what is the shape of the multinationality …...diversification through foreign direct investment...

89

Volume 12 Number 1Spring 2004T

h e M u l t i n a t i o n a l B

u s i n e s s R e v i e w

What is the Shape of theMultinationality-Performance

Relationship?

Douglas E. Thomas, University of New MexicoLorraine Eden, Texas A&M University

Dr. Douglas E. Thomas is an Assistant Professor in the Anderson Schools ofManagement at the University of New Mexico. His research interests includeemerging market firms, strategic management in Latin America, and theinternationalization of firms. E-mail: [email protected] Page: http://www.unm.edu/~thomasdDr. Lorraine Eden is Professor of Management at Texas A&M University. Herresearch interests focus on the political economy of MNEs, specializing in transferpricing, international taxation and regional integration. E-mail: [email protected] Page: http://cibs.tamu.edu/leden

A previous version of this paper was presented at the 2000 Academy ofManagement meeting in Toronto, Canada.

ABSTRACT: Previous theoretical explanations and empirical analyses of themultinationality-performance relationship have produced mixed arguments andresults. Linear and inverted U-shaped relationships have been theorized andconfirmed empirically. Recent research has theorized that there is a three-stage,sigmoid relationship between multinationality and performance. We contributeto the debate by showing that the impact of multinationality depends on thetime dimension incorporated in the performance measure; that is, the netbenefits from multinationality are likely to be higher in the longer term. Theresults from our sample of US manufacturing multinationals indicate that there isa three-stage, sigmoid multinationality-performance relationship.

INTRODUCTIONOver the past three decades,

international business andmanagement scholars have sought tounderstand how internationaldiversification through foreign directinvestment (FDI) affects firmperformance. However, previousresearch on the multinationality-performance relationship hasproduced mixed arguments and results

(e.g., Buhner, 1987; Kotabe,Srinivasan, & Aulakh, 2002; Tallman& Li, 1996). The most recent researchhas focused on explaining the shapeof the relationship: linear, U- orinverted-U, or sigmoid; however,theoretical and empirical gapscontinue to bedevil researchers.

This paper provides a freshperspective on the multinationality-performance debate. We first review

90

Volume 12 Number 1 Spring 2004T

h e

M u

l t i

n a

t i o

n a

l B

u s

i n

e s

s R

e v

i e

wrecent literature on the debate, findingfive different hypotheses about theshape of the multinationality-performance relationship. We arguethat there are three partialexplanations for the confusing resultsin the literature (Annavarjula &Beldona, 2000). First, the term‘multinationality’ has differentmeanings. Conflicting results arepartly due to different understandingsand proxy measures. Second, thetheoretical benefits and costs ofmultinationality to an individual firm,and how they are reflected in firmperformance, are not well understood.Third, the temporal dimension of themultinationality-performance hasgenerally been ignored. For example,the differences between short-run andlong-run performance have not beenfully explored.

In the following sections, wedevelop a theoretical explanation ofthe multinationality construct, arguingit has three dimensions: foreignproduction presence, foreign marketpenetration and country scope. Next,we provide a clear outline of thebenefits and costs of multinationalityto firms in general. Finally, we arguethat the impact of multinationalitydepends on the temporal dimensionincorporated in the performancemeasure, that is, the net benefits frommultinationality are likely to be higherin the longer term. We test our modelon a sample of US manufacturingmultinationals over the 1990-94 period,and present the somewhat surprisingresults. Finally, we offer someconclusions and guidance for futureresearch.

LITERATURE REVIEWThe literature provides several

theoretical explanations of the natureof the multinationality-performancerelationship. Five general models -positive and linear, positive but withdiminishing returns, inverted-Ushaped, and sigmoid - explaining therelationship have been presented byprevious researchers. Empirically,these models have been tested tovarying degrees. We review thesemodels briefly below.

Positive and Linear Model One stream of research oninternational diversification hashypothesized and found support for apositive, linear relationship (e.g., Grant,1987; Grant, Jammine, & Thomas,1988; Han, Lee, & Suk, 1998; Kim,Hwang, & Burgers, 1989; Tallman &Li, 1996). The theory and empiricalresults indicate that as firms expandinternationally, there is a positive,linear impact on firm performance.More recent research finds that thispositive, linear relationship ismoderated by other factors (e.g., thefirm’s R&D and marketingcapabilities (Kotabe et al., 2002)).

Positive But DiminishingReturns Model

From this perspective,multinationality has an initial positiveimpact on performance; however,over time, diminishing marginal returnsset in. The slope of the curve is stillpositive (i.e., multinationality continuesto have a positive impact on firmperformance) but at a decreased rate(i.e., the benefits are not as great as

91




they were initially) (Gomes &Ramaswamy, 1999). Very littleempirical research has specificallytested this relationship.

U-shaped RelationshipOther researchers have theorized andfound evidence for a U-Shapedrelationship between multinationalityand firm performance (Ruigrok &Wagner, 2003). Using organizationallearning theory, Ruigrok and Warner(2003) find that firms initiallyexperience negative performancewhen expanding internationally. Overtime, however, as firms learn fromtheir international experience, theirperformance becomes positive. Verylittle empirical research has examinedthis relationship.

Inverted-U RelationshipAnother stream of research hasargued (Hitt, Hoskisson, & Ireland,1994; Sullivan, 1994b) and found(Geringer, Beamish, & daCosta, 1989;Gomes & Ramaswamy, 1999; Hitt,Hoskisson, & Kim, 1997) that thereis a curvilinear (inverted-U)relationship between multinationalityand firm performance. Over time, thepositive impact on performance isoutweighed by the costs ofcoordinating a widely dispersednetwork of international operations. Inother words, the slope of the curve isinitially positive and after reaching anapex, becomes negative. Hitt et al.(1997) argue that the relationship isinverted U-shaped because greatergeographic dispersion increases thecosts of coordinating, integrating andmanaging the multinational

enterprise’s (MNE) overalloperations. They find an inverted U-shaped relationship betweenmultinationality and performance, withthe slope initially positive but turningnegative at high levels ofmultinationality. Gomes andRamaswamy (1999) argue thatmultinationality offers the ability toleverage scale economies, utilizehome-based skills, competencies andresources, arbitrage differences infactor costs across countries, andspread the costs of productdevelopment. However, once the levelof foreign operations becomes large,the MNE is forced to adopt morecomplex and costly organizationalstructures and move to less familiarsettings where higher cultural diversityraises transactions costs. The resultsof their study indicate that themul t inat ional i ty-performancerelationship is nonlinear, first rising andthen declining as costs eventuallyovertake the benefits ofmultinationality.

Sigmoid RelationshipHitt et al. (1994) contend that the

relationship is in the shape of aninverted-U. However, they point outthat if one magnifies this relationship,there are peaks and valleys in theupward-sloping portion of the curve.As the firm expands internationally, itis involved in a process of minordownturns of performance and thenupswings. Thus, they argue, “theactual shape of the curve is moreaccurately depicted as multiplewaves” (p. 312).

Contractor et al. (2003) argue

92


h e

M u

l t i

n a

t i o

n a

l B

u s

i n

e s

s R

e v

i e

wand find, using a sample of servicefirms, that there is a three-stage,sigmoid relationship betweenmultinationality and performance. Thenonlinear relationship is also examinedby Riahi-Belkaoui (1998) (using thespline technique) to test therelationship between the foreign salesratio and return on assets. The authorfinds a negative relationship at lowforeign sales ratio levels, whichbecomes positive and then finallynegative as the foreign sales ratiorises.

What is the shape of themul t inat ional i ty-performancerelationship? We contend that thereare three partial explanations for thedifferent theories and confusingresults in the literature. First, the term‘multinationality’ means differentthings to different people. This meansthat conflicting results are partly dueto different understandings and proxymeasures. Second, the theoreticalbenefits and costs of multinationalityto firms, and how they are reflectedin firm performance often conflict.Finally, we suggest that the temporaldimension of performance has notbeen adequately captured empirically.The multinationality-performancerelationship is different for short-run,as compared to long-run, performanceand must be accounted for in empiricalmeasures.

The paper is organized in thefollowing manner. First, we developtheory and hypotheses related tomultinationality, its costs and benefitsand the erstwhile ignored timedimension. Next, we test thesehypotheses and discuss the empirical

results. Finally, we offer someconclusions and guidance for futureresearch.

THEORY DEVELOPMENTWhat is Multinationality?

Incomplete theoretical analysisof the multinationality construct hascontributed to the mixed empiricalresults in studies of themul t inat ional i ty-performancerelationship. There are three keycomponents to multinationality(Sullivan, 1994a, 1996). 1. Foreignmarket penetration, the dependenceof the firm on foreign markets. 2.Foreign production presence or thedegree to which the enterprise isengaged in production-based activitiesacross borders (Annavarjula &Beldona, 2000). 3. Country scope,which is the geographic range orbreadth of the firm’s internationalpresence (Goerzen & Beamish, 2003).The first two components, foreignmarket penetration and foreignproduction presence, represent thedepth of the MNE’s involvementabroad. Together they answer thequestion: What percent of the MNE’sactivities are conducted outside thehome country? The third component,country scope, captures the breadthof multinationality. Country scopeaddresses the question: How wide isthe global reach of the multinationalenterprise? Because there has beenan incomplete theoretical treatment ofthe multinationality construct,researchers have operationalized it inmany different ways and generallypaid inadequate attention to theparticular dimension that their proxy

93




may only be capturing.For example, multinationality has

most often been measured using theforeign sales-total sales ratio (Sullivan,1994a) which only captures onedimension of multinationality, foreignmarket penetration. Other commonmeasures of multinationality includethe foreign assets-total assets ratioand foreign employment-totalemployment ratio. We argue thatforeign assets and employment ratioscapture the foreign productionpresence dimension ofmultinationality. Finally two othervariables that have been commonlyused are number of foreign countriesand number of foreign affiliates (Allen& Pantzalis, 1996; Gomes &Ramaswamy, 1999; Mishra & Gobeli,1998; Morck & Yeung, 1991). Thesevariables are good proxies for the thirddimension of multinationality, countryscope. Most previous research has

used only one of these classes ofvariables and, therefore, limitedmultinationality to one dimension(exceptions include Allen & Pantzalis,1996; Daniels & Bracker, 1989;Goerzen & Beamish, 2003; Qian &Li, 2002).

Figure 1 provides a visual,comparing two firms on differentdimensions of multinationality.Lockheed is not very multinationalwhen all three dimensions areconsidered. Exxon scores highly onboth the foreign market penetrationand foreign production presencedimensions but has relatively limitedcountry scope. However, Gillettescores very highly on all threedimensions. This simple examplemakes it clear that all dimensions ofmultinationality must be consideredtheoretically and empirically in orderto produce clear results in studies ofthe multinationality-performance

94


h e

M u

l t i

n a

t i o

n a

l B

u s

i n

e s

s R

e v

i e

wrelationship.

Hence, to truly capturemultinationality all three dimensionsshould be measured: foreign marketpenetration, foreign productionpresence and country scope. A givenfirm may score high on one dimensionbut may be less multinational thananother that scores moderately on allthree dimensions. For example, Figure1 provides some examples of threemajor U.S.-based manufacturingmultinationals and their levels ofmultinationality using the threedimensions and appropriate proxies(foreign sales-total sales ratio forforeign market penetration, foreignassets-total assets ratio for foreignproduction presence, and number offoreign countries for country scope).

We argue that firms will seek tooperate at their optimal degree ofbreadth and depth of multinationalinvolvement (Allen & Pantzalis, 1996).Given the existing breadth of the MNEnetwork, the enterprise can choose toallocate a higher or lower percent ofits total activities to foreign operationsand vice versa. Goerzen and Beamish(2003) find that multinational depth isnegatively related to firm performancewhile breadth is positively related tofirm performance. In contrast, Allenand Pantzalis (1996) find that theremay be a tradeoff between breadthand depth. Their study found thatperformance levels were highest infirms with broad but not deepmultinational networks and theyconcluded that “negative returns frommultinationality can be attributed tooverinvestment in ‘within country’ asopposed to ‘across country’

international expansion” (Allen &Pantzalis, 1996: 645). In terms of ourmodel, their work suggests thatbreadth and depth may have differentimpacts on firm performance. Hence,we argue that:

H1: The individualcomponents of multinationalitywill have differing impacts onMNE performance; inparticular, breadth ofmultinational involvementshould be more positivelyrelated to performance thandepth.

Benefits and Costs ofMultinationality

MNEs have the ability to exploitsources of competitive advantage notavailable to domestic firms. These arepotential benefits from multinationalitybecause not all firms are able tocapture them. The first general benefitis the ability to use foreign directinvestment (FDI) to leveragedifferences and take advantage ofopportunities between countries. Asthe level of international diversificationincreases the potential opportunitiesfor the MNE also increase. We outlinethem below.

! Differences in tastes, demandsand income levels. MNEs can shiftsales from low-income to high-incomemarkets, generating higher profits onthe firm’s resources. Market-seekingFDI is designed to exploit profitableopportunities in higher incomemarkets. As products becomeobsolete in high-income markets, FDI

95




can be used to shift sales towards low-income markets, extending the life ofan obsolete product line. Ininternational trade theory, these gainsare called gains from exchange sincethey are generated by country-baseddifferences in demand. We arguesuch differences can also beinterpreted as offering FDI potentialgains from exchange.! Differences in factorendowments. Through FDI, a firm canreap gains from specialization byshifting production to lower costlocations and altering productiontechniques to take advantage ofdifferences in factor abundancesbetween countries. In internationaltrade theory, these gains are calledgains from specialization.Specialization gains are generated byeither horizontally integrated FDI(producing the same product lines intwo or more countries) or verticallyintegrated FDI (segmenting the stagesalong the value chain acrosscountries). We call such investmentsresource seeking FDI, and these gainscan be seen as the specialization gainsfrom FDI.! Differences in knowledge-based assets. Acquisition ofknowledge-based assets alsomotivates FDI (Dunning, 1993).Researchers have argued (Kostova,1999; Kostova & Roth, 2002) andshown that (Hitt et al., 1997) MNEswith business units in more than onecountry can draw on the knowledgebase in their foreign affiliates toimprove their products and processes,generating worldwide organizationallearning within the MNE’s network.

! Differences in governmentregulations. Investment shunting canbe used to shift production to locationswith lower taxes, higher subsidies oreasier regulations. MNEs can set upfinancial affiliates in tax havens ormove investments to avoid anti-dumping duties.! Multinational flexibility.Managing multinational risk has longbeen considered a motivation for FDI(Rugman, 1976). In addition,researchers have argued that MNEscan take advantage of their multiplelocations to flexibly adapt to changes/shocks in the external environment(Kogut, 1984). For example, duringthe Mexican peso crisis in the 1990s,US subsidiaries in Mexico were ableto shift their production from thedepressed local market to the USmarket. Domestic Mexican firms,without access to US marketing anddistribution networks, were not ableto expand exports nearly as quicklyas US subsidiaries.! Bargaining power. LargeMNEs, due to their ability to moveassets quickly between countries,have more bargaining power relativeto location-bound actors, such asgovernments, trade unions anddomestic firms.

Multinationality, however, is nota one-sided proposition. There are alsocosts because cross-borderopportunities are not risk free. As thenumber of foreign countries in whichthe MNE operates rises, we expectthe following firm-specific costs frommultinationality.

96


h e

M u

l t i

n a

t i o

n a

l B

u s

i n

e s

s R

e v

i e

w! Costs and risks of multiplesources of value. Theoreticalresearch indicates that one of the keycharacteristics differentiating MNEsfrom domestic firms is that MNEsoperate with multiple sources of value(Sundaram & Black, 1992). Differentexchange rates between countriescreate foreign exchange risks for theMNE. Foreign sales, the market valueof the firm, and its ability to raisecapital are negatively affected bytranslation exposure. The greater thenumber of foreign countries in whichthe MNE operates, the more foreignexchange risks to which the MNEmay be sensitive. On the other hand,international diversification of marketsand production locations may cushionthe MNE from exchange rate shocksthat are region specific (e.g., Asiancurrency crisis).! Costs and risks of multiplelevels of authority. As the MNEexpands into more countries, it isfaced with higher cross-bordertransactions costs and higherinteraction costs with a wider varietyand number of governments. Empiricalresearch indicates that political risksmay increase (Chase, Kuhle, &Walther, 1988). Additionally, it hasbeen argued that there are increaseddifficulties to establishing legitimacy(Kostova & Zaheer, 1999). On theother hand, as time passes and MNEsbecome established actors in hostcountries, the liability of foreignnessshould decrease at the individualaffiliate level.! Costs of greater culturaldiversity. As the number of foreignmarkets and production locations

increase, the MNE is faced with thecosts of adapting to new and moreheterogeneous cultures. Empiricalresearch indicates that the liability offoreignness (Zaheer, 1995) increasesas firms move to more culturallydistant countries (Gomes &Ramaswamy, 1999).

We concur with previousempirical research, which finds thatthere is a curvilinear relationshipbetween multinationality and firmperformance (Gomes & Ramaswamy,1999; Hitt et al., 1997). The benefitstaper off while the costs rise as thedegree of multinationality increases;the relationship betweenmultinationality and firm performanceshould be non-linear, first rising andthen declining. The costs and risks ofmultiple authorities, values andcultures should rise as their numberand diversity increases. Therelationship between multinationalityand performance depends on a firm’sability to manage the complexitiesinherent in the internationalizationprocess. We therefore hypothesizethat:

H2: The relationshipbetween multinationality andfirm performance iscurvilinear, first rising andthen declining (an inverted-Ushape), reflecting the fact thatmultinationality carries bothbenefits and costs.

The Role of TimeFinally, the mixed results can at

least be partially explained by

97




neglecting the temporal dimension ofthe multinationality-performancerelationship (exceptions includeAleson & Escuer, 2001). Both thebenefits and costs of multinationalitycan have different impacts in the shortversus long term. Investments inR&D, particularly in basic research,have a negative impact on short-runperformance because costs areincurred well in advance of benefits.The anticipated returns frominvestments in intangible assets arebetter reflected in long-runperformance. First mover advantagesfrom entering foreign markets at anearlier stage than later entrants arereflected better by long-run anticipatedreturns than by short-run gains (e.g.,Luo, 1998; Mascarenhas, 1992).Though empirical results are notcompletely consistent in this streamof literature, research generallyindicates that First movers incur higherinitial costs, reducing short-termfinancial performance but theytypically gain higher market sharesthan latecomers.

As the enterprise becomes moremultinational, it will have to adoptmore sophisticated controlmechanisms and organizationaldesigns, raising the costs of operatingthe enterprise as a whole. Thus, costsof multinationality may also increaseover time as shown in previousresearch (Hitt et al., 1997). At thesame time, as firms internationalizethey increasingly learn how to manageinternational operations (Barkema &Vermeulen, 1998). We thereforehypothesize that the multinationality-performance relationship has a time

dimension. Because the benefits aremore likely to be longer term in nature,relative to the costs, we hypothesizethat:

H3: The benefits frommultinationality for a firm arehigher when measured in termsof long-run performance thanshort-run performance.

METHODOLOGYSample

We limited our sample to U.S.headquartered manufacturing firms(i.e., primary SIC between 2000-3999)included in the S&P 500 (inCOMPUSTAT) during the 1990-1994period. Further, because we arefocusing on the internationaloperations of these companies, weeliminated firms that did not have atleast one foreign affiliate. Missing dataon several important variables for afew firms precluded their inclusion.Our final pooled cross-sectional, time-series consisted of 151 firms followedover five years (i.e., sample size = 755observations). It should be noted thatwe are focusing on the impact ofrelative degree of multinationality onfirm performance since all of oursample firms are multinationalenterprises (i.e., they all have at leastone foreign affiliate). Ourmultinationality variables arecontinuous ranging from zero (purelydomestic firms) to one (100 percentinternationalized firms). The averagefirm in our data set has approximatelyone-third of its sales and one-quarterof its assets overseas and operatingaffiliates in 16 foreign countries (see

98


h e

M u

l t i

n a

t i o

n a

l B

u s

i n

e s

s R

e v

i e

wFigure 1).

Firm Performance.Our analysis uses four measures

of MNE performance: return onassets, return on equity, excessmarket value and average marketvalue. Initially, we started with twoadditional performance measures,return on sales and earnings per share.The Cronbach’s alpha for the set ofsix measures was .8155, suggestingthey represent one common construct,firm performance. (Complete resultsfrom the factor analysis are availableupon request from the authors.)However, a principal componentsfactor analysis shows that firmperformance loads on two dimensions:the first dimension we call short-runfinancial performance (return onassets, return on equity, return onsales, and earnings per share) and thesecond, long-run expected marketperformance (excess market valueand average market value), with nounique factors. We chose to reportreturn on assets and return on equityas our two measures of short-runprofitability, rather than all four, toconserve space and because returnon assets and return on assets arecommonly used profitability measures.Return on assets (ROA) is the mostfrequently used dependent variable inperformance-multinationality studies(Gomes & Ramaswamy, 1999;Sullivan, 1994a). Return on equity(ROE) is an alternative measure offinancial performance, focusing onshareholders’ equity. The numeratorin both ratios is based on net incomeafter tax and interest expenses.

Excess Market Value (EMV) isthe ratio of market value plus the bookvalue of debt minus total assets, alldivided by total net sales (Allen &Pantzalis, 1996). It measures theexcess market valuation of the firmover and above the value of itsphysical assets. High ratios suggestthe existence of intangible assets (e.g.,technology, brand loyalty, superiormanagerial skills) and the implicitassessment of the firm’s ability tocapture the benefits of theseintangibles in its long-runperformance. Average market value(AMV), measured as market valuedivided by total assets, similarlyfocuses on market valuation.

EMV and AMV are futureoriented, while ROA and ROE focuson last year’s profitability. In thissense, EMV and AMV better capturethe potential benefits frommultinationality, particularly when thefirm also has high ownershipadvantages. If this were the case, thenone would expect that the returns tomultinationality might be more closelytied to market-based performancemeasures than to financial ratios.

Multinationality.We have argued above that

multinationality has three components.The first is foreign marketpenetration, which we measure usingthe foreign sales-total sales ratio. Thesecond component is foreignproduction presence measured bythe foreign assets-total assets ratio.We considered both the total numberof foreign affiliates and the numberof countries where the MNE has

99




foreign affiliates as a country scopevariable. Since foreign affiliates canbe concentrated in particularcountries, the number of foreigncountries provides a betterrepresentation of the globalconfiguration of MNE activities.

We construct our index ofmultinationality, following Gomes andRamaswamy’s (1999) technique ofemploying a principal componentsanalysis of the foreign sales ratio, theforeign assets ratio and the numberof foreign countries to find theireigenvectors, and using theseeigenvectors as weights. Given thatthese three components use differingscales it was important to standardizethis variable (which we did) (Gomes& Ramaswamy, 1999). TheCronbach’s alpha for the threevariables is .8172, suggesting that ourcomposite measure is valid.

Control Variables.In order to isolate the impacts of

multinationality on firm performance,it is important to include all othervariables likely to affect performance.In the resource-based view of the firm(Barney, 1991), ownership advantagesare intangible, inimitable and rareassets owned by the firm. In the OLIparadigm (Dunning, 1980, 1993),ownership advantages are similarlydefined. Therefore, in our analysis wecontrolled for the impact thatresources or ownership advantageshave on firm performance.

We include three other firm-levelcontrols for MNE resources. Our firstproxy is the firm’s technologicalintensity, measured by its R&D

expenditures to sales ratio. As a proxyfor resources arising from marketingexpertise and brand loyalty, we usegeneral administrative expensesdivided by total sales. Sinceadministrative costs can also proxyfor the overhead or fixed costsassociated with the firm’s globalactivities, if there are firm-leveleconomies of scale, administrativecosts should decline as global salesincrease. Hence, we included the ratioof administrative costs to total salesas a control variable. Our third controlfor ownership advantage is firm size,a proxy for economies of scale andscope. We use the natural log of totalassets as our measure of firm size.We did a factor analysis of these threemeasures of ownership advantages;the resulting Cronbach’s alpha valuewas .6538, suggesting that these threevariables do load on a commonconstruct, the competitive/ownershipadvantage of the MNE. In addition,the firm’s debt-equity ratio is includedas a control variable to capture aportion of firm’s value and financialindebtedness. Finally, we alsocontrolled for industry effects byincluding a set of industry dummyvariables, based on the primary SICcode reported by each MNE: Food,Wood, Metal and Chemicals. Weomitted one industry category,machinery.

Data AnalysisBecause our sample is a panel

data set, we tested our hypothesesusing pooled, cross-section time-seriesregression techniques. First, we ran aseries of F-tests in order to determine

100


h e

M u

l t i

n a

t i o

n a

l B

u s

i n

e s

s R

e v

i e

wif there were significant year or firm-specific effects for which we neededto account. The results indicatedsignificant period effects; hence, weincluded four year dummy variablesin our regressions. We alsostandardized all variables (except thedependent and dummy variables) tomean 100 to correct for potentialmulticollinearity. Standardizingreduced the variance inflation factors(VIFs) for most variables toapproximately 1.0. Our initial tests alsoindicated evidence ofheteroscedasticity but notautocorrelation; therefore, we ranOLS regressions with White-corrected standard errors to correctfor heteroscedasticity providing robustvariance estimates.

RESULTS AND DISCUSSIONDescriptive statistics and

correlations for the standardizedvariables are presented in Table 1. Ourregression results are presented inTable 2, grouped by performancemeasures. The first column in eachgroup shows the results for each ofthe individual multinationalitycomponent measures (foreign salesratio, foreign assets ratio, and numberof foreign countries). The secondcolumn replaces these three variableswith the single Multinationality index,and the third column introduces thesquared term of the multinationalityindex (Multinationality Squared).

Previously, we argued that ROAand ROE are measures of short-termperformance while EMV and AMVare long-term measures. In general,our results can be grouped based on

these two sets of performancemeasures. Because our independentvariables are centered at 100, eachbeta coefficient should be interpretedas the conditional effects of aparticular predictor at the mean of allthe other predictors (Aiken & West,1991)

Hypothesis 1 focuses on thediffering impacts that breadth anddepth should have on multinationality.Table 2 shows that only Country Scopeis positively and significantly relatedto performance across all fourmeasures. The signs on ForeignMarket Penetration are not significantin the short-run performance runs, andchange direction in the long-runregressions. Foreign ProductionPresence is significant, and negative,only in the ROA regression. Thissuggests that breadth has more impacton performance than does depth,supporting hypothesis 1.

We individually regressedquadratic forms of the threeMultinationality variables on our fourperformance measures to seewhether the individual Multinationalityvariables were linear in theirrelationships with firm performance.The results show that Foreign MarketPenetration is significant and has a U-shaped relationship for all fourperformance measures (i.e., thecoefficient on Foreign MarketPenetration is significant and negativewhile the coefficient on ForeignMarket Penetration-Squared issignificant and positive). This suggeststhat higher foreign market penetrationratios are positively related to firmperformance, at least after some

101

Volum

e 12 Num

ber 1S

pring 2004T h e M u l t i n a t i o n a l B u s i n e s s R e v i e w

TABLE 1Descriptive Statistics and Pairwise Correlation Coefficients

Mean S.D. Min. Max. 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17

1 Return on Assets 5.89 6.73 -24.45 36.192 Return on Equity 13.25 26.78 -319.27 184.18 0.64*3 Excess Market Value 1.57 3.24 -1.82 40.6 0.3* 0.12*4 Average Market Value 2.51 3.53 0 36.17 0.37* 0.13* 0.96*5 Foreign Market Penetration 100.00 0.17 99.65 100.57 0.08* 0.07 0.02 0.046 Foreign Production Presence 100.00 0.14 99.69 100.5 0.02 0.07* -0.04 -0.00 0.82*7 Country Scope 100.00 9.74 83.32 128.32 0.15* 0.17* -0.01 -0.02 0.51* 0.46*8 Multinationality 100.00 4.91 91.39 114.2 0.15* 0.17* -0.01 -0.02 0.54* 0.49* 1.00*9 Multinationality Squared 10024.07 1000.99 8352.59 13039.83 0.15* 0.17* -0.00 -0.01 0.541* 0.49* 1.00* 1.00*10 R&D/Sales 100.00 0.04 99.95 100.15 0.11* -0.03 0.46* 0.40* 0.26* 0.04 0.13* 0.13* 0.13*11 Administrative Costs 100.00 0.13 99.77 100.32 0.29* 0.19* 0.37* 0.37* 0.34* 0.21* 0.28* 0.28* 0.28* 0.60*12 Firm Size 100.00 1.25 96.83 103.98 -0.11* 0.02 -0.34* -0.39* 0.14* 0.12* 0.45* 0.45* 0.44* -0.22* -0.33*13 Debt/Equity 100.00 96.82 -341.88 1169.62 -0.18* 0.16* -0.18* -0.21* -0.17* -0.09* 0.07* 0.07 0.07 -0.23* -0.17* 0.32*14 Food Industry 0.09 0.28 0 1 0.19* 0.19* -0.00 0.05 -0.04 0.01 -0.03 -0.03 -0.03 -0.26* 0.11* 0.00 0.11*15 Wood Industry 0.07 0.26 0 1 -0.11* -0.07* -0.11* -0.11* -0.13* -0.03 -0.14* -0.14* -0.14* -0.21* -0.20* -0.01 0.09* -0.09*16 Chemicals Industry 0.32 0.47 0 1 0.20* 0.16* 0.05 -0.00 0.11* 0.07 0.24* 0.24* 0.24* -0.01 0.05 0.24* -0.08* -0.21* -0.19*17 Metal Industry 0.06 0.24 0 1 -0.10* -0.13* -0.07* -0.07 -0.04 0.05 -0.00 -0.00 0.00 -0.19* -0.24* -0.05 -0.02 -0.08* -0.07 -0.17*18 Machinery Industry 0.46 0.50 0 1 -0.19* -0.16* 0.05 0.06 0.01 -0.08* -0.13* -0.13* -0.13* 0.36* 0.11* -0.21* -0.03 -0.27* -0.26* -0.64* -0.23*

Note: *Indicates significance at the .05 level.

102


h e

M u

l t i

n a

t i o

n a

l B

u s

i n

e s

s R

e v

i e

wminimum level of foreign involvement.Foreign Production Presence has thetraditional inverted-U shape for theROA and ROE regressions, but isinsignificant in the EMV and AMVregressions. The coefficients onCountry Scope and Country ScopeSquared, on the other hand, are notsignificantly different from zero in allfour regressions, implying thatCountry Scope is linearly related toperformance. Thus, we find limitedsupport for our hypothesis that breadth(Country Scope) is more important forfirm performance than depth (ForeignMarket Penetration and ForeignProduction Presence).

Hypothesis 2 argues that thereis an inverted-U relationship betweenmultinationality and firm performance.Multinationality is significant andpositive in regressions using each ofthe four performance measures.However, when the squared term isadded in the ROA and ROE runs,neither Multinationality norMultinationality Squared is significant.The change in F-statistic is notsignificant in either regression whereMultinationality Squared is added. Wealso tested (available on request), thechange in F statistic when squaredterms for the three individualmultinational components are addedto the ‘Multinationality Variables’regression (column 1). The change inthe F-statistic is significant at the 5%level for both the ROA and ROEregressions. Taken together, theseresults suggest that the relationshipbetween the multinationality index andfinancial performance is positive, butlinear, even though individual

components may have a nonlinearrelationship with financialperformance.

In the EMV and AMVregressions, when the squared termis added, the sign on Multinationalityswitches to negative, whileMultinationality Squared is positive.The change in F statistic is significantat the .001 level, supporting hypothesis2 that the relationship is nonlinear.However, the relationship appears tobe U-shaped for EMV and AMV, firstfalling and then rising. In other words,multinationality initially has a negativeimpact on firm performance and then,as firms continue to expandinternationally, the impact becomespositive. The change in F statisticwhen squared terms for the threeindividual multinational componentsare added to the ‘MultinationalityVariables’ regression (column 1) issignificant at the .001 level for bothEMV and AMV.

Looking specifically at the thirdregression for EMV, the coefficienton Multinationality is -2.2487 and forMultinationality Squared is +0.01188,implying a slope of dEMV/dMultinationality = -2.2487 +2(.01188) Multinationality. Substitutingin different values of Multinationality,we see that the relationship is U-shaped, bottoming out atMultinationality = 94.6. Hence, our U-shape is increasing, not decreasing, inMultinationality, contrary to otherstudies finding an inverted-Urelationship and contrary toHypothesis 2. Employing the sametechnique for the AMV regressionsuggests a similar pattern. We

103

Volum

e 12 Num

ber 1S

pring 2004T h e M u l t i n a t i o n a l B u s i n e s s R e v i e w

TABLE 2MULTINATIONALITY AND FIRM PERFORMANCE

SHORT-TERM PERFOMANCE LONG-TERM PERFORMANCEReturn on Assets Return on Equity Excess Market Value Average Market Value1 2 3 1 2 3 1 2 3 1 2 3

Foreign Market

Penetration 1.38 1.07 2.52* -2.96*Foreign Production

Presence -6.99* -1.03 0.64 1.81Country Scope 0.14*** 0.39*** 0.04** 0.05***Multinationality 0.2*** -1.44 0.76*** -5.21 0.04† -2.25*** 0.07** -1.66**Multinationality Squared 0.01 0.04 0.02*** 0.01**R&D/Sales -1.21 1.95 -2.71 -51.68† -50.94* -64.09* 34.44*** 33.52*** 33.26*** 30.84*** 29.34*** 28.88***Administrative Costs 7.93** 7.03** 7.55** 30.48** 30.36** 31.06** -0.68 -1.07 -1.28 -0.78 -1.08 -1.18Firm Size -0.85** -0.82** -0.95*** -2.83** -2.82** -3.33*** -0.79*** -0.82*** -0.84*** -1.04*** -1.08*** -1.11***Debt/Equity -0.01*** -0.01*** -0.01*** 0.06* 0.06* 0.06* -0.01 0.01 0.01 -0.01* -0.01 -0.01Constant -32.91 -827.71 -335.15 2370.95 2273.64 3807.84† -3111.65*** -3166.62*** -3007.52*** -2789.18*** -2724.83*** -2582.25***

No of Obs 755 755 755 755 755 755 755 755 755 755 755 755R-squared .22 .21 .22 .17 .17 .18 .32 .31 .32 .31 .31 .31F statistic 20.20*** 22.69*** 20.61*** 14.31*** 15.86*** 13.85*** 15.34*** 16.52*** 19.33*** 13.84*** 15.28*** 15.63***Chg. in F 1.34 0.62 16.57 *** 8.26 **

Note: Asterisks show significance levels using a 2-tailed t-test where † < .10, * < .05, ** < .01, *** < .001. Industry dummy variables are omitted from table to conserve space.The coefficients are available upon request from the authors.

104


h e

M u

l t i

n a

t i o

n a

l B

u s

i n

e s

s R

e v

i e

wlinear, but inverted-U shaped,relationship between multinationalityand performance. However, they areconsistent with hypothesis 3, whichargues that the impact ofmultinationality on firm performanceis time dependent. We argued thatEMV and AMV measure long-runexpected performance of the firm,taking into account both the intangibleand tangible assets of the firm. Bothare proxies for Tobin’s q, whichcaptures the potential rents from thefirm’s intangible assets. ROA andROE, on the other hand, are moredirectly related to short-run financialperformance based on the firm’scapital assets. That is, EMV andAMV are future oriented, while ROAand ROE focus on last year’sprofitability. In this sense, EMV andAMV better capture the potentialbenefits from multinationality,particularly when the firm also hashigh ownership advantages. Weillustrate this argument in Figure 3below.

Considering the results from allof the tests together, we find thatthere is a non-linear, three-stagesigmoid relationship betweenmultinationality and performance.Thus, the relationship suggested inFigure 2 is only part of the picture (thelast two stages). Figure 3 indicates thetotality of the multinationality-performance relationship (all threestages). Initially, multinationality hasa positive impact on performance;however, in time, the costs ofmultinationality outweigh the benefits.But in the long-run, multinationalityleads to positive performance. Our

illustrate the EMV-Multinationalityrelationship in Figure 2.

Hypothesis 3 focuses on thediffering impact that multinationalityshould have on firm performance withrespect to time. Some evidence of adifference is already suggested above,where the ROA and ROE regressionsshow a positive but linear relationshipbetween multinationality andperformance, while the EMV andAMV regressions show a U-shapedrelationship. We also conducted a testof Hypothesis 3 using spline analysis.Table 3 shows the beta coefficientsfor the four performance variables,with Multinationality split into threesegments (Riahi-Belkaoui, 1998).

The spline coefficients in theROA and ROE regressions aresignificant (and positive) in only onesegment, suggesting a linearrelationship between multinationalityand these performance ratios. Thespline coefficients in the EMV andAMV regressions, on the other hand,are significant and first positive, thennegative, and then positive. Thissuggests a sigmoid multinationality-performance relationship. Since thecoefficients are the slopes of theEMV-Multinationality relationship,conditional on all other variables beingat their mean values, these EMVregressions provide further evidencethat multinationality positively affectsmarket performance in the long-term.In other words, multinationality is non-linear and increasing in its relation withEMV and AMV, but not with ROAor ROE.

Our results are surprising giventhat other studies have found a non-

105




Table 3Spline Analysis of Multinationality-Performance Relationship

Return on Return on Excess Market Average Market Assets Equity Value Value

MultinationalitySP1 0.08 0.56 0.13 0.25*MultinationalitySP2 0.38* 1.35* -0.16† -0.17†MultinationalitySP3 0.1 0.42 0.14*** 0.18***R&D/Sales 2.41 -50.08† 33.18*** 28.65***Administrative Costs 6.96** 30.27** -1.03 -0.97Firm Size -0.82** -2.86** -0.81*** -1.08***Debt/Equity -0.01*** 0.06* 0.01 -0.01Constant -856.31 2219.72 -3145.31*** -2683.08***

No of Obs 755 755 755 755R-squared .21 .18 .32 .31F statistic 20.16*** 15.02*** 19.54*** 16.72***

Note: Asterisks show significance levels using a 2-tailed t-test where † < .10, * < .05, ** < .01,*** < .001. Industry dummy variables are omitted from table to conserve space. Thecoefficients are available upon request from the authors.

106


h e

M u

l t i

n a

t i o

n a

l B

u s

i n

e s

s R

e v

i e

wresults suggest that the long-runexpected market returns might bemuch larger than the short-run actualfinancial returns to multinationality,and that even the largest USmanufacturing multinationals may nothave exhausted the returns fromincreasing the depth and breadth oftheir international operations.

CONCLUSIONThis study has focused on

unraveling the debate over the shapeof the multinationality-performancerelationship. Previous research hasproduced mixed results, with somestudies finding a positive, linearrelationship and more recent ones asigmoid relationship. We argued thatthere were three partial explanationsfor the confusing results in theliterature. First, the term‘multinationality’ means differentthings to different people, so thatconflicting results are partly due todifferent understandings and proxymeasures. Second, the theoreticalbenefits and costs of multinationalityto firms, and how they are reflectedin firm performance, are not wellunderstood. Third, we argued that timematters. The multinationality-performance relationship is differentfor short-run, as compared to long-run, performance.

Our results provide limitedsupport for the hypothesis that theconstruct is composed of severaldimensions and that breadth has moreimpact than depth on firmperformance. Future research on themul t inat ional i ty-performancerelationship should be careful to

consider and explain carefully thedimensions of multinationality in bothempirical and theoretical studies.

Perhaps the most importantfinding of our study is that themul t inat ional i ty-performancerelationship is non-linear but mostimportantly in the long term. Indeed,our results indicate that there are initialbenefits from multinationality that arethen outweighed by rising costs ofgoing abroad; however, over time, thelong-run benefits dominate the costs,suggesting a significant, positiverelationship between multinationalityand long-run market performance.

Based on our empirical results,we speculate that most USmanufacturing multinationals may nothave reached their optimal degree ofmultinationality; there are stilladditional gains from internationaldiversification to be garnered by USmultinationals. To the degree that thisis true, the future is very bright forthe internationalization of firms. At thesame time, we do not know, based onthe results of this study, how long itmay actually take for firms to receivethe greatest benefits fromglobalization. Much more empiricaland theoretical research is needed tobetter understand this relationship. Weneed to know, for example, how longit might take for the benefits tooutweigh the costs. In order to gainthis understanding, longitudinal studiesover a much longer time horizon arestill needed. In addition, we do notknow how future protectionistmeasures and backlashes toglobalization may affect the ability ofmultinationals to fully reap the benefits

108


h e

M u

l t i

n a

t i o

n a

l B

u s

i n

e s

s R

e v

i e

wof internationalization.

Our results make an importantcontribution to the internationalbusiness and management literatureby demonstrating that themul t inat ional i ty-performancerelationship is much different thanwhat has traditionally been theorized.Instead of being a simple linear oreven curvilinear relationship, themul t inat ional i ty-performancerelationship contains multiple peaksand valleys as organizations adapt tothe constant challenges andopportunities that internationaldiversification presents. This hasbroader implications for strategicmanagement because so often thetemporal dimension of strategies is notwell emphasized.

These results have importantimplications for managers. Thebusiness press, managementconsultants, and others havetrumpeted the benefits of globalizationand international diversification fortwo decades now. On the other hand,strong opposition to the globalizationof firms and markets has beenexpressed as well. Our studyhighlights that it is over the longer termthat the returns from multinationalityto firms is shown in stronger marketperformance. These results indicatethat harvesting real benefits frominternational diversification mayrequire patience on the part ofmanagers.

Future research should continueto explore the temporal dimension ofthe multinationality-performancerelationship. For example, studying theeffects of international diversification

on firm performance over longer timehorizons (e.g., thirty or more years)could further validate the results ofthis study. In addition, researchersshould explore this relationship incontexts other than U.S. basedmultinationals. For example, it isimportant to determine if the resultsof this study are applicable toemerging market multinationals.

REFERENCESAiken, L. S., & West, S. G. 1991. Multiple

regression: Testing andinterpreting interactions. NewburyPark, CA: Sage Publications.

Aleson, M. R., & Escuer, M. A. E. 2001.The effect of internationaldiversification strategy on theperformance of Spanish—basedfirms during the period 1991-1995.Management International Review,41: 291-315.

Allen, L., & Pantzalis, C. 1996. Valuationof the operating flexibility ofmultinational corporations. Journalof International Business Studies,27: 633-653.

Annavarjula, M., & Beldona, S. 2000.Multinationality-performancerelationship: A review andreconceptualization. InternationalJournal of OrganizationalAnalysis, 8: 48-67.

Barkema, H. G., & Vermeulen, F. 1998.International expansion throughstart-up or acquisition: A learningperspective. Academy ofManagement Journal, 41: 7-26.

Barney, J. B. 1991. Firm resources andsustained competitive advantage.Journal of Management, 17: 99-120.

Buhner, R. 1987. Assessing internationaldiversification of West Germancorporations. StrategicManagement Journal, 8: 25-37.

109




Chase, C. D., Kuhle, J. L., & Walther, C. H.1988. The relevance of political riskin direct foreign investment.Management International Review,28: 31-38.

Contractor, F. J., Kundu, S. K., & Hsu, C.C. 2003. A three-stage theory ofinternational expansion: The linkbetween multinationality andperformance. Journal ofInternational Business Studies, 34:5-18.

Daniels, J. D., & Bracker, J. 1989. Profitperformance: Do foreign operationsmake a difference? ManagementInternational Review, 29: 46-56.

Dunning, J. H. 1980. Toward an eclectictheory of international production:Some empirical tests. Journal ofInternational Business Studies, 11:9-31.

Dunning, J. H. 1993. Multinationalenterprises and the globaleconomy. Harlow, UK: Addison-Wesley.

Geringer, J. M., Beamish, P. W., &daCosta, R. C. 1989. Diversificationstrategy and internationalization:Implications for MNE performance.Strategic Management Journal, 10:109-119.

Goerzen, A., & Beamish, P. W. 2003.Geographic scope and multinationalenterprise performance. StrategicManagement Journal, 24: 1289-1306.

Gomes, L., & Ramaswamy, K. 1999. Anempirical examination of the form ofthe relationship betweenmultinationality and performance.Journal of International BusinessStudies, 30: 173-188.

Grant, R. M. 1987. Multinationality andperformance among Britishmanufacturing companies. Journalof International Business Studies,18: 79-89.

Grant, R. M., Jammine, A. P., & Thomas,H. 1988. Diversity, diversification,and profitability among Britishmanufacturing companies, 1972-84.Academy of Management Journal,31: 771-801.

Han, K. C., Lee, S. H., & Suk, D. Y. 1998.Multinationality and firmperformance. MultinationalBusiness Review, 6(2): 63-70.

Hitt, M. A., Hoskisson, R. E., & Ireland,R. D. 1994. A mid-range theory ofthe interactive effects ofinternational and productdiversification on innovation andperformance. Journal ofManagement, 20: 297-326.

Hitt, M. A., Hoskisson, R. E., & Kim, H.1997. International diversification:Effects on innovation and firmperformance in product-diversifiedfirms. Academy of ManagementJournal, 40: 767-798.

Kim, W. C., Hwang, P., & Burgers, W. P.1989. Global diversification strategyand corporate profit performance.Strategic Management Journal, 10:45-57.

Kogut, B. 1984. Normative observationson the international value-addedchain and strategic groups. Journalof International Business Studies,15: 151-167.

Kostova, T. 1999. Transnational transferof strategic organizational practices:A contextual perspective. Academyof Management Review, 24: 308-324.

Kostova, T., & Roth, K. 2002. Adoptionof an organizational practice bysubsidiaries of multinationalcorporations: Institutional andrelational effects. Academy ofManagement Journal, 45: 215-233.

Kostova, T., & Zaheer, S. 1999.Organizational legitimacy underconditions of complexity: The caseof the multinational enterprise.

110


h e

M u

l t i

n a

t i o

n a

l B

u s

i n

e s

s R

e v

i e

wAcademy of Management Review,24: 64-81.

Kotabe, M., Srinivasan, S. S., & Aulakh,P. S. 2002. Multinationality and firmperformance: The moderating role ofR&D and marketing capabilities.Journal of International BusinessStudies, 33: 79-97.

Luo, Y. 1998. Timing of investment andinternational expansionperformance in China. Journal ofInternational Business Studies, 29:391-407.

Mascarenhas, B. 1992. Order of entry andperformance in internationalmarkets. Strategic ManagementJournal, 13: 499-510.

Mishra, C. S., & Gobeli, D. H. 1998.Managerial incentives,internalization, and market valuationof multinational firms. Journal ofInternational Business Studies, 29:583-597.

Morck, R., & Yeung, B. 1991. Whyinvestors value multinationality.Journal of Business, 64: 165-187.

Qian, G., & Li, J. 2002. Multinationality,global market diversification andprofitability among the largest USfirms. Journal of Business Research,55: 325-335.

Riahi-Belkaoui, A. 1998. The effects of thedegree of internationalization on firmperformance. InternationalBusiness Review, 7: 315-321.

Rugman, A.-M. 1976. Risk reduction byinternational diversification.Journal of International BusinessStudies, 7: 75-80.

Ruigrok, W., & Wagner, H. 2003.Internationalization andperformance: An organizationallearning perspective. ManagementInternational Review, 43: 63-83.

Sullivan, D. 1994a. Measuring the degreeof internationalization of a firm.Journal of International Business

Studies, 25: 325-342.Sullivan, D. 1994b. The “threshold of

internationalization”: Replication,extension, and reinterpretation.Management International Review,34: 165-186.

Sullivan, D. 1996. Measuring the degreeof internationalization of a firm: Areply. Journal of InternationalBusiness Studies, 27: 179-192.

Sundaram, A. K., & Black, J. S. 1992. Theenvironment and internalorganization of multinationalenterprises. Academy ofManagement Review, 17: 729-757.

Tallman, S., & Li, J. 1996. Effects ofinternational diversity and productdiversity on the performance ofmultinational firms. Academy ofManagement Journal, 39: 179-196.

Zaheer, S. 1995. Overcoming the liabilityof foreignness. Academy ofManagement Journal, 38: 341-363.

what is the shape of the multinationality …...diversification through foreign direct investment...

Documents