C,Sahlin-Andersson K, Suddaby R. (eds). Sage: Lon-don, U.K.; 243–275.

Luke DA. 2004. Multilevel Modeling. Sage: ThousandOaks, CA.

Martinez JI, Jarillo JC. 1989. The evolution of researchon coordination mechanisms in multinational corpo-rations. Journal of International Business Studies 19:489–514.

Martinez JI, Jarillo JC. 1991. Coordination demandsof international strategies. Journal of InternationalBusiness Studies 21: 429–444.

Meyer J, Rowan B. 1977. Institutionalized organizations:formal structure as myth and ceremony. AmericanJournal of Sociology 83: 340–363.

Miller D, Chen M. 1995. Nonconformity in competitiverepertoires. In Academy of Management Proceedings ,Moore D. (ed). Academy of Management: BriarcliffManor, NY; 256–260.

Parsons T. 1960. Structure and Process in ModernSocieties . Collier-MacMillian: London, UK.

Phillips N, Tracey P. 2009. Dialogue: institutional theoryand the MNC. Academy of Management Review 34:169–171.

Pingle R. [CEO, PMD International ]. 2009. Howweak institutions affect firms and managers indeveloping and emerging markets. Presentation madeat George Washington University CIBER Workshopon Succeeding in Emerging and Developing Markets.20 June 2009.

Preacher KJ, Curran PJ, Bauer DJ. 2006. Computationaltools for probing interaction effects in multiplelinear regression, multilevel modeling, and latentcurve analysis. Journal of Educational and BehavioralStatistics 31: 437–448.

Rodriguez P, Uhlenbruck K, Eden L. 2005. Governmentcorruption and the entry strategies of multinationals.Academy of Management Review 30: 383–396.

Scott WR. 1991. The adolescence of institutional theory.Administrative Science Quarterly 32: 493–511.

Smarzynska BK, Wei S. 2002. Corruption and crossborder investment: firm level evidence. WilliamDavidson Working paper #494. DOI: 08.02/wp.

Stinchcombe, AL. 1965. Social structure and organiza-tions. In Handbook of Organizations , March JG (ed).Rand-McNally: Chicago, IL; 142–194.

Svensson J. 2003. Who must pay bribes and how much?Evidence from a cross section of firms. QuarterlyJournal of Economics 118(1): 207–230.

Transparency International. 2000, 2003, 2004. Corruptionperceptions index. Transparency International : BerlinGermany. http://www.transparency.org/ (Accessed 13October 2009).



Transparency International. 2008. Frequently asked ques-tions. http://www.transparency.org/policy research/surveys indices/cpi/2008/faq#general2. (Accessed 13October 2009).

Uhlenbruck K, Rodriguez P, Doh J, Eden L. 2006. Theimpact of corruption on entry strategy: evidence fromtelecommunication projects in emerging economies.Organization Science 17: 402–417.

Wei SJ. 1997. Why is corruption so much more taxingthan tax? Arbitrariness kills. NBER Working paper#6255. National Bureau of Economic Research,Cambridge, MA.

Westney DE. 1993. Institutionalization theory and themultinational corporation. In Organization Theoryand the Multinational Corporation , Ghoshal S, West-ney DE (eds). St. Martin’s Press: New York; 53–76.

Widjajabrata S, Zacchea NM. 2004. International corrup-tion: the republic of Indonesia is strengthening theability of its auditors to battle corruption. Journal ofGovernment Financial Management 53(3): 34–43.



APPENDIX 1: SURVEY MEASURES

Dependent variable in BEEPS surveyof Eastern European firms: (1–6 scale)

How often do firms like yours nowadays need tomake extra, unofficial payments to public officialsfor any of the following?

1. To get connected to public services (electricity,telephone)

2. To get licenses and permits3. To deal with taxes and tax collection4. To gain government contracts5. When dealing with customs/imports6. When dealing with courts7. To influence the content of new laws, decrees,

or regulations.

Dependent variable in surveyof Ghanaian firms

This was measured using a five-point Likert scalewith anchors ‘no obstacle’ and ‘major obstacle.’

Please report whether the following issues createobstacles/problems to your company’s efforts toincrease profits:

1. A tendency for government administrators torequest unofficial payments

2. A tendency for my company’s competitorsto offer unofficial payments to governmentadministrators

3. Corruption among government officials

Decentralized decision making in survey ofGhanaian firms

This was measured using a five-point Likert scalewith anchors from ‘strongly disagree’ to ‘stronglyagree.’

Please indicate the extent of your agreementwith the following statements regarding your com-pany’s strategy.

1. In my company, managers in each countrymake major strategic decisions about operationsin their country.

2. My company gives managers in Ghana author-ity to make decisions about our strategy inGhana.

3. Managers of our Ghanaian subsidiary have dis-cretion to do what they think is best withoutgetting approval from headquarters.

APPENDIX 2 EQUATIONS FROMHIERARCHICAL LINEAR MODELS:

Model for Eastern Europe dataset (with firmand country main effects):

Level-1 modelY = P0 + P1 × (partial local ownership) +

P2 × (manufacturing industry) + P3 × (construc-tion industry) + P4 × (mining industry) + P5× (import/export orientation) + P6 × (500 +employees) + P7 × (100–499 employees) + P8× (50–99 employees) + e

Level-2 modelP0 = theta(0) + b00 + c00

+ (G01 + c01) × CPI home country+ (G02) × Home country OECD participation+ (B01) × CPI host country

P1 = theta(1)P2 = theta(2)P3 = theta(3)P4 = theta(4)P5 = theta(5)P6 = theta(6)P7 = theta(7)P8 = theta(8)

Model for Eastern Europe dataset (full modelspecification for local ownership interaction)

Level-1modelY = P0 + P1 × (partial local ownership) +

P2 × (manufacturing industry) + P3 × (construc-tion industry) + P4 × (mining industry) + P5× (import/export orientation) + P6 × (500 +employees) + P7 × (100–499 employees) + P8× (50–99 employees) + e

Level-2 modelP0 = theta(0) + b00 + c00

+ (G01 + c01) × CPI home country+ (G02) × Home country OECD participation+ (B01) × CPI host country

P1 = theta(1) + (G11) × CPI homecountry

P2 = theta(2)P3 = theta(3)P4 = theta(4)P5 = theta(5)P6 = theta(6)P7 = theta(7)P8 = theta(8)



Model for Ghana dataset (with firm-level maineffects and home country CPI main effect)

Level-1 modelY = B0 + B1 × (localized decision making) +

B2 × (employees in Ghana) + B3 × (constructionindustry) + B4 × (manufacturing industry) + B5× (service industry) + B6 × (Ghanaian managingdirector) + B7 × (local alliance) + R

Level-2 modelB0 = G00 + G01 × (Home country CPI)

+ U0B1 = G10B2 = G20B3 = G30B4 = G40B5 = G50B6 = G60B7 = G70

Model for Ghana dataset (full modelspecification for alliance interaction)

Y = B0 + B1 × (localized decision making) +B2 × (employees in Ghana) + B3 × (constructionindustry) + B4 × (manufacturing industry) + B5× (service industry) + B6 × (Ghanaian managingdirector) + B7 × (local alliance) + R

Level-2 modelB0 = G00 + G01 × (Home country CPI)

+ U0B1 = G10B2 = G20B3 = G30B4 = G40B5 = G50B6 = G60B7 = G70 + G71 × (Home country CPI)


MNES AND CORRUPTION: THE IMPACT OF NATIONAL … · 2017-09-09 · MNEs and Host Country Corruption...

Documents

Transcript of MNES AND CORRUPTION: THE IMPACT OF NATIONAL … · 2017-09-09 · MNEs and Host Country Corruption...