You are on page 1of 23

Academy of Management Journat 1982, Vol.25, No. 3, 510-531.

Strategic Attributes and Performance in the BCG Matrix A PIMS-Based Analysis of Industrial Product Businesses^
DONALD C. HAMBRICK IAN C. MacMILLAN DIANA L. DAY Columbia University

This paper empirically explores the performance tendencies and strategic attributes of businesses in the four cells of the Boston Consulting Group product portfolio matrix. Businesses differed in their performance and strategic attributes, according to the two dimensions of the BCG matrixproduct life cycle stage (growth rate) and market share. Most discussions of business-level strategy fall into one of three groups. First are normative propositions about which strategic actions make sense under different conditions. These prescriptions typically are set forth by seasoned observers of organizations (Andrews, 1971; Glueck, 1976; Katz, 1970), but, so far, creation of these ideas has substantially outpaced empirical tests of their validity. A second category of literature is empirically based, but aimed at demonstrating universal "laws" of strategy. Findings on the pervasive positive effects of market share (Chevalier, 1972; Schoeffler, Buzzell, & Heany, 1974) and the experience curve (Boston Consulting Group, 1968) are primary examples. The third group also is empirical but concludes that so many contingent factors exist that strategy must be highly situational (Hatten, Schendel, & Cooper, 1978). In the latter vein, Hofer (1975) set forth what he considered to be a manageable list of 20 contingent factors (narrowed down from 54) that affect strategy for
The authors gratefully acknowledge sponsorship by the Strategy Research Center, Columbia University Graduate School of Business, and generous support from the Strategic Planning Institute, Cambridge, Mass. Thomas Lenz, William Newman, Max Richards, Sidney Schoeffler, and Michael Tushman made helpful suggestions on earlier drafts. 510

1982

Hambrick, MacMittan, and Day

511

mature businesses. All possible combinations of these (assuming only two values per variable) result in over a million possible configurations. What is needed are empirically based "mid-range" theories about business-level strategy. As Bourgeois noted, "The solution is for the researcher to abstract a smaller number of more encompassing conceptual categories with a broader range of generalizability" (1980, p. 29). This paper pursues that advice by focusing on only two key contingent variablesthe product life cycle and market shareand identifying their i'elationships with different strategic attributes and performance. The term "strategic attributes" is preferred to "strategic action" because this essentially is a cross-sectional study in which an array of strategic variables, including those that are controllable only in the longer term (e.g., capital intensity, productivity) are examined. In the literature on business-level strategy, probably no constructs have been deemed more significant than market share and the product life cycle. However, there is little empirical research treating these as contingent factors. The choice of these two constructs has the added advantage that, taken together, they form the framework for a widely known model for analyzing corporate strategythe Boston Consulting Group (BCG) product portfolio matrix (Henderson, 1979). (Purely speaking, the vertical dimension of the BCG matrix is "market growth." For most products, growth rates closely correspond with certain stages of the life cycle. The conceptual distinction is that each stage typically is attributed with characteristics in addition to growth rate, for example, customer adoption rates and the nature of competition. The emphasis here on life cycle stage is not inconsistent with BCG's strict emphasis on market growth.) If both the product life cycle and market share have major significance, and if many firms actually conceive of businesses along these two dimensions, then this study has the potential for generating empirically strong and managerially meaningful results. Two broad questions are addressed in this paper: 1) How do businesses in the four cells of the BCG matrix tend to differ in their performance on various key criteria (profitability, risk, cash flow, market share change)? 2) How do businesses in the four cells of the BCG matrix tend to differ in their strategic attributes? A third research question, which follows from the first two, is addressed in a companion paper (MacMillan, Hambrick, & Day, 1982). 3) Which strategic attributes are associated with the various performance measures in each cell of the matrix? Theoretical Review Business-Level Strategy The distinction between corporate-level strategy (what businesses to be in) and business-level strategy (how to compete in a given business) is

512

Academy of Management Journal

September

established (Hofer & Schendel, 1978; Vancil, 1976). This study focuses on business-level strategy, which is of interest to strategic units (typically divisions) in multibusiness firms, or to single business firms. Although business-level strategy centers on the concept of "competing," it is important to stress that it encompasses all the functional areas of the business. Options in operations, marketing, distribution, R&D, finance, and personnel all determine the business's basis for competing. Contingency Theories Contingency models, in which the appropriateness of certain actions are deemed contingent on particular given conditions, abound in the field of organizational theory (Ford & Slocum, 1977). Relatively little progress has been made in investigating contingent models of strategy. Hofer (1975) made an eloquent call for such research, and he also provided his own notions of what types of relationships might exist. However, since his paper appeared, essentially all empirical investigators of business-level strategy have had goals other than identifying and testing contingencies (Hatten et al., 1978; Lenz, 1978; Miles & Snow, 1978). A notable exception is Harrigan's (1981) research on factors affecting the appropriateness of different strategies in declining industries. It is not clear which contingent variables will lead to the strongest theory. Hofer (1975) lists 10 to 20 variables (depending on life cycle stage) in several categories: market, consumer, industry structure, competition, suppliers, macroenvironmental, and organizational characteristics. Hofer surmised about the importance of each variable within its category but did not suggest an overall priority. He did propose that "the most fundamental variable in determining an appropriate business strategy is the stage of the product life cycle" (1975, p. 798). Product Life Cycle The concept of the product life cycle is well established (Fox, 1973; Hofer, 1975; Levitt, 1965; Wasson, 1974). Although scant empirical research has been done on the life cycle, theorists have set forth an abundance of prescriptions about which strategic behaviors lead to success at each stage. For the early stages (introductory and growth), theorists generally lay emphasis on strategic actions aimed at gaining a strong competitive foothold, such as aggressive pricing, building capacity, heavy marketing expenditures, and product R&D. For the later stages (maturity and decline), the emphasis is on extending/expanding the product category and seeking efficiencies via adding channels, broadening the product line, vertically integrating, avoiding price cuts, and so on (Clifford, 1971; Fox, 1973; Henderson, 1979; Wasson, 1974). Again, these prescriptions have been based more on seasoned observation than on systematic research.

1982

Hambrick, MacMillan, and Day

513

There are common threads, but there also are various prescriptive inconsistencies among life cycle theorists. For example, Wasson (1974) calls for cost cutting accompanied by price cutting in the maturity stage, whereas Fox (1973) encourages price maintenance. Clifford (1971) claims that "vigorous" advertising and sales efforts are crucial in the growth stage, yet Patton says that "marketing steps to the center of the stage" during the maturity stage (1959, p. 12). Not only are there unresolved differences among theorists, but there also are confusing stances within given theorists' prescriptions. For example, Wasson encourages mature businesses to seek new markets, product expansions, product improvements, and cost reductions. This array of suggestions is so encompassing as to leave the strategist with little sense of priorities, or any sense of what might actually work. Market Share Market share was selected as the second contingent factor in this study, on the rationale that market share has been demonstrated empirically to be a key factor affecting the performance of business units. Although it is possible for businesses to have (or buy) more market share than is optimal (Fruhan, 1972), the weight of evidence indicates that high share businesses have significantly higher earnings than do low share businesses (Chevalier, 1972; Schoeffler et al., 1974). Hofer (1975) endorses the importance of market share by listing it as dominant among all the organizational attributes he would include in contingency models for all except brand new businesses. Very little systematic research has been conducted on different strategies for different share positions. Bloom and Kotler (1975), drawing on anecdotal evidence, counseled high share companies to evaluate the risks (primarily regulatory and public pressure) of their dominant positions and to adjust their shares in light of those risks. Conversely, Hamermesh, Anderson, and Harris (1978) outlined strategies for low share businesses. Based on anecdotes, they offered these suggestions to underdogs: conduct narrowly targeted R&D; limit growth; diversify cautiously; and recruit a strong-willed, hands-on chief executive. These authors' prescriptions have an intuitive allure, yet they are neither precise nor systematically derived. Foremost, there is no indication why they have any more applicability to high share (or low share) businesses than to others. A recent, more systematic study by Woo and Cooper (1980) identifies discrete clusters of high performing, low share businesses, and it contrasts them with clusters of low performing, low share businesses and high performing, high share businesses. Their paper is a step toward more accurate understanding of issues associated with market share. Combining the Two Dimensions: The BCG Matrix The choice of the product life cycle and market share as the two contingent variables has an added attraction. When combined, they form the

514

Academy of Management Journal

September

framework for the BCG's product portfolio matrix (Henderson, 1979). Although the matrix has limitations (Hofer & Schendel, 1978), it is widely cited in academic and popular discussions of strategy. Other more refined portfolio matrices, such as those used by Arthur D. Little (Patel & Younger, 1978), Royal Dutch Shell (Robinson, Hickens, & Wade, 1978), and General Electric (Taylor, 1976) generally are consistent with the BCG framework, because they all incorporate an "attractiveness" dimension (broader than, but encompassing, life cycle) and a "competitive position" dimension (broader than, but encompassing, market share). There is not unanimity as to what the dividing points on the two dimensions of the matrix should be. BCG favors 10 percent real market growth as the point of distinction between high and low market growth. The group views market share in terms of the ratio of the share held by the business relative to the share held by its leading competitor. A ratio of 1.0, indicating highest market share, is commonly considered the dividing line between high and low share. BCG also states that a ratio of 1.5 is necessary to claim and exploit true dominance in a market (Hedley, 1977). In the absence of systematic data on alternatives, the present study relies on the 10 percent market growth rate and 1.0 relative market share as the dividing points in the matrix. Although a great deal has been written about the BCG matrix (and less so about other portfolio matrices), the emphasis has been on how to allocate resources among the four cells and what kinds of performance patterns to seek for each. High share/low growth businesses (Cash Cows), for example, should be managed for maximum generation of cash, and that cash should be directed to newer, higher growth businesses (Wildcats and Stars). Low growth/low share Dogs are seen as serious cash drains that should be promptly harvested, liquidated, or divested. No systematic evidence exists about whether the performance tendencies of businesses in the four cells actually allow or warrant these prescriptions. No information exists about the strategic tendencies of businesses in the four cells. The goal of the present paper is to fill these voids in part. In doing so, it will form the backdrop for a companion paper on the relationships between strategic attributes and performance for each of the four cells. At this stage, it is not appropriate to set forth all plausible hypotheses. However, two broad propositions may be stated: (1) Businesses in the four cells will differ systematically in their performance (including in their cash flow tendencies as argued by BCG); (2) Businesses in the four cells will differ in their strategic attributes. Previous literature which should contribute to more specific hypotheses has been speculative, imprecise, and even contradictory. And the range of strategic variables and performance measures to be examined make an inventory of hypotheses unwieldy. Thus, the approach will be to treat this as an exploratory study to enhance understanding of a widely recognized but little-documented, strategic framework.

1982

Hambrick, MacMillan, and Day

515

Method The PIMS Data Base The data used in this study were drawn from the Profit Impact of Market Strategies (PIMS) project, an ongoing study of environmental, strategic, and performance variables for individual business units. About 200 corporations submit data annually on a total of about 2,000 of their business units. Each business, often a division, is a distinct product-market unit. For a technical summary of the PIMS data base, see Schoeffler (1977). Anderson and Paine (1978) provide a comprehensive critical review of the PIMS data base. Most of their concerns are about how the PIMS data had been statistically analyzed and presented in previous studies. Although they raise limited concerns about the quality of the data, they generally acknowledge the data base to be of high quality and reliability. Two factors that previous critics have not noted especially commend the data. First, PIMS staffers help each business interpret and answer the questions, thus assuring a high degree of data comparability. This feature is missing from conventional questionnaire studies. Second, each company pays a substantial sum to participate in PIMS, and the software is oriented such that their ability to derive meaningful conclusions from the data is particularly a function of the accuracy of their own data. The businesses would appear to have a commitment to thoroughness and accuracy that is missing in most survey studies. No conventional tests of the reliability of the data base (e.g., test-retest, multiple respondent consistency) are known to have been conducted. The data base does have some limitations, including those noted by Anderson and Paine. Foremost, the businesses in the data base cannot be viewed as typical of business units in general. On average, the participating businesses probably are more sophisticated, more dominant within their markets, and more effective in general than the total population of business in the United States. Categorizing the Businesses The study reported here is based on the most recent four years of PIMS data for businesses that manufacture industrial products. Further research, in progress, will extend the analysis to the consumer product sector. The businesses in the sample were classified into the four cells of the BCG matrix according to their life cycle stage (market growth rate) and relative market share. PIMS respondents indicate the business's life cycle stage by answering this question: "How would you describe the stage of development of the types of products or services sold by this business during the last three years?"

516

Academy of Management Journal

September

Introductory stage: Primary demand for product just starting to grow; product/services still unfamiliar to many potential users Growth stage: demand growing at 10 percent or more annually in real terms; technology or competitive structure still changing Maturity stage: Products or services familiar to vast majority of prospective users; technology and competitive structure reasonably stable Decline stage: Products viewed as commodities; weaker competitors beginning to exist. Responses to this straightforward categorizing question were used in the analysis. A more elaborate index for measuring life cycle stages from PIMS data has been shown to yield strikingly equivalent results (Christensen, 1977). The data base has relatively few businesses in the introductory and decline stages, so the study included only those in the growth and maturity stages, a total of 1,028. Even though the label "life cycle" is being used, the classification is fully consistent with BCG's view of this as the "market growth" dimension, because part of the PIMS definitional distinction between a growth business and a mature business is a real growth rate in primary demand of 10 percent. Relative market share is defined as the ratio of the unit's market share relative to the share of its leading competitor (using a four-year average). A relative market share figure of greater than 1.00 is considered high relative market share; 1.00 or below is classified as low relative market share. The average relative share for the entire sample is 1.30 (supporting the speculation that PIMS businesses are relatively dominant), and the standard deviation is 1.68, indicating that the businesses are not too heavily clustered around the 1.0 dividing line. The four subsamples are portrayed in Figure 1. Figure 1 The Subsamples of Industrial Products Businesses Studied
Growth Product Life Cycle Real Growth 10% Year Mature Stars N=114 Wildcats N=181

Cash Cows

N=315

Dogs Af=418

High 1.0 Low Relative Market Share

Variables

Two types of variables are distinguished in this studystrategic attributes and performance. Environmental variables are not included.

1982

Hambrick, MacMillan, and Day

517

As uncontrollables, they amount to additional contingent factors that must be systematically treated in future studies. The PIMS data base includes data on dozens of strategic attributes. Only variables that have been prominent in previous PIMS findings or in other strategy studies (e.g., product quality, capital intensity, capacity utilization) were included in this study. The strategic attributes, as grouped for analysis and discussion, are as follows: Expense Structure Resources and Resource Usage Manufacturing/revenue Investment/revenue Product R&D/revenue Plant and equipment newness Process R&D/revenue Capacity utilization Sales force/revenue Capacity/market size Advertising and promotion/revenue Sales/employee Competitive Devices Working Capital Management Sales from new products Receivables/revenue Relative sales from new products Inventory/revenue Relative prices Domain Relative direct costs Relative product line breadth Relative image Relative customer type breadth Relative services Relative number of customers Relative advertising expenses Customer fragmentation Relative promotion expenses Vertical Integration Relative sales force expenses Value/added revenue Relative integration backward Relative integration forward The definitions of the strategic attributes can be obtained from the authors. In all cases, four year averages were used. Four performance measures were examined: (1) Return on investment (ROI) (average of last two years): Pretax net income minus allocated corporate overhead costs, as a percent of average investment including fixed and working capital at net book value. (2) Cash flow on investment (CFOI) (average of last two years): Aftertax income (estimated at 50 percent of pretax income) minus changes in net investment, as a percentage of average investment. (3) Return per risk (RPR): Average ROI divided by the variability of ROI (calculated as the sum of the absolute differences between the fouryear average ROI and each year's ROI). (4) Market share change (MSC): The change (annualized via least squares) in this business's average share of the market (expressed as a percentage of the market) for the four-year period. The rationale for using two-year averages for ROI and CFOI bears primarily on the analyses to be reported in the companion paper (MacMillan et al., 1982) and will be discussed there. The two-year vs. four-year timeframe has little effect on the tendencies reported in this paper.

518

Academy of Management Journal

September

Method of Analysis

Reporting of means, with two-way analysis of variance, was used to identify any tendencies for the four types of businesses to differ in their performance or strategic attributes, either according to life cycle stage or market share. Zero-order correlations among performance measures for each cell also are reported. Results and Discussion Performance Across the Four Ceils This study empirically corroborates the primary theme espoused by the originators of the BCG matrix (see Table 1): namely, the four types of businesses have significantly different tendencies to consume or generate cash. The average Wildcat has a negative cash flow (-2.67 percent) ["requires large cash inputs that it cannot generate itself" (Henderson, 1979, p. 164)]. The average Star essentially generates as much cash as it uses (.74 percent) ["may or may not generate all of its own cash" (Henderson, 1979, p. 166)]. Cash Cows are net cash generators (10.10 percent) ["generate large amounts of cash" (Henderson, 1979, p. 164)]. These results are preinterest charges (which are not reported in the PIMS data), and so must be regarded as suggesting patterns rather than absolute figures. The results present a somewhat more positive view of Dogs than that expressed by Henderson, who said that a Dog gives "no cash throwoff. The product is essentially worthless, except in liquidation" (1979, p. 164). The Dogs examined in this study had average net positive cash flow on investment of 3.4 percent, while holding their market shares. This cashflow rate from Dogs is more than is required to meet the cash needs of the average
Table 1

Performance Levels of Businesses in the Four Cells of the BCG Matrix (Means Reported, with Standard Deviations in Parentheses)
2-Way Anova Performance Measure Return on investment Cash flow on investment ROl/ROI variability (return per risk) Market share change *p< 05 *p< 001 Wildcats (N= 181) 20.55 (24.53) -2.67 (18.79) 2.37 (3.53) .39 (1.76) Stars (N= 114) 29.58 (22.59) .74 (18.26) 3.96 (5.20) .72 (2.97) Cash Cows

(N=315)
30.00 (22.67) 10.01 (17.03) 4.57 (4.15) .38 (2.30)

Dogs (N=418) 18.48 (21.68) 3.41 (16.17) 2.80 (4.68) .14 (1.55)

(Main Effects) Life Cycle Market


Stage Share

*
** ** * * *

1982

Hambrick, MacMillan, and Day

519

Wildcat (-2.67percent). If one assumes that the absolute size of the average Dog is larger than the typical Wildcat (often fledgling operations), then the absolute cash throw-off from some Dogs may be sufficient to fund two or more promising new ventures. It also should be noted that the standard deviation of CFOI for Dogs is 16 percent, indicating that some Dogs generate substantial amounts of cash. And the modest correlation (r=-.O7) between CFOI and Market Share Change (Table 2) for Dogs indicates that cash flow can come other than through "harvesting." It appears that Dogs outperform the expectations placed on them by BCG. Instead of focusing solely on harvesting or liquidation strategies for Dogs, researchers should start dealing more positively and creatively with this type of business (e.g., Hamermesh et al., 1978; Woo & Cooper, 1980). Consistent with previous PIMS findings, ROI (reflecting accrual profits, not cash flow) was higher for high share than for low share businesses (Buzzell, Gale & Sultan, 1975). There were no significant ROI differences across the two life cycle stages. Specifically, Stars (29.58) and Cows (30.00) were roughly equal in outperforming Wildcats (20.55) and Dogs (18.48) on the ROI measure. When ROI is adjusted for risk (4-year ROI variability), differences across the four cells reflect both the profitability of high market share and the profit stability of maturity. Cows score highest (4.57), due to their dominance in relatively stable markets. Stars follow (3.96), due to their dominance, but in turbulent markets. Dogs are next highest (2.80). Wildcats are lowest (2.37), reflecting their moderate profitability and unsettled markets. These results support the established concept that growth businesses are more uncertain than mature businesses. They also reemphasize the soundness of carefully tending mature businessesincluding Dogs. Market share change was one of the performance measures examined. As expected, there were indications of greater share increases in the growth businesses than in the mature businesses, reflecting both the relatively entrenched positions in mature industries and the presumed charter of mature businesses to seek primarily cash, not share. Also, high share businesses tended to gain more share than did their low share counterparts. Apparently, "the strong get stronger." Yet, there is no indication that businesses attempt to gain or succeed in gaining truly large share increases. In absolute terms, the average share increases are modest (highest is .72 percent for Stars), as are the standard deviations (highest is 3.0 percent for Stars). A quest for a 5 percent to 10 percent share increase apparently is unusual, unrealistic, or both. The results (Table 2) also shed light on the tradeoffs presumed to exist between performance measures, particularly cash flow and market share gain. Nil correlations between these two performance measures suggest that no real tradeoff exists (except for Stars, for which r=-.27, p < .01) or that unaccounted factors may be masking indications of such a tradeoff. The results also indicate nil relationships between ROI and share change for each cell. It is beyond the scope of this paper to identify circumstances

520

Academy of Management Journal

September

Table 2 Relationships Among Performance Measures (Pearson Correlation Coefficients)


Wildcats (N= 181)
ROI

ROI .68** .45** .13 ROI .61** .39** -.15 ROI

CFOI .54** -.07 CFOI .53** -.27* CFOI .46** -.07 CFOI _ .45** -.05

RPR

MSC

CFOI RPR MSC Stars (N= 114) ROI CFOI RPR MSC Cash Cows (N = 315) ROI CFOI RPR MSC Dogs (N= 418)
RDI

-.03 RPR MSC

-.07 RPR MSC

.61** .38** -.04 ROI .57** .39** -.06

-.12 RPR

MSC

CFOI RPR MSC

-.02

*p<.01 **p<.O0l

or strategies that will allow market share and cash flow or profits to increase together. Factors such as the magnitude of an attempted share increase, competitors' strategies and goals, and product differentiation no doubt are among those that future research should include as moderating variables in studying tradeoffs between profits and share changing. The present results suggest that a tradeoff may not always exist. The findings also raise questions about the extent to which managers should be relieved of profitability or cash flow goals when they also are charged with gaining share. Strategic Attributes of the Four Cells The results indicate a host of significant differences in the strategic attributes of the four types of businesses. Some of the attributes, including almost all those expressed in terms "relative to competitors," varied primarily according to market share. Several others varied according to life cycle stage. Still others varied according to both dimensions of the matrix. Care must be taken in trying to interpret the attributes as either causes or effects of the business's position in the matrix. For example, it will be observed that high share businesses have relatively broader product lines and customer types than do low share businesses. Because the data are cross-sectional, there is no way to disentangle the extent to which broad product/market bases precede, accompany, or follow market dominance.

1982

Hambrick, MacMitlan, and Day

521

Future longitudinal treatment of the PIMS data base could yield such information. Resources and Resource Usage It comes as no surprise that high share businesses have substantially more of their markets' production capacity than do low share businesses (Table 3). High share businesses also utilize their capacity at a higher rate than do low share businesses, probably reflecting the leaders' relative ease in securing orders to fill capacity. The lower utilization rates of low share businesses indicate that those businesses either built capacity in earlier days when they held or anticipated higher shares or built capacity recently as part of a plan to gain share. That plant and equipment newness (net P&E/gross P&E) varies according to the product life cycle, but not according to market share, suggests that low share businesses typically have not engaged in recent capacity buildups, but rather they simply do not have the market power to fill earlier built capacity the way a high share business can. Table 3 Strategic Attributes of Businesses in the Four Cells of the BCG Matrix (Means Reported, with Standard Deviations in Parentheses) Resources and Resource Usage
2-Way Anova (Main Effects) Life Cycle Market Stage Share ***

***

Strategic Attributes Capacity/market Capacity utilization Plant and equipment newness Investment/revenue Sales/employee *p<.05 **p<.Ol ***p<.OOl

Wildcats (N=181) 20.37 (16.37) 73.36 (16.88) 59.24 (15.80) 63.93 (36.12) 63.22 (44.29)

Stars (N=114) 56.61 (25.85) 75.28 (15.60) 58.93 (14.09) 60.05 (27.66) 63.92 (39.50)

Cash Cows (N=315) 52.03 (25.96) 78.09 (14.23) 50.58 (14.26) 51.35 (24.72) 55.37 (36.49)

Dogs (N = 418) 21.27 (15.40) 75.14 (16.04) 51.40 (14.84) 56.06 (27.33) 59.06 (40.68)

***

**

Mature businesses have higher capacity utilization rates than do growth businesses. This may reflect the relative stability and equilibrium that exists in mature industries (as indicated by relative market share stability as discussed above), and perhaps also an efficiency orientation in mature businesses. Because the numerator of investment/revenue (capital intensity) is measured in terms of net investment, it is not surprising that growth businesses

522

Academy of Management

Journal

September

have higher scores on this attribute than do mature businesses. Growth businesses, on average, have newer, less depreciated assets [the correlation between plant and equipment newness and investment/revenue was .34 (p>.01) for the entire sample, indicating a strong relationship between newness and capital intensity]. The high capital intensity figure for growth businesses also may reflect their presumably smaller revenue bases over which they must spread investment. The disguised dollar figures in the data base prevent any test of this speculation. To some extent, the finding of high capital intensity for growth businesses runs counter to conventional thought of businesses becoming more capital intensive over time, as competition shifts to a volunie or efficiency orientation. The high sales/employee figure for growth businesses, compared to mature businesses, suggests that mature businesses have not dramatically replaced labor with capital assets. The higher employee productivity figure for growth businesses may be partly attributable to the relatively stronger pricing structure that exists in growth businesses (gross margins in growth businesses average 30 percent, compared to 27 percent in mature businessesa significant difference at the .05 level), which inflates their revenue figures. Or Parkinson's Law may be operating: mature businesses are choked with people (presumably administrators/stafO who, on average, generate relatively little salable output. Capital intensity is less for high share than for low share businesses. This may reflect the smaller revenue bases of the low share businesses, or it may suggest that low capital intensity provides flexibility and market responsiveness which, in turn, can lead to high market share. Working Capital Management Wildcats have a significantly higher receivables/revenue ratio than do any of the other types of businesses (Table 4). This may reflect their attempts to attract customers via liberal credit terms. This relatively Table 4 Strategic Attributes of Businesses in the Four Cells of the BCG Matrix (Means Reported, with Standard Deviations in Parentheses) Working Capital Management
2-Way Anova

(Main Effects)
Strategic Attributes Wildcats (N= 181) 17.62 (11.77) 20.90 (11.71) Stars (N=114) 15.62 (8.80) 18.61 (10.54) Cash Cows (N=315) 15.15 (7.47) 19.87 (10.59) Dogs (N=418) 15.13 (8.89) 22.01 (11.45) Life Cycle Stage Market Share

Receivables/revenue Inventories/revenue *p<.Ol **p<.OOl

1982

Hambrick, MacMittan, and Day

523

mundane way of competing as a low share business apparently is not the norm for Dogs. A possible explanation is that rigid credit norms build up over an industry's life cycle, discouraging unusual credit practices. An indicting explanation for the relatively high inventories held by low share businesses is that these businesses are less adept at managing their inventories, which in turn raises the question of whether they are more poorly managed in general. Their low shares could prompt such a speculation, but the data from this study do not necessarily support it. Another, more charitable, explanation for the high inventories held by Wildcats and Dogs is that they attempt to compete, using those inventories for ready delivery. This is another example of a relatively mundane competitive device, aimed at what Katz (1970) would call "going for the crumbs"those customers in which the dominant players have little interest. Expense Structure This study also included an examination of the expense structures of businesses, or how they add value (Table 5). Results indicate differences primarily according to life cycle stage, and less so by market share. Table 5 Strategic Attributes of Businesses in the Four Cells of the BCG Matrix (Means Reported, with Standard Deviations in Parentheses) Expense Structure
2-Way Anova (Main Effects) Life Cycle Market Stage Share

Strategic Attributes Manufacturing/revenue Product R&D/revenue Process R&D/revenue Sales lorce expenditures/ revenue Advertising and promotion expense/revenue

Wildcats (N= 181) 26.31 (12.06) 2.63 (2.66) (1.24) 6.48 (4.17) 1.02 (1.15)

Stars (N=II4) 25.08 (9.99) 2.76 (2.51) 1.10 (1.32) 5.45 (4.21)

Cash Cows (N=315) 28.45 (11.50) 1.68 (2.08) (.81) 4.99 (4.15)

Dogs (N=418) 28.30 (11.33) 1.76 (2.28) (.84) 5.32 (4.35)

* ** **
** *

.87

.53

.52

.85
(.93)

.71
(.92)

.81
(1.07)

*p<.05 **p<.Ol ***p<.001

Growth businesses tend to spend proportionately more on what might be called "future-oriented" expenses, that is, product R&D, process R&D, advertising, and sales force, than do mature businesses. Three factors may be creating this difference. First, managers of growth businesses may tend to view their businesses as having longer, brighter futures than do the managers of mature businesses, and thus they are more willing to

524

Academy of Management Journat

September

incur costs that will have an impact only in the future. This is not a convincing rationale. Sales force and advertising expenses presumably have some important near and current term payoffs even for mature businesses. A second possibility is that many of these mature businesses are being managed for cash throwoff, according to BCG prescriptions, and therefore nondirect expenses are minimized. A third interpretation stems from the semifixed, rather than variable, nature of most of these "futureoriented" expenses. Thus, in growth businesses, which may have smaller revenue bases than mature businesses, these expenses take on disproportionate magnitude when expressed as a percentage of sales. This line of reasoning also may explain why low share businesses (with relatively small revenue bases) have higher sales force and advertising expenses than do high share businesses. Mature businesses add more value through manufacturing than do growth businesses. This is an indication of the relative emphasis of mature businesses on the "core technology" (Thompson, 1967) or "engineering" (Miles & Snow, 1978) aspects of the business rather than on the "domain" or "entrepreneurial" aspects. In light of this emphasis, it could be expected that mature businesses would spend a relatively heavy amount on process R&D, in an attempt to make their throughout functions even more efficient (Utterback & Abernathy, 1975). As already observed, the opposite is true. Mature businesses, on average, spend about half as much of their sales dollars on process R&D as do growth businesses. One explanation is that organizational structures, climates, and technological orientations either foster R&D in general, or they do not. That is, product R&D and process R&D tend to go hand in hand. This speculation pales in light of the only moderate correlation (/= .24, p > .01) between the two types of R&D expenses for the entire sample. Another possible explanation for the relatively low process R&D expenditures by mature businesses is that these businesses are being managed for cash throwoff in industries with severe price competition, such that even expenditures on process R&D are viewed as profit detractors. Domain The term "domain" is used as Thompson (1967) did, to refer to the array of products and markets a business stakes out for itself. Because the rnain domain variables in the PIMS data base are expressed in terms relative to the competition, no differences were expected across stages of the life cycle and, in fact, none were observed (Table 6). Differences in the domain breadths of low and high share businesses were significant. Stars and Cash Cows reported more relative product line breadth, customer type breadth, and relative number of customers than did their low share counterparts. Because this is a cross-sectional study, there is no way of determining whether a broad domain is a means of gaining share, or whether domain broadening is an activity typically pursued

1982

Hambrick, MacMillan, and Day

525

Table 6 Strategic Attributes of Businesses in the Four Cells of the BCG Matrix (Means Reported, with Standard Deviations in Parentheses) Domain
2-Way Anova (Main Effects) Life Cycle Market Share Stage

Strategic Attributes Relative product line breadth^ Relative customer type breadth Relative number of customers Customer fragmentation

Wildcats (N= 181) 1.81 (.81) 1.81 (.62) 1.47 (.65) 12.91 (10.32)

Stars (N= 114) 2.39 (.75) 2.28 (.65) 2.35 (.81) 13.94 (12.14)

Cash Cows (N=315) 2.42 (.72) 2.29 (.66) 2.47 (.69) 13.33 (11.51)

Dogs (N=418) 1.85 (.75) 1.89 (.59) 1.68 (.74) 13.54 (11.53)

^For the sake of brevity and interpretability, means and ANOVA results are reported for ordinal variables. A display of response distributions and chi-square statistics for the ordmal variables does not suggest different patterns or conclusions. p<001

by businesses that have already achieved dominance in a segment of a market. What is clear is that the low share businesses tended toward "market concentration" (Hofer & Schendel, 1978) or "focused" (Abell, 1980) strategies. They concentrated their efforts either because they recognized their weak positions or were constrained by their weak positions. There were no significant differences across the four cells in the amounts of customer fragmentation, the variable that taps the extent to which the business avoids relying on a few key customers. It could have been expected that Wildcats, in particular, would have a small set of customers with which they were establishing themselves. And, in fact, the customer fragmentation score for Wildcats is the lowest (though not significantly) of the four types. On average, though, none of the types has extraordinarily more or less customer fragmentation than the others. Vertical Integration Just as high share businesses have broader domains than low share businesses, so do they also tend to be more vertically integrated (Table 7). Their value added/revenue figures are higher than for low share businesses. And they indicate significantly more vertical integration (both backward and forward), relative to their competition, than do low share businesses. As with domain breadth, there is no way of knowing from these data whether vertical integration is a cause or an effect of high market share. A reasonable speculation is that high share businesses tend to integrate vertically to perpetuate their growth and that they integrate because their

526

Academy of Management Journal

September

Table 7 Strategic Attributes of Businesses in the Four Cells of the BCG Matrix (Means Reported, with Standard Deviations in Parentheses) Vertical Integration
2-Way Anova (Main Effects) Life Cycte Market Stage Share *

Strategic Attributes Value added/revenue Relative V.I. backward Relative V.I. forward

Wildcats (N=18I)
56.13 (16.58) 1.83 (.60) 1.89 (.52)

Stars (N= 114) 61.14 (13.63) 1.98 (.62) 1.97 (.50)

Cash Cows Dogs (N=315) (N=418)


59.57 (14.19) 2.04 (.60) 1.99 (.49) 54.77 (14.84) 1.78 (.59) 1.92 (.47)

*p<.05
**p<.OQl

scale of operations makes it relatively difficult to be assured of outside supplies in the quantities and at the prices they desire (Williamson, 1975Kreiken, 1980). Competitive Devices In examining the tendencies of the four types of businesses to use various competitive devices, some striking differences are observed (Table 8). Understandably, growth businesses have higher sales from new products than do mature businesses. (Businesses in all four cells claim to have, on average, higher sales from new products than their competitors. This can be reconciled only by returning to the earlier contention that the PIMS businesses are likely to be more aggressive, and hence perhaps more prone to new product activity, than are their non-PIMS competitors.) Wildcats have the highest new product sales, and Dogs the lowest. This may reflect a kind of self-fulfillment of the BCG doctrine, in which Wildcats are vie\yed, either by their own managers or their parent firms' strategists, as having the potentially longest and most rewarding horizons of any of the four types, and thus most deserving of a new product orientation. Dogs, typically viewed as having no promising future (Henderson, 1979), are viewed as not warranting the outlays associated with new products. High share businesses indicate the relatively low direct costs that should accrue to them due to their accumulated experience (Henderson, 1979; Hofer & Schendel, 1978). The typical prescription for high share businesses in the growth stage is for them to drive costs down and to price at discouragingly low levels. However, the Stars in the data base had relatively high prices, perhaps reflecting that they already were established leaders instead of struggling for leadership. More broadly, it warrants noting that both Stars and Cash Cows are reaping double benefits from their market power: relatively low costs and relatively high prices.

1982

Hambrick, MacMillan, and Day

527

Table 8 Strategic Attributes of Businesses in the Four Cells of the BCG Matrix (Means Reported, with Standard Deviations in Parentheses) Competitive Devices
2-Way Anova (Main Effects) Life Cycle Market Stage Share * * * * * *

Strategic Attributes Sales from new products Relative sales from new products Relative prices Relative direct costs Relative product quality Relative image Relative services Relative advertising expenses Relative sales promotion expenses Relative sales force expenses

Wildcats (N= 181) 18.66 (20.69) 4.01 (11.80) 103.20 (7.02) 104.30 (7.68) 22.03 (29.18) 3.21 (.86) 3.27 (.80) 2.19 (.99) 2.39 (.93) 2.76 (1.01)

Stars (N=1I4) 18.16 (20.73) 2.08 (10.19) 105.00 (7.40) 99.52 (8.38) 45.12 (29.77) 4.06 (.71) 4.00 (.83) 2.75 (1.11) 3.03 (1.07) 3.09 (1.04)

Cash Cows (N=315) 7.31 (12.97) .79 (6.21) 104.30 (6.37) 100.20 (7.13) 34.25 (28.56) 3.96 (.78) 3.83 (.82) 2.82 (.96) 2.99 (.89) 3.12 (1.02)

Dogs (N=418) 7.82 (13.68) .50 (7.08) 102.70 (5.13) 103.20 (6.93) 17.64 (25.38) 3.26 (.81) 3.28 (.79) 2.29 (.93) 2.51 (.86) 2.79 (.93)

Stars and Cows apparently command their high prices through a broad array of superiorities. They claim to have higher average product quality, image, and services than their low share competitors. They claim to spend relatively more on advertising, sales promotions, and sales forces than their lesser adversaries. All of these measures are ordinal and somewhat impressionistic, so there is some likelihood that the market leaders falsely attribute to themselves strength in all categories (Nisbett & Wilson, 1977). Still, some of the apparent differences are substantial, and they tend to square with expectations. Businesses with various strengths would be expected to gain market share and, once dominant, they would be expected to reinforce and add strengths with the slack generated from market leadership (Cyert & March, 1963). Conclusions This paper has attempted to test and extend the BCG product portfolio matrix. The primary theme of BCGthat the four cells of the matrix have quite different tendencies to generate or consume cashhas been corroborated. Significant differences among the four cells on other performance measuresreturn on investment, return per risk, and market share changealso were observed. Thus, each of the four types in the matrix

528

Academy of Management Journat

September

Wildcats, Stars, Cows, and even Dogscontributes in its own way to the balanced performance of the corporation. Of particular note is the finding that the average Dog has a positive cash flow, even greater than the cash needs of the average Wildcat. BCG's schematic (Henderson, 1979, p. 165) portraying optimal cash flows could be revised, as in Figure 2, on the basis of these results. Figure 2 Cash Flows Within the BCG Matrix
Stars

t>6gs BCd Prescription (Henderson, 1979) Revised, based on study ^

The results do not support BCG's advice that Dogs should be promptly harvested or liquidated. This should come as a relief to many managers, because more and more of their industries are maturing and because all but the market leaders qualify as Dogs. What is needed is creative, positive research and thinking about how Dogs can be managed for maximum long term performance. Another key, tentative conclusion to come from the study is that businesses may not always face sharp tradeoffs between share building and cash flow or profitability. Only among Stars was there an inverse relationship between market share change and any of the measures of returns. Otherwise the relationships were nil, suggesting that multiple, seemingly incompatible objectives can be pursued in tandem. More research is needed on the circumstances that favor such "well-rounded" effectiveness and on the internal features that can promote or stymie it. The four types of businesses differ markedly in thfir strategic attributes. Some attributes (e.g., R&D expenses, plant and equipment newness) vary according to life cycle stage. Some (e.g., domain breadth, vertical integration, relative marketing expenditures) vary according to market share position. Still others (e.g., capacity utilization, sales/employee) vary according to both dimensions. What emerges is an expanded understanding of the strategic profile of each type of business: Relative to the other cells. Wildcats tend to have low capacity utilization, new plant and equipment, high current asset levels, high capital intensity, high R&D expenses, high marketing expenses, narrow domains, heavy new product activity, high direct costs, and competitive devices that lag Star competitors on all fronts.

1982

Hambrick, MacMiltan, and Day

529

Stars tend to have new plant and equipment, high capacity utilization, high R&D expenses, broad domains, high sales per employee, high value added, and superiority on a number of competitive devices. Cows tend to have very high capacity utilization, dated plant and equipment, low capital intensity, low sales per employee, low R&D and marketing expenditures, broad domains, and superiority on essentially all competitive devices examined. Dogs tend to have dated plant and equipment, medium capital intensity, high inventory levels, low R&D expenses, moderate marketing expenses, narrow domains, low value added, and competitive devices that lag Cow competitors on all fronts. Overall, there is a clear indication that businesses differ in their performance and strategic attributes, according to their life cycle stages and market shares. The importance attached to these two key constructs by Hofer (1975) and others appears not to have been ill-placed. This paper sheds empirical light on an important, but heretofore illdocumented strategic framework, but it also raises many questions. For example, would consumer products or service businesses yield findings different from the industrial businesses studied here? Would a sample of non-PIMS businesses yield similar findings? The only contingency variables included were life cycle stage and market share. Many others have been suggested in the literature (Hofer, 1975). If businesses were subdivided further within the four BCG cells according to additional contingent factors (e.g., concentration rates, frequency of purchase, or advertising intensity), what new findings would emerge? The present study should serve as a springboard for adding contingent factors toward the goal of full scale contingency models as advocated by Hofer. The cross-sectional nature of this study poses an obvious problem. How businesses move among the four cells of the matrix or how their strategic attributes tend to change as they move from cell to cell has not been examined. This shortcoming highlights a key opportunity for future PIMS research. The data base has been in existence long enough that some longitudinal analyses should now be possible. Empirical analysis of the BCG matrix has long been overdue, as have analyses of many other normative and conceptual devices in the field of strategy. This paper has corroborated and extended the published ideas of the Boston Consulting Group. It serves as an important backdrop to an accompanying study of the relationships between strategic attributes and performance in each of the four cells (MacMillan et al., 1982). References
Abell, D. Defining the business. Englewood Cliffs, N.J.: Prentice Hall, 1980. Anderson, C. R., & Paine, F. I. PIMS: A reexamination. Academy of Management Review, 1978, 3, 602-612. Andrews, K. R. The concept of corporate strategy. Homewood, 111.: Dow Jones-Irwin, 1971.

530

Academy of Management Journal

September

Bloom, P. E., & Kotler, P. Strategies for high market-share companies. Harvard Business Review. 1975, 53(6), 63-72. Boston Consulting Group. Perspectives on experience. Boston: The Boston Consulting Group, 1968. Bourgeois, L. J. Performance and consensus. Strategic Management Journal, 1980, I, 227-248. Buzzell, R. D., Gale, B. T., & Sultan, R. G. M. Market share: A key to profitability. Han/ard Business Review, 1975, 53(1), 97-106. Chevalier, M. The strategy spectre behind your market share. European Business, 1972, 34, 63-72. Christensen, K. H. Product, market, and company influences upon the profitability of business unit research and development expenditures. Unpublished doctoral dissertation, Columbia University 1977. Clifford, D. K., Jr. Managing the product life cycle. In R. Mann (Ed.), The arts of top management: A McKinsey anthology. New York: McGraw Hill, 1971, 216-226. Cyert, R., & March, J. A behavioral theory of the firm. New York: Prentice-Hall, 1963. Ford, J. D., & Slocum, J. W., Jr. Size, technology, environment, and the structure of organizations. Academy of Management Review, 1977, 2, 561-575. Fox, H. W. A framework for functional coordination. Atlanta Economic Review, 1973, 23, 8-11. Fruhan, W. E., Jr. Pyrrhic victories in fights for market share. Harvard Business Review; 1972, 50(5),
1UU* IU /.

Glueck, W. G. Business policy. New York: McGraw-Hill, 1976. Hamermesh, R. G., Anderson, M. J., & Harris, J. B. Strategies for low market share businesses. Harvard Business Review, 1978, 56(3), 95-102. Harrigan, K. R. Strategies for declining industries. Journat of Business Strategy, 1981, 2(2), 20-34. Hatten, K. J., Schendel, D. E., & Cooper, A. C. A strategic model for the U.S. brewing industry: 1952-1971. Academy of Management Journat, 1978, 21, 592-610. Hedley, B. Strategy and the "business portfolio." Long-Range Planning, 1977, 10, 9-15. Henderson, B. D. Henderson on corporate strategy. Cambridge, Mass.: Abt Books, 1979. Hofer, C. W. Toward a contingency theory of business strategy. Academy of Management Journal 1975, 18, 784-810. Hofer, C. W., & Schendel, D. Strategy formulation: Anatyticat concepts. St. Paul: West, 1978. Katz, R. L. Cases and concepts in corporate strategy. Englewood Cliffs, N.J.: Prentice-Hall, 1970. Kreiken, J. Effective vertical integration and disintegration strategies. In W. Glueck (Ed.), Business poticy and strategic management. New York: McGraw Hill, 1980, 256-263. Lenz, R. T. Strategic interdependence and organizationatperformance: Patterns in one industry. Unpublished doctoral dissertation, Indiana University, 1978. Levitt, T. Exploit the product life cycle. Harvard Business Review, 1965, 43(6), 81-94. MacMillan, I. C , Hambrick, D. C , & Day, D. L. The association between strategic attributes and profitability in the four cells of the BCG matrixA PIMS-based'analysis of industrial-product businesses. Academy of Management Journat, 1982, forthcoming. Miles, R. E., & Snow, C. C. Organizational strategy, structure, and process. New York: McGrawHill, 1978. Nisbett, R. E., & Wilson, T. D. Telling more than we know: Verbal reports on a mental process. Psychological Review, 1977, 84, 231-259. Patel, P., & Younger, M. A frame of reference for strategy development. Long Range Planning, 1978,11,6-12. Patton, A. Stretch your products' earning years. Management Review, 1959, 48(6), 9-14. Robinson, S. J. Q., Hickens, R. E., & Wade, D. t h e directional policy matrixTool for strategic plannins. Long Range Ptanning, 1978, 10, 17-27. Schoeffler, S. Cross-sectional study of strategy, structure, and performance: Aspects of the PIMS program. In H. Thorelli (Ed.), Strategy and structure performance. Bloomington, Ind.: Indiana University Press, 1977, 108-121.

1982

. Hambrick, MacMittan. and Day

531

Schoeffler, S., Buzzell, R. D., & Heany, D. F. Impact of strategic planning on profit performance. Harvard Business Review, 1974, 52(2), 137-145. Taylor, B. Managing the process of corporate development. Long-Range Planning, 1976, 9, 81-100. Thompson, J. D. Organizations in action. New York: McGraw-Hill, 1967. Utterback, J., & Abernathy, W. A dynamic model of process and product innovation. OMEGA, 1975, 3, 639-656. Vancil, R. F. Strategy formulation in complex organizations. Sloan Management Review, 1976, 17, 83-90. Wasson, C. R. Product management. St. Charles, 111.: Challenge Books, 1974. Williamson, O. Markets and hierarchies: Analysis and antitrust implications. New York: Free Press, 1975. Woo, C. Y. Y., & Cooper, A. C. Strategies of effective low market share businesses. Academy of Management Proceedings, 1980, 21-25. Donald C. Hambrick is Associate Professor, Graduate School of Business, Columbia University. Ian C. MacMitlan is Associate Professor, Graduate School of Business, Columbia University. Diana L. Day is a doctoral student at the Graduate School of Business, Columbia University.

You might also like