Corporate-Level Strategy, Business-Level Strategy, and Firm Performance

Author(s): Donald W. Beard and Gregory G. Dess

Source: The Academy of Management Journal , Dec., 1981, Vol. 24, No. 4 (Dec., 1981), pp.
663-688
Published by: Academy of Management

Stable URL: https://www.jstor.org/stable/256169

REFERENCES
Linked references are available on JSTOR for this article:
https://www.jstor.org/stable/256169?seq=1&cid=pdf-
reference#references_tab_contents
You may need to log in to JSTOR to access the linked references.

JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide
range of content in a trusted digital archive. We use information technology and tools to increase productivity and
facilitate new forms of scholarship. For more information about JSTOR, please contact support@jstor.org.

Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at
https://about.jstor.org/terms

Academy of Management is collaborating with JSTOR to digitize, preserve and extend access to
The Academy of Management Journal

This content downloaded from

45.77.104.39 on Thu, 23 May 2024 13:51:30 +00:00
All use subject to https://about.jstor.org/terms
i Academy of Management Journal
1981, Vol. 24, No. 4, 663-688.

Corporate-Level Strategy,
Business-Level Strategy,
and Firm Performance
DONALD W. BEARD
University of Washington
GREGORY G. DESS
University of South Carolina

Corporate-level strategy and business-level strategy

are operationalized in terms of interindustry and intra-
industry variation, respectively. Variables representing
both levels of strategy are used in a regression model to
explain variation infirm profit performance. Both kinds
of variable are found to be significant in explaining var-
iation in firm profitability.

Theoretical literature in the business policy area has increasingly empha-

sized distinctions between two levels of organizational strategy: (1) corpo-
rate-level strategy, concerned with questions about what businesses to
compete in, and (2) business-level strategy, concerned with questions of
how to compete within a particular business. Differing conceptual
schemes and associated analytic techniques have been proposed to aid top
managers in making decisions about the two different kinds of strategy.
Hofer and Schendel (1978) provide a recent literature review and a ra-
tionale for separating and sequencing these, two kinds of strategic deci-
sions.
Although business policy theory has been -evolving in this direction, at
least since Ansoff (1965), empirical research to test propositions derived
from this theory has been limited. It is the purpose of this paper to present
the results of such an empirical test. More specifically, the research pre-
sented here provides evidence about the relative importance of corporate-
level strategy and business-level strategy in determining firm profit perfor-
mance.
Conceptually, corporate-level strategy and business-level strategy are
seen as corresponding, respectively, to interindustry and intraindustry var-
iations in business firms' strategies. The research design used single-
industry firms as the unit of analysis. In this respect, the research is similar
to the Profit Impact of Marketing Strategy (PIMS) project at Harvard

663

This content downloaded from

45.77.104.39 on Thu, 23 May 2024 13:51:30 +00:00
All use subject to https://about.jstor.org/terms
664 Academy of Management Journal December

University, except that single-industry corporations rather than single-

industry subunits of multi-industry firms are the primary unit of analysis.
This design has the advantage that interindustry and intraindustry strate-
gic variation can be clearly distinguished and operationalized.

CONCEPTS OF STRATEGY AND UNITS OF ANALYSIS

Organizational strategy is one of the broadest and most complex con-

cepts used in studying organizations. Because the concept of strategy has
been evolving rapidly in the business policy literature, it should be made
clear at the outset what concepts of strategy have been used and what units
of analysis have been studied. Figure 1 sets forth a framework that helps
to specify the concern of this paper. Figure 1 cross-classifies three hierar-
chical levels of strategy with four hierarchical units of analysis. On the left
in Figure 1, the concepts of strategy are arranged in hierarchical order
from top to bottom, ranging from the most general concept, corporate-
level strategy, to the least general concept, functional strategy. Across the
top of Figure 1, the units of analysis are arranged in reverse hierarchical
order, ranging from the smallest unit, management decision makers, to
the largest unit, the indirectly linked environment. The concepts of direct
and indirect linkages in an organization's environment as used in Figure 1
are analogous to the concepts of direct and indirect requirements in a
Leontief-type input-output model. Leontief (1953), Chenery and Clark
(1959), and Miernyk (1965) provide thorough discussions of the theory
and applications of this model.
The research discussed below deals with only two concepts of strategy in
Figure 1, corporate-level strategy and business-level strategy. The focus of
the research is primarily on the organization as a whole, the second unit of

FIGURE 1
Three Concepts of Strategy and Four Units of Analysis

Four Units of Analysis

Organizational Units Environmental Units

1. 2. 3. 4.
Three Management The The Directly The Indirectly
Concepts of Decision Organization Linked Linked
Strategy Makers as a Whole Environment Environment

1. Corporate-Level l _
Strategy l l

2. Business-Level
Strategy
LI -- - - - - -1I- - - - -

3. Functional-Level
Strategy

This content downloaded from

45.77.104.39 on Thu, 23 May 2024 13:51:30 +00:00
All use subject to https://about.jstor.org/terms
 1981 Beard and Dess 665

analysis in Figure 1, and the organization's directly linked environment,

the third unit of analysis in Figure 1.
Although space limitations prevent much discussion of the distinctions
made in Figure 1, it is important to elaborate briefly on the differences be-
tween the characteristics of strategic decisions or decision makers on one
hand and the characteristics of an organization's strategy on the other
hand. It is believed that organizations originate and change on the basis of
creative, strategic decisions by individuals or groups occupying key orga-
nizational roles. Weick (1969), Child (1972), and Miles, Snow, Meyer, and
Coleman (1978) develop this view more thoroughly. Strategic decision
making is seen as a crucial part of the process by which organizations
adapt to their environments.
It also is believed that those decisions that actually succeed in creating or
changing organizations do so via complex iterative processes, which policy
theorists subsume under the concept of strategy implementation. Andrews
(1971) provides a broad theoretical overview of the strategy implementa-
tion process. Empirical work documenting the complex nature of strategic
decision and implementation processes includes Cyert and March (1963),
Bower (1970), Carter (1971), Pfeffer and Salancik (1974), and Mintzberg,
Raisinghani, and Theoret (1976).
For purposes of the present study, a key assumption has been made
about the relationship between organizational strategy and organizational
performance. It was assumed that the effects of strategy on performance
at a particular point in time, or during a particular period of time, are best
studied in terms of the organization's implemented strategy at or during
the relevant time. This means that the concept of strategy here is based on
organizational characteristics that embody earlier strategic decision and
implementation processes. Actual outcomes of decision and implementa-
tion activities are, of course, determined both by complex, iterative pro-
cesses among decision makers within the organization and by interaction
between the organization and its environment.
Concern will be only with the concepts of strategy enclosed within the
dotted line of Figure 1-corporate-level strategy and business-level stra-
tegy. Corporate level strategy is conceived in terms of variation in the
portfolio of industries in which a firm does business. Business-level stra-
tegy is conceived in terms of variation in the firm's strategic characteristics
relative to the population of firms within the industries in which it does
business. The verbal definitions employed closely follow Hofer and Schen-
del's (1978) concepts of corporate-level and business-level strategy. The
operational measures of these concepts are discussed in more detail later in
this paper.
Corporate-level strategy is defined in terms of variation in the deploy-
ment of a firm's resources among the portfolios of industries within which
all business firms compete. Hofer and Schendel propound this view:
"corporate-level strategy is concerned primarily with answering the ques-
tion of what set of businesses should we be in. Consequently, scope and

This content downloaded from

45.77.104.39 on Thu, 23 May 2024 13:51:30 +00:00
All use subject to https://about.jstor.org/terms
 666 Academy of Management Journal December

resource deployments among businesses are the primary components of

corporate strategy" (1978, p. 27). Thus, a firm's corporate-level strategy
can be operationalized in terms of the distribution of firm assets, sales,
employment, capital-budget, or other indexes of firm resources among the
range of existing industries.
Most firms have simple corporate-level strategies, in these terms. They
compete in only one industry among the hundreds that are possible. Other
firms, however, such as the Fortune 500 largest United States industrial
firms, typically participate in several industries, and their top managers
must contend with the varied and conflicting demands of their industrially
specialized subunits.
Because research interest in this paper is primarily in the question of
how important variation in corporate-level strategy is relative to business-
level strategy in explaining firm performance, the complex differences
among industries will be represented quite abstractly. Differences in the
average profitability among industries will be used to represent the overall
differential in profit making opportunity among industries.
It is true that decisions about corporate-level strategy in multi-industry
firms are based on a wide variety of information other than industry prof-
itability and that these decisions affect many variables other than the dis-
tribution of firm assets among industries. Springer and Hofer (1980), for
example, document the rich variety of decisions that have attended the
evolution of General Electric's strategic planning process, which now in-
cludes distinct responsibilities and procedures at the corporate level and at
the business level. Berry (1975) has identified General Electric as the sec-
ond most diversified firm as of 1965 among the Fortune 500 largest indus-
trials.
Lieberson and O'Connor (1972) have used variation in the average prof-
itability of a subject firm's primary industry to assess the impact of dif-
ferences in the firm's competitive environment upon firm performance-
an approach similar to that of the present study. Other researchers have
measured variation in a firm's corporate-level strategy in different ways.
For example, Gort (1962) used the number of industries in a firm's port-
folio to measure the diversity of a firm's corporate-level strategy, and
Rumelt (1974) used a measure of the technical relatedness of the industries
in which multi-industry firms competed. In another vein, Pitts (1977) has
shown that marked structural differences exist at the corporate level be-
tween firms that have diversified via internal growth and those that have
diversified by acquisition.
Business-level strategy is defined in terms of variation in firm character-
istics relevant to competitive success or failure within a given industry. In
this paper, a firm's competitively relevant, business-level characteristics
are conceived exclusively in relative terms. That is, a firm would have a
separate business-level strategy for each industry in which it competed,
and the relevant characteristics of the firm's business-level strategy would

This content downloaded from

45.77.104.39 on Thu, 23 May 2024 13:51:30 +00:00
All use subject to https://about.jstor.org/terms
 1981 Beard and Dess 667

be measured relative to the range and norms on each characteristic in each

of its industries.
Hofer and Schendel again provide a succinct definitional statement:
At the business level, strategy focuses on how to compete in a particular industry or
product-market segment. Thus, distinctive competences and competitive advantage are
usually the most important components of strategy at this level (1978, pp. 27, 28).
In the selection of variables to represent business-level strategy, focus
was on variables that have been shown empirically to effect firm competi-
tive performance. In this respect, less emphasis was put on a firm's mix of
market segments and product line items within a particular industry than
that used by Hofer and Schendel (1978). Instead, emphasis has been on
variables that demonstrably tend to confer competitive advantage or dis-
advantage.
As with corporate-level strategy, business-level strategy can be opera-
tionalized in terms of a rich variety of measures. In two of the widest rang-
ing studies, Schoeffler, Buzzell, and Heany (1974) and Schendel and
Patton (1978), firm size relative to competitors and firm resource alloca-
tions to capital investment, advertising, and research relative to competi-
tors were studied as strategic determinants of firm profitability. Taking a
more financially oriented view, Hall and Weiss (1967) and Fisher and Hall
(1969) found two risk factors, unpredictability of firm profitability and
debt leverage, respectively, to explain considerable variance in firm profit-
ability.

RELEVANT THEORY AND RESEARCH

The main research question addressed in this paper is degree to which

variation in a firm's corporate-level strategy and in its business-level stra-
tegy explains variation in its profit performance. Although this may seem
to be a simple question, rarely has theory and research combined both the
interindustry and intraindustry perspectives in terms of definitions of
corporate-level strategy and business-level strategy used here.
A great deal of theory and supporting research on the economics of in-
dustrial organization leaves little doubt that interindustry differences in
structure and profitability are persistent over time and are similar among
industrialized nations. This also means, of course, that there are differ-
ences in the average profitability of the firms competing in different indus-
tries. Scherer (1970), Weiss (1974), and Caves (1977) review this literature.
The industrial organization field has focused largely on industrial aggre-
gates of firms, however, rather than on the firms themselves. The indus-
trial organization framework thus says little about either the range of vari-
ation in firm performance across or within industries or about other rela-
tively large differences among firms in general or among firms competing
within a single industry. Although business policy and other areas of busi-
ness administration have focused on business firms as the unit of analysis,
until recently they have produced little systematic or comprehensive re-
search on variation in the environments in which individual firms com-
pete.

This content downloaded from

45.77.104.39 on Thu, 23 May 2024 13:51:30 +00:00
All use subject to https://about.jstor.org/terms
668 Academy of Management Journal December

The literature reviewed below is organized in three sections. First, a

brief review of pivotal works contributing to the evolution of separate,
hierarchical concepts of strategy will be presented. These works provide
more background on the concepts of corporate-level strategy and
business-level strategy set forth above. The remaining two sections cover
the empirical research from which the hypotheses of the present research
derive. These sections discuss evidence on relationships between firm
profit performance and both corporate-level strategy and business-level
strategy, respectively.

Corporate-level and Business-level Concepts of Strategy

Ansoff was among the first to conceptualize different levels of organiza-

tional decision making. Ansoff saw three levels of decisions facing the or-
ganization's decision makers: strategic decisions- "the selection of prod-
uct mix and markets ... an impedence match between the firm and the en-
vironment," administrative decisions-"structuring a firm's resources to
maximize performance potential," and operating decisions-"maximize
the efficiency of the firm's resource conversion process" (1965, pp. 5, 6).
Ansoff's first two types of strategy roughly approximate the concepts of
corporate-level and business-level strategy, respectively, that are used in
the present research.
Authors in the Harvard Business School tradition (Levitt, 1960; An-
drews, 1971; Uyterhoeven, Ackerman, & Rosenblum, 1977; Christensen,
Andrews, & Bower, 1978) have recognized two similar levels of strategy.
Andrews, for example, defined corporate strategy as "the pattern of
major objectives, purposes, or goals and essential policies and plans for
achieving those goals stated in such a way as to define what business the
company is in or is to be in and the kind of company it is or is to be"
(1971, p. 25). The decision on what business the company is in or is to be
in clearly approximates the concept of corporate-level strategy used here.
The decision on what kind of company it is or is to be is too vague to be
easily interpreted, but it could be seen as incorporating the concept of
business-level strategy.
Vancil and Lorange (1975) define three levels of strategy that parallel
those of the present study. They view strategic planning in diversified com-
panies as moving through three cycles: setting corporate objectives at the
top, setting consonant business objectives and goals in the divisions, and
establishing the required action programs at the functional level.
Miles et al. (1978) identify three broad types of problems facing organi-
zations: the entrepreneurial problem, the engineering problem, and the
administrative problem. Solving the entrepreneurial problem in their
model is equivalent to decisions on corporate-level strategy in presently
used terms, and the latter two types of problems fit loosely with the con-
cept of business-level strategy.

This content downloaded from

45.77.104.39 on Thu, 23 May 2024 13:51:30 +00:00
All use subject to https://about.jstor.org/terms
 1981 Beard and Dess 669

Hofer and Schendel (1978) prescribe different analytic strategic tasks at

the corporate level and the business level. They see the principal task of
analysis at the corporate level as evaluating the relative attractiveness of
business(es) in the firm's portfolio and the principal tasks of analysis at
the business-level as assessing the stage of the product life cycle and the
firm's competitive position-within each relevant business. Hofer and
Schendel's (1978) definitions are the most specific, and theirs are followed
closely in this paper.

Corporate-level Strategy and Firm Performance

Empirical research on relationships between corporate-level strategy

and firm performance is discussed in two parts below. The first part con-
cerns effects of the quantity and type of diversity in a firm's business port-
folio on its profit performance. The second part concerns effects of varia-
tion in industry on firm profit performance.
Although some theoretical reasons can be advanced that the quantity of
industrial diversification per se may affect business firms' profitability,
empirical research thus far indicates that little relationship exists between
diversity and profitability. Rhoades (1973) suggested that diversified firms
might create barriers to entry to various industries in two ways: first, by
using profits from one industry to subsidize predatory pricing in another
industry and, second, by obscuring attractive returns in one or more of
their industries through consolidated fmancial reporting. Rhoades' (1973)
initial research, based on 1963 data for a sample of 244 manufacturing in-
dustries-four digit Standard Industrial Classifications (SIC)-showed
some modest support for this view. However, Rhoades (1974) subse-
quently developed three additional measures of industry diversity using
improved detail in data for 1967 published by the U.S. Census Bureau.
The data allowed measurement of firm diversilfication in terms of both
(a) the number of industries in which firms competed and (b) the propor-
tion of firms' sales outside their primary industry. In the second study, he
found a modest negative relationship between diversity and profitability.
Rhoades (1974) attributed the contradictory results of the two studies
more to differences in their levels of industrial aggregation than to their
differences in diversification measurement.
Several studies using large U.S. manufacturing firms as the unit of anal-
ysis have found no relationship between diversity and profitability. Gort's
(1962) work is one of the most comprehensive studies available on this
subject. In a sample of 100 of the 200 largest manufacturing firms in the
United States in 1954, including data for the years 1947 through 1954, he
found virtually no correlation between return on net worth and two mea-
sures of firm diversification.
Gort (1962) did find a minimally significant positive correlation between
firm growth in assets between 1939 and 1954 and diversification in the
latter year. Berry (1975) supported this result in a sample including nearly
all of the 500 largest U.S. manufacturing firms and including data for

This content downloaded from

45.77.104.39 on Thu, 23 May 2024 13:51:30 +00:00
All use subject to https://about.jstor.org/terms
 670 Academy of Management Journal December

1960 and- 1965. Berry found low positive associations between growth in
corporate assets in this 5-year period and several measures of growth in
corporate diversification. However, the regressions as a whole explained
little variance in corporate growth. Using a sample of approximately 300
of the largest U.S. manufacturing firms, Rumelt (1977) regressed Berry's
measures of firm diversity in 1960 and in 1965 against firm profitability in
these years and found no significant relationship.
A second kind of evidence about corporate-level strategy is especially
relevant to the present research design. This is evidence as to the effect of
variation in the average profitability of industries on the profitability of
fi'rms competing within them. On the average, of course, the weighted
average for all firms in an industry gives the industry's profitability, and
much of this variation in industry profitability can be explained by varia-
tion in industrial market structure. However, individual firms within a
given industry clearly vary markedly in their profitability, and thus varia-
tion in profitability among firms can be explained only partially by varia-
tions in the industry or industries in which they compete. The main re-
search interest here concerns both how much of an individual firm's prof-
itability can be explained by its industry compared to other industries and
how much can be explained by the firm's strategy compared to other
firms' strategies within its particular industry.
Rumelt (1977) found specialized firms to be the most profitable, rela-
tively speaking, when his sample firms' performance was controlled for
the profitability of their differing industries. Firms with technically related
portfolios dropped to average relative to their industries, and firms with
unrelated portfolios remained the least profitable in both relative and ab-
solute -terms. The latter results add insight on Rumelt's earlier results as
well as provide evidence of the positive effects of industry profitability on
firm profitability.
Lieberson and O'Connor (1972) studied a sample of listed firms over the
period 1946 to 1965. They found that variation in firms' primary industry
explained 20 to 30 percent of the variation in their profitability and
growth. Lieberson and O'Connor's additional finding that variation in the
firms themselves accounted for much of the remaining variation in firm
perfornmance is also important. Because of Lieberson and O'Connor's
(1972) unorthodox method of partitioning variance, the validity of their
findings is difficult to assess.
Beard and Dess (1979) obtained results similar to those of Lieberson and
O'Connor. Both industry return on assets and industry return on equity
proved to be significant predictors of the corresponding measures of firm
profitability. In addition, intraindustry variables were found to be signifiL-
cant.

Business-level Strategy and Firm Performance

The review of research on relationships between business-level strategy

and firm performance is selective. Business-level strategic variables for

This content downloaded from

45.77.104.39 on Thu, 23 May 2024 13:51:30 +00:00
All use subject to https://about.jstor.org/terms
 1981 Beard and Dess 671

literature review and

four criteria. The fir
level strategy variables have an empirical tradition showing a relationship
with firm performance. This is consistent with the definition of business-
level strategy above, which stresses differences conferring competitive ad-
vantage or disadvantage among the competitors within a given industry. A
second criterion stemmed from practical resource constraints on the re-
search. This was that data on the variables must be available in secondary
sources for both firms and industries and that comparable measurement
of profit performance and other variables be available for both units of
analysis. A third criterion was that the variables be amenable to manage-
ment control. A flnal criterion was that the variables must be characteris-
tics of the organization as a whole that can be observed objectively across
organizations in a given industry. This restriction eliminated perceptual or
judgmental variables such as the uncertainty felt by management decision
makers.
In applying these four criteria, three business-level strategy variables
have been identified as most significant: relative size, debt leverage, and
capital intensiveness.
Firm size in either absolute or relative terms is one of the most validated
correlates of firm profit performance. For this reason it was chosen as the
first business-level strategy variable. Research generally has shown a
positive association between either absolute or relative firm size and firm
profitability. This relationship is consistent with a large body of theory
and research that demonstrates a wide variety of economies of scale.
Scherer (1970) provides an extensive review of the literature in the indus-
trial economics tradition, as of the date of publication. More recently, the
Boston Consulting Group (1972) has documented the ubiquity of log-
linear declines in unit costs and prices as cumulative output experience in-
creases.
Relative firm size within a specific industry is the main concern of this
paper and will be used as a measure of firm business-level strategy. Studies
using market share as an independent variable in explaining firm proflt
performance include Shepherd (1972), Gale (1972, 1974), Schoeffler et al.
(1974), Buzzell, Gale, and Sultan (1975), Winn (1975), and Bass, Cattin,
and Whittink (1978). All of these studies included other independent vari-
ables as controls in addition to the market share variable, and all found a
significant positive correlation between firm market share and firm profit-
ability.
Shepherd (1972) was one of the first researchers to specify firm market
share as an independent structural variable in attempting to explain firm
profltability. In a study of over 200 firms among the Fortune 500 largest
U.S. industrial flrms during the period 1960 through 1969, Shepherd
found firm market share to explain as much or more variance in these
flrms' profitability than the more traditional market structure variables.
The latter included leading-firm group share of the market, firm asset size,

This content downloaded from

45.77.104.39 on Thu, 23 May 2024 13:51:30 +00:00
All use subject to https://about.jstor.org/terms
672 Academy of Management Journal December

firm advertising to sales ratio, firm growth rate, and industry barriers to
entry.
Gale (1972) published a theoretically more complex study than Shep-
herd's (1972) of the relationships among firm profitability, firm market
share, and several interaction and control variables. Gale's data base was a
sample of over 100 firms from the Standard and Poors Compustat, An-
nual Industrial Tapes (1979) for the five years 1963 through 1967. Gale
(1972) also found that firm market share exhibited a positive association
with firm profitability, but that this association was quite variable due to
interaction between market share and other independent variables, among
which industry concentration was the strongest.
Gale's (1972) study is especially relevant to the research reported in this
paper because his theoretical discussion and research on variability in mar-
ket share's effects on profitability are important in explaining the present
findings about this relationship. His theoretical treatment and results on
the relationship between firm debt leverage and firm profitability are also
germane to the present discussion of firm debt-leverage as a business-level
strategy variable.
Winn (1975) conducted a study of firm profitability similar in design to
the two just discussed except that firm size was measured absolutely rather
than relatively. Winn's sample included nearly 800 firms in 79 industries
from the Standard and Poors Compustat, Annual Industrial Tapes (1979)
for the years 1960 and 1968. Winn's findings of a strong positive associa-
tion between firm size and profitability supports the findings on market
share cited above.
The studies of Gale (1974), Schoeffler et al. (1974), and Buzzell et al.
(1975) are based on data gathered as part of the Harvard Business School's
Profit Impact of Marketing Strategies (PIMS) project. The PIMS project
data base as of 1972 included data from 57 large North American com-
panies, about 620 of their single-industry subunits. As Buzzell et al. (1975)
indicate, an advantage of this data base for studying market share is that
businesses or markets are defined more narrowly than the U.S. SIC system
usually allows.
The well known result of the PIMS research on market share and firm
profitability is a strong positive association among the sample of single-
industry subunits. However, the relative importance of market share com-
pared to other independent variables has not been precisely quantified in
published form. Schoeffler et al. (1974) report that a regression model de-
veloped from the PIMS data base explained 80 percent of the variance in
return on investment among the 620 single-industry subunits. Gale (1974)
includes regression results, but the coefficients are not standardized. It ap-
pears from the latter results that market share and capital intensiveness ac-
count for most of the variance in profitability and are about equal in im-
portance among over 35 independent variables reported in Gale (1974).
The PIMS project approach of isolating business units of analysis com-
peting within only one product-market strongly influenced the present

This content downloaded from

45.77.104.39 on Thu, 23 May 2024 13:51:30 +00:00
All use subject to https://about.jstor.org/terms
 1981 Beard and Dems 673

study's design. However, the present design has relied on the SIC system
in the United States Office of Management and Budget (1972) to define in-
dustries or markets. The research of Shepherd (1972), Gale (1972), and
Winn (1975) reviewed above shows that this method is serviceable.
A final study using firm performance as a dependent variable and firm
relative size as an independent variable, Bass et al. (1978), suggests
another important qualification of the general market share-profit asso-
ciation. Overall, Bass et al. (1978) confirmed this association among a
sample of 63 manufacturers of food, tobacco, and cosmetics. However,
when the sample was grouped in 10 more internally homogeneous industry
classifications, the market share variable was statistically significant and
positive in only about half of the groups.
The second business-level strategy variable, capital intensiveness, also is
well validated as a correlate of firm profitability. In this case, the relation-
ship is generally negative. The theoretical context and explanation of this
phenomenon are not always consistent, however. Winn (1975) presented
and tested the hypothesis that the relationship between firm capital inten-
siveness and firm profitability is positive, not negative. His reasoning in-
cluded two major points. First, capital intensiveness implies a relatively
large mninimum efficient scale, a barrier to entry Second, consistent with
the first, firm size and capital intensiveness are associated positively, and
the latter relationship has a strong theoretical and empirical relationship
with firm profitability, as discussed above.
However, Winn (1975) found a negative regression coefficient for firm
capital intensiveness, as measured by the assets to sales ratio, in relation to
profitability. Not only was this result statistically significant, but it ex-
plained 20 to 30 percent of the variance in firm profits. As mentioned
above, this is precisely the result Schoeffler et al. (1974) and Gale (1974)
obtained in their analyses of the PIMS data base. Further supporting evi-
dence is provided by Rumelt (1974), who found predominantly vertically
integrated firms to be among the most capital intensive and the least prof-
itable.
Winn (1975) explained his finding of a negative association between
firm capital intensiveness and fLrm profitability in terms of higher fixed
operating costs that the former variable implies. He reasoned that rela-
tively capital intensive firms were more subject to operating losses in times
of cyclical downturn. Schoeffler et al. (1974) reasoned, in addition, that
capital intensive firms tend to compete in markets with relatively stan-
dardized products where price cutting to obtain volume is frequent.
Hatten and Schendel (1977) provide further evidence of a negative asso-
ciation between firm capital intensiveness and profitability. In a sample of
13 major brewers, covering 20 years of data for most of them, they found
a negative association which remained consistent and significant among a
number of subgroups within their sample.
The final business-level strategy variable is debt leverage. Empirically,
this variable has had a fairly consistent negative association with firm

This content downloaded from

45.77.104.39 on Thu, 23 May 2024 13:51:30 +00:00
All use subject to https://about.jstor.org/terms
674 Academy of Management Journal December

profitability, as was the case with capital intensiveness. However, in the

case of firm debt leverage, the theoretical context is much more complex.
The main complexity is 'that more than one source of risk can be identified
empirically and that these separate risk elements appear to interact.
Gale (1972) and Baker (1973) distinguish between business risk and fi-
nancial risk. Business risk tends to be a function of variability resulting
from rather stable aspects of industry structure and technology. It thus is
best studied in terms of interindustry variation. Interindustry variation of
this kind is viewed in this paper as relevant primarily to corporate-level
strategy. Financial risk, as measured by financial leverage, is then corre-
spondingly best stidied in terms of intraindustry variation. Thus, finan-
cial leverage and attendant risk should be measured relative to the norms
and range within a particular industry. This is the approach taken in the
present study.
Most studies of risk have focused on either business risk or financial
risk. Only one study, to the authors' knowledge, has included both and
considered them separately. Studies focusing on the variability and unpre-
dictability of profits generally have found a positive relationship between
this kind of business risk and rates of return. These include Conrad and
Plotkin (1968) and Fisher and Hall (1969). Winn (1975) pursued a similar
design in studying business risk. Rate of return among the almost 800
firms he studied was negatively related to the standard deviation and posi-
tively related to the skewness of this sample, just the opposite of what he
had hypothesized. The coefficient of determination was small in this case,
in contrast to earlier studies.
Studies focusing on financial risk as measured by the debt to equity
ratio have found a negative association between this kind of risk and firm
profitability. Arditti (1967), Hall and Weiss (1967), and Gale (1972) fall
into this category. Baker (1973) obtained similar results using a single
equation, ordinary least squares model. However, when Baker used a two
equation, two stage least squares model, the relationship between debt
leverage and rate of return became positive, as classical theory suggests it
should. One could feel more confident in Baker's resolution of apparently
contradictory findings if it had been replicated.

HYPOTHESES AND METHOD

The research aims to provide a balanced test of the power of variation in

firm corporate-level strategy and in firm business-level strategy in explain-
ing variation in firm profitability. The correlational research design used
involves testing the statistical significance and explanatory power of a
linear regression model.
As mentioned in the introduction, the research is limited to single-
industry firms. Thus, first the model will be specified as it was actually
tested, i.e., the single-industry version of the model shown in equation (1).
A brief discussion will show how the model can be generalized to include

This content downloaded from

45.77.104.39 on Thu, 23 May 2024 13:51:30 +00:00
All use subject to https://about.jstor.org/terms
 1981 Beard and Dess 675

multi-industry firms. Such generalization is much easier in theory than in

practice. Variation in corporate-level strategy has been measured in terms
of the average profitability of the industry in which a firm does business.
Variation in business-level strategy has been measured in terms of the
firm's relative position within its particular industry on the three business-
level strategic variables discussed above: sales size, capital intensiveness,
and debt leverage.

The Model

The hypotheses tested are specified in terms of an additive linear regres-

sion model:

Yi = bo + b1X1j - b2X2i- b3X3i+ b4X4i + U (1)

where:
Yi= the before tax return on total investment or on equity of the ith
firm,
Xli = the before task return on total investment or on equity of the in-
dustry in which the ith firm competes,
X2j = the debt to equity ratio computed as the ith firm's ratio relative to
the average ratio of the industry in which the ith firm competes,
X3j = the assets to sales ratio computed as the ith firm's ratio relative to
the average ratio of the industry in which the ith firm competes,
X4j = the sales size of the ith firm relative to the average firm's sales size
in the industry in which the ith firm competes,
U = an error term accounting for unspecified variables,
i = 1 through n, and
n = the number of firms in the sample or population.
The signs of the coefficients in equation (1) indicate the direction of the re-
lationships hypothesized to exist between the independent variables and
the dependent variables.
To generalize the above regression model to include multi-industry
firms one would need to substitute weighted averages on the independent
variables for the single-industry variables shown in equation (1). The
weights required to compute weighted averages on the independent varia-
bles could be specified as the proportions, Pj, of a given multi-industry
firm's assets or other resources assignable to the various industries in
which it does business. Algebraically the weights can be expressed as:

Pj =Aj/A (2)

where:
Pi =the propor
Aj = the absolu

This content downloaded from

45.77.104.39 on Thu, 23 May 2024 13:51:30 +00:00
All use subject to https://about.jstor.org/terms
676 Academy of Management Journal December

A = total firm resources = 2Aj,

j
j = 1 through m, and
m = the number of industries in which investment is possible.
Given the weights, Pj, any business-level strategy variable can then b
expressed as follows:

Xk = 2Pj Xkj, or (3)

j

Xk = PlXkl + P2Xk2 +... + PmXkm (4)

where:
Xk = the kth business-level strategy variable,
k = i through q,
q = the number of business-level strategy variables,
Pj = the proportions used as industry weights as defined in equation (2
above,
j = 1 through m, and
m = the number of industries in which investment is possible.
The relatively large increase in resources required for actual implemen-
tation of the more general model in equations (3) and (4) compared to the
single-industry model in equation (1) was judged not to be worthwhile.
In practice, accurate estimates of the proportion of firm assets allocated
among the industries in which diversified firms did business were found to
be difficult to obtain by survey procedures. This source of measurement
error was judged to be major on the basis of a pilot survey. Therefore, the
research was limited to a sample of listed firms that competed in only a
single industry. The single-industry firms studied here are similar in many
respects to the single-industry subunits of larger firms included in the Har-
vard Business School's PIMS project, as reported in Schoeffler et al.
(1974), Gale (1974) and Buzzell et al. (1975).

The Sample

The population sampled in the present study was the single-industry

manufacturing firms included in Standard and Poors (1979). All firms in-
cluded in the final sample were required to have been in one and the same
industry for the years 1969 through 1974. A firm was considered to be a
single-industry firm if, and only if, during the 1969-1974 period, a sub-
stantial majority, and in most cases all, of its sales could be clearly classi-
fied within one three digit SIC as defined by the U.S. Office of Manage-
ment and Budget (1972).
The process of identifying the single-industry corporations as specified
above was painstaking. Standard and Poors (1979) gives only the primary
enterprise industrial classification of each firm in the Compustat file. Thus

This content downloaded from

45.77.104.39 on Thu, 23 May 2024 13:51:30 +00:00
All use subject to https://about.jstor.org/terms
 1981 Beard and Dess 677

each firm drawn at random from this file was checked manually in Stan-
dard and Poors (1969 through 1974) which gives each four digit SIC in
which a corporation does business. Only about one in six of the Compu-
stat firms were found to be bona fide single-industry firms in terms of the
four digit SIC codes. A final sample of 40 single-industry Compustat firms
was studied.

Data, Measurement, and Analysis

Firm-level data required to compute the appropriate coefficients in

equation (1) above were obtained from Standard and Poors (1979).
Industry-level data required to compute the appropriate coefficients in
equation (1) were obtained from U.S. Internal Revenue Service (1974
through 1979) and Troy (1973 through 1978). These sources of firm-level
and industry-level data, respectively, provide a consistent set of account-
ing classifications across the reporting units and across the six years
studied.
The years 1969 through 1974 were chosen for analysis because this was
the most recent six year period for which the Internal Revenue Service
data were published and because this period included an equal number of
recession years and relatively full employment years. In 1971, 1972, and
1974, U.S. unemployment was between 5.5 and 6.0 percent. In 1969, 1970,
and 1973 unemployment was between 3.5 and 5.0 percent.
Operational measurement of the variables specified in equation (1)
above is straightforward. Two measures of the dependent variable, firm
profitability, were analyzed in parallel fashion. The first, firm return on
equity (ROE), was measured as the ratio of profits before income taxes
and extraordinary items to equity. The second, firm return on total invest-
ment (ROI), was measured as the ratio of profits before income taxes and
extraordinary items plus interest to year-end total investment.
Concerning measurement of the independent variables, the appropriate
industry profitability measure (X1, the corporate-level strategy variable),
either return on equity (ROE) or return on total investment (ROI), was
computed from data in the U.S. Internal Revenue Service (1974 through
1979) exactly as the corresponding firm profitability measure was com-
puted. The remaining three independent variables, all business-level stra-
tegy variables, were measured in the same way regardless of which firm
profitability measure was used as the dependent variable. All three were
measured as firm-characteristics relative to industry norms. The relative
debt leverage measure (X2) used was the ratio of firm total debt to equity
divided by the corresponding average ratio for all firms (corporate tax re-
turns) in the appropriate industry. Similarly, the relative capital intensive-
ness measure (X3) was the firm total assets to total sales ratio divided by
the average ratio for all firms in the appropriate industry, and the relative
sales size measure (X4) was firm sales divided by the average sales of all
firms in the appropriate industry.

This content downloaded from

45.77.104.39 on Thu, 23 May 2024 13:51:30 +00:00
All use subject to https://about.jstor.org/terms
678 Academy of Management Journai December

The statistical analysis and hypothesis testing were done using a stepwise
linear regression procedure. Independent variables were entered at each
step in the order of their squared partial correlation with the dependent
variable when all other independent variables were controlled. The mini-
mum level of acceptable statistical significance for the regression equa-
tions tested was p < .05. The model was tested for each of the six years in-
cluded in the study as well as for the six year average, with respect to both
of the dependent variables, return on equity and return on investment.

Reliability and Validity Issues

The pooling of firms in a sample, where these firms are in some impor-
tant respects heterogeneous, has been accorded increasing critical atten-
tion recently. Hatten and Schendel (1977) and Bass et al. (1978) are the
most germane to our study. The basic point of both studies is that, if re-
gression coefficients of subgroups of firms within a population or sample
differ significantly from those of the population or sample as a whole, the
reliability of the latter coefficients is subject to question.
These studies make a useful methodological point and also serve to re-
mind one of the complexities in the areas of organization-environment re-
lationships and business policy. Nonetheless, it is believed that this meth-
odological point leaves the reliability of representative samples such as the
present one at a viable level. The above two critical studies additionally
raise the important judgmental issue of what populations are most rele-
vant in studying firm performance. This question clearly has many accept-
able answers depending on the purposes and interests of the researchers.
Two important trade-offs are seen between the external validity of a
sample of firms and the kind of internal homogeneity that Hatten and
Schendel (1977) and Bass et al. (1978) have persuasively raised as a reliabil-
ity criterion. First, as such homogeneity is sought, the size of the popula-
tion of firms that the sample represents, and thus the generality of the re-
sults, diminishes. Second, the parsimony of the empirically supportable
theory resulting is diminished, a cost of contingency or situational theories
in general. Both articles discuss other, more statistically technical advan-
tages and disadvantages of pursuing internal homogeneity in samples of
firms. The present authors have opted to weigh the above kinds of exter-
nal validity more heavily than the added reliability that internal homo-
geneity admittedly provides.
The present research design is believed to have a number of strengths.
The sample is drawn from a large population of listed manufacturing
firms. The single-industry firms are relatively homogeneous with respect
to industrial diversity. Schoeffler et al.'s (1974) experience with single-
industry subunits of large firms suggests that larger, diversified firms can
usefully be represented as aggregates of units similar to the present ones.
Thus the generality of current results is relatively wide.

This content downloaded from

45.77.104.39 on Thu, 23 May 2024 13:51:30 +00:00
All use subject to https://about.jstor.org/terms
 1981 Beard and Dess 679

In using only single-industry firms, in using industry profitability to

control for interindustry variation, and in measuring firm variables exclu-
sively in relation to industry norms, it is believed that the present research
has done much to distinguish clearly between interindustry and intra-
industry (i.e., firm) sources of variation, a significant confounding prob-
lem in most earlier studies of firm performance. The sample of 40 single-
industry firms represents 38 separate industries as well. Because of this
interindustry heterogeneity, the sample provides an adequate range for
variation at this level of analysis.
Six years of data that are about equally split in terms of high and low
points on the business cycle have been used, and reasonably consistent re-
sults for each year separately and in the aggregate have been developed.
The results also are developed for two different performance measures.
The replicability of the design leaves its reliability and validity open to rel-
atively easy future testing.

RESULTS

Results of the stepwise regression analysis are summarized in Tables 1

and 2. The former gives results when firm return on equity (ROE) is the
dependent variable, and the latter gives parallel results when firm return
on total investment (ROI) is the dependent variable.
The results shown in Tables 1 and 2 indicate that both corporate-level
strategy and business-level strategy, as defined and measured here, are im-
portant in explaining variations in firm profitability. With respect to the
measure of corporate-level strategy (X1, industry profitability) the sign is
positive in all equations in Tables 1 and 2. In a large majority of equa-
tions, XI is either first or second in explanatory power as indicated
either the standardized regression coefficients or the stepwise change in
multiple R2.
With respect to the three measures of business-level strategy (relative
size, debt leverage, and capital intensiveness), one encounters both confir-
mation of some relationships symbolized in equation (1) and some unex-
pected results. The explanatory power of firm relative debt to equity, X2,
is surprising. X2 has a negative sign in virtually all equations, as hypothe-
sized in equation (1). It is also first or second in explanatory power in the
large majority of equations. The sign of firm relative capital intensiveness,
X3, is negative as hypothesized in equation (1). Overall, X3 rivals X2 in ex-
planatory power, but it is not as consistently high in this respect. The al-
most universally low explanatory power of firm relative sales size is a
major surprise. Because the regression coefficients for X4 are so close to
zero, their sign is of no interest.
Turning now to the overall magnitude and statistical significance of the
regression results, one can see in Table 1, where return on equity (ROE) is
the dependent variable, that the multiple Rs and associated R2s exceed the
p < .05 criteria on one or more steps in four of the seven equations. Results
for two other equations, including the equation for the six year averages,

This content downloaded from

45.77.104.39 on Thu, 23 May 2024 13:51:30 +00:00
All use subject to https://about.jstor.org/terms
680 Academy of Management Journal December

1972
197 1970 196

TABLE1

StepwisRgronAalysifFrmRetunoEqiy(RO),196-74

CoeficntsFalSepndtiscalSgnf e

b4,FirmRelatvSsizeX4-.0 742.19806 b2,FirmRelatvDboEquityX2-.065184.072531 b3,FirmRelatvCpiInesv X3-.09263.7805143 b1,IndustryRe onEquityX1.835924 .0685 bo,Cnsta-.25103 b3,FirmRelatvCpiInesv X3.1258- .30478913 b4,FirmRelatvSsizeX4.01 283.60 12 b1,IndustryRe onEquityX1.8253 60.275 31 b2,FirmRelatvDboEquityX2-.130 9.2306 boCnsta-.270 - b4,FirmRelatvSsizeX4-.0 73564.081 5 b2,FirmRelatvDboEquityX2-.0614 59.0316 b3,FirmRelatvCpInsieX3-.19076253.07 b1,IndustryRe onEquityX1.5042871.0423 bo,Cnsta-.189 - b4,FirmRelatvSsizX4.0 254.0172 b2,FirmRelatvDboEquityX2-.0351 2.0973 b3,FirmRelatvCpiInesvX3-.1507294.3609 bl,IndustryRe onEquityXI1.3957401.872 bo,Cnsta-.21 CoeficntsadVribleXCofcintsEroCeficntsR 2 RegrsionEquatSepwisMultCoreainfcets

RegrsionEquatRegrsionStadRegrsionMultpef ilMutpe il UnstadrizeSnad tepwisSgnfacetpwisChange

This content downloaded from

45.77.104.39 on Thu, 23 May 2024 13:51:30 +00:00
All use subject to https://about.jstor.org/terms
1981 Beard and Dess 681

1974 1973

6-YearAvg,1974

TABLE1(cont.)

CoeficntsFalSpditcgnfae

b3,FirmRelatvCpInsieX3.0 14965.0 bI,ndustryReoEqiXl.4083126795.0 b2,FirmRelatvDoEquyX2-.073491685 bo,Cnsta-.0964 b3,FirmRelatvCpInsiX3-.076815943.0 b4,FirmRelatvSszX.0167593482 bl,IndustryReoEqiXI1.2603875 .9 b2,FirmRelatvDoEquyX2-.30854912 boCnsta-.16249 b3,FirmRelatvCpInsX3.0 574218.03 b2,FirmRelatvDoEquyX2.0375-1648 2 b4,FirmRelatvSszX4-.06918 35.09 bl,IndustryReoEqiX1.486039521 bo,Cnsta-.145086 CoeficntsadVrblXCoeficntsE R2
RegrsionEquat SdrRegsionMultpf kile
RegrsionEquatSpweMliCoratnfces

UnstadrizeS pwsgnifcaeStCh

This content downloaded from

45.77.104.39 on Thu, 23 May 2024 13:51:30 +00:00
All use subject to https://about.jstor.org/terms
682 Academy of Management Journal December

1972 197 196

1970

TABLE2

StepwisRgronAalyfFmetuToInvs(RO),196-74

bo,Cnsta.25067-
coientsadSl

b4,FirmelatvszA.0239741 b1,IndustryeoqiXA.5273490168 b3,Firmelatvcpns2-.0793486 b2,FirmelatvdoquyX-.0813562 b4,Firmelatvsz.09736214 b1,IndustryeoqiX.67820135 b3Firmelatvcpns-.09841327 b2,FirmelatvdoquyA-.0975328 bo,Cnsta-.25014 b4,FirmelatvszT.05263817 b3,FirmelatvcpnsX3-.10482956 b2,FirmelatvdoquyX-.06483219 b1,IndustryeoqiX.5049321 bo,Cnsta-.163059 b4,FirmelatvszX.02531786 bl,IndustryeoivmX.980542376 b2,FirmelatvdoquyXT-.06431279 b3,FirmelatvcpnsX-.14063278 bo,Cnsta-.234068 CoeficntsadVrblX EoCeficntsR2 RegrsionEquatSpwMlCeficns

RegrsionEquat SdRegrsionMultpf ie UnstadrizeS pwgfcntiseCha

This content downloaded from

45.77.104.39 on Thu, 23 May 2024 13:51:30 +00:00
All use subject to https://about.jstor.org/terms
1981 Beard and Dess 683

1974 1973

6-YearAvg,1974

TABLE2(cont.)

CoeficntsFalSpdg

b4,FirmelatvszX.0561723 b3,FirmelatvcpnsX-.04315682 bl,IndustryeoqiX1.673584029 b2,FirmelatvdoquyX-.10374985 bo,Cnsta-.195068 b4,FirmelatvszX.081627 b3,FirmelatvcpnsX-.0234951 bi,IndustryeoqX1.57634902 b2,FirmelatvdoquyX-.104536879 bo,Cnsta-.17408 b3,FirmelatvcpnsX-.027459318 b2,FirmelatvdoquyX-.06345712 b1,IndustryeoqiX.25630491 bo,Cnsta-.10372 CoeficntsadVrblXER2 RegrsionEquatSpwMlCfc

RegrsionEquat Sd Mlipeofut UnstadrizeS pwgfcsChanei

This content downloaded from

45.77.104.39 on Thu, 23 May 2024 13:51:30 +00:00
All use subject to https://about.jstor.org/terms
684 Academy of Management Journal December

approach this level of significance, and the one remaining equation is

clearly not significant.
The magnitude and significance of the results shown in Table 2, where
return on total investment (ROI) is the dependent variable, are moderately
stronger than those for ROE. In Table 2, the multiple Rs and associated
R2s exceed the p < .05 criteria on one or more steps in five of the seven
equations, including the equation for the six year averages. Of the remain-
ing two equations, both approach significance on one step, but the latter
two also lag the other five equations in the magnitude and significance of
their multiple Rs. The appropriate industry profitability variable (X1) is
stronger in explaining variance in ROE than it is in explaining variance in
ROI.
The instability from year to year in the top three independent variables'
relative explanatory power is noteworthy. This is, in part, a result of the
statistical criterion used to sequence variables for stepwise entry into the
regression equations. A pattern with respect to the general business cycle is
also apparent, however. In years in which U.S. unemployment, a major
coincident business cycle indicator, was below 5 percent (1969, 1970, and
1973), the appropriate industry profitability measure has the greatest ex-
planatory power. In years in which U.S. unemployment exceeded 5 per-
cent (1971, 1972, and 1974), relative firm debt to equity and relative firm
capital intensiveness have the greatest explanatory power.

DISCUSSION AND CONCLUSIONS

On the question of the importance of corporate-level strategy and

business-level strategy in explaining firm profitability, the results indicate
that both are important. In 10 of the 14 regression equations described in
Tables 1 and 2, the appropriate industry return variable and either firm
relative leverage or firm relative capital intensiveness (and most often both
of the latter two) contribute appreciably to statistically significant multiple
correlation coefficients.
As to which variable, corporate-level or business-level, is the more im-
portant in explaining firm profitability, one must exercise caution. If one
looks at the two six year average equations, the standardized regression
coefficients in both equations show consistent rankings of the independent
variables. Firm relative debt to equity is the most important, and this vari-
able combined with firm relative capital intensiveness exceeds the appro-
priate industry profitability index in explanatory power.
In Table 2, where firm ROI is the dependent variable, the single-year
equations show that the two major business-level variables combined gen-
erally explain more variance in performance than does the industry return
variable. However, this is not so clearly the case in Table 1, where firm
ROE is the dependent variable. In three of the six years shown in Table 1,
the industry ROE variable exceeds the two major business-level strategy
variables in explanatory power. All three such years are relatively low

This content downloaded from

45.77.104.39 on Thu, 23 May 2024 13:51:30 +00:00
All use subject to https://about.jstor.org/terms
 1981 Beard and Dess 685

unemployment years, suggesting that the general business cycle affects the
relative importance of the independent variables.
The universal absence of any relationship between X4, firm relative sales
size, and either measure of profitability is an unanticipated result. How-
ever, this result fits well with Gale's (1972) discussion of the relationship
between firm relative size and firm profitability, and it is consistent with
his finding of a strong, positive interaction between firm market share and
industry concentration. When concentration was high in Gale's study,
market share was strongly correlated with profit. When concentration was
low, market share was not strongly correlated with profit. These results
are consistent with the theoretical view that collective monopoly power
among major competitors is a major source of the profits associated with
market share.
The present study's sample is certain to have smaller firms on the aver-
age than those based upon the Fortune 500 largest industrial firms such as
Shepherd's (1972) study. Size data on the PIMS project participants sug-
gest that their single-business subunits are also likely to be among the top
oligopolists in the industries in which they compete. Although the present
sample is drawn from the Compustat Annual Industrial Tapes, like Gale's
(1972) and Winn's (1975) samples, the exclusion of multi-industry firms in
the current study no doubt would result in the sample's containing smaller
firms than theirs, on the average.
The great majority of the sample of single-industry firms were profit-
able during the six years studied. It thus appears that these firms generally
had reasonably successful strategies. It appears likely, however, that they
would compete on a more selective basis than do leading oligopolists.
They thus would seem likely to specialize in serving particular market seg-
ments, producing only selected products, or serving restricted geographic
areas. Hamermesh, Anderson, and Harris (1978) suggest that firms with
small market shares must follow these kinds of specialization in order to
succeed. If one could define the relevant competitive environments for
these firms more precisely than the industry data allow, their relative size
then would be likely to appear more important to their success.
On the question of how much of the variance in firm profit perfor-
mance can be explained by the independent variables taken together, the
results are encouraging but not entirely persuasive. In the 8 equations, out
of the total of 14, in which statistical significance is high, more than a
quarter of the variance in the dependent variable is explained. The next
three most significant equations, which exceed or approximate the p < .05
significance level, explain between 15 and 20 percent of the variance in the
dependent variable. Because of the aggregated nature of the three digit
SICs used to provide industry data, one would expect higher coefficients
of determination if more precise industry classifications could be
employed.
The multiple coefficients of determination in Tables 1 and 2 are gen-
erally stronger in years of relatively low unemployment (1969, 1970, and

This content downloaded from

45.77.104.39 on Thu, 23 May 2024 13:51:30 +00:00
All use subject to https://about.jstor.org/terms
686 Academy of Management Journal December

1973), although 1974, a high unemployment year, shows stronger results

than 1973. The effects of the general business cycle on the regression re-
sults can be rationalized as follows. In years of rising employment and
GNP, firms in any given industry are relatively equal in their ability to
command increases in financial, human, and material resources needed to
expand output. Under these conditions, the distribution of firm profitabil-
ity around the industry mean is narrowed, and the mean provides a better
approximation of each firm's profitability. In years of increasing unem-
ployment and falling GNP, firms with relatively high fixed costs asso-
ciated with higher debt and capital intensiveness suffer disproportionate
drops in profitability compared to competitors with lower fixed costs.
Under these conditions, the distribution of firm profit around the industry
mean is wider, and the mean provides a less satisfactory approximation of
each firm's profitability.
At any rate, it appears that business cycle effects are not transmitted
proportionately among firms in a given industry. It also seems likely that
controllable strategic variables, such as capital structure and capital inten-
siveness, account for wide variance in the business cycle's effects on the in-
dividual firms in a particular industry.
The present study's results support several tentative conclusions. First,
variation in a firm's corporate-level strategy and in its business-level stra-
tegy both help to explain variation in firm profitability. Second, the rela-
tive importance of variation in corporate-level compared to business-level
strategy in explaining firm profitability remains somewhat ambiguous on
the basis of present results. On the face of it, the relative debt leverage and
relative capital intensiveness dimensions of business-level strategy appear
stronger than industry return. However, the latter is measured more
crudely than the former two variables. Thus, a more discriminating mea-
surement of industry-level variation might possibly tip the balance of ex-
planatory power in favor of corporate-level strategy.
Third, relative firm size within a given industry does not hold up here as
a powerful predictor of firm profitability. Differing populations of firms
studied seem likely to account for the difference between the research re-
sults and the results of several studies discussed above. In competition
among the few, the relative size proposition looks valid. In competition
among the many, it does not.
Fourth, the average level of the multiple correlation coefficients and the
statistical significance of the regression equations suggest that under study
are variables important to understanding and predicting firm profitability.
Nevertheless, these results suggest that room remains both for better mea-
surement of our variables and for specification of additional explanatory
variables.
Finally, the variability of the results over time argues for more attention
in future research to sources of temporal variation. The effect of some
strategic variables on a firm's profitability appears to vary with business

This content downloaded from

45.77.104.39 on Thu, 23 May 2024 13:51:30 +00:00
All use subject to https://about.jstor.org/terms
 1981 Beard and Dess 687

cycle conditions or with other longitudinal changes in the business envi-

ronment. Firm differences in business-level strategy such as firm capital
structure and capital intensiveness appear to account for widely varying
effects of environmental change on individual competitors within a given
industry.

REFERENCES

1. Andrews, K. The concept of corporate strategy. Homewood, Ill.: Dow-Jones, 1971.

2. Ansoff, H. Corporate strategy. New York: McGraw-Hill, 1965.
3. Arditti, F. Risk and the required return on equity. Journal of Finance, 1967, 22, 19-36.
4. Baker, S. Risk, leverage and profitability. Review of Economics and Statistics, 1973, 55, 503-507.
5. Bass, F. Cattin, P., & Whittink, K. Firm effects and industry effects in the analysis of market
structure and profitability. Journal of Marketing Research, 1978, 15, 3-10.
6. Beard, D., & Dess, G. Industry profitability and firm performance: A preliminary analysis of the
business portfolio question. Academy of Management Proceedings, 1979, 123-127.
7. Berry C. Corporate growth and diversification. Princeton, N.J.: Princeton University Press,
1975.
8. The Boston Consulting Group, Inc. Perspectives on experience. Boston: The Boston Consulting
Group, 1972.
9. Bower, J. The resource allocation process. Homewood, Ill.: Richard D. Irwin, 1970.
10. Buzzell, R., Gale, B., & Sultan, R. Market share-a key to profitability. Harvard Business
Review, 1975, 53 (1), 97-106.
11. Carter, E. Behavioral theory of the firm and top level corporate decisions. Administrative Sci-
ence Quarterly, 1971, 16, 413-429.
12. Caves, R. American industry: Structure, conduct, performance. 4th ed. Englewood Cliffs, N.J.:
Prentice-Hall, 1977.
13. Chenery, H., & Clark, P. Interindustry economics. New York: John Wiley & Sons, 1959.
14. Child, J. Organization structure, environment and performance: The role of strategic choice. So-
ciology, 1972, 6, 1-22.
15. Christensen, C., Andrews, K., & Bower, J. Business policy: Text and cases. 4th ed. Homewood,
Ill.: Irwin, 1978.
16. Conrad, G., & Plotkin, I. Risk/return: U.S. industry pattern. Harvard Business Review, 1968, 46
(2), 90-99.
17. Cyert, R., & March, J. A behavioral theory of thefirm. Englewood Cliffs, N.J.: Prentice-Hall,
1963.
18. Fisher, I., & Hall, G. Risk and corporate rates of return. Quarterly Journal of Economics, 1969,
83, 79-92.
19. Gale, B. Market share and rate of return. The Review of Economics and Statistics, 1972, 54,
412-423.
20. Gale, B. Impact of strategic position relative to rivals on business profitability. Unpublished
working paper, Harvard Graduate School of Business Administration, Boston, 1974.
21. Gort, M. Diversification and integration in American industry. Princeton, N.J.: Princeton Uni-
versity Press, 1962.
22. Hall, M., & Weiss, L. Firm size and profitability. The Review of Economics and Statistics, 1967,
69, 319-331.
23. Hamermesh, R., Anderson, M., & Harris, J. Strategies for low market share businesses. Harvard
Business Review, 1978, 56 (3), 95-102.
24. Hatten, K., & Schendel, D. Heterogeneity within an industry: Firm conduct in the U.S. brewing
industry, 1952-71. The Journal of Industrial Economics, 1977, 26, 97-113.
25. Hofer, C., & Schendel, D. Strategy formulation: Analytical concepts. St. Paul, Minn.: West
Publishing Co., 1978.
26. Leontief, W. Studies in the structure of the American economy. New York: Oxford University
Press, 1953.
27. Levitt, T. Marketing myopia. Harvard Business Review, 1960, 38 (4), 45-56.
28. Lieberson, S., & O'Connor, J. Leadership and organizational performance: A study of large cor-
porations. American Sociological Review, 1972, 37, 117-130.
29. Miernyk, W. The elements of input-output analysis. New York: Random House, 1965.

This content downloaded from

45.77.104.39 on Thu, 23 May 2024 13:51:30 +00:00
All use subject to https://about.jstor.org/terms
688 Academy of Management Journal December

30. Miles, R., Snow, C., Meyer, A., & Colemen, H. Organizational strategy, structure and process.
Academy of Management Review, 1978, 3, 546-562.
31. Mintzberg, H., Raisinghani, D., & Theoret, A. The structure of "unstructured" decision pro-
cesses. Administrative Science Quarterly, 1974, 19, 135-151.
32. Pfeffer, J., & Salancik, G. Organizational decision-making as a political process: The case of a
university budget. Administrative Science Quarterly, 1974, 19, 135-151.
33. Pitts, R. Strategies and structures for diversification. Academy of Management Journal, 1977,
197-206.
34. Rhoades, S. The effect of diversification on industry profit performance in 241 manufacturing
industries: 1963. The Review of Economics and Statistics, 1973, 55, 146-155.
35. Rhoades, S. A further evaluation of the effect of diversification on industry profit performance.
The Review of Economics and Statistics, 1974, 56, 557-559.
36. Rumelt, R. Strategy, structure and economic performance in large industrial corporations. Bos-
ton: Harvard University Press, 1974.
37. Rumelt, R. Diversity and profitability. Unpublished paper presented at the Annual Meeting of
the Western Division, Academy of Management, Sun Valley, Idaho, 1977.
38. Schendel, D., & Patton, G. A simultaneous equation model of corporate strategy. Management
Science, 1978, 24, 1611-1621.
39. Scherer, R. Industrial market structure and economic performance. Chicago: Rand-McNally and
Co., 1970.
40. Schoeffler, S., Buzzell, R., & Heany, D. The impact of strategic planning on profit performance.
Harvard Business Review, 1974, 52, 137-145.
41. Shepherd, W. The elements of market structure. The Review of Economics and Statistics, 1972,
54, 25-37.
42. Springer, C., & Hofer, C. General Electric's evolving management systems. In C. Hofer, E.
Murray, R. Charan, & R. Pitts (Eds.), Strategic management: A casebook in business policy and
planning. St. Paul, Minn.: West Publishing Co., 1980, 454-470.
43. Standard and Poor's Corporation. Standard and Poor's register of corporations, directors and
executives: United States and Canada. 1969 ed. through 1974 ed. New York: Standard and
Poor's Corporation, 1969 through 1974.
44. Standard and Poor's Corporation. Standard and Poor's compustat, annual industrial tapes. New
York: Standard and Poor's Corp., 1979.
45. Troy, L. Almanac of business and industrial financial ratios. 1973 ed. through 1978 ed. Engle-
wood Cliffs, N.J.: Prentice-Hall, 1973 through 1978.
46. U.S. Internal Revenue Service. Statistics of income, corporate income tax returns. 1969 ed.
through 1974 ed. Washington, D.C.: U.S. Government Printing Office, 1974 through 1979.
47. U.S. Office of Management and Budget. Standard industrial classification manual. 1972 ed.
Washington, D.C.: U.S. Government Printing Office, 1972.
48. Uyterhoeven, E., Ackerman, R., & Rosenblum, J. Strategy and organization: Text and cases in
general management (Rev. ed.). Homewood, Ill.: Irwin, 1977.
49. Vancil, R., & Lorange, P. Strategic planning in diversified companies. Harvard Business Review,
1975, 53 (1), 81-90.
50. Weick, K. The social psychology of organizing. Reading, Mass.: Addison-Wesley, 1969.
51. Weiss, L. The concentration-profits relationship and antitrust. In H. Goldschmid, H. Mann, &
R. Weston (Eds.), Industrial concentration: The new learning. Boston: Little, Brown, 1974,
184-245.
52. Winn, D. Industrial market share and performance. Ann Arbor, MI: University of Michigan,
1975.

This content downloaded from

45.77.104.39 on Thu, 23 May 2024 13:51:30 +00:00
All use subject to https://about.jstor.org/terms