Size and marketing centralization in multinational companies

Journalof Market FocusedManagement,1.231-247 (1996) @ 1996Kluwer Academic Publishers,Boston. Manufacturedin The Netherlands.

Size and Marketing Centralization in Multinational Companies PASCALE G. QUESTER Department of Commerce,

[email protected] University

of Adelaide,

Adelaide,

SA Australia

5005

JODIE CONDUIT Monash University

Abstract

A large-scale mail survey of some 200 Australian subsidiaries of multinational organizations aimed to identify a possible relationship between organizational size-related variables and the degree of centralization enforced by the multinational in relation to marketing decision making. The findings-that size and centralization are not correlated for each marketing mix variable-enhance previous literature on this issue, whose studies combined all marketing tasks with inconsistent results. This suggests that using organizational size-related variables as means of assigning the appropriate level of decision-making autonomy to each of the traditional marketing areas is not appropriate. Keywords:

centralization, organizationalsize, multinational companies,Australia

1. Introduction Centralization is one of the structural elements of any multinational organization. It refers to the division of the decision-making authority between a local subsidiary and an outside organization controlling it (parent company). In marketing terms, centralization refers to the degree to which the local management of a subsidiary operation is involved in, or precluded from, the decision-making process relating to marketing strategies, plans, and tactics. In an era of increasing competition, it is essential to assign the appropriate decision-making autonomy to subsidiaries to allow reactions to changes in the local environment to be timely, without losing sight of the corporate strategic direction. Centralization has been an area of considerable research interest (Picard, 1977; Goehle, 1980; Hedlund, 1981; Garnier, 1982; Gates and Egelhoff, 1986). To date, however, researchers have demonstrated contradictory hypotheses and reported inconsistent results regarding centralization (Gates and Egelhoff, 1986). These inconsistencies may be attributed to previous researchers who studied marketing as an entirety, without considering the possibility that the nature of the size and centralization relationship may differ for each marketing mix variable. With the emerging globalization trend, particularly market expansion into Asia, the relationships associated with centralization need to be clarified.

232

PASCALE G. QUESTER AND JODIE CONDUIT

Centralization is not an end in itself. Rather, it enablesthe organization to mobilize its resourcesin order to reachgiven objectives. Although theseobjectivesare many, they ultimately relate either to the maximization of profits or to the minimization of risk. Whereas multinational companiesstandardizewith the aim of maximizing their profits, centralization is often undertaken to minimize risk. Three elements of risk are important in the headquarters-subsidiaryrelationship and have varying impacts on the degreeof centralization (Garnier, 1982): l

l

l

The greaterdependencea parent companyhason its foreign operations,the greaterthe risk to the parent company and the stronger the tendencyto centralize all decisions at headquarters. The greaterthe degreeof foreignnessof the host country environment,the larger the risk of misinterpreting messagesfrom customersand suppliers or of misunderstandingthe technical languageand so forth at headquarters,andthereforethe strongerthe tendency to leave decision making to the executiveswithin the subsidiary. The greatervariability and unpredictability of changesin the subsidiary’s environment, the more important it is that it react quickly and that decisions be made at subsidiary level.

Other benefitsarisefrom centralizing a multinational company’sdecision-makingauthority as resourcescan be pooled, and synergy among operationscan be gained. Other benefits, similar to thoseachievedthrough standardization,may also be more likely (Daniels, 1987). Adversariesof centralizationbelievethat the lack of clear-cutresponsibility at a subsidiary level can be detrimental. If all decisions are made by headquarters,logical thinking and rationality by the local managerswill be discouraged. As decisions are not discussed, managersfeel isolated, and the perceived need for interaction disappears. Although top managersat headquarterslevel arefree to be assertive,the lower managershavelittle input, and this may breedresentment(Miller, 1987). Aylmer (1970) concluded that marketsgenerally tended to be decentralized. However, in more recent times, technology hasfacilitated a trend toward centralization, just as it has for standardization(Daniels, 1987). Trends in centralization have often reflected those in the level of standardization. Decisions related to the product itself are more centralized while advertising, pricing, and distribution decisions are relatively decentralized(Aylmer, 1970; Brandt and Hulbert, 1977). Theseare similar to the relative rankings found for the level of standardization,and this is why it is generally acceptedthat the two are correlated, even in the absenceof empirical evidence. It is this assumedcorrelation that has led to the relationships apparentto the level of standardizationsometimesbeing applied to the level of centralization without question. 2.

Factors affecting centralization

Although previous researchhas highlighted trends in the level of centralization adopted by organizations(Aylmer, 1970;Brandt and Hulbert, 1977),they provide little insight into

SIZE AND MARKETING

CENTRALIZATION

IN MULTINATIONALS

233

the reasons why firms settle on a given level of centralization. Mixed results obtained when examining the effect of size on the level of centralization for a multinational company suggeststhat other factorsmust havean influence on the level adopted.Thesefactorscan be categorizedasrelation to (1) the multinational companyasan entirety and (2) the subsidiary itself. Someof thesefactors are outlined below.

2.1. Multinational company-relatedfactors The extent to which the workflow of the parentcompanyand its subsidiariesare integrated has a definite impact on the level of centralization. A significant positive relationship was found betweenthe level of centralization and the degreeof integration of activities of memberswithin the group (Gamier, 1982). The ownership policy and legal dependence, representedby the percentageof the subsidiaries’equity held by the parent and the number of parent company representatives on the affiliates board, were also found to be significantly

and positively correlatedwith centralization (Gamier, 1982;Gatesand Egelhoff, 1986). Conversely,multinational companiesmay decentralizeasthe extent of foreign acquisition increases,for one of two reasons:(1) the parentcompanyencountersdifficulties in gaining control over a subsidiary with an independenthistory and structure,or (2) acquisitions are often undertakenby theparentcompanyto obtainexpertisetheydo not haveandthereforeare reluctant to interfere with (Hedlund, 1981;GatesandEgelhoff, 1986). Thereis alsonegative correlation betweenthe level of centralization and both the extent of foreign acquisitions as well as the number of years a corporation has controlled overseassubsidiaries(Gatesand Egelhoff, 1986).

2.2. Subsidiary-related factors The operational independenceand ageof a subsidiary havebeenfound to havea significant negative relationship with the level of centralization applied to a subsidiary (Gates and Egelhoff, 1986; Garnier, 1982; Sim, 1977). Environmental factors and product-related factorshave,surprisingly, beenfound to havelittle impact overthe degreeto which decisionmaking authority is delegatedto foreign subsidiaries(centralization) (Garnier, 1982; Gates and Egelhoff, 1986; Kirpalani, 1988). This is of particular interest as these factors are dominant in determining the level of standardizationbut not the level of centralization, Product type (industry), country of origin, and number of product lines were also among variables found to have little or no effect on the level of centralization (Kirpalani, Laroche, and Darmon, 1988; Sim, 1977; Gamier, 1982). It is important to determine the effect that somesubsidiary-relatedvariables have on the overall degreeof centralization because,ultimately, eachspecificcompanyshould be able to deducethe desirablelevel of centralization for their situation. Moreover, it is important for the benefit of local managersto determinewhat type of size and centralization relationship is exhibited within Australian firms.

234 3.

PASCALE G. QUESTER AND IODIE CONDUIT

Centralization

and size

It is essentialto determinethe appropriatelevel of centralization within an organization to ensurereactionsto changesin the local environment are timely, without losing sight of the corporation’s strategicdirection. To date,therehasbeenno consensusasto the relationship between the size-related variables of organizations and the level of centralization they implement. Centralization is generally believed to have a negative relationship with the size of a company, although no conclusive evidence has been presentedto support this (Pugh, Hickson, Hinings, and Turner, 1969; Blau and Schoenherr, 1971). Organization theorists commonly agree that (1) increasedsize leads to more structuring of activities, which then facilitates decentralization (Pugh et al., 1969) or (2) size forces managersto decentralize by overloading the company with too many decisions (Blau and Schoenherr, 1971). The absolute size of a subsidiary, absolute size of a multinational corporation, relative size of a subsidiary, and relative size of foreign operations have been previously testedto determinetheir impact on the level of centralization (Picard, 1977;Goehle, 1980; Gamier, 1982; Hedlund, 1981; Gatesand Egelhoff, 1986). Below is a summary of the general beliefs held by researchersrelating the size of the companyto the level of centralization andthe reasoningbehind their beliefs. Also presented is a brief summary of more relevant and comprehensiveprevious researchfindings, which highlights the inconsistenciesthat have emergedwhen thesebeliefs have been tested (see Table 1). Reportedcentralization in theseinstancesrelatesto marketing variablesasa whole and has not beendisaggregatedinto various marketing mix variables. Table 1. Previousresearchon the centralizationand size relationship.

Researcher

Relative Size of Foreign Operations

Gatesand Egelhoff (1986)

-

Gamier (1982)

+

Hedlund (1981) Goehle (1980)

lncl

Picard (1977)

n.s.

m.r = mixed results n.s. = not significant

3.1.

Absolute Size of a Multinational

Relative Size of Subsidiary

Absolute Size of Subsidiary

m.r

m.r

-t

+

ns.

n.s.

Incl

-

-

n.s.

ns.

n.s.

-

-

- = negativerelationship + = positive relationship

Incl = inconclusive

Absolute size of the subsidiary

Centralization is negatively correlated with the absolutesize of a subsidiary. The larger the subsidiary, the greater ability it has to support a full management staff (Picard, 1977). The

SIZE AND MARKETING

CENTRALIZATION

IN MULTINATIONALS

235

increasedresourcesenable it to be possible for more decisions to be made at subsidiary level. 3.2. Absolute size of a multinational Centralization is negatively correlated with the size of a multinational company. Larger multinational companieshavemany more managers,departments,and contingencies,and, therefore, becausethe operationsare more difficult to control, a high degreeof autonomy is granted (Gamier, 1982;Miller, 1987). However,the barriers to centralization-namely, the inability to monitor the subsidiary’s behavior and the inability of managersto manage large conglomerates-are being erodedasmoderntechnologyimprovesandmanagersskills increase. 3.3. Relative size of subsidiary Centralization is negatively correlated with the relative size of the subsidiary. The larger the size of the subsidiary relative to the parent firm, the betterbargaining position it has. Its bargaining strength can lead to a more decentralizedpattern of decision making (Garnier, 1982; Aylmer, 1970). 3.4. Relutive size of foreign operations Centralization is positively correlatedwith the relative size of foreign operationsin multinational companies. As parent companiesbecomemore reliant on revenue from foreign operations, the dependencerisk increases,and the parent maintains control through increasedcentralization (Garnier, 1982;Picard, 1977; Goehle, 1980;Aylmer, 1970). The weak and nonsignificant relationships and mixed results often found in previous researchsuggestthat the relationship betweensize and centralization is not simple (Gates and Egelhoff, 1986). The equivocal findings of previous researchers,however,may be due to the aggregatednature of the research. By disaggregatingthe marketing responsibilities into decisions relating to the various marketing mix elements(product, promotion, price, and distribution), further insight into the relationship betweenthe size of an organization and its level of centralization may be gained. Organizationalsize (salesvolume and number of employees)may exert different influences on various aspectsof the firms’ management processes.It may be that thesevariablesdo not sufficiently explain the different levels of centralization for firms in similar environmentsandmay only be a proxy for someunknown variable (Sim, 1977; Douglas and Wind, 1973-1974). 3.5. The dichotomous relationship of size variables The size of a multinational company and the level of centralization they undertakemay be interrelating variables. This can be illustrated by analysing the relationships the size of a multinational companyhas with the level of centralization.

236

PASCALE

G. QUESTER

AND JODIE CONDUIT

The measurementof the size of a companyis rather arbitrary. However,in keeping with previous research(Aylmer, 1970;Gatesand Egelhoff, 1986),the size variables usedin the following hypotheseswill be defined as follows: Relative size offoreign operations: Percentageof a company’ssalesoccurring outside

the parent company. Absolute size of a multinational:

Total number of employeeswithin the corporation.

Relative size of a subsidiary: A single subsidiary’s salesas a percentageof total sales. Absolute size ofa subsidiary: Number of employeeswithin the subsidiary. General size of the company: A compoundmeasureof the four variables above.

3.6. Size-relatedhypotheses To clarify the natureof therelationship that sizevariableshavewith the level of centralization a firm adopts,separatehypothesesmust be developedto test for correlation. Hl: The level of centralization between a parent company and its subsidiary is negatively correlated with the general size of the company.

It is necessaryto examine the impact on the level of centralization of each size-related contingency variable as a relationship may exist for one variable and not the others, which would not be found if only the generalsize was examined. The effect of each size-related contingency variable, therefore,must also be examinedfor centralization. H 1.1: The level of centralization between a parent company and its subsidiary is positively correlated with the relative size offoreign operations of the corporation.

H1.2: The level of centralization between a parent company and its subsidiary is negatively correlated with the absolute size of the multinational corporation.

H1.3: The level of centralization between a parent company and its subsidiary is negatively correlated with the relative size of the subsidiary.

H1.4: The level of centralization between a parent company and its subsidiary is negatively correlated with the absolute size of the subsidiary.

The testing of the proposedhypotheseswill contributeto expandingthe current knowledge of the concept of centralization.

SIZE AND MARKETING CENTRALIZATION IN MULTINATIONALS

237

4. Methodology 4.1. Sample The study examined 104 Australian companies that were multinational subsidiaries and had their parent company based overseas. This represented a 52 percent response rate from the 200 surveys posted. Another fifty-one companies replied, but their questionnaires were either insufficiently completed or they notified their desire not to participate in the survey. Firms in the service industry were reluctant to respond to the questionnaire, causing a slight nonresponse bias. This may have been caused by the questionnaire design, which was tailored more specifically for manufacturing companies. A description of the nature of the companies included in the sample can be seen in Table 2. The sampling frame used for selecting these companies was obtained from Mucmillun Directory of Multinationals (Stafford and Purkis, 1989). This directory profiles the world’s 450 largest industrial corporations with sales of over $1 billion and significant world investments during 1987. Only those with Australian subsidiaries were selected for the sample. In an effort to reduce bias toward lqge firms and industrial corporations, these companies were supplemented with other multinational companies obtained from lists provided by the Reserve Bank of Australia, the Bureau of Industry Economics, and the Department of Industry and Technology, and the Regional Development Departments of Aerospace, Automotive, Information Industries, Marine and Forest Products, Pharmaceuticals, and Service Industries. Companies were then chosen at random from these lists. Despite an attempt to reduce bias, the sample may have provided some overrepresentation of manufacturing industries in comparison to service industries, as only one service industry firm was among the respondents. Large and successful companies (sales of over $1 billion and significant world investments) were also slightly overrepresented in the sample. Any bias arising from examining only the views of subsidiaries and not that of parent companies was consistent across the sample. Since the hypotheses testing involved correlations and comparisons rather than tests of absolute levels, the results were unaffected by such a bias. As a result, however, conclusions should not be drawn regarding the absolute levels of standardization and centralization in this study.

4.2. Data collection To enable information on the level of centralization a company undertakes to be compiled, company employees were required to make a subjective judgment. Data were collected through a pretested questionnaire addressed to the marketing manager of the firms. A mail survey, including a prepaid return envelope, was used as it had proven successful in previous Australian studies. A first reminder for unreturned responses was sent three weeks after the original survey, followed by a further reminder by way of telephone, and a third questionnaire, three weeks after that. The reminder included another copy of the questionnaire and an accompanying prepaid envelope. These reminders and the inclusion of prepaid envelopes were initiated in an endeavor to reduce the nonresponse bias in the sample.

238

PASCALEG. QUESTERAND JODIECONDUIT

Table 2. Summary of sample statistics.

Country of Parent North America Western Europe Asia United Kingdom

Number 46 22 20 16


Number

O-50 51-100 101-150 151-200 201-250 250-300 301-350 351~00 >400

16 22 12 5 3 10 7 3 26


Number

O-l% 1-2 2-5 5-7.5 7.5-10 >lO N/A

34 33 23 3 5 4 2

Percentage Indust@ 44.23% 21.12% 19.23% 15.38%

Percentage 15.38% 21.15 11.54 4.81 2.88 9.62 6.73 2.88 25.00

Percentage 32.69% 31.73 22.12 2.88 4.81 3.85 1.92

Electrical equipment Mechanical equipment

Chemicals Automotive Other manufacturing Other consumer products Food and beverage Raw materials Services

Absolute Size of Multinational 0-10000 1001-20000 20001-40000 4000160000 60001-80000 SOOOl-lOOdo >100000 N/A

Relative Size of Foreign Operations o-10% 11-25 26-50 51-75 76-95 N/A

28 23 18 10 8 7 6 3 1

Percentage 26.92% 22.12% 17.31% 9.62% 7.69% 6.73% 5.77% 2.88% 0.96%

Percentage 32 20 13 14 7 3 14 1

30.77% 19.23 12.50 13.46 6.73 2.88 13.46 0.96

Number

Percentage

11 15 43 22 10 3

10.58% 14.42 41.35 21.15 9.62 2.88

The questionnaireconsistedof four major parts,two of which are directly relevant to this article (seeAppendix). The first part askedmanagersto provide generalinformation on the company regarding its industry, parent company’scountry, size, and product lines. This information wasneededto assessthe characteristicsof the sampleandto act asindependent variables when testing the hypothesesdetailed above. Parts 2, 3, and 4 concentratedon the marketing mix variables. These variables were representativeof the four P’s of the marketing mix-product, promotion, price, and distribution (placement). The marketing elementsexaminedwere determinedafter a review of previous research(Ozsomer,Bodur, andCavusgil, 1991;SorensonandWiechmann, 1975; Brandt and Hulbert, 1977; Terpstra and Sarathy, 1990; Rau and Preble, 1987), and they

SIZE AND MARKETING CENTRALIZATIONIN MULTINATIONALS

239

were representativeof the major decisions madeby firms in relation to the marketing mix (listed in Table 3). However,this choice may be a sourceof bias within the study. Table 3. Marketing mix variables. Promotion Product Creative expression Brand name Basic advertising message Packaging Media allocation Warranty Salespromotion Productcharacteristics Productpositioning

m Wholesaleprice Retail price Discountsgiven

Distribution Degreeof penetration Managementof salesforce Role of intermediaries Type of intermediaries Role of salesforce

It is important to note that these categoriesare not mutually exclusive. For example, product positioning is included as a product-related variable, yet the positioning of the product will have an effect on the other marketing mix elements. The variables were surveyedin a random order to reduceany bias that may haveotherwise occurred. Specifically, Part 2 required managersto expressan opinion regarding the level of guidance and direction (centralization) their subsidiary received from its parent company. A Likert-type scalewasprovided for eachof the abovemarketing variables,and the manager wasrequired to indicate the appropriatelevel for his or her companybetween1 (no direction received) and 7 (all direction from headquarters). A N/A option was provided for those variables that were unrelated to the specific company. This measurementtechnique had been used by previous researchersto account for the subjective nature of the information required (Sorensonand Wiechmann, 1975;Brandt and Hulbert, 1977;Gatesand Egelhoff, 1986; Hoffman, 1988; Ozsomer,Bodur, and Cavusgil, 1990).

4.3. Statistical analysis of data All the information collected from the returned questionnaireswas enteredonto a spreadsheetusing Excel. The data were then subjectedto various statistical testsusing the Excel program. Size-relatedvariableswherethe questionnairerequired the selection of an appropriate range (such as 0 to 1 percent) were enteredin the spreadsheetas the midpoint of this range (0.5 percent). Open-endedrangeswere enteredas the averageof the elementswithin that range (> 10 percententeredas 2 1 percent). Before testscould be conductedon therelationship betweenthe sizevariablesandcentralization, it wasnecessaryto ensurethat the level of correlation betweenthe sizevariableswas sufficiently low sothat eachvariablecould be viewedasrepresentinga different contingency condition. Therefore, a correlation matrix for the contingency variables was established

240


Table 4. Correlation matrix for contingency variables. Absolute Slze of Subsidiary Absolute Absolute Relative Relative

size of subsidiary size of multinational size of subsidiary size of foreign operation

1.0000 0.3523 0.0980 -0.0912

Absolute Size of Multinational


1.oooo -0.1686 0.0602

1.oooo 0.1359

Relative Size of Foreign Operation

1.oooo

(seeTable 4). The correlations generally were found to be extremely low, and therefore eachvariable was viewed as representinga different contingency variable. Table 5 illustrates the mean and standarddeviation of the level of centralization a subsidiary receivesfrom their parent company for eachof the marketing mix variables. Decisions relating to the product-relatedvariableswere found to be highly centralized, whereas distribution-related decisions were found to be more often the responsibility of the subsidiary. Table 5. Mean rankings for marketing mix variables.

Mean

Standard Deviation

Brand name Product design Packaging Warranty Product positioning Advertising message Wholesale price Creative expression Type of intermediaries Role of intermediaries Degree of penetration Sales promotion Role of salesforce Retail price Discount given Media allocation Management of salesforce

5.5096 5.1287 4.5938 4.4130 3.4700 3.2700 3.1881 2.7857 2.4268 2.3810 2.3457 2.3100 2.2524 2.2245 1.9697 1.9677 1.9515

2.0288 1.8968 2.1986 2.3115 1.6274 1.7615 2.0907 1.7326 1.3154 1.2022 1.4562 1.4562 1.4089 1.1468 1.4728 1.3309 1.3672

Product Promotion Price Distribution

4.6349 2.5934 2.4664 2.2561

2.0896 1.6411 1.7544 1.3472

To test whether the size of a multinational companyhasany effect on the level of centralization, two testswere conducted. Regressionswere run to measurethe proportion of the variation in eachmarketing mix variable that could be explained by the regressionequation (size-relatedvariables). The coefficients and their t-statistics for each size variable were


241

comparedto suggestwhich variable explained the level of centralization to a greaterextent than the others. An analysisof variancewasalsoperformedon eachof the marketing mix variables. To do this, the samplewasdivided accordingto surveyresponsesto the independentsize variables (such as absolute size of a multinational). Whether the meansof the subgroupsdiffered from each other significantly for each of the marketing mix variables was subsequently examined:

Ho: pl =p2=...=p,,. These results were supportedby running a correlation analysis betweeneach of the four size variables and the level of centralization. A high coefficient of correlation would cause the hypothesesto be accepted. 5.

Size and centralization

relationship

5.1. Regression Hypothesis 1 examines the size-related variables and their correlation with the level of centralization. Consequently,a regressionanalysis was run, a summaryof which results is shown in Table 6. Previous researchersexaminedthe relationships betweenthe size variables and the level of centralization and often provided contradictory conclusions. The results suggestthat there was not a significant relationship betweenthe size of a companyor its subsidiary and the level of centralization. The R-squaredvaluescalculatedweretoo low for the size-related variable to be consideredto have explanatory powersregarding the level of centralization. Only one variable (warranty) had an R-squaredgreaterthan 0.1, and thereforeHypothesis 1 (that the two variables are correlated) was not accepted. This finding was almost to be expectedconsidering the inconsistenciesbetweenprevious researchand the predominance of inconclusive and nonsignificant results (seeTable 1). Some trends emergedfrom an examination of the coefficient of the size variable in the regressionanalysisperformedon the level of centralization. The coefficient for the relative size of the subsidiary was often the largestcoefficient of the four variables. Conversely,the absolute size of a multinational had the lowest coefficient. However,the t-statistics for all of the size variableswere low, implying that the perceiveddifferencein explanatory powers was not significant. 5.2. Analysis of variance An analysis of variance, with a confidence level of 0.95, was performed to determine whether the size-relatedvariables, when examined in isolation, were correlated with the level of centralization (seeTable 7 for summary). An analysis of variance determined the effect on the level of centralization of eachsize variable in isolation, whereasin a regression

242


Table 6. Size and centralization regression analysis.

Marketing Mix Variable

R squared



Coefficients Relative Size of Subsidiary


Brand name Packaging Warranty Product design Product positioning Product

0.0079 0.0388 0.1690 0.0120 0.0187 0.0491937

-0.0002 -0.0004 -0.0020 5.94 E-05 0.0003

- 1.08 E-06 -3.04 E-06 2.58 E-06 -1.21 E-06 4.08 E-08

1.827 -4.964 -2.284 3.498 -2.614

-0.220 -0.865 -1.331 -0.352 -0.468

Creative expression Advertising message Media allocation Sales promotion Promotion

0.0525 0.0301 0.063 1 0.0165 0.0405474

-0.0004 0.0003 0.0007 -2.33 E-05

4.33 E-06 2.69 E-06 -2.210 E-06 -1.13 E-07

2.375 -0.049 -4.465 0.693

-1.154 0.479 -0.217 -0.792

Wholesale price Retail price Discounts given Price

0.0515 0.0190 0.0343 0.0349364

0.0002 0.0003 -0.0003

-3.39 E-07 1.31 E-06 -7.09 E-08

12.641 -0.404 5.170

-0.539 -0.179 -0.641

Degree of penetration Management of salesforce Role of intermediaries ‘Qpe of intermediaries Role of salesforce Distribution

0.0348 0.0689 0.0499 0.0541 0.0240 0.0463355

-0.0002 -0.0003 0.0003 0.0002 6.02 E-05

- 1.92 E-06 2.32 E-06 -2.05 E-06 -1.54 E-06 1.85 E-06

5.000 7.258 3.509 6.531 4.781

-0.370 - 1.005 -0.383 -0.496 -0.310

analysis the direction and magnitude of the effect of individual variables may negate each other. The average probability value for each of the size variable was greater than 0.45, implying that the data was derived from populations with the same mean. This lack of significance indicates that the level of centralization was not explained by, or correlated with, any of the size variables in isolation, and, therefore, Hl. 1, H1.2, H1.3, and H1.4 were all not accepted. Only three relationships were significant at the 5 percent level (labeled *), and one at the 1 percent level (labeled **). This was not sufficient to accept any of the hypotheses. Even though the relative size of a subsidiary expectably had the lowest average probability value (supporting the findings of the regression analysis), the probability value for the absolute size of a subsidiary was low, and the value for the relative size of foreign operations was high, conflicting with previous trends. 5.3.

Coe&ient

of correlation

Coefficients of correlation were determined to emphasise the weak relationship between the size variables and the level of centralization for each of the marketing mix elements

243


Tuble 7. Size and centralization, analysis of variance probability and values.

Marketing

Mix Variable





Brand name Packaging Warranty Product design Product positioning Average product

0.289 0.5887 0.0014** 0.4423 0.4902 0.3623

0.9502 0.7744 0.5417 0.5115 0.8482 0.7252

0.5254 0.0565 0.1310 0.7036 0.3295 0.3492

0.8379 0.0385’ 0.0399” 0.8688 0.2422 0.4055

Creative expression Advertising message Media allocation Sales promotion Average promotion

0.1275 0.1462 0.4802 0 4733 0 3068

0.6719 0.7353 0.0431* 0.7023 0.5382

0.7306 0.7884 0.9037 0.5448 0.7419

0.1870 0.4067 0.6715 0.6153 0.470 1

Wholesale price Retail price Discounts given Average price

0.6746 0.4321 0.7063 0.6043

0.6820 0.7376 0.6335 0.6844

0.5563 0.5674 0.3121 0.4819

0.7145 0.3334 0.4587 0.5022

Degree of penetration Management of salesforce Role of intermediaries Type of intermediaries Role of salesforce Average distribution

0.7063 0.8737 0.2572 0.3858 0.9868 0.6420

0.6459 0.7760 0.2438 0.1260 0.8184 0.5220

0.0753 0.4811 0.203 1 0.3991 0.5796 0.3476

0.8938 0.5809 0.5297 0.7442 0.8680 0.7233

Average probability

0.4742

0.6142

0.4638

0.5289

* Insignificant at the 5% level. ** Insignificant at the 1% level.

(see Table 8). It can be seen that the coefficients determined suggested a weak correlation between any of the four size variables and the level of centralization. None of the combined average coefficients of correlation, calculated for each of the size variables, were greater than 0.1. Therefore, neither Hypothesis 1 nor any of its subparts could be accepted, and it was found that the absolute size of a subsidiary, the absolute size of a multinational, the relative size of a subsidiary, and the relative size of foreign operations were not correlated with the level of guidance a subsidiary receives from its parent (centralization). Three of the four combined average corrklation coefficients (absolute size of subsidiary, absolute size of multinational, and relative size of foreign operations) exhibited negative correlation (see Table 8 below). This combined tendency for the correlations to be negative could be seen as support for the views of organizational theorists that as the size of the company increases, the parent company is forced to decentralize its operations (Blau and Schoenherr, 1971; Pugh et al. 1969). However, because the relationship examined in this study was not significantly correlated, the results could not be seen to be sufficient support for these ideas. Also, the direction of correlation evident for the relative size of a subsidiary

244


Table 8. Centralization and size.variables coefficients of correlation. Absolute Size of Subsidiary


Brand name Packaging Warranty Product design Product positioning Average product

-0.0918 -0.1232 -0.4066 -0.0877 0.0916 -0.1235

-0.0558 -0.1403 -0.0755 -0.1135 0.0411 -0.0688

-0.0411 -0.0770 -0.1401 -0.007 1 -0.0778 -0.0686

0.0034 -0.0928 -0.0978 -0.0729 -0.0777 -0.0676

Creative expression Advertising message Media allocation Sales promotion Average promotion

-0.0426 0.0634 0.1868 -0.0305 0.0443

0.1478 0.1277 -0.0524 -0.0361 0.0463

-0.0600 -0.0825 -0.0440 -0.0424 -0.0572

-0.1370 0.0518 -0.0737 -0.1514 -0.0776

Wholesale price Retail price Discounts given Average price

0.0239 0.0919 -0.0951 0.0069

-0.0573 0.0670 -0.0769 -0.0224

0.1287 -0.0096 0.1191 0.0794

-0.0895 -0.0167 -0.0761 -0.0607

Degree of penetration Management of salesforce Role of intermediaries Type of intermediaries Role of salesforce Average distribution

-0.0767 -0.0570 0.1000 0.0772 0.0371 0.0161

-0.1433 0.0283 -0.1288 -0.1129 0.648 -0.0584

0.1177 0.1082 0.1378 0.1873 0.0461 0.1194

-0.0634 -0.1321 -0.0950 -0.1133 -0.0583 -0.0915

Average correlation

-0.0200

-0.0305

0.0155

-0.0758

Marketing

Mix Variable



and the relative size of foreign operationsdid not support the direction of correlation and specific conclusions generally acceptedby previous researchers.

5.4. Summary of results The analysis of variance and regressionanalysis performed on the level of centralization suggestthat size has little or no impact (Hl was not accepted). The level of correlation found to exist betweenany of the size variablesand the level of centralization was so small that it would not be worth including in a general model. HypothesesH1.l, H1.2, H1.3, and H1.4 were not accepted,as no relation was found betweenthe relative size of foreign operations, absolute size of the multinational, relative size of a subsidiary, absolute size of a subsidiary, and level of centralization, respectively. Previous researchhas shown that other variables-such asownershippolicy, legal dependence,extent of foreign acquisitions, operationalindependence,andageof a subsidiary-explain the level of centralization more efficiently (Garnier, 1982; Gatesand Egelhoff, 1986).

SJZEANDMARKETINGCENTRALIZATIONINMULTINATIONALS

245

Even though the sampleof this study was slightly biased toward successfulcompanies, no model could indicate the level of centralization that a multinational company should undertake, It could, however,give managersan idea of the norm for their company. This is particularly important for Australian firms that have not had accessto information that takesinto accountthe characteristicsof the Australian market. The notion that the level of centralization a firm adoptsdoes not necessarilychange as the different size variables changealso constitutesuseful information for management.A company may not concern itself with modifying the level of centralization it undertakesas the size of the company changes.

6.

Future research

This study explores the issue of whether multinational companiesof various sizes exhibit a different degreeof centralization in their marketing management.Future researchmay examine the profitability of specific levels of marketing centralization, independentlyof the size factor or within specific size categories. The impact of centralization on the product-relatedand host country factors within a firm could be also established. And attemptscould be madeat elaborating a model that would incorporate many other interactions betweenthe contingency variables. Furthermore, different relationships may exist in different industries. The Japaneseelectronic industry may show more correlation betweencentralization and size than the U.S. automotive industry. The size variablesexaminedmay actually be correlatedwith the levels of centralization in someindustriesbut not in others,which would not havebeendetectedin this study. It is necessaryfor future researchto examinevariablesin industry-basedclusters rather than individually. To further support the findings in this article, a similar study could be conductedusing a less subjective measurefor the levels of centralization. A samplemore representativeof specific industries and different sized multinational companiesin Australia could be used before results are compared. A comparison could also be made with a similar study in which data from the parent companyrather than from the subsidiary is obtained in order to ensuregreaterinternal validity of the findings. Overall, however, the study presentssome interesting and worthwhile findings, and, in particular, provides insights into the possible relationship betweensize and centralization that would allow managementin multinational Australian subsidiairies to better manage this process.

Acknowledgments

This article is basedon the honors thesis completedby Jodie Conduit at the University of Adelaide under the supervision of PascaleQuester.

246

PASCALEGQUESTERANDJODIECONDUIT

Appendix:

Questionnaire part 2

This section investigates your opinion about the level of guidance and direction your subsidiary receives from its parent company. When making decisions on these marketing aspects, how much actual correspondence and input is received from overseas headquarters? No Direction Received

All Direction from Headquarters

Brand name

1

2

3

4

5

6

7

N/A

Wholesale price

1

2

3

4

5

6

7

N/A

Retail price

1

2

3

4

5

6

7

NIA

Basic advertising message

1

2

3

4

5

6

7

N/A

Creative expression (advertising)

1

2

3

4

5

6

7

N/A

Role of salesforce

1

2

3

4

5

6

7

N/A

Managementof salesforce

1

2

3

4

5

6

7

N/A

Packagingdesign

1

2

3

4

5

6

7

N/A

Media allocation

1

2

3

4

5

6

7

N/A

Warranty or guarantee

1

2

3

4

5

6

7

N/A

Discounts given

1

2

3

4

5

6

7

N/A

Degreeof penetration (number of intermediaries per thousandpeople)

1

2

3

4

5

6

7

N/A

Product design and characteristics

1

2

3

4

5

6

7

N/A

Salespromotion

1

2

3

4

5

6

7

N/A

Role of intermediaries

1

2

3

4

5

6

7

N/A

Type of intermediaries

1

2

3

4

5

6

7

N/A

Product positioning

1

2

3

4

5

6

7

N/A

SIZE AND MARKETING CENTRALIZATIONIN MULTINATIONALS

247

References Aylmer, R. J. (1970). “Who Makes Marketing Decisions in the Multinational Firm?” Journal OfMurketing 34 (October), 25-30. Blau, P.,and R. Schoenherr.(1971). The Structure of Orgunizations. New York: Basic Books. Brandt, W. K., and J. M. Hulbert. (1977). “HeadquartersGuidancein Marketing Strategy in the Multinational Subsidiary.” Columbia Journal of World Business 11 (Winter), 7-14. Daniels, J. D. (1987). “Bridging National and Global Marketing Strategies Through Regional Operations.” International Marketing Review 4 (Autumn), 29944. Douglas, S., and Y. Wind. (1973-1974). “Environmental FactorsandMarketing Practices.”Journal ofMarketing 7(3) (Winter), 155-165. Gamier, G. (1982). “Context and Decision Making Autonomy in the Foreign Affiliates of U.S. Multinational Corporations.” Acudemy of Management Journal 25(4), 893-908. Gates,S. R., and W. G. Egelhoff. (1986). “Centralization in Headquarters-SubsidiaryRelationships.” Journal of International Business Studies 17 (Summer),71-92. Goehle, D. G. (1980). Decision Making in Multinational Corporations. Ann Arbor: UMI ResearchPress. Hedlund, G. (1981). “Autonomy of Subsidiariesand Formalization of HeadquartersSubsidiary Relationships in Swedish MNCs.” In Lars Otterbeck (ed.), The Management of Headquarters-Subsidiary Relationships in Multinational Corporations. New York: St. Martin’s Press. Hoffman, R. C. (1988). “The GeneralManagementof Foreign Subsidiariesin the U.S.A.: An Exploratory Study.” Management International Review 28.41-55.

Keegan,W. J. (1970). “Five Strategiesfor Multinational Marketing.” European Business (January),35-40. Kirpalani, V. H., M. Laroche, and R. Y. Darmon. (1988). “Role of HeadquarterControl by Multinationals in International Advertising Decisions.” International Journal of Advertising 7, 323-333. Miller, D. (1987). “Strategy Making and Structure: Analysis and Implications for Performance.” Academy of Management Journal 30(l), 7-32. Ozsomer,A., M. Bodur,andS. T. Cavusgil. (1991). “Marketing Standardisationby Multinationals in an Emerging Market.” European Journal of Marketing 25(12), 50-64. Picard, J. (1977). “Factorsof Variancein Multinational Marketing Control.” In L. G. Mattsonand F. WidersheimPaul (eds.), Recent Research on the Internationalization of Business. Uppsala: Almqvist & Wiksel. Pugh, D., D. Hickson, C. Hinings, and C. Tuner. (1969). “The Context of OrganizationStructures.”Administrative Science Quarterly 14,91-l 14. Rau, P A., and J. F. Preble. (1987). “Standardizationof Marketing Strategy by Multinationals.” International Marketing Review 3 (Autumn), 18-28. Sim, A. B. (1977). “Decentralized Managementof Subsidiariesand Their Performance.”Management International Review 17(2),45-5 1. Sorenson,R. Z., and U. E. Wiechmann.(1975). “How Multinationals View Marketing Standardization.”Harvard Business Review 53 (May-June), 398-56,166167. Stafford, D. C., and R. H. A. Purkis. (1989). Macmillan Directory of Multinationals (Vols. l-2). New York: Macmillan. Terpstra,V., and R. Sarathy. (1990). International Marketing. Orlando, FL: Dryden Press.