20 HRMpractices inMNCs

8/2/2019 20 HRMpractices inMNCs

1/23

INFLUENCES ON HUMAN RESOURCE MANAGEMENTPRACTICES IN MULTINATIONALCORPORATIONSPhilip M. Rosenzweig*and Nitin Nohria**

Harvard Business SchoolAbstract. A study of human resource management practices in249 U.S. affi'liates of foreign-based multinational corporations(MNCs) shows that in general affi'liate HRM practices closelyfollow local practices, with differences among specific practices.The degree of similarity to local practices is significantly influencedby the method of founding, dependence on local inputs, thepresence of expatriates, and the extent of communication withthe parent. In addition, sharp differences are revealed amongaffiliates of Canadian, Japanese and European MNCs, suggestingstrong country effects. Together, these findings support the viewof MNCs as composed of differentiated practices, which in turnare shaped by forces for local isomorphism and for internalconsistency.

One of the central questions in the literatureon MNCs is the extent to whichtheir various foreign affiliates (or subsidiaries) act and behave as local firmsversus the extent to which their practices resemble those of the parentcorporationor some other global standard. In a seminal paper on this topic,Perlmutter [1969] described three types of MNCs: ethnocentric, polycentricand global. According to this typology, the managementpractices in foreignaffiliates of MNCs could resemble those of the MNC's home country(ethno-centric), could conform to local practices of the affiliate's host country(polycentric), or could adhere to a worldwide standard (global).

*PhilipM. Rosenzweigis AssistantProfessorat theHarvardBusiness School. Hisresearch nterests enteron themanagement f multinationalirmsand haverecentlyfocused on foreign-based irms operating n the United States. Publishedarticleshave appearedn Management cience,CaliforniaManagementReview,Academyof ManagementReview,andEuropeanManagement ournal.**Nitin Nohria is Associate Professorat the HarvardBusiness School. He hasextensivelystudiedthe managementof MNCs and has publishedarticles on thesubjectin StrategicManagementReviewand Sloan ManagementReview. He hascoauthoredwo books:Beyond heHype:Rediscoveringhe Essenceof ManagementandNetworksandOrganization, othpublishedby HarvardBusiness SchoolPress.Support or this researchwas generouslyprovidedby the Division of Research,HarvardBusinessSchool. The authorsextendtheirthanks o Mike Stevenson,PhilipHamiltonand Steve Andrews ortheir contributionso thisstudy.Received:July 1992;Revised:July& November1993; Accepted:November 1993.

229


2/23

230 JOURNAL OF INTERNATIONALBUSINESS STUDIES, SECOND QUARTER 1994

Perlmutter's typology described the management practices of the MNC interms of an overall orientation. More recently, some scholars have arguedthat viewing multinationals in terms of an overall orientation obscures theinternal differentiation of management practices within an MNC [Hedlund1986; Bartlett and Ghoshal 1989; Ghoshal and Nohria 1989]. Instead, theyargue, an MNC is properly viewed as a nexus of differentiated practices.According to this view, MNC affiliates are composed of many separatepractices, ranging from manufacturingto finance to human resources, eachof which faces distinct pressuresfor global efficiency andfor local responsive-ness [Doz, Bartlettand Prahalad1981; Prahaladand Doz 1987]. In organizationtheoretic terms, the profile of management practices in an MNC is shapedby the interplay of opposing pressures for internal consistency and forisomorphism with the local institutional environment,with specific practicesshaped by these opposing pressures to varying degrees [Rosenzweig andSingh 1991]. As a consequence, Perlmutter's three-parttypology may notbe sufficient to describe the complexity within MNC affiliates. As Soenenand Van den Bulcke [1988] found in their study of Belgian firms, someaffiliate practices may tend more closely to resemble the MNC's home-country practices, while others may more closely resemble host-countrypractices, and still others may be global in nature.In this study, we examine the forces that shape HRM practices in MNCaffiliates. HRM practices, no less than others, as Evans and Lorange [1989]have noted, are subject to the dual pressuresfor local adaptationand internalconsistency. Accordingly,we startby examiningthe resemblanceof a numberof affiliate HRM practices to local practices and to parent practices. Fromthatbaseline, we consider a set of relatedhypotheses, testing whether factorssuch as the extent to which the affiliate is embeddedin the local environment,the extent of flows between the parentand the affiliate, the characteristicsof the parent, and the nature of the industry, influence HRM practices.Finally, we examine differences among affiliates according to their parentnationality.Ourguiding objective is to providea more thoroughunderstandingof the forces that influence human resource management in MNCs, and,more generally, to shed light on the determinantsof management practicesin MNCs.

HUMAN RESOURCEPRACTICESIN MNCS:THEORY AND HYPOTHESESOf the two logics that shape HRM practices in MNCs, which is stronger:the pressure for internal consistency or the pressurefor local isomorphism?One view argues thatthe pressurefor internalconsistency is stronger.Becauseemployees may move among subunits, MNCs seek to develop a consistentapproach to compensation and benefits, and to maintain consistent policiesand procedures. Even in the absence of extensive employee mobility acrossborders, MNCs may wish for reasons of internalequity to adopt a consistent


3/23

HRM PRACTICESIN MNCS 231

set of HRM practices around the world. Consequently, we should expectaffiliate HRM practices to be shaped mainly by the pressure for internalconsistency.Counteractingthe pressurefor internalconsistency is a pressurefor affiliateHRM practices to resemble local practices. Some practices may have tocomply with local laws and regulations, and may thereforebe compelled toresemble local practices through "coercive isomorphism" [DiMaggio andPowell 1983]. In other instances, affiliates may seek to resemble local finns,perhaps to compete more effectively in local labor markets, or to "fit in"by imitating local practices. Adopting local practices for these reasons maybe explained as "normative isomorphism" and "mimetic isomorphism,"respectively [DiMaggio and Powell 1983]. These pressures, separately orin combination, would lead MNC affiliates to adhere to HRM practicesfound in the local environment rather than those of the foreign parent.While theoretically it can be argued that HRM practices are shaped byequally compelling counteracting logics, previous empirical research hastended to show that of all functions, human resource management tendsmost closely to adhereto local practices,whereas finance and manufacturingtend more closely to resemble parentpractices (see, for example, Kobayashi[1982]). This is because HRM practicesare often mandatedby local regulationor shaped by strong local conventions. Moreover, since an MNC affiliate,in most instances,has little choice but to hire its employees from competitivelocal labor markets, it is hard for it to diverge too much from local norms.Finally, as indicatedby recent concems aboutthe rising levels of employmentof U.S. nationals in foreign-owned firms (e.g., Fucini and Fucini [1990]),HRM practices can be politically sensitive and affiliates may prefer toconform to local practices ratherthan invite unnecessary scrutiny. Thus ourfirst hypothesis is that the overall tendency of HRM practices to moreclosely resemble local practices, as reported in earlier studies, will hold inour data as well:

Hypothesis 1: Overall, affiliate HRM practices will more closelyresemble those of local competitors than those ofthe parent in its home country.While we have hypothesized that in general affiliate HRM practices willmore closely resemble those of local competitors, we expect this tendencyto vary across HRM practices. In keeping with the precepts of institutionaltheory, practices that are mandated by local laws, or are highly visible, oraffect a large number of locals, or entail a high degree of interaction withlocals, will face stronger pressures to conform to local practices [Meyer andRowan 1977; DiMaggio and Powell 1983]. Conversely, practices that aremost likely to raise concerns of internalequity or that involve a high degreeof interaction with the parent will face stronger pressures to conform toparent practices. To explore this variation, we examined six different HRMpractices.


4/23

232 JOURNALOF INTERNATIONALBUSINESS STUDIES, SECOND QUARTER 1994

Walton and Lawrence [1985] identified four major areas of HRM policy:reward systems, including compensation and benefits; employee influencemechanisms such as participation;job design and work organization; andemployee selection and development. Seeking to capture a number of theseelements, we identifiedsix HRM practicesfor study:(i) the extentof employeebenefits (Benefits), (ii) the extent of annualpaid time off (Time Off), (iii) theuse of bonuses to compensate managers (Executive Bonus), (iv) the degreeof participation in executive decisionmaking (Participation), (v) gendercomposition of management (Gender Composition), and (vi) the amount ofemployee training (Training).For these six HRM practiceswe offer a hypothesis followed by our underlyingrationale:

Hypothesis 2: HRM practiceswill tendto conformto local practicesin the following order,startingwith the one that willmost closely resemble local practice: (1) Time Off;(2) Benefits; (3) Gender Composition; (4) Training;(5) Executive Bonus; and,(6) Participation.

We expect Time Off to most closely resemble local practices because normsabout this practiceare often set by local legislationor determinedby collectivebargaining agreements. Thus an MNC affiliate often has very limited choicebut to offer at least as much Time Off as its local competitors. And thoughit has the option of offering more Time Off than is customary in the localenvironment, it is unlikely that a profit-oriented firm would do so.Most of these arguments also apply to the practice of employee benefits. Inthe case of Benefits, however, local firms usually have greater discretion,giving the MNC more room to choose. Also, when it comes to benefits,there are at least some, especially long-term benefits such as retirementandpension plans, on which the MNC is likely to want to maintain internalconsistency and equity to the extent that is possible.When it comes to Gender Composition, local norms tend to be even morediffuse than is the case with Benefits. While there certainly are laws in-somecountries against discrimination on the base of gender, these laws are hardto enforce. Also, in some countries it is socially and politically correct tohave broad-based participation by women in the workforce, whereas inothers, women are still excluded from participatingin certain jobs. In mostsituations, an MNC affiliate is likely to try and adhere to local norms. Yetparentpractice will continue to serve as an importantinfluence because theparentmay be uncomfortablewith placingwomen in positionsof responsibility,especially if they are not used to that in their own environment. Moreover,even if the parent wanted to change its practice to more closely resemblelocal practice, it may take some time to do so.With respect to Training, firms again will feel pressures in both directions.Trainingneedsof differentworkforcesmay vatydependinguponlocalconditions.


5/23

HRM PRACTICES IN MNCS 233

Also, in order to compete effectively, MNC affiliates will need to provideopportunities for their employees to enhance their human capital that arecomparable to what local competitorsoffer. Both these considerationscreatepressures to conform to local practices. However, there are countervailingforces because many fims have corporate-widepolicies regardingemployeetraining,as well as managementphilosophiesthat favorcertan types of training,creating pressuresfor their affiliates to adhere to parent practice.MNC affiliates are less likely to conform to the practices of their localcompetitors when it comes to the extent to which bonuses are used inexecutive compensation (Executive Bonus). This is because there are likelyto be greater concerns of internalequity when it comes to the compensationof senior executives. Executives are also the class of employees most likelyto be transferredor rotated across the various affiliates of the MNC, makingit more important to maintain consistency across the MNC. Of course, anMNC affiliatewill face some pressureto confonn to local practices,especiallyif it wishes to recruittalented executives in a competitive local labor market.But here too, as in the case of Training and Gender Composition, localnorms tend to be diffuse and broad, leaving the MNC with greaterflexibilityto respond to the need for internal consistency.With respect to the extent of Participation in executive decisionmaking, thepressure to conform to local practices is likely to be the weakest. No doubtif the MNC is operating in an environment where executives are used tohaving a great deal of voice in key decisions, they must to some extentrespond to this expectation in order to attract and keep satisfied the bestlocal talent. But executive decisionmaking is not an area in which the MNCis likely to deviate too much from its parent practice because it speaks tothe most critical processes of strategic choice and resource allocation, topicsabout which the parent is likely to attach a great deal of importance, andon which it often retains direct influence.In sum, HRM practices for which there are well-defined local norms andwhich affect the rank-and-file of the affiliate organization are likely toconform most closely to the practices of local competitors. Practices forwhich there are diffuse and poorly defined local norms, or which are seenas being critical to maintaining internal consistency or arriving at criticaldecisions, are less likely to conform to local norms.The foregoing discussion has hypothesizeddifferencesamong HRM practices,but has treated all affiliates alike. In fact, not all affiliates should be expectedto adhere equally to local practices; some may face stronger pressures toconform to local practices while others face strongerpressuresto conform toparentpractices.Affiliatesarelikely to varysystematically n the degreeto whichthey resemble local practicesaccording to specific contextualvariables.Thesevariablescan be conceptualizedas shown in Figure 1.One set of contextual factors that can influence an affiliate's conformanceto local practices is the extent to which the affiliate is embedded in the local


6/23

234 JOURNALOF INTERNATIONALBUSINESS STUDIES, SECONDQUARTER 1994

environment. The extent of local embeddedness depends upon factors suchas the history of the affiliate including its method of founding (whethergreenfield or by acquisition) and its age. Also importantare factors such asthe size of the affiliate, the extent to which it is dependent upon local inputs,and the extent to which it is subject to pressure from local institutions suchas unions and governmental bodies. We would expect that the greater theextent to which the affiliate is embedded in the local environment,the moreit will take on the practices that prevail locally.A second set of factors that can influence an affiliate's conformance to localpractices are the flows of various resources between the parent and theaffiliate, such as capital, informationand technical experts.We would expectthat the greater these flows, the stronger will be the influence of the parent,and the less the affiliate will conform to local practices.Characteristics of the parent are a third set of factors that will affect theextent to which an affiliate will conform to local HRM practices. One suchfactor is the culture of the country in which the parent is located, and howthat differs from the culture in which the affiliate is embedded. The greaterthe cultural distance between the parentand affiliate countries, the less theaffiliate will conform to local practices. On the other hand, the greater theparent's internationalexperience and exposure to operatingabroad, the morelikely it will be to adopt a cosmopolitan attitude and let its affiliates adoptlocal practices.Another parentcharacteristic hatis important s its orientationtowards control. As Egelhoff [1988] has observed, some MNC parentscontrol their subsidiaries more tightly thanothers. We would expect thatthemore controlling the parent, the less likely it will be that affiliate practiceswill resemble local practices.Finally, the nature of the industry in which the affiliate operates can alsohave a bearing on the extent to which its practices resemble those of localcompetitors. One would expect, for instance, that the pressure to conformto local norms would be greater if the affiliate were doing business in amultidomestic industryas opposed to a global industry [Porter1986; Prahaladand Doz 1987].In order to test some of these theoretical propositions we investigated anumber of hypotheses, as discussed below.Factors Related to Local EnmbeddednessMethod of Founding. Affiliates are founded as greenfield investments orthroughacquisition.Affiliates founded as greenfieldinvestmentsare typicallyfounded by MNC employees who seek to replicate key featuresof the parentcompany [Brooke and Remmers 1970; Bartlett and Ghoshal 1989]. On theother hand, affiliates founded as independent firms and only later acquiredby the MNC may remain relatively similar to local firms. Entry strategytherefore serves as a proxy for the effects of organizational imprinting


7/23


FIGURE 1Contextual Factors that Affect Affiliate HRMPractices

{ mbeddedness f ) CharacteristicsLocalAffliateX~~~~~~~* People* Method f Founding I apital * ParentCountry ulture? Age 9~~~~~~nformation * InternationalxperienceAize * ControlOrientation* Dependenceon LocalInputs Nature fBusiness* Unionization * Global* LocalRegulations * Multidomestic

[Stinchcombe 1965] and organizational inertia [Hannanand Freeman 1977,1984], as MNC affiliates initially molded in the image of their parent areexpected to retain that similarity in subsequent years. The hypothesis canbe stated as follows:Hypothesis 3: Resemblance to local HRM practices is greater foracquired affiliates than for greenfield affiliates.

Affiliate Age. The effect of imprinting, even if it is initially strong, may beexpected to diminish over time, as affiliates increasingly come to resemblethe local environment. We can thereforehypothesize a relationshipbetweenthe resemblance to local HRM practices and time since founding:Hypothesis 4: Resemblance to local HRM practices is positivelyrelated to the age of the affiliate.

Affiliate Size. The effect of an affiliate's size on its adherence to local HRMpractices is indeterminate. On one hand, large affiliates may be better able,given their resources, to withstand pressure to embrace local practices. Onthe otherhand, largeraffiliatesmay find themselves more in need of resourcesfrom the local environment, and may be more dependent upon hiring localemployees, placing them under greater pressure to adhere to local practices.We will test the effect of affiliate size, measuredby the numberof employees,as follows:Hypothesis 5: Resemblance to local HRM practices is positivelyrelated to the size of the affiliate.

Local Resource Dependency. Affiliates receive inputs, such as raw materialsand intermediate parts, from other units of the MNC, as well as from thelocal environment.If an affiliate depends heavily upon the local environmentfor needed inputs, it may be more likely to adopt management practicescharacteristic of the local environment.


8/23


Hypothesis 6: Resemblance to local HRM practices is positivelyrelated to the affiliate's dependenceuponlocal inputs.Union Representation.Affiliate HRM practicesmay be affected by a numberof local institutions. Of these, one of the strongest influences may comefrom labor unions, which can force an affiliate to conformto specific practices.We would therefore expect that if an affiliate's employees are representedby a union, its HRM practices may be very close, or even identical, to thoseof other local firms.

Hypothesis 7: Resemblance to local HRM practices is greater foraffiliates with local union representation.Regulatory and Other Local Pressures. Affiliate HRM practices may beaffected by a numberof local institutionsotherthanunions, includinggovern-ment regulations, trade associations, consumer interest groups and othercommunity organizations.

Hypothesis 8: Resemblance to local HRM practices is positivelyrelated to the degree of pressureplaced on affiliatesby local institutions.

Factors Related to Parent-Affiliate FlowsPresence of Expatriates.MNC affiliatesarecomposedof host-countrynationalsand expatriates.Researchershave foundthat expatriatesoften act as "carriers"of culture in MNCs [Edstrom and Galbraith 1977; Lincoln, Olson andHanada 1978], tending to introduce in the affiliate some features of theparent country culture. Accordingly, we would expect MNC affiliates witha high presence of expatriates to tend relatively more to resemble theirparent firns, and conversely, for affiliates composed almost exclusively ofhost-countrycitizens to adhereclosely to the practicesof the local environment.

Hypothesis 9: Resemblance to local HRM practices is negativelyrelated to the presence of expatriates.Dependence on the Parent. MNC affiliates may be more or less dependenton the parent for various types of inputs, which include different types ofknow-how [Rugman 1979; Guptaand Govindarajan1991]. We would expectaffiliates that are more reliant on the parent for technical and managerialknow-how to be more heavily influenced by parentpractices, and hence lessclosely to resemble local firms.

Hypothesis 10: Resemblance to local HRM practices is negativelyrelated o thefrequencywith whichtheaffiliatedependsupontheparent or technicalandmanagerialknow-how.

Communication with the Parent. The executives of some affiliates may befrequently in communication with members of the parent organization,whereas others may tend to communicate less often. We would expect thatthe more frequent the communication, the more likely it is that affiliateexecutives will be influenced by parentpractices[Bartlettand Ghoshal 1989].


9/23


Hypothesis 11: Resemblance to local HRM practices is negativelyrelated to the frequency with which the members ofthe affiliate communicate with the parent.

Factors Related to Parent CharacteristicsParent Country Culture. A substantial body of research has shown thatmanagement practices are influenced by national culture [Hofstede 1980;Laurent 1983; D'Iribarne 1989]. Similarly, the culture of an MNC parentmay affect the management of its foreign subsidiaries, as has been shownin the case of Japanese MNCs [Tsurumi 1986] and Swedish MNCs[Hedlund and Aman 1984]. Just as national culture has been shown to havea signiflcant effect on entry strategy [Kogut and Singh 1988], we mightexpect a similar effect on the affiliate's management practices subsequentto entry.Hofstede [19801 developed four indices that operationalized dimensions ofnational culture.Following Kogut andSingh [1988], we have used a compositeindex of these four dimensions as a proxy for the overall distance betweenthe parentcountry and the host-country culture (Culture Distance). Becauseaffiliates face a dual pressure to resemble both local practices and parentpractices,we hypothesize that when CultureDistance is high, affiliate practiceswill be less like those in the local environment.

Hypothesis 12: Resemblance to local HRM practices is negativelyrelated to the Culture Distance between the parentcountry and the host country.

International Experience of Parent. As Franko [1976] and Bartlett andGhoshal [1989] have noted, an MNC thatestablishes an affiliate abroadmayat first seek simply to replicate its home country practices. Stopford andWells [1972] suggest, on the other hand, that with greater internationalexperience,MNC parentsmay develop a more cosmopolitan andless parochialperspective. Such an attitude, we hypothesize, may well lead to a greateracceptance on the part of the parent for affiliates to adopt practices thatresemble those of their local competitors.

Hypothesis 13: Resemblance to local HRM practices is positivelyrelated to the extent to which the parent has interna-tional experience.ControlOrientationof the Parent. A considerablebody of literaturedocumentsthe differences among MNCs in terms of the control orientation of theparent (see Egelhoff [1988] for an extensive review). The headquarters ofsome MNCs exercise much tighter control over their subsidiariesthan others,thereby influencing more directly decisionmaking in strategy, marketingandotherfunctions including humanresourcemanagement. We would expect thataffiliates with greater autonomy are more likely to adopt local practices.Conversely, we would expect that if the parent exercised tight control over


10/23


the affiliate,perhaps akingan active part in decisionmakingor having arigid set of formal rules andguidelines,the affiliatewill be less likely toadoptlocal practices.Hypothesis14: Resemblance o local HRMpractices s negativelyrelatedto the extentto which the parentexercisestightcontrolover the affiliate.

Factors Relatedto the Natureof the BusinessGlobalor MultidomesticndustryScope.The distinction hatPorter 1986]has drawnbetween global and multidomestic ndustrieshas an importantbearingon whatwe mightexpectwith respect o an affiliate'smanagementpractices.In global industries,where demand s commonacrosscountries,where herearestrong orces orworldwidentegration,ndwherecompetitionis amongglobalplayers,we wouldexpectthe affiliateto be less concernedwithfollowinglocal custom.Conversely,n multidomesticndustries,wherenationalresponsiveness s key and whereone must confrontstronglocalcompetitors, t is more likely that an affiliatewill embrace ocal practices[Prahalad nd Doz 1987].

Hypothesis15: Resemblance o local HRM practices s negativelyrelatedto the extentto which the affiliateoperatesin a global industry.RESEARCHDESIGN AND METHODOLOGY

Testing these hypothesesrequiresan instrument hat can measureintra-organizationalphenomena n a large sample of MNC affiliates.Archivaldatararelycapture ntra-organizationalractices,andtypicallydo not offerany insightinto the relativesimilarityof affiliatepractices,eitherto MNCparent ractices, rto localpractices. ield-based esearch, asedoninterviewswithmanagersat theaffiliatesite, wouldbe exceedinglydifficult o conducton the scale requiredo test thesehypotheses.The research nstrument estsuited to this design is a questionnairehat can gathercomparativedataaboutspeciflc practices n a largenumberof firms.Since this research esigncalledforcomparativeudgments,t wasimperativethat the instrument sk for comparisons hatrespondents ould makewitha high degreeof accuracy.We solicited hecounselof fouraffiliatemanagers,who agreedto serve as pilots. Based on their comments,we alteredthescales of some questions,droppedsome questionsentirely,and amendedthe wordingof others.Thefinal questionnairewas deemedby ourpilots topose questions hataffiliategeneralmanagers houldbe able to answerwithaccuracy.Foroursample,we examinedaffiliates n the UnitedStatesof foreign-basedMNCs. This samplewas ideal for several reasons: t was abundant,t waslocated nearby,and it was complex enoughto providevariationon each


11/23


independent variable of interest. An initial list of U.S. companies with aforeign-based parent was developed from MacMillan & Co.'s CorporateAffiliationsdatabase.We reducedthe list by keeping only the most abundantlyrepresented industries and parent countries, so as to permit meaningfulcomparisons across these variables. The final population included 1,055affiliates in ten 2-digit SIC industries, ranging from food processing toelectronics to wholesale trade,and of eight parentcountries:Canada, France,Germany, Japan, the Netherlands, Sweden, Switzerland and the UnitedKingdom. Surveys were sent to the top officer of each affiliate.' We received264 completed questionnaires for a 23.6% response rate. Of these, fully 249were complete and deemed to be usable for purposesof analysis. The compo-sition of these respondents and rates of response are provided in Table 1.As noted earlier, we identified six HRM practices for study: the extent ofemployee benefits (Benefits), the extent of annual paid time off (Time Off),the use of bonuses to compensateexecutives (Executive Bonus), the degree ofparticipation n executive decisionmaking(Participation),gender compositionof management (Gender Composition),and the amountof employee training(Training). For each of these practices we posed three questions. First, weasked respondents to identify the affiliate's practice on a 5-point scale(Affiliate Practice). Second, we asked for a comparison of the affiliate'spractice to that of the parent in its home country, using a 5-point scaleranging from "Much Less" to "Much More." This variable was calledComparison to Parent. Third, we asked for a comparison of the affiliate'spractice to the typical practice of firms with which it competes in the localenvironment. This variable used the same 5-point scale and was calledComparison to Local. Based on responses to these questions, we were ableto determine the relative similarity of a given practice both to the parentand to local competitors.In order to measure differences among practices, we undertook a criticaltransformation. The need for this transformation is illustrated by looking atone practice: annual days of paid time off to new employees (Time Off).When asked to compare their Time Off policy with that of their parent,68.9% of affiliates reported that they offer either Much Less (73) or Less(91) than theirparent.The mean reply for Comparison to Parent was 2.09 ona scale from 1 (MuchLess) to 5 (Much More). Comnparisono Local provideda very different picture: fully 200 of 239 affiliates (83.7%) reported thattheir Time Off policy was "Same" as local firms, producinga mean score of3.06. Overall, affiliates reportedoverwhelmingly that their TimeOff practiceswere very close to local practice, and were frequently very different fromparent practice.2That affiliate TimeOff practices are substantiallydifferent from parent TimeOff is accurately captured by the mean score of Comparison to Parent(2.09), but only because the great majority of replies were to one side of"Same." In other situations, reliance on the mean of comparison scores


12/23

240 JOURNALOF INTERNATIONALBUSINESS STUDIES, SECONDQUARTER 1994

TABLE 1Distribution of Respondentsa. Distribution of Respondents by Country

Population Respondent Rate ofDistribution Distribution Response (%)Canada 136 28 20.6France 66 15 22.7Germany 145 51 35.2Japan 262 55 21.0Netherlands 61 10 16.4Sweden 68 16 23.5Switzerland 46 18 39.1Britain 271 56 20.7Total 1055 249 23.6 (average)

b: Distribution of Respondents by IndustryPopulation Respondent Rate ofSIC Description Distribution Distribution Response (%)

20 Food Products 57 11 19.327 Printingand Publishing 60 15 25.033 PrimaryMetals 49 15 30.634 Fabricated Metals 54 17 31.435 IndustrialEquipmentand 168 39 23.2Computers36 ElectricalEquipmentand 134 30 22.3Electronics37 TransportationEquipment 56 10 17.838 Instruments 83 22 26.550 Wholesale Trade-Durables 315 71 22.560 Depository Institutions 79 19 24.0Total 1055 249 23.6 (average)

could be misleading: one reply of "More" (4) and one reply of "Less" (2)produces the same mean [(4+2)/2=3] as two replies of "Same" (3)[(3+3)/2=3]. To assess more accurately the relative similarity of affiliatepractice to parentand local practices, we transformedComparison to Localand Comparison to Parent as follows: "Same" was scored as 1; "Less"or "More" as 2; and "Much Less" or "Much More" as 3. These transformnedvariables were called TransformedDifferencefrom Local and TransformedDifference from Parent, respectively. They could range from a low of 1(identical) to a high of 3 (very different), and provided a way to capturethesimilarity to or difference from parentand local practices.In addition to our 5-point scale questions regarding management practices,we also posed a number of questions that served as independent variables.These included such things as method and date of founding, number ofexpatriates,presence of a union, locus of decisionmakingauthority n various


13/23


functions, and many others. The survey thus provided data for most of ourindependent variables. In a few cases, as will be outlined below, we lookedto other sources of data to operationalize our independent variables, as withnational cultureand whether the industryin which the affiliate operated wasglobal or multidomestic.

FINDINGSTesting Hypotheses I and 2: Evidence of Local IsomorphismThe transformedmeasuresof differencefor each of the six HRM practicesareshown in Table 2. Taking, for example, Time Off, we note that TransformedDifferencefrom Parent has a score of 2.09, whereas Transformed Differencefrom Local has a score of 1.16. The difference between these scores ispositive (2.09-1.16=.93), meaning that affiliate Time Off practices muchmore closely resemble those of local firms than those of the parent. Thethird column shows the probabilityat which we can reject the null hypothesisthat the difference between scores is zero. For Time Off, the differencebetween scores is .93, and the null hypothesis can be rejected at p


14/23


TABLE 2Transformed Differences:Difference from Parent and Difference from LocalN Mean Probability

Time OffDifference from Parent 219 2.09Difference from Local 220 1.16.93 .0001 *BenefitsDifference from Parent 215 1.80Difference from Local 212 1.30.49 .0001 *

Gender CompositionDifference from Parent 212 1.83Difference from Local 203 1.38.45 .0001TrainingDifference from Parent 212 1.75Difference from Local 208 1.54.21 .0001Executive BonusDifference from Parent 206 1.68Difference from Local 199 1.64.05 .3844ParticipationDifference from Parent 220 1.76Difference from Local 191 1.74.02 .8653Average of HRM PracticesDifference from Parent 229 1.83Difference from Local 226 1 45.38 .0001****p


15/23


TABLE 3The Effect of Contextual Factors on the Extent to Which AffiliateHRM Practices Resemble Those of Local CompetitorsDirection f SignificanceofContextualFactors HypothesizedEffect ANOVAResults

I. Local EmbeddednessHyp3: Foundedby Acquisition Positive .02*Hyp4: Age Positive .42Hyp5 Size Positive .71Hyp6: LocalDependence Positive .05*Hyp7: Unionization Positive .06+Hyp8: RegulatoryPressures Positive .56II.FlowsHyp9: Presence of Expatriates Negative .01**Hyp10: Resource/Know-how Negative .12Flows fromParentHyp11: Communication lows Negative .01**III.Parent CharacteristicsHyp12: CulturalDistance romAffiliate Negative .09+Hyp13 Internationalxperience Positive .58Hyp14: ControlOrientation Negative .48IV.Nature of BusinessHyp15: Industry cope Negative .76+p


16/23


to which affiliates felt pressuredby three institutional orces: local governmentregulations,trade associations and consumer groups. We predicteda positiverelationship,but no significant effect was found. Whereasa labor union mightbe effective in stipulating at least some HRM practices, the more diffuseeffect of pressure from outside organizations had no apparent relationshipwith an affiliate's resemblance to local practices.The fact that Hypothesis 7 showed only marginal significance, and thatHypothesis 8 revealedno significanteffect, suggests that most affiliates arenotcoerced into adopting local practices. Rather, the overwhelming tendencyof affiliateHRM practices to resemble local practicesmay betterbe understoodas normative or mimetic in nature:affiliates consistently seek to adopt localHRM practices although they are not compelled to do so. As an example,consider Novotel New York, a large hotel in midtown Manhattanowned bythe French lodging giant, ACCOR. Novotel New York is one of very fewnon-union hotels in Manhattan;yet, in orderto hire talented employees andto retain their services, it adopted many of the practices of union hotels[Rosenzweig and Raillard 1992]. In this case, and perhaps in many otherslike it, an affiliate chose for competitive reasons to adopt local practiceseven in the absence of coercion.Shifting our analysis from embeddedness within the local environment toflows between the MNC parent and the U.S. affiliate focused attention ona different set of dynamics. Hypothesis 9 tested the effect of expatriates,operationalizedas the percentageof affiliate employees who were expatriates.As shown on Table 3, the effect was in the predicted direction and washighly significantatp


17/23


Drawing on the work of Kogut and Singh [1988], we used a scale thatmeasured the difference among countries based on an index of GeertHofstede's four dimensions of work-related values. The effect was in thepredicteddirectionbut only significant atp


18/23


affect internaldecision processes are relatively closer to the parent.In addition,we have found thatspecific independent variables affect the degree to whichaffiliate practices resemble local practices. Affiliates founded by acquisitionand dependent upon the local environment for inputs tend more closely toresemble other local firms in their HRM practices. By contrast, affiliateswith high levels of expatriatesand in close communication with their parenttend, on the margin, more closely to share the practices of their parents.A Further Exploration of DifferencesAccordingto Parent NationalityOf the independent variables tested above, only Hypothesis 12, which testedthe effect of parent country culture, allowed us to compare affiliates ofdifferent parent nationality, and the result was only marginally significant.It is possible that the nationality of the parent firm is not, by itself, animportant nfluence on affiliate HRM practices;alternately,we might questionwhether parent nationality is adequately captured by this scale. To pursuethe matter more directly, we set aside the proxy of Culture Distance andexamined directly affiliates according to their parent nationality.Table 4a shows the TransformedDifference from Local for each practice,broken down by parentcountry. Scheffe's tests were performedto determinewhich countries varied significantly from the others. For the most part,countries were narrowly distributed around the mean, with relatively fewstanding apart.The only exception was Japan:its average score of 1.56 wassignificantly higher (p


19/23


TABLE 4Comparison of Mean DifferenceGender Executive

Time Off Benefits Cmp Training Bonus ParticipationAveragea. Comparison of Mean Difference from LocalCanada 1.20 1.29 1.25 1.68 1.70 1.62 1.46France 1.15 1.17 1.42 1.20+ 1.60 1.67 1.39Germany 1.25+ 1.34 1.32 1.49 1.47+ 1.77 1.45Japan 1.26* 1.24 1.61** 1.62 1.79 1.79 1.56**Netherlands 1.00 1.13 1.25 1.67 1.75 1.75 1.39Sweden 1.06 1.31 1.19 1.60 1.64 1.64 1.37Switzerland 1.07 1.46 1.21 1.54 1.42 1.58 1.34U.K. 1.07+ 1.37 1.42 1.48 1.69 1.81 1.43Average 1.16 1.30 1.38 1.54 1.64 1.74 1.45

b. Comparison of Mean Difference from ParentCanada 1.40** 1.56* 1.60+ 1.48* 1.25** 1.50* 1.47France 2.23 1.77 1.77 1.77 1.82 1.69 1.85Germany 2.56** 1.96* 1.83 1.81 1.50+ 1.98* 1.98**Japan 1.61** 1.52** 1.98+ 1.83 1.82 1.56* 1.72*Netherlands 2.75** 2.25* 1.75 1.78 2.00 1.75 1.98Sweden 2.63** 2.19** 1.93 1.93 1.88 1.81 2.06**Switzerland 2.25 1.93 2.00 1.64 1.79 1.80 1.92U.K. 2.08 1.83 1.75 1.75 1.76 1.90+ 1.86Average 2.09 1.80 1.83 1.75 1.68 1.76 1.83+p


20/23


Swedish, with scores of 1.98 and 2.06, respectively. At first these scoresseem odd-if Canadian affiliates show the lowest difference from parent,might Japanese affiliates not be expected to show the highest? But at 1.72,Japanese affiliates are actually below the mean, indicating relatively greatersimilarity to the parent country. The explanationfor these scores is twofold.First, as noted above when discussing Table 4a, Japanese affiliates tend leastto adhere to U.S. practices. They instead tend to retain a numberof Japanesepractices, as reflected by the lower scores on Table 4b. Second, the highscores of German and Swedish affiliates reflect in large part two practices:Benefits and Time Off. For these practices,adherence to U.S. practices makesGerman and Swedish affiliates far different from the very generous policiesof their home countries, whereas adherence to U.S. practices by Japaneseaffiliates demands relatively less deviation from parent country practices.

CONCLUSIONSWe began this study with an interest in understandingthe forces that shapespecific management practices in the affiliates of MNCs. Focusing on HRMpractices in U.S. affiliates, we determined, as a first conclusion, that theytend primarily o resemblelocal practices.More importantly,we demonstratedthat human resource management is not a monolithic function, but consistsof practices which differ in their relative resemblance to local practices andto parent practices. These findings offer support for the view of MNCs asa nexus of differentiated practice, with specific practices shaped, to varyingextents,by different orces.Inaddition,we demonstratedhatspecific independentvariables have significant effects on HRM practices overall, with the mostimportantbeing founding, dependence upon the local environment or inputs,presence of expatriates, and frequency of communication. Finally, we haveshown specific tendencies according to parent nationality.To what extent can these findings be generalized? All too often, empiricalfindingsaretreatedas implicitly universal,withoutappropriate ircumspection.Because this study has examined HRM practices of MNC affiliates in theUnited States, we should ask ourselves two questions: first, would theseresults hold for other functions as well as human resource management?And second, would these results hold for affiliates in countries other thanthe United States?Regarding functions other than human resource management, one of theauthors has elsewhere presented findings about MNC affiliate practices infour functions: human resource management, marketing, financial control,and manufacturing [Rosenzweig 1994]. Whereas HRM practices are moststrongly influenced by local isomorphism, it was found that marketing andmanufacturing practices also tend to resemble local practices, although to adistinctly lesser degree than HRM, but that financial control practices moreclosely adhere to parent practices, and are therefore shaped mainly by aneed for internalconsistency in the MNC. As in the presentstudy, the mode


21/23


of entry was an importantinfluence, and the dependence upon local sourcesof inputs was especially importantfor marketing and manufacturing,but itwas not found that the presence of expatriates had a significant effect forany practice other than HRM. Thus, independent variables were found toexert differing pulls on various functions, lending furthersupportto the viewof a multinational corporationas a nexus of differentiated practices.The data from this study do not allow us to determine whether MNCaffiliatesin countriesotherthanthe UnitedStatesexhibitthe same characteristicsconcerning HRM practices. We are left to speculate: does the United Statesfor some reason emit an especially strong isomorphicpull for HRM practices,or is pressure to adhere to local practices robust across countries? Becausethe United States is relatively less replete with HRM regulations than manyother industrialized countries (for example, betriebsratwork councils inGermany, work rules in Sweden, and extensive national holidays in France),we may hypothesize that there is at leastas muchpressure orlocal isomorphismin other industrialized countries as that which we have found in the UnitedStates. Conversely, would the foreign affiliates of American MNCs adoptlocal HRM practices to the same extent that MNC affiliates in the U.S. haveadopted local U.S. practices? Without advancing reasons why AmericanMNCs should behave differently than the MNCs examined in the presentstudy, it would not be prudent to generalize beyond what the data in thisstudy can support. Indeed, conducting a study of management practices inforeign affiliates of U.S. MNCs would be a valuablecomplement to this study.As a final note, this study enables Americans to comparetheirHRM practiceswith those prevalent in other industrializedcountries. An interesting pictureemerges as we infer from these data:compared to MNCs in other countries,U.S. firms overall offer less paid time off and provide a lower percentageof benefits, but have a higher proportion of women in management, thanfirms from several other countries.Although institutionalpressureand inertiasurely make rapid change difficult, managers in U.S. firms should at leastbe moved to question the practices they often assume to be "normal," and,where appropriate,would do well to ask whether the practices that prevailelsewhere could be of use to them.These findings also provide an important perspective to the continuingdebate about the growing levels of foreign direct investment in the UnitedStates. According to the U.S. Department of Commerce, more than 5% ofthe labor force now works for affiliates of foreign-based firms. In recentyears, Americans have voiced ambivalence about this trend and expressedconcern about "working for foreign bosses" (e.g., Fucini & Fucini [1990]).Althoughthis studyhas notaddressed he full rangeof concerns thatAmericansface when working for affiliates of foreign-based MNCs, which includeissues of career advancement as well as racial discrimination, the evidencepresented here makes plain that for a number of importantHRM practices,working for the affiliate of a foreign-based MNC is much like working for


22/23


a typical U.S. company. Concerns about conformingto "foreign" workrules and HRM practicesappear o be exaggerated-for the several humanresourcepractices ncluded n this study,at least, adherenceo local practicesis the dominant nfluence.

NOTES1. Humanresourcemanagementwas one of fourfunctionsaboutwhichwe soughtinfornation.Theaffiliate'stop managerwas identifiedas the personbest placedto speakto all of them.2. Thisfinding s noteworthy n lightof recentstudiesthathave shown thatAmericanworkers eceiveless paid timeoff than heircounterpartsn other ndustrializedountries Schor 1992].Giventhe largeandincreasingnumberof Americansworking or affiliatesof foreign-basedMNCs,it is worthwhileto askwhether hese affiliates endto follow localpracticeswhenit comesto paidtime off, or whethertheyadhere o the practicesof theirparent.Datafromthis studyshow thataffiliatesoverwhelminglyfollow local U.S. practices.3. For affiliates n serviceindustries,productandprocessknow-howwere deleted.4. Strictlyspeakingtheseare not "Canadian ffiliates"or "Japaneseaffiliates"but "U.S. affiliatesof CanadianMNCs"and "U.S. affiliatesof JapaneseMNCs,"' espectively.We have used the formerterms to avoid the cumbersome hrasingof the latter.

REFERENCESBartlett,ChristopherA. & SumantraGhoshal. 1989. Managingacross borders:The transnational

solution. Boston: Harvard Business School Press.Brooke,MichaelZ. & H. Lee Remmers.1970. The trategyof mnultinationalnterprise:Organizationandfinance. London:Longman.DiMaggio,Paul J. & WalterW. Powell. 1983. The ironcage revisited: nstitutionalsomorphism ndcollective rationalityn organizationalields. AmnericanociologicalReview,48: 47-160.D'Iribame,Philippe.1989. Lalogitque e l'honneur.Paris:Seuil.Doz, Yves L., ChristopherA. Bartlett& C.K. Prahalad. 981. Globalcompetitivepressuresand hostcountrydemands:Managing ensions n MNCs. CaliforniaManagementReview,23: 63-73.Edstrom,Anders& JayGalbraith. 977. Transfer f managersas a coordination ndcontrolstrategyin multinationalorporations. dmiinistrativecienceQuarterly, 2: 248-63.Egelhoff,WilliamG. 1988. Organizinghe mtrultinationalnterprise:An informiationrocessingper-spective.Cambridge,Mass.:Ballinger.Evans, Paul & Peter Lorange.1989. The two logics behind humanresourcemanagement. n PaulEvans,Yves Doz & AndreLaurent, ditors,Humanresourcemanagemenetn internationalirls:Change,globalization, nnovation.London:Macmillan.Franko,LouisG. 1976. TheEuropeanmultintationals.reenwich,Conn.:Greylock.Fucini,JosephJ. & Suzy Fucini. 1990. Workingor theJapanese.New York:TheFree Press.Ghoshal,Swnantra& Nitin Nohria. 1989. Internaldifferentiationwithin multinationalorporations.StrategicManagement ournal,10: 323-37.Gupta,Anil & VijayGovindarajan.991. Knowledgelows andthestructuref controlwithinmultina-tionalcorporations. cadenmyf ManagementReview,16:768-92.Hannan,Michael T. & John Freeman.1977. The populationecology of organizations.AmllericanJournalof Sociology,82: 929-64.

1984.Structuralnertiaandorganizationalhange.AmericanSociologicalReview,49: 149-64.Hedlund,Gunnar. 986.Thehypermodern NC:A heterarchy? umnanesourceManagement,5:9-35.______ & Per Aman. 1984. Mantagitgrelationshipswithforeigntsubsidiaries:Organizationandcontrolin SwedishMNCs.Stockholm:Mekan.Hofstede,Geert.1980.Culture's onsequences:nternationalifferencesn work-relatedalues.BeverlyHills, Calif.:Sage.


23/23


Kobayashi, Noritake. 1982. The present and future of Japanese multinational enterprises. InternationalStudiesof Manand Organization,12(1):38-58.Kobrin, Stephen J. 1991. An empirical analysis of the determinants of global integration. StrategicManagement ournal 12: 17-31.Kogut, Bruce & Harbir Singh. 1988. The effect of national culture on choice of entry mode. Journalof InternationalBusinessStudies,49: 411-32.Laurent, Andre. 1983. The cultural diversity of Western conceptions of management. InternationalStudiesof Management nd Organization, 3(1-2):75-96.Lincoln, James R., Jon Olson & Mitsuyo Hanada. 1978. Cultural effects on organizational structure:

The case of Japanese firms in the United States. American Sociological Review, 43: 829-47.Meyer, John W. & Brian Rowan. 1977. Institutionalized organizations: Formal structure as myth and

ceremony. AmiericanJournal of Sociology, 83: 343-63.Negandhi, Anant R., Golpira S. Eshgi & Edith C. Yuen. 1985. The management practices of Japanese

subsidiaries overseas. California Management Review, 27(4): 93-105.Nohria, Nitin & Sumantra Ghoshal. 1992. Distributed innovation in the differentiated network multi-national. Working paper 92-040. Boston, Mass.: Harvard Business School.

Perlmutter,Howard. 1969. The tortuous evolution of the multinational corporation. Columtibiaournalof WorldBusiness,4: 9-18.Porter, Michael E. 1986. Changing patterns of international competition. Californtia Managemrlelnt

Review, 28: 2, 9-40.Prahalad,C. K. & Yves L. Doz. 1987.Themtiultiniationialttissioni:alanicinigocal demtianidsnid lobal

vision. New York: The Free Press.Rosenzweig, Philip M. 1994. Management practices in U.S. affiliates of foreign-owned firms: Are

'they' just like 'us?' The Initerniationial xecutive, July-August._ & Benoit Raillard. 1992. ACCOR (A), HarvardBusiness School Case, 9-393-012.

Rosenzweig, Philip M. & Jitendra V. Singh. 1991. Organizational environments and the multinationalenterprise. Acadeomyof Man1agemtieniteview, 16(2): 340-61.Rugman,Alan M. 1979. Initernationialiversificationanid he multiniationialniterprise.Lexington,

Mass.: Lexington Books.Schor,JulietB. 1992.Theoverworked mtiericani:heuniexpectedeclinie f leisure.New York:Basic

Books.Singh, JitendraV., Robert J. House & David J. Tucker. 1986. Organizational change and organizational

mortality. Admiiinistrative cientceQuarterly, 31: 587-611.Soenen, Luc A. & Daniel Van den Bulcke. 1988. Belgium's largest industrial companies: A compari-son of size and financial performance between foreign and Belgian owned firms. MantagemtientInternationalReview,28: 51-63.Stinchcombe, Arthur L. 1965. Social structure and organizations. In James G. March, editor, Hanidbook

of organizations, 142-93. Chicago: Rand McNally.Stopford,John M. & Louis T. Wells. 1972. Man?agin1ghe mtiultinational enterprise. New York:Basic Books.Tsurumi, Yoshi. 1986. Japanese and European multinationals in America: A case of flexible corporate

systems. In Klaus Macharzina & Wolfgang H. Staehle, editors, European approaches to initerna-tional managemtienit. ew York: Walter de Gruyter.Walton, Richard E. & Paul R. Lawrence. 1985. Hiumllaniesource mtianagemtienit:rends anidchalleniges.Boston: Harvard Business School Press.

Documents

20 HRMpractices inMNCs