Organizing for dual strategies of product diversity and international expansion

Strategic Managemznt Journal, Vol. 6,223-23 7 (1985)

L Organizing for Dual Strategies of Product Diversity and International Expansion JOHN D. DANIELS AND ROBERT A. PITTS College of Business Administration, The Pennsylvania State University, University Park, Pennsylvania, U. S. A.

MARIETTA J. TRETTER Department of Business Analysis and Research, Texas A & M University, College Station, Texas, U. S. A.

Summary This study examines organizafion sfrucfures of 37 large U.S. multinationals which are both highly diverse and heavily involved abroad. The findings show that the international division is still the most popular structure for this form of enterprise. However, several other structures are also found to be in common use. An attempt is made to relate use of these structures to such strategic variables as exfent of vertical integration. R&D intensify, level of foreign sales and method of diversifying.

Stopford and Wells (1972), in their seminal study on how large U.S. multinational firms organize for handling their foreign operations, segmented their analysis of companies with the dual characteristics of high product diversity and high foreign sales. They relied on data collected between 1963 and 1968 and encountered no clear-cut preference for a given structure. They found the 49 firms they examined with these characteristics to be using five different types of structure of-which none accounted for more than a third of those in use. Since then, there has been general agreement that the pursuit of these dual strategies (i.e. high product diversity and high foreign sales) creates organizational problems. There have also been numerous proclamations explaining which structures these firms should adopt. Yet, in spite of several more recent studies on organization structures for international operations, none of these studies focused separately on those firms which have both high foreign sales and high product diversity. Given these remaining questions and the length of time that has transpired since the Stopford and Wells study, we decided to look specifically at this type of company. We were influenced as well by the evidence of considerable organizational change during this period because of such factors as the emergence of newer structural forms (Vernon, 1980).

DEFINITIONS

Several studies have classified companies’ organization structures by the way they place their international operations (Alpander, 1978; Egelhoff, 1982; Franko, 1973; Lovell, 1966; Stopford and Wells, 1972). They found it useful to classify firms according to their dominant structural types, even though a few of the firms’ operations may lie outside the pure structure. There are basically five classifications which we have used throughout our 0143-2095/85/030223-15$01.50 0 1985 by John Wiley & Sons, Ltd.

Received 24 January 1984

224 J. D. Daniels, R . A. Pitts and M. J. Tretter

study: world-wide functional, world-wide product, international division, area and matrix structures.

In a world-wide functional structure, top-level line executives have world-wide responsibility for separate functions, e.g. for manufacturing, sales, engineering. In a world- wide product organization, top-level line executives are responsible for one or more world- wide business(es). In an international division form of organization, two kinds of line executives report to the chief operating officer. All but one of these are managers of domestic activities, whereas the remaining executive is in charge of all the company’s foreign business. Executives reporting to the chief operating officer in an area organization are responsible for all company businesses within specific geographic regions of the world. The defining characteristic of a matrix organization is the simultaneous reporting by middle-level line executives to two or more bosses who do not themselves have exclusive line authority.

THEORY AND HYPOTHESES

Several studies have argued a relationship between strategy and structure, particularly a tendency to move away from functional and towards product divisions as companies increase their diversity (Chandler, 1966; Franko, 1974). This tendency is claimed to be caused by technical and market complexities brought about by increased product diversity. By breaking down responsibilities to include only part of the product line, a company can decrease the complexities handled by a divisional head.

Stopford and Wells’ data on the 49 firms with both high foreign sales and diversity were actually a subset of their larger study, which examined 187 firms from the 1963 and 1964 list of Fortune’s 500 largest manufacturing enterprises in the U.S. Of the many strategic variables they examined, foreign sales and diversity accounted for much of the variability in structure. When these were examined simultaneously along two axes, certain concentrations of structures emerged. For example, 10 out of the 13 firms which had very low diversity coupled with high dependence on foreign sales were using area divisions. Of the firms with both very low diversity and low foreign sales, 30 out of 30 were using international divisions. Although they did not find a similar concentration by structure among firms with both high foreign sales and diversity, their overall results indicated a probable linkage between strategy and structure.

Hypothesis 1. Few, if any, firms handle foreign operations through functional structures.

Since the functional structure has been shown to become less prevalent as product diversity increases, the fact that we are examining diverse firms should itself eliminate most functional structures from our sample. By adding in the constraint of significant foreign operations, we would expect even less a likelihood of finding functional structures. This is because the problem of complexity is compounded for each functional division (Chandler, 1975). For instance, international activities usually entail communications and information processing over larger geographical and cultural distances. Decision makers may be faced with new constraints which vary on a country-to-country basis because of government restrictions on, for example, ownership and resource transmission. Technical complexity may increase because of differences in factor costs and availabilities. Past performance and

Organizing for Dual Strategies 225

future allocations may be more difficult to evaluate because of such uncertainties as exchange rate movements and political changes.

Hypothesis 2. Few, if any, firms handle foreign operations through matrix structures.

Some writers have recently argued that dual reporting paths may be appropriate in certain situations (Mee, 1964), especially for multinational enterprises which need to reconcile business and geographical viewpoints (Galbraith, 1971; Davis, 1974). They reason that the matrix structure might provide an advantage of a better balance between product and geographical perspectives. Recent writers have also reported their observations of many multinational firms’ turning toward structures in which foreign subsidiaries report to more than one division at headquarters (Stopford and Wells, 1972; Prahalad, 1976; Davis, 1976; Franko, 1975; Egelhoff, 1980).

In spite of these recent advocacies and observations about matrix structures among multinationals, we feel there are two reasons for expecting to find few, if any, firms with matrix structures. The first is that the unity of command principle, advocated by Henri Fayol as early as 1916 (Fayol, 1949), is too established to abandon easily. Secondly, there is evidence that firms are employing a variety of mechanisms to balance geographical and product perspectives without having to forsake the unity of command doctrine. These include the greater use of specialized staff expertise (Berg, 1973; Pitts, 1977a), of managerial rotation among diverse product and geographical areas (Galbraith and Edstrom, 1976; Pitts, 1977b), and of locating functional, business, and area representatives in close proximity to each other (Beer and Davis, 1976; Bartlett, 1979). These practices may enhance a more balanced perspective by giving decision makers a better understanding of other viewpoints. Yet, after gaining the understanding, they do not have the additional problem of gaining agreement with someone who shares the decision making responsibility.

Hypothesis 3. As either foreign sales or diversity increase, the global product structure will be found more and the international division structure less prevalently.

Although Stopford and Wells found no clear-cut preference for either the product or the international division structure among firms with both high foreign sales and diversity, they and others (Chandler, 1975; Channon and Jalland, 1979; Brooke and Remmers, 1978) have argued against the suitability of international divisions for these firms. The arguments against the international division have been based largely on the dependence of that division on domestic product divisions for the resources necessary for international expansion. The domestic product groups are said to control product specific resources, such as research and development and personnel with product specific expertise. They also, because of age and size, are politically powerful at gaining the major financial allocations. Since the domestic product groups are evaluated primarily on the performance of their own divisions, they may be reluctant or slow at ceding these resources to an international division. Foreign operations may initially grow under an international division because there is a high level spokesman championing foreign expansion. However, they may reach a plateau above which it is difficult to expand because of the schism with product divisions.

In spite of the arguments against the suitability of the international division for these firms, Stopford and Wells found 15 of their 49 respondents using that structure. One of

226 J. D. Daniels, R. A. Pith and M. J. Tretter

their explanations was that any boundary which might be drawn between low and high dependence on foreign sales and product diversity is bound to be an arbitrary one. Thus firms, although placed above the boundary, may not yet have reached the limitations described above. If these firms continued to become more dependent on foreign sales or product diversity, Stopford and Wells considered them likely candidates to abandon their international divisions-probably in favour of global product structures.

Our hypothesis is based, therefore, on the arguments favouring global product structures over international divisions for the type firm in our sample. Further evidence comes from a more recent study by Egelhoff (1982). In a study which included 34 multinational firms, he found those with international divisions to have a significantly lower dependence on foreign sales than those using global product structures. He reasons that this limited use of international divisions was due to information processing overload brought about by increased communications between domestic product and international divisions. Although our sample is drawn very differently than Egelhoff s (he included both U.S. and European firms and did not limit inclusion only to those with either high foreign sales or high diversity), we did expect that the same type of problem would exist for firms in our study.

Hypothesis 4. Some firms will continue to use international division structures even though they have reached very high levels of foreign dependence and product diversity.

In spite of expecting a reduction in the portion of firms using international divisions as foreign sales dependence and product diversity increase, we did not expect that all firms would abandon this structure. Our reasoning is based on Stopford and Wells’ second possible explanation for finding so many firms with international divisions among their subgroup of 49 firms. They reasoned that companies might forgo the demise of the international division by developing mechanisms to mitigate against its inherent dependence on domestic divisions for the resources necessary for continued growth.

There is considerable evidence that firms with international divisions have adopted mechanisms to gain needed resources for foreign operations. Indirect evidence comes from Davidson and Haspeslagh (1982) who found in a study of 57 large U.S. multinationals that those with an international division had a faster diffusion of technology abroad and a faster foreign sales growth than did firms using global product divisions.

One may infer from this either that the dependence of international divisions on domestic divisions does not really impose the constraints to foreign growth which theorists have claimed or that firms with international divisions have found means to overcome the problems of dependence. Based on Bartlett’s study (1983), the latter explanation is He examined 10 large multinational firms with international divisions and found them using a number of mechanisms to enhance the movement of resources from domestic groups. These included the hiring of more managers with a strategic as opposed to product specific perspectives within domestic product divisions, the altering of managerial systems so that managers have a wider base of information, the development of informal channels of communication, the pulling together of people from different areas on committees and task forces, the greater articulation of corporate global objectives, changes in reward systems, and the provision of corporate role models of co-operation between domestic and international executives. Although neither of these two studies looked specifically at firms following the dual strategies of high foreign sales and product diversity, it is probable that


some of the firms they examined do fit these parameters. Even if they do not, it seems feasible that firms in our sample could have adopted similar mechanisms.

Hypothesis 5. Conglomerates (i.e. firms which have diversi3ed into unrelated businesses primarily by acquisition) tend to use global product structures, whereas diversiped non-conglomerates tend to separate foreign operations into international or area divisions.

The argument against separate foreign operations has largely assumed that domestic product divisions control key resources. There is evidence, however, that this is not. always the case. In some multinationals, R&D (Berg, 1973; Pitts, 1977a), marketing (Aylmer, 1970), and finance (Stobaugh, 1970) are centralized at the corporate level. Even with respect to resources turned over to divisions, corporate management frequently retains considerable decision making prerogative. For example, Galbraith and Edstrom (1976) found top management of a number of multinationals to be exercising such a prerogative with respect to personnel resources in order to transfer personnel between domestic and foreign units.

Such centralization does not appear to. be uniform across multinationals, however. Considerable evidence suggests that it is much more common among diversified non- conglomerates than among congolerates. For example, both Berg (1973) and Pitts (1977b) found the former to centralize more activities at the corporate level than the latter. Pitts also found corporate management to exercise more decision making influence over interdivisional transfer of personnel (1977a) and manufacturing resources (1980) in the former as compared to the latter.

Those conducting these studies explain this difference between conglomerates and diversified non-conglomerates as follows. First, they point out that the latter operate businesses which are more closely related functionally. As a result, the latter have more opportunity to benefit from sharing resources among businesses and, consequently, more incentive to institute the centralizing mechanisms to effect such sharing. Secondly, a fundamental difference in the need to ensure autonomy for operating divisions may also play a role. This need is felt to be far greater for conglomerates, whose divisional managers have often become habituated to independent operation prior to acquisition, than for diversified non-conglomerates whose managers are generally long-term company employees.

Hypothesis 5 is supported by yet another piece of evidence. In their study of 187 multinationals, Stopford and Wells found 24 firms which, contrary to the usual sequence, had adopted a global product structure without moving through an international division stage. Although they provide no explanation for this deviation, they do report that most of the aberrant firms were diversifiers which had expanded through acquisition and merger rather than internal development.

Hypothesis 6. Firms using international division structures invest more in R&D than those using global product structures.

Research indicates that successful internal diversification usually comes from commercializing a series of technological innovations (Chandler, 1966; Berg, 1973), whereas conglomerate diversification comes simply from acquisition of already established

228 J. D. Daniels, R. A. Pitts and M. J. Tretter

firms. It seems reasonable, therefore, to expect internal diversifiers to invest more heavily in R&D than their acquisitive counterparts. Since we concluded in our discussion of the preceding hypothesis that internal diversifiers prefer an international division, it follows that international division firms will tend to invest more in R&D than global product firms.

A recent study by Davidson and Haspeslagh (1982) provides indirect support for this connection between high R&D and an international division structure. They examined the speed at which products initially introduced in the U.S. were transferred abroad in 57 U.S. multinationals. Their results show such transfers to be significantly more rapid in firms using an international division structure than in those using a global product structure. One would therefore expect multinationals heavily involved in R&D-for which the transfer abroad of products first commercialized in the U.S. is particularly critical to corporate success-to prefer an international division structure.

Hypothesis 7. Firms with area division structures depend more on foreign sales than firms with either international or product division structures.

There are three reasons to expect companies with area structures to be more dependent on foreign sales than firms with either global product or international division structures. The first is behavioural., As long as an international division is small relative to domestic product divisions, it poses little threat to the power and autonomy of the domestic groups. Stopford and Wells noted, though, that heads of domestic product divisions begin to feel threatened when the international division approaches the size of domestic divisions. At that point pressures are created to split the international division into two or more units in order not to upset the relative power position of domestic divisions. If the area structure is the outgrowth of this type situation, then it will exist only when foreign sales are quite large.

The second reason is administrative. The international division may be divided into area divisions in order to balance the power of people at the same operating level. Many firms tend to prefer that operating divisions are reasonably similar in size in order to facilitate the balance of resources, titles and responsibilities within the organization. The separation of the international division may, thus, take place to keep it from becoming too big relative to domestic product divisions. Similarly, growth of the international division may lead to internal pressures to split the division into two or more parts. In so doing, divisional heads may reduce the number of countries and people reporting to them, thus maintaining a more workable span of control.

The third factor concerns the pressures to reduce costs. They may often be reduced, for example, by developing larger production runs than for a single country market or by standardizing products for different markets in order to avoid development duplication. Pressures have been growing because of increased international market homogeneity, the existence of global competitors, and reduced trade barriers (Doz, 1980). The longest production runs would result from single factory outputs of each standardized component, a strategy which would probably necessitate control through global product divisions. If sales grow sufficiently, however, optimum plant size may be achieved on a regional as opposed to a global basis. Growth and optimum plant size may, thus, create pressures to divide the product structure geographically. As similar growth takes place within several product groups, the geographical presence of the company may be so large that it becomes advantageous to pull the units together on an area basis in order to do such things as deal collectively with government authorities or to manage financial activities.


RESEARCH METHODOLOGY

Sample selection Since the Stopford and Wells study is the most extensive examination of U.S. multinational structure and since that study drew a sample from the Fortune 500, we decided to draw our sample from the Fortune 500 as well. We relied on Standard and Poor’s Compustat I1 tapes to determine which firms had both high diversity and a high dependence on foreign sales. The 1978 Fortune list was first pared to 256 to include only those firms which are required under FASB 14 to segment foreign operations. (Segmentation is required if foreign sales, assets or profits are at least 10 per cent of corporate total.) We then arrayed these 256 firms on a matrix bounded on one axis by the number of two digit Standard Industrial Classification (SIC) codes of operations and on the other axis by foreign sales as a percentage of total sales. Next we divided companies as closely in half as possible along each axis in order to develop a quadrant of companies which were high in both diversity and dependence on foreign sales. The dividing points were seven or more SICS for high diversity and 20 per cent or more foreign sales for high foreign sales dependence. This yielded a population of 93 firms. We sent questionnaires to all 93, receiving usable responses from 37, which represented a 40 per cent usable response rate.

Organization structure To determine the nature of each firm’s approach to organizing foreign activity, respondents to the questionnaire survey were asked to trace the path or paths along which a ‘typical’ non-North American production or processing facility reported up through their organizations as financial results were consolidated at successively higher levels, up to the level of chief operating officer. The use of financial consolidation as an indicator of organization structure was based largely on studies showing this to parallel the primary line of superior-subordinate relationships (Reece and Cool, 1978). We sought this information from the chief financial officer of each firm because we believed he would most likely be knowledgeable about the overall consolidation process. Respondents were asked to supply two kinds of information for each reporting level: kind of responsibility (e.g. for a function only or for one or more businesses) and geographical domain over which responsibility extended. Three classifications for the latter variable were established: global (the entire world), international (the globe excluding the U.S. and Canada) and area (an international subregion). A firm’s top-level structure was determined by the responsibility of the last individual along this path immediately below the level of chief operating officer. Levels below this were categorized as well by kind and geographical domain of responsibility. The resulting classifications were checked against official organization charts or ones which respondents sketched for us. Telephone interviews were used to clear up inconsistencies.

Although it is fairly easy to define pure types of organization structure, it is often difficult in practice to classify firms within one of the structures. Most companies’ structures are mixed to some extent because of growth and changes in managerial assignments. There may be uncertainty about who has the authority over certain decisions, especially where dual relationships exist with line and staff personnel. Managers who are new to a post may temporarily continue to be responsible for some activities from their prior positions. Managers may continue to use old labels even though the organization has changed; or a structural type may be referred to before it is completely in place. Because of these difficulties, we acknowledge that our classification of firms corresponds only to their major

230 J. D. Daniels, R . A . Pitts and M. J . Tretter

organizational thrust. We found none of our respondents to be so mixed, though, that we were not confident in our labelling of them.

Company characteristics We used two sources of information to identify company characteristics. These were Standard and Poor’s Compustat I1 tapes and Forbes industrial classifications for 1980. We depended on the tapes for data on foreign sales as a percentage of total sales and for the number of two digit SIC categories. These were our measures, respectively, of foreign dependence and of diversity. The tapes also gave us our proxy of research intensity which we measured by research and development as a percentage of sales.

Forbes Industrial Classification was used to identify conglomerates. Forbes has established a separate ‘Conglomerate’ category for firms which have diversified widely into unrelated businesses primarily by acquisition or merger. Eight of the 37 firms in our sample were included in this Forbes category in 1980, and all but one of the rest were classified under one of the Forbes industry categories-generally under the one representing a firm’s largest related group of businesses. The one remaining firm was classified in Forbes ‘Multi- industry’ category, representing firms which have diversified widely primarily by internal development into related .businesses.

RESULTS

Functional structures Table 1 shows that we found two firms (5 per cent) using a functional structure. That we found even this many was at first surprising to us. We feel that there are two possible explanations for this finding.

The first is that these firms should not have been included in our analysis. It is tempting to accept this explanation. Any line is an arbitrary one to separate high and low foreign sales or high and low diversity. Had we drawn our sample by defining borders at a slightly higher or lower point, we might certainly have altered somewhat the percentages of firms using each type of structure. That would not, however, have eliminated the problem of dealing with firms being near the border. Tables 1 and 2, nevertheless, show that the two functional respondents are at the lower margin in our sample. The mean for both diversity and for foreign sales is lower than for any other organization type. We might conclude, therefore, that our functional respondents are borderline firms which will abandon their present

Table 1. Foreign sales and diversity by organization type

Number of two digit SICS Foreign sales as percentage of total sales

Type of structure n % Mean Standard deviation Mean Standard deviation

Global product 13 35 18.46* 8.59 International division 16 43 11.75* 5.16 Area 5 14 10.80 1.64 Functional 2 5 10.00 4.24 Matrix 1 3 11.00 0

35.04 11.35 35.18 11.70 40.42 15.35 28.97 1 .oo 46.89 0

Global product greater than international division, ~ ~ 0 . 0 5 .

Organizing for Dual Strategies 23 1

Foreign sales to

total sales

40%

30%

20 070

Table 2. Organizational types* by foreign sales and diversity

A = l F=O 1 = 2

M = 1 P = 1

A = l F=O 1 = 3

M=O P = 1

A=O F= 1 1 = 4

M=O P = 2

A = l F=O I=O

M=O P = 1

A = l F = 1 1 = 3

M=O P = 1

; M=O M=O

P=O P = 3

A=O F=O I = 1

M=O P = 1

A=O F=O I = 1

M=O P = 3

7-1 I 12-19 20 + Number of two digit SIC categories

* A =area; F = functional; I = international division; M = matrix; P= product.

structure if they move to become more dependent on foreign sales or if they increase their diversity more.

Although it is tempting to accept the above explanation, a second possibility is that some other characteristics set the functional respondents apart from others in our study. On close examination we found that both of these firms were highly dependent on one product, petroleum. For one of the companies, the combination of refined products, crude oil, natural gas and their distribution made up 85 per cent of total revenue; and for the second firm, the figure is over 90 per cent. These were the only two petroleum firms in our sample and the only firms so dominated by a so closely related group of products. We conclude, therefore, that a functional structure can be used by firms which combine the strategies of product diversity and international expansion, provided that a group of vertically integrated products dominate the companies’ operations. One is hard pressed, though, to think of firms, other than those in integrated extraction, which fit these constraints.

Petroleum firms may, therefore, be an exception to the generalizations which one might draw about the overall relationships between strategies and structures. Stopford and Wells felt likewise about oil firms and excluded them from their analysis.

Matrix structures Table 1 , showing only one matrix response, supports our hypothesis that we would find few, if any, matrix structures. That we found only one seems at first glance to run counter to the many observations that firms are turning to dual reporting channels between subsidiaries and headquarters. We can only speculate why our results are different.

Few writers have specified the characteristics of the firms that they have observed with matrix structures. It is possible, therefore, that most of their examples are of firms with lower levels of foreign sales or diversity than the companies we studied. Companies with both high foreign sales and high product diversity face the highest environmental and technical complexity. Given this complexity, they may perceive more risk to making a

232 J. D. Daniels, R . A. Pitts and M. J. Tretter

revolutionary structural change. After all, the publicity about the benefits of using dual reporting channels for international firms is fairly recent. The type of firms we surveyed may be minimizing risk by waiting to see what happens to early matrix adopters. Or some of the companies we surveyed could already be moving, intentionally or unintentionally, towards matrix designs on an evolutionary rather than revolutionary basis. By implementing such mechanisms as those reported by Bartlett (1 983), companies may be altering corporate environments so that group decisions and consensus may be more easily reached in the future.

Another possibility is that we have defined matrix differently than some other observers. Authors of studies indicating the portion of firms using matrix structures have not specified how they have made their classifications. This may account for the fairly wide range of matrix responses, from a low of zero (Alpander, 1978) to a high of 17 per cent (Franko, 1975), in prior studies. We based our classification on two criteria proposed by Galbraith (1971). These were that there be a dual authority relationship and that there be a power balance between the superiors to whom matrix subordinates report. In two cases, we had responses showing single line channels for consolidating financial data but dual lines in the organization charts. Through telephone interviews, we determined that the former were the real decision authorities, whereas the additional channel shown in the latter was to a person who was de facto staff for the given subordinate. (In one case we classified the firm as a global product and in the other as an international division structure.) Had we relied only on organization charts or even on respondents’ classifications, we might have had a very different breakdown of results.

Increasing foreign sales and diversity We expected that as either foreign sales or diversity increased, we would find more global product structures and fewer international division structures. We found support in terms of diversity but not in terms of foreign sales.

Table 1 shows that firms with global product structures had a mean of 18.46 two digit SICs, a figure significantly higher (p < 0.05) than the 11.75 for companies with international divisions. Figure 1 also shows that the portion of firms with global product structures increases relative to those with international divisions as diversity increases. At a minimum of seven SICs, for example, the number of international division structures exceeds global product forms by two; however, by nine SICs, the number of global product respondents is higher. At 23 SICs, there are no more international divisions and six product structures .

Both Table 1 and Figure 2 indicate no difference between firms with global product and international division structures in terms of dependence on foreign sales. Although we had hypothesized a difference based on several theoretical treaties, our finding seems consistent with that of Davidson and Haspeslagh (1982). They found foreign sales growth to be faster in firms with international divisions than with global product structures. It is probable, therefore, that multinational firms have recently managed to adopt practices which allow them to overcome the earlier foreign growth constraints of international divisions

Continued use of international divisions Although we expected the incidence of international divisions to decline with a combination of increased foreign sales and diversity, we did not expect to find a complete abandonment of international divisions. Just as we drew an arbitrary line to define our original


Number

of

Respondents

by Structure

20

15

10

7 5 INTERNATIONAL \

\ j-

\- 5 7 10 15 20 25 30

Number of Two Digit SIC Products

Figure 1 . Number of structures at different levels of diversity

20

Number

of

Respondents

by Structure

f\\ \ INTERNATIONAL

-_ --- -- --- J MATRIX

J . . . . . . . . . . . . . . . . -.-.-.-. L...T.- .-.-.-.---.-.-.-.

-\

20 25 30 35 40 45 50 55 60 65

Foreign Sales as a Oh of Total Sales

Figure 2. Number of structures at different levels of foreign sales

234 J. D. Daniels, R . A. PittsandM. J. Tretter

population, we drew additional lines to divide our sample into high, medium and low for each of the two variables. Table 2 illustrates the divisions and supports our hypothesis. The four upper right hand sections include those firms which are highest in both of the parameters which were used to draw a sample. Since the sample included only firms which were high in both dimensions relative to other large U.S. multinational firms, we are comparing high versus low within a preselected high group. Below the major dividing line (five sections), there are 27 responses. Of these, 13 (48 per cent) are international divisions and eight (30 per cent) are global product structures. The 10 firms above the line include three international divisions and five global product structures. We conclude, therefore, that companies can and do find means to make an international division succeed even when the dual strategies of product diversity and international expansion are pursued at very high levels relative to other companies.

Conglomerates versus non-conglomerates We have hypothesized that conglomerates would tend to use global product structures, whereas non-conglomerates would tend to use other structures. Our findings (Table 3) strongly support this assertion. All of the conglomerates in our sample were using global product structures, whereas only 17 per cent of the non-conglomerates were doing so. Among the latter, the international division was a strong favourite.

Table 3. Conglomerates versus non-conglomerates

Organizational structures* Non-conglomerates Conglomerates

Global product International division Area Function Matrix

5 16 5 2 t 1$

8

Total 29 8

Refers to responsibility of executives reporting directly to chief operating officer. t Both were international oil companies. These were the only oil companies in the sample.

A drug company.

Research intensity To test Hypothesis 7, R&D expenditures as a percentage of sales were measured for firms using global product and international division structures. As hypothesized, this percentage did indeed tend to be greater for the former (3.06) than for the latter (2.76). The difference was not significant at the 0.1 level, however. The failure of this difference to reach significance may in part be explained by the inclusion in our global product subgroup of several conglomerates formed by the merger of very high technology firms each of which continues to carry out substantial R&D activity on a decentralized basis following acquisition.

Area structures We hypothesized that companies using area structures would be more dependent on foreign sales than firms with either international or product division structures. Our results, however, gave only directional support. The mean value of foreign sales as a percentage of


total sales was higher for firms with an area structure (Table 1). Table 2 and Figure 2 also show that area structures comprise an increased portion of the total sample as foreign sales increase. For example, of the 17 firms with foreign sales of less than 30 per cent of their corporate total, only one (6 per cent) had an area structure. Above 30 per cent, area structures accounted for four out of 20 responses (20 per cent). In spite of this directional support, none of the relationships are statistically significant. We are reluctant, therefore, to accept our original hypothesis. We conclude, instead, that, although most companies using area structures have relatively high foreign sales as a portion of total sales, most firms which have this same characteristic have chosen not to adopt an area structure.

IMPLICATIONS

We recognize some inherent limitations to our study which we wish to mention explicitly. First, the study is based on a group of firms which may not be representative of the total population of companies which are both highly diverse and highly dependent on foreign operations. We looked only at very large U.S. firms. This had some positive and negative implications. By concentrating only on the larger firms, we were able to target our analysis on those firms which account for the bulk of U.S. multinationals’ assets abroad. (The 62 largest U.S. non-bank firms account for less than 2 per cent of the number of affiliates abroad but 50 per cent of the assets (Barker, 1981).) We were also able to draw some inferences based on other studies which relied on data drawn from the largest companies. On the negative side, our sample selection may have resulted in skewness toward examination of how very large foreign affiliates are handled organizationally. Drake and Caudill(1981), in a study of nine multinationals, found, for example, that the absolute size of a given foreign country operation was important in determining the autonomy allowed in the operation.

By including only U.S. firms, we are not able to infer whether the relationships we found between strategy and structure are universal. We suspect that they are not in terms of the use of international divisions. For example, Egelhoff (1982) pooled samples of 17 U.S. and 17 European firms and found those with international divisions to have significantly lower foreign sales as a percentage of total sales than firms using other structures. We did not find this in our sample. We believe that this is due to the fact that U.S. firms typically derive a far larger percentage of their sales and profits from their domestic market than do foreign multinationals. Furthermore, we have been unable to find any example of a large U.S. firm whose domestic market is not much larger than its single largest foreign market; whereas the opposite is true for many large non-U.S. firms. We believe, therefore, that the problem of balancing business, functional and geographical perspectives is likely to be different for the U.S. firms than for their foreign counterparts. As a result U.S. firms are more apt to use an international division because it creates a spokesman for geographical expansion at the same level as spokesmen for product and functional interests. Our conjecture is also consistent with Egelhoffs study. He found only one of 17 European firms with an international division but six of 17 U.S. firms with one.

Respondents, in describing foreign operations, had to use personal discretion to determine what is a typical foreign operation. We, in turn, had to use further discretion in order to classify their responses. Our methodology forced us to place each firm into a single organizational category. Such ‘pure’ structures are rare. In future research, it may prove useful to score each firm on the degree to which it uses various multinational structures.

236 J. D . Daniels, R . A. Pitts and M. J. Tretter

Finally, our perspective is limited. We have examined a cross-section of firms at a given time and inferred how structures evolve; historical analysis might have yielded different results.

Despite these weaknesses and limitations, we feel that our findings shed light on important factors influencing the choice of organization structure of large U.S. firms which are both highly diversified and which have significant foreign operations. Although we found examples of five different structures, two (global product and international division) accounted for 78 per cent of responses. Since these are by far the most popular forms, it is useful to distinguish among the firms using one versus the other. First, all eight conglomerates have global product structures. Among non-conglomerates, however, five have global product structures and 16 have international divisions. We found no variable to explain which of the two structures the non-conglomerates would choose. When both conglomerates and non-conglomerates were combined, companies with global product structures were significantly more diverse; however, when only the non-conglomerates were examined there was no statistical difference. Nor were non-conglomerates using the two structures statistically different with respect to dependence on foreign operations or research intensity.

Area structures were next in importance, accounting for 14 per cent of responses. Their diversity ranged only from nine to 12 product groups, which was toward the low end of our sample. They had a higher dependence on foreign sales than any other group. Since these relationships were not significant, however, we are hesitant to conclude that these characteristics help to set off firms using area structures. There may be other strategic variables which we did not examine which help to explain which firms organize by area.

As we hypothesized, we found very few functional and matrix structures, a combined 8 per cent of the responses. The two functional firms in our sample are in vertically integrated extraction, probably the only type which can succeed with a functional structure at so high a level of diversity. That we found only one matrix structure may be indicative either that this form is not well suited to firms with so complex a strategy or that some firms might operate mechanisms outside their formal hierarchies to encourage decision makers to consider multiple (e.g. product, geographical, functional) perspectives.

REFERENCES

Alpander, G. G. ‘Multinational corporations: homebase-affiliate relations’, California Management

Aylmer, R. J. ‘Who makes marketing decisions in the multinational firm?’ Journal of Marketing,

Barker, B. L. ‘A profile of U.S. multinational companies in 1977’, Survey of Current Business,

Bartlett, C. A. ‘MNCs: get off the reorganization merry-go-round’, Harvard Business Review, 61(2),

Bartlett, C. A. ‘Multinational structural evolution: the changing decision environment in

Beer, M. and S. Davis. ‘Creating a global organization: failures along the way’, Columbia Journal of

Berg, N. A. ‘Corporate role in diversified companies’, in Taylor, B. and N. McMillan (Eds), Business

Brooke, M. Z . and H. L. Remmers. The Strategy of Multinational Enterprise, 2nd edn, Pitman

Review, 20(3), 1978, pp. 47-56.

34(4), 1970, pp. 17-29.

6l(lO), 1981, pp. 38-57.

1983, pp. 138-146.

international divisions’, Doctoral Thesis, Harvard Business School, Cambridge, Mass., 1979.

World Business, 11(2), 1976, pp. 72-84.

Policy: Teaching and Research, Wiley, New York, 1973, pp. 298-347.

Publishing Ltd., London, 1978.


Chandler, A. D., Jr. ‘The multi-unit enterprise: a historical and international comparative analysis and summary’, in Williamson, H. F. (Ed.) Evolution of International Management Structures, University of Delaware Press, Newark, Del., 1975, pp. 225-254.

Chandler, A. D., Jr. Strategy and Structure, Anchor, Garden City, New York, 1966. Channon, D. F. and M. Jalland. Multinational Strategic Planning, Macmillan Press Ltd., London,

Davidson, W. H. and P. Haspeslagh. ‘Shaping a global product organization’, Harvard Business

Davis, S . M. ‘Two models of organization: unity of command versus balance of power’, SIoan

Davis, S. ‘Trends in the organization of multi-national corporations’, Columbia Journal of World

Doz, Y. L. ‘Strategic management in multinational companies’, Sloan Management Review, 21(2),

Drake, R. L. and L. M. Caudill. ‘Management of the large multinational: trends and future

Egelhoff, W. G. ‘Matrix strategies and structures in multinational corporations’, Paper presented at

Egelhoff, W. G. ‘Strategy and structure in multinational corporations: an information-processing

Fayol, H. General and Industrial Administration, Sir Isaac Pitman and Sons Ltd., London, 1949. Franko, L. G. ‘Organization change in European multinational enterprises’, unpublished

manuscript, Center for Education in International Management, Geneva, Switzerland. Cited in Williamson, H. F. (Ed.) Evolution of International Management Structures, University of Delaware Press, Newark, Del., 1975, p. 237.

Franko, L. G. ‘The move toward a multi-divisional structure in European organizations’. Administrative Science Quarterly, 19(4), 1974, pp. 495-506.

Franko, L. G. ‘Who manages multinational enterprises?’, Columbia Journal of World Business,

Galbraith, J. and A. Edstrom. ‘International transfer of managers: some important policy

Galbraith, J. R. ‘Matrix organization design-how to combine functional and project forms’,

Lovell, E. B. The Changing Role of the International Executive, National Industrial Conference

Mee, J. F. ‘Matrix organization’, Business Horizons, 7(2), Summer 1967, pp. 70-72. Pitts, R. A. ‘Strategies and structures for diversification’, Academy of Management Journal, 20(2),

Pitts, R. A. ‘Unshackle your “comers”’, Harvard Business Review, 55(3), 1977b, pp. 127-136. Pitts, R. A. ‘Correlates of intra-company sales in large multinational business firms’, Proceedings of

the Seventeenth Annual Conference, Eastern Academy of Management, Buffalo, New York, May

Prahalad, C. K. ‘Strategic choices in diversified MNCs’, Harvard Business Review, 54(4), 1976,

Reece, J . S. and W. R. Cool. ‘Measuring investment center performance’, Harvard Business Review,

Stobaugh, R., Jr. ‘Financing foreign subsidiaries of U.S. controlled multinational enterprises’,

Stopford, J. M. and L. T. Wells, Jr. Managing the Multinational Enterprise, Basic Books, New

Vernon, R. ‘Gone are the cash cows of yesterday’, Harvard Business Review, 58(6), 1980, pp. 150-

1979.

Review, 60(4), 1982, pp. 125-132.

Management Review, 16( l), 1974, pp. 29-40.

Business, 11(2), 1976, pp. 50-71.

1980, pp. 27-46.

challenges’, Business Horizons, 24(3), 1981, pp. 83-91.

Academy of International Business Annual Meeting, New Orleans, October 1980.

approach’, Administrative Science Quarterly, 27, 1982, pp. 435-458.

8(2), 1973, pp. 30-42.

considerations’, Columbia Journal of World Business, 11(2), 1976, pp. 100-1 12.

Business Horizons, 14(1), February 1971, pp. 29-40.

Board, New York, 1966.

1977a, pp. 197-208.

1980, pp. 59-62.

pp. 67-78.

56(3), 1978, pp. 28-46.

Journal of International Business Studies, 1( l) , 1970, pp. 43-64.

York, 1972.

155.

Organizing for dual strategies of product diversity and international expansion

Documents