Business unit manager influence on corporate-level strategy formulation.

The distinction between corporate-and business-level strategy is a distinction as old, and as durable, as the field of strategic management itself (Bowman and Helfat, 2001; Hofer and Schendel, 1978). This distinction between levels of strategy is reflected in the multidivisional (M-form) structure

of multibusiness corporations. The M-form, according to Chandler (1962), allows the Chief Executive Officer (CEO) and the "corporate office" to set corporate strategy, while delegating more detailed matters to the managers of business-specific subunits. It might thus appear that, although these business unit managers may be formulators of business strategy, they are implementers of corporate strategy.

Such appearances, however, may well be deceptive. The business unit managers traditionally responsible for implementing corporate strategy may have more to do with its formulation than conventional wisdom suggests. Bower's (1970) classic account of the resource allocation process in multibusiness corporations includes examples of "upward influence." Indeed, upward influence is the foundation of a stream of literature demonstrating the importance in strategy formulation of bottom-up processes as well as top-down processes (Burgelman, 1983; Floyd and Wooldridge, 1992; Wooldridge and Floyd, 1990).

In this study, we focus on a specific form of upward influence. Our dependent variable is upward influence on the formulation of corporate strategy by managers traditionally associated with its implementation: managers of specific businesses within the corporation. This form of upward influence has received little or no systematic research attention, perhaps because it cuts across the traditional distinction between corporate- and business-level strategy.

This article, and hence our contribution to the understanding of corporate strategy formulation, proceeds as follows. We start by developing theoretically-grounded arguments about specific antecedents of business unit manager influence. These arguments have intuitive, as well as theoretical, grounding. For example, one would expect business unit size to be positively related to the upward influence of the business unit manager on corporate strategy, since the manager of a larger unit is responsible for a larger part of the corporation. One would also expect business unit performance to be positively related to upward influence, since unit performance may lend credibility to its manager. Indeed, to the extent that managerial judgment that enhances business unit performance also enhances corporate performance, one would hope, as well as expect, to see a positive relationship between business unit performance and upward influence.

Having developed our hypotheses, we then go on to describe the methods by which we tested them, including the mail survey of business unit managers by which we gathered most of our data. We then present the results, which show support for some, but not all, of our hypotheses. Finally, we discuss our results, acknowledge limitations of the current study, and identify directions for further research.

THEORY AND HYPOTHESES

Theoretical Foundations

Prior to developing specific hypotheses, we locate the influence on corporate strategy of business unit managers in four theoretical contexts: levels of strategy, upward influence, organizational communication, and the attention-based view of the firm.

First, describing our research relative to levels of strategy is mainly a matter of establishing terms. The distinction between the corporate and business levels is well stated by Hofer and Schendel. At the former level, the question is: "what set of businesses should we be in?" (Hofer and Schendel, 1978: 27). The question of resource allocation among these businesses is also vital to corporate-level strategy.

Business-level strategy, on the other hand, "focuses on how to compete in a particular industry or product/market segment" (Hofer and Schendel, 1978: 27). We use the term business unit manager to describe a manager heading a unit with this focus. It is worth noting at this point that "business unit" is not the only term used to describe a business-specific subunit of a corporation. Chandler (1962) uses "division," while Gupta and Govindarajan (1986) use "strategic business unit" (SBU). This latter term is sometimes reserved for groupings of businesses.

Although we have been careful to make the distinction between the corporate and business levels of strategy, we locate our research at a third, less traditional level of strategy: the level termed intracorporate by Galunic and Eisenhardt (1996). At this level, focus is on the linkages between the corporate and business levels. Perhaps the most substantial contribution to intracorporate research to date is that made by Gupta and Govindarajan (see, for example, their 1986 study). An important contrast between our research and theirs is that the focus of much of their research was on the top-down processes by which the corporate office manages business units, whereas our focus in this study is on the upward influence of business unit managers on corporate strategy.

This brings us to the second of our theoretical contexts: upward influence. The study of upward influence can be traced back to Bower's (1970) study of the resource allocation process in multibusiness firms. Bower found that although resource allocation may culminate in a decision taken by corporate management, it is more accurately viewed as a process involving multiple tasks and multiple levels of the organization. Burgelman (1983) built on this insight, and used his own field research, to describe two categories of strategic behavior, which he termed induced (top-down) and autonomous (bottom-up). We focus here on a particular form of the latter category of behavior.

Floyd and Wooldridge (1992) described the strategic behavior, including the autonomous strategic behavior, of middle managers and linked this behavior to the "strategic types" (defender, prospector, etc.) identified by Miles and Snow (1978). They report that different patterns of middle-manager behavior are associated with different strategic types. Hence, they described upward influence on business strategy. Our contribution is complementary to theirs, in that it focuses on upward influence on corporate strategy. Our business unit managers are similar to Floyd and Wooldridge's middle managers in that each set of managers may exert upward influence. It may be helpful to note that we do not claim that every business unit manager is a middle manager. Some business units are in themselves large organizations, with billions of dollars in annual sales, and so the managers of these units may be more accurately regarded as top managers than as middle managers. Hence our research on business unit managers draws on, but is not a subset of, the research on middle managers.

The third theoretical context in which we place such influence is that of organizational communication. In this context, upward influence is a specific form of communication between business unit managers and corporate managers such as the CEO. It is useful, then, to describe some of the basic components of communication. Krone, Jablin, and Putnam (1987) include among these components the sender, the receiver, and the message.

For our purposes, the sender is the business unit manager attempting to influence corporate strategy, the receiver is the CEO, and the message is the specific influence attempt. For example, a business unit manager (sender) may contact the CEO (receiver) with the suggestion (message) that the corporation make a specific diversifying acquisition. During the same week, the CEO may receive related messages from others: the chair of the board has her own views on diversification; unions are wary of acquisitions, fearing that jobs will be lost in post-acquisition consolidation; and so on. As this example shows, CEOs are receivers of messages from many senders--not only business unit managers, but also board members, and myriad stakeholders--and some of these messages will receive higher priority than will others. Hence messages from business unit managers seeking to influence corporate strategy must compete for CEO attention with the influence attempts of other senders, and with the other messages directed at CEOs.

This brings us to the last of our four theoretical contexts: the attention-based view of the firm proposed by Ocasio (1997), building on the work of Simon (1947). In this context, the emphasis is on "the socially structured pattern of attention ... within an organization" (Ocasio, 1997: 188). In order for business unit managers to influence corporate strategy, it is necessary (although far from sufficient) that their attention is on matters of corporate strategy (as well as on matters of the strategy, structure, and processes for their specific businesses), and that their influence attempts gain the attention of the CEO. Upward influence, then, requires the attention of both the sender (business unit manager) and the receiver (CEO). Hence, managerial attention will be a continuing theme as we turn to the task of developing specific hypotheses.

Hypotheses

Our first three hypotheses focus on the place of the business unit within the corporation. The first of these centers on the formal channel of communication between the business unit manager and the CEO. In some cases, this channel is direct, in that the business unit manager reports directly to the CEO. In others, it is indirect, in that reporting takes place via another member of the corporate office, such as the Chief Operating Officer (COO).

Of course, formal channels and organization charts tell only part of the story of communication within organizations. There are also informal networks of advice, of trust--and of communication--within organizations (Krackhardt and Hanson, 1993). While not denying the importance of such networks, we note that a direct report to the CEO provides many opportunities, such as written reports and staff meetings, for influence on corporate strategy. We also suggest that managers prominent in the informal networks of the CEO may be promoted to the level of business unit manager, thus making their formal channels to the top congruent with their informal channels. We, therefore, predict a link between direct reporting and upward influence.

Hypothesis 1: Business unit managers reporting directly to the CEO of the corporation exert greater influence on corporate strategy than do business unit managers reporting to the CEO through one or more intermediate layers of management.

In moving from our first hypothesis to our second, we move from a structural predictor of upward influence to a strategic predictor. To be more specific, we invoke one of the central concepts in diversification research: relatedness. It is likely that the extent of upward influence on corporate strategy will differ depending on whether the business managed is related to the core business of the firm. The question is: will relatedness tend to increase or to decrease the extent of upward influence? On the one hand, we could argue that relatedness will decrease attempts at upward influence on the grounds that business unit managers are likely to be content with a corporate strategy of which their businesses are an integral part. It is the managers of businesses unrelated to the current corporate core who are likely to attempt to change corporate strategy, probably towards a strategy to which their businesses are more central.

On the other hand, we could argue that that relatedness will increase attempts at upward influence. Our hypothesis in fact rests on this latter argument, which we now make and illustrate with reference to the potential sender and receiver of an upward influence message. If the unit's business and the corporation's core business are related, the business unit manager will be strongly motivated to influence corporate strategy, since this relatedness affords opportunities for economies of scale and scope not available to managers of unrelated businesses. Hence, the manager of a related business is likely to become the sender of upward influence messages. Relatedness will also influence the receiver of such messages, the CEO. The business unit manager's experience in managing a business related to the strategic core of the corporation will lend credibility to, and encourage attention to, the manager's recommendations or other attempts to influence corporate strategy.

In sum, relatedness is likely to be associated with influence attempts that fall within the corporation's "dominant logic." If we think of dominant logic as a filter (Bettis and Prahalad, 1995), less related influence attempts may be perceived as less relevant, and hence be filtered out. Based on this consideration of sender, receiver, and dominant logic, we make the following prediction.

Hypothesis 2: Managers of businesses related to the core business of the corporation exert greater influence on corporate strategy than do managers of businesses unrelated to the core business of the corporation.

The third and last of our hypotheses concerning the place of the business unit in the corporation concerns size, rather than structure or strategy. In particular, it concerns the size of the business unit relative to that of the corporation. The performance of a business unit comprising a large fraction of the corporation has strong influence on corporate performance. Messages from the unit manager are therefore likely to receive particular attention from the CEO.

It is helpful in this context to consider how the manager came to head this large business unit. The manager may have been appointed to head what was already a large business; the CEO is unlikely to have made or permitted the appointment unless she or he had considerable respect for the manager. Alternatively, the manager may have been appointed to head a relatively small business, and have grown the business; this growth is likely to have earned the respect of the CEO. In either case, it is likely that the manager of a large business unit will be particularly able to influence corporate management with respect to strategy.

Hypothesis 3: Business unit manager's influence on corporate strategy is positively associated with business unit size relative to corporation size.

The CEO's opinion of the business unit manager remains central to our account of upward influence as developed in our fourth and last hypothesis. Of all the attributes of a business unit, perhaps the one most likely to be salient to the CEO is performance. A business unit manager whose unit is making a loss, or otherwise performing poorly, is unlikely to be credible as a source of suggestions for corporate strategy. If the manager's judgment is not producing positive results for her or his business unit, it may be deemed unlikely to produce positive results for the whole corporation. On the other hand, high-performing business units, and their managers, are likely to appear prominent and positive to the CEO. This will enhance the ability of such managers to influence corporate strategy.

There will, of course, be exceptions to this reasoning. For example, a highly-regarded manager may recently have been appointed to head a troubled business unit. In such a case, the business unit's low performance may be attributed, not to its new manager, but to the state in which the previous manager left it. The newly-appointed manager's credibility as a source of input to corporate strategy, then, will not be tarnished by the performance of the business unit recently taken over. However, we consider that, for a sample of business unit managers, the following relationship will hold.

Hypothesis 4: Businesses unit manager's influence on corporate strategy is positively associated with business unit performance.

METHODS

Data Collection

Our primary source was a mail survey of business unit managers at Fortune 500 manufacturing corporations. Prior to the mailing described below, we assembled a database of business unit managers, using corporate web sites and annual reports. We sent a pre-test version of the survey instrument to the managers from one of these industries (Forest and Paper Products). On the basis of this pretest, we revised the instrument for clarity. The revised instrument is available from the first author upon request.

We sent surveys to 534 business unit managers. Twenty-five of these 534 proved non-deliverable. Some three weeks later, we remailed the survey (not just a reminder letter) to the managers whom we had not yet heard from. In total, we received 100 responses. The raw response rate--calculated by taking the number of responses and dividing it by the number of pieces mailed, minus the number of non-deliverables (Alreck and Settle, 1995: 206)--is 19.6%.

The net response rate--calculated by taking the number of usable responses and dividing it by the number of pieces mailed, minus the number of non-deliverables (Alreck and Settle, 1995: 206)--is 16.1%, since 82 of the responses were usable. Hence the response rate for this survey exceeds "the 10 to 12 percent typical for mailed surveys to top executives in large American corporations" (Hambrick el al., 1993: 407).

Measures

We did not develop any new measures for this study. Some of our variables were simple and objective: whether the business unit manager reported directly to the CEO is an example of such a variable. For more complex and subjective variables, our review of the relevant literatures yielded measures requiring relatively minor adaptation for use in the current study. We are aware that the use of a single-respondent survey may raise suspicion of social desirability bias, particularly when subjective measures are involved. Hence we were careful in developing the survey to include objective, as well as subjective measures. Moreover, we were ready when interpreting the data to be suspicious of associations between subjective predictor variables (such as business unit performance) and our subjective dependent variable.

Dependent Variable. In order to measure our dependent variable--business unit manager influence on corporate strategy--we adapted Wooldridge and Floyd's (1990) five-item measure of upward influence. We adapted it by emphasizing that our question was about influence on corporate-level strategy (Wooldridge and Floyd's measure was not specific as to level of strategy). We provide our adapted measure in the Appendix. Strictly speaking, it is a measure of managers' reports of their upward influence. The reliability of this measure in terms of Cronbach's alpha was a more than satisfactory .95.

Independent Variables. Business unit managers indicated on the survey how many levels separated them from the CEO. We set a dummy variable according to the response, setting the variable to 1 to indicate a direct report, and to 0 otherwise.

Following Bergh (1995), we operationalized relatedness by comparing the business unit and the corporation at the two-digit SIC level. We used the COMPUSTAT database to determine the SIC code of the corporation. In order to determine the SIC code of the business unit, we referred to the name of the business unit, and to the descriptions of the unit and its product/market domain found in sources such as annual reports and web sites. If the first two digits of the SIC codes (corporate and business unit) were identical, we coded a dummy variable to 1, indicating relatedness; otherwise, we set the variable to 0.

The survey asked business unit managers the size of their organizations, in terms of number of employees. It emphasized that the question referred to the size of the organization (what we term business unit), and not to the larger corporation of which it was a part. We divided each response by the number of employees of the corporation (according to the COMPUSTAT database), thus computing our measure of business unit size relative to corporation size.

The survey also asked the manager for the organization's annual sales. Seven of the respondents did not report annual sales; some of them indicated that the amount was confidential. The responses of these seven managers would have been unusable had we based our measure of business unit size on sales, rather than on number of employees. Hence, in the interests of sample size, we chose number of employees. Since the employees measure and the sales measure were very highly correlated (.84, p <.001), we consider it unlikely that our results would have been different had we used the sales measure, rather than the employees measure.

We measured business unit performance using Gupta and Govindarajan's (1986) measure. This measure, which involves two questions and ten dimensions of performance, is in the Appendix. For each dimension, respondents were asked to indicate the performance of the business unit relative to its industry competitors on a seven-point scale, and the importance of the dimension on a five-point scale. From these responses, we computed a weighted average of business unit performance.

RESULTS

Table 1 shows descriptive statistics for, and correlations between, the variables. We measured the dependent variable in the hypotheses, business unit manager influence on corporate strategy, on a seven-point scale ranging from 1 (involved in corporate strategy not at all) to 7 (involved to a great extent). We note that the mean value for the response exceeds 4.00 (the mid-point of the scale). This may indicate a greater, rather than a lesser, degree of involvement on average, but we must acknowledge that it may also indicate social desirability bias. Values of this upward influence measure varied greatly--from a minimum of 1.60 to a maximum of 7.00. We now discuss the extent to which our predictions of patterns in this variance were supported by the data.

Regression Results

Table 2 shows the test statistics and coefficients from our model. The coefficients are consistent with the corresponding correlations in Table 1. Although the model itself was statistically significant, results were mixed with respect to specific predictors and hypotheses.

Business unit managers reporting directly to the CEO had stronger upward influence on corporate strategy than did their counterparts who reported through at least one intermediate layer of management. Hence, Hypothesis 1 was supported. Hypothesis 2, on the other hand, was not. We predicted that relatedness of the business to the corporation would be positively associated with upward influence; in fact, the association was negative, although not significantly so. Business unit size relative to corporation size was positively and significantly associated with upward influence, as predicted in Hypothesis 3. The association between business unit performance and upward influence was positive, but not significant; hence, Hypothesis 4 was not supported.

We noted from Table 1 a marginally significant correlation between two of the predictor variables: business unit managers reporting directly to the CEO tend to manage larger units than those who do not report directly. This multicollinearity between these two predictors of business unit manager influence on corporate strategy is not surprising, given that we measure business unit size relative to corporation size, rather than in absolute terms. If a corporation includes a few large business units, the CEO may feel able to be the direct manager of the business unit managers, as well as of the CFO, COO, and so on. On the other hand, if a corporation includes many small business units, the CEO may elect to have the business unit managers report to her or him via the COO or other corporate manager, thus avoiding overload in terms of direct reports.

Following Cohen and Cohen (1975: 100-102), we used hierarchical regression to check whether each of the collinear predictors added uniquely to explaining variance in upward influence. We found that relative size was a significant addition to a model already including direct reporting ([DELTA]R-sq = 0.05 *), but not vice versa ([DELTA]R-sq = .03, n.s.).

DISCUSSION

Findings

Our focus in this article is on the influence of business unit managers on corporate strategy formulation. Our results suggest that this influence is associated with the size of the business unit relative to the corporation and, to a lesser extent, with the nature of the reporting relationship between the business unit manager and the CEO. On the other hand, it does not appear to be significantly associated with either relatedness or business unit performance.

The result of testing our "relatedness" hypothesis (H2) may be particularly surprising, since the sign of the association was, contrary to prediction, negative (albeit not significant; see Tables 1 and 2). In the light of this finding, and of the fact that relatedness is one of the fundamental concepts of corporate strategy, from Rumelt (1974) onwards, we pause to discuss this concept and its operationalization.

Our assessment of relatedness between the business unit and the corporation was based on SIC codes. This is the traditional basis for measuring relatedness, but it is by no means the only such basis. While the Standard Industrial Classification (SIC) is based on physical assets deployed, these are not the only assets used in businesses. Farjoun (1998) notes that businesses can be related on the basis of skills, and that skill-based clusters of industries by no means correspond to SIC groups. Had we interpreted relatedness in terms of human assets rather than in terms of physical assets, our results might well have been very different.

We might also have measured upward influence differently. The scale we used made no distinction between influencing corporate strategy to change radically (i.e., shifting the corporation's strategic core) and influencing it to change less dramatically (e.g., entering a business new to the corporation but related to its current core). For the reasons we outlined in our development of Hypothesis 2, it may well be that managers of unrelated businesses are more interested in radical change than are managers of related businesses. In other words, had we measured the nature, rather than the extent, of upward influence, we might have found a more systematic difference between the managers of related and unrelated businesses.

We used the concept of dominant logic in our development of the hypothesis. If managers of unrelated businesses are able to influence corporate strategy, they may be able to shift the dominant logic of the firm. Bettis and Prahalad (1995) illustrate the importance of such shifts. For example, it was vital, if painful, for IBM to shift its dominant logic away from mainframe computing.

Limitations of the Current Study and Directions for Future Research

In a single study, it is only possible to examine a limited number of facets of a phenomenon as complex as upward influence on strategy formulation. We now review the facets we have examined, and discuss facets left unexamined in the current study.

We should immediately acknowledge that we gathered our data via a self-reported, cross-sectional mail survey of managers, and that this method has inherent limitations. Since it is a "cold call" request for the time of busy managers, such a survey produces more non-responses than responses. Since it is cross-sectional, it can prove only association, not causality (Tegarden et al., 2003). For example, we found support for our prediction of positive association between business unit size and unit manager upward influence. Our prediction rested on an argument that business unit size causes upward influence, with the causal relationship being moderated by CEO perceptions. This argument embodies further limitations of our study.

First, our cross-sectional research design does not allow us to rule out the possibility that causality flows in the opposite direction, and that business unit manager upward influence influences business unit size. A cynic could posit that a manipulative business unit manager may be able to influence corporate strategy in such a way that her or his business receives more resources, and thus is able to grow. Second, our data comprise the reports of business unit managers, and do not include the perceptions of CEOs or other corporate-level managers. In other words, we gathered data from the senders of messages aimed at influencing corporate strategy, but did not gather data from the receivers. Thus, the dependent variable in our study is, strictly speaking, business unit manager reports of influence on corporate strategy. It is possible that the perceptions of the receivers (CEOs) may differ from those of the senders (business unit managers). In particular, it is possible that business unit managers' reports of their influence on corporate strategy may be exaggerated.

This would be an example of social desirability bias--it is "good" for a manager to exert influence on strategy. Our measure of business unit performance might also be prone to this bias. This might be seen as a serious problem with the study, since performance was one of our predictors of upward influence. Had performance been stronger as an antecedent of influence than our more objective predictors, we would have grave concerns about the validity of the study. However, as Tables 1 and 2 show, performance was one of the weaker predictors. Each of the stronger predictors was an objective measure. Of the stronger predictors, one was provided by the respondents, when they indicated whether they reported directly to the CEO. The other was computed by us from a survey response (number of business unit employees) and a value obtained from outside the survey (total number of employees in the corporation, obtained from COMPUSTAT). Since our significant results come from objective predictor variables, one of which was not directly supplied by the respondent, we feel confident that these results are not an artifact of the single-respondent survey method we used.

We examined four antecedents of business unit manager influence on corporate strategy. We can relate these antecedents to some of the major topics in strategic management as follows. Two of the antecedents are structural, in that they describe the place of the business unit manager and of the unit in the organizational structure of the corporation. Each of these structural antecedents was significantly associated with our dependent variable; in other words, direct reporting to the CEO (H1) and large business unit size (H3) are each positively associated with upward influence on corporate strategy. Another of the antecedents is strategic, in that it describes the strategy of the corporation in terms of relatedness. The last of the antecedents is performance-related, in that it describes the performance of the business unit. Neither the strategic nor the performance-related antecedent was significantly associated with upward influence.

Although strategy, structure, and performance have been central to the discipline of strategic management for decades (Miles and Snow, 1978; Rumelt, 1974), they are by no means the only constructs that can be related to upward influence. We now identify other constructs that might in future studies be linked to our dependent variable. As we do so, we note that future work might best complement the current study by using different research methods, preferably methods that do not produce the low response rate of mail surveys (including this one). For example, interviews with business unit managers and CEOs would yield a rich set of data. Furthermore, studies set in service firms would complement the current study, for which the sample comes from manufacturing firms.

Some of the further constructs of relevance to upward influence on corporate strategy are prominent in the literature on strategic leadership. The first of these is manager demographics. For example, CEOs who have shorter tenure in their job may be more open to strategic suggestions from business unit managers than more established CEOs. This rests on the premise that manager demographics may be acceptable proxies for managerial attributes that are more difficult to measure (Hambrick and Mason, 1984); in this case tenure may serve as a proxy for openness to suggestions. To acknowledge that demographic variables may provide proxies for other variables does not mean that they always should be. Markoczy's (1997) warning that demographic variables have severe limitations as substitutes for psychological variables is well taken.

Individual-level variables, such as tenure, may be complemented by group-level variables. For example, much strategy research on top managers focuses not only on the CEO, but also on the top management team (TMT). The independent variable in our first hypothesis described the reporting relationship between the business unit manager and the CEO. Further research might examine the relationship between the business unit manager and the TMT.

Further research might also examine informal, as well as formal, relationships. We acknowledged the importance of informal relationships as we developed our hypotheses. Future studies of business unit manager upward influence might describe the place of the manager in informal networks, such as those described by Krackhardt and Hanson (1993). We expect that managers' influence on corporate strategy would be influenced by the nature of their connections to members of the TMT, and to their centrality within networks. Hence research following from the current study might usefully incorporate social network analysis.

Concluding Remarks

This study focuses on business unit managers. It is obvious that they are important agents in strategy, since they are the agents mainly responsible for the strategy and performance of the business units that comprise large corporations. In closing, we emphasize just how vital they are. One pillar of support for this emphasis on business unit managers comes from the "performance effects" research, which seeks to identify the drivers of economic performance. For example, Bowman and Helfat (2001), when they ask whether corporate-level strategy matters, discuss the extent to which performance is a corporate-level effect. Reviews of the performance effects research, such as the recent one by Hoopes et al. (2003) show that the largest single component in performance variance is at the level of the business unit. Other levels, and in particular the corporate level and the industry level, are important, but analysis of performance data places the business unit, and hence its managers, at the center of the stage.

Our contribution in this article has been to show that the importance of business unit managers is even greater, in that they may influence corporate strategy. We say "may" because our data show considerable variance in the extent to which business unit managers exert upward influence on corporate-level strategy. We find that the managers who exert most influence are those who manage the largest business units, and who report directly to the CEO. In particular, they are not those managers who manage the best-performing business units. This finding suggests an insight into the inertia for which large corporations are notorious--influence on corporate strategy may come more from established interests, in the form of managers of already-large business units, rather than from managers of smaller but better-performing business units.

Our findings have implications for managers at both the corporate and business unit levels. CEOs might make more of an effort to heed, and even to seek, advice from managers of business units that are high in performance, even if they are small in size. Managers of such business units might make more of an effort to apply their strategic acumen to the issues facing the corporation as a whole, as well as to the issues facing their particular business units.

Turning finally and briefly from managerial to research implications, we note that Dess et al. (1995), in identifying directions for strategy research, call for more research cutting across the traditional levels of strategy. In that we examine the effects on corporate strategy formulation of variables including business unit performance, our intracorporate research answers this call. While conceptual lines such as the one between corporate- and business-level strategy have enabled strategic management researchers to accomplish much, our contribution spans this line and thus, we hope, adds to understanding of strategy as a multilevel phenomenon.

Appendix: Survey Scales

We measured business unit manager influence on corporate strategy using the question below. The scale for responses ranged from 1 (not at all) to 7 (to a great extent). The scale had a reliability of .95.

   This question concerns decisions about corporate
   (as opposed to business) strategy,
   such as which businesses the corporation
   should be in. To what extent are you involved
   in each of the following aspects of
   such decisions: (a) identifying problems
   and proposing objectives, (b) generating
   options, (c) evaluating options, (d) developing
   details about options, and (e) taking
   the necessary actions to put changes into
   place?

We measured business unit performance by generating a weighted average from the following two questions.

   How important is each of the following dimensions
   of the performance of your organization:
   (a) return on investment, (b)
   profit, (c) cash flow from operations, (d)
   cost control, (e) development of new products,
   (f) sales volume, (g) market share, (h)
   market development, (i) personnel development,
   and (j) political-public affairs?

   How effective is your organization on each of
   the following dimensions of performance:
   (a) return on investment, (b) profit, (c)
   cash flow from operations, (d) cost control,
   (e) development of new products, (f) sales
   volume, (g) market share, (h) Market development,
   (i) personnel development, and
   (j) political-public affairs?

Table 1: Descriptive Statistics and Correlations

     Variable          Mean    s.d.    1               2

1. Upward influence    4.89    1.47
2. Direct report        .46     .50     .24 *
3. Related              .72     .45    -.14      .09
4. Relative size        .13     .13     .26 *    .20 ([dagger])
5. Performance         5.12     .71     .15      .01

     Variable           3       4

1. Upward influence
2. Direct report
3. Related
4. Relative size        .03
5. Performance         -.11    .09

* p < .05, ([dagger]) < .10, 2-tailed test

N = 82

Variables 2 and 3 are dummies; hence correlations involving one or both
are Spearman.

Table 2: Regression Analysis

                                                      Standardized
Predictor variable (hypothesis)                       coefficient

Direct report to CEO? (H1)                                .19 *
Related to core business of corporation? (H2)            -.14
Size of business unit relative to corporation (H3)        .21 *
Business unit performance (H4)                            .11

* p < .05, one-tailed test

N = 82

Model: Y = b0 + b1.X1 + b2.X2 + b3.X3 + b4.X4,
where Y is upward influence, X1 is the dummy for direct reporting, X2
is the dummy for relatedness, X3 is the relative size of the business
unit, and X4 is business unit performance.

Model statistics: F = 2.86 (p < .05), R-square = .129.

References

Alreck, Pamela L. and R. B. Settle. 1995. The Survey Research Handbook (2nd edition). Chicago, IL: Irwin.

Bergh, Donald D. 1995. "Size and Relatedness of Units Sold: An Agency Theory and Resource-Based Perspective." Strategic Management Journal 16: 221-239.

Bettis, Richard A. and C. K. Prahalad. 1995. "The Dominant Logic: Retrospective and Extension." Strategic Management Journal 16: 5-14.

Bower, Joseph L. 1970. Managing the Resource Allocation Process. Boston, MA: Harvard Business School.

Bowman, Edward H. and C. E. Helfat. 2001. "Does Corporate Strategy Matter?" Strategic Management Journal 22: 1-23.

Burgelman, Robert A. 1983. "A Process Model of Internal Corporate Venturing in the Diversified Major Firm." Administrative Science Quarterly 28: 223-244.

Chandler, Alfred D. 1962. Strategy and Structure: Chapters in the History of the American Industrial Enterprise. Cambridge, MA: MIT Press.

Cohen, Jacob and P. Cohen. 1975. Applied Multiple Regression/Correlation Analysis for the Behavioral Sciences. Hillsdale, NJ: Lawrence Erlbaum Associates.

Dess, Gregory G., A. Gupta, J.-F. Hennart and C. W. L. Hill. 1995. "Conducting and Integrating Strategy Research at the International, Corporate, and Business Levels: Issues and Directions." Journal of Management 21: 357-393.

Farjoun, Moshe. 1998. "The Independent and Joint Effects of the Skill and Physical Bases of Relatedness in Diversification." Strategic Management Journal 19: 611-630.

Floyd, Steven W. and B. Wooldridge. 1992. "Middle Management Involvement in Strategy and its Association with Strategic Type." Strategic Management Journal 13: 153-167.

Galunic, D. Charles and K. M. Eisenhardt. 1996. "The Evolution of Intracorporate Domains: Divisional Charter Losses in High-Technology, Multidivisional Corporations." Organization Science 7: 255-282.

Gupta, Anil K. and V. Govindarajan. 1986. "Resource Sharing Among SBUs: Strategic Antecedents and Administrative Implications." Academy of Management Journal 29: 695-714.

Hambrick, Donald C., M. A. Geletkanycz and J. W. Fredrickson. 1993. "Top Executive Commitment to the Status Qno: Some Tests of its Determinants." Strategic Management Journal 14: 401-418.

--and P. A. Mason. 1984. "Upper Echelons: The Organization as a Reflection of its Top Managers." Academy of Management Review 9: 193-206.

Hofer, Charles W. and D. E. Schendel. 1978. Strategy Formulation: Analytical Concepts. St. Paul, MN: West.

Hoopes, David G., T. L. Madsen and G. Walker. 2003. "Guest Editors' Introduction." Strategic Management Journal 24: 889-902.

Krackhardt, David and J. R. Hanson. 1993. "Informal Networks: The Company Behind the Chart." Harvard Business Review 71 (4): 104-111.

Krone, Kathleen J., F. M. Jablin and L. L. Putnam. 1987. "Communication Theory and Organizational Communication: Multiple Perspectives." In Handbook of Organizational Communications. Eds. Fredric M. Jablin, L. L. Putnam, K. H. Roberts and L.W. Porter. Newbury Park, CA: Sage.

Markcozy, Livia. 1997. "Measuring Beliefs: Accept No Substitutes." Academy of Management Journal 40: 1228-1242.

Miles, Raymond E. and C. C. Snow. 1978. Organizational Strategy, Structure, and Process. New York, NY: McGraw-Hill.

Ocasio, William. 1997. "Towards an Attention-based View of the Firm." Strategic Management Journal 18 (Summer): 187-206.

Rumelt, Richard P. 1974. Strategy, Structure, and Economic Performance. Boston, MA: Division of Research, Harvard University.

Simon, Herbert A. 1947. Administrative Behavior: A Study of Decision-Making in Organizations. Chicago, IL: MacMillan.

Tegarden, Linda F., Y. Sarason and C. Banbury. 2003. "Linking Strategy Processes to Performance Outcomes in Dynamic Environments: The Need to Target Multiple Bull's Eyes." Journal of Managerial Issues 15 (2): 133-153.

Wooldridge, Bill and S. W. Floyd. 1990. "The Strategy Process, Middle Management Involvement, and Organizational Performance." Strategic Management Journal 11: 231-241.

Andrew Watson

Assistant Professor of Management

Northeastern University

Bill Wooldridge

Associate Professor of Management

University of Massachusetts