Download BCG Portfolio Model: Maximizing Profitability through Business Unit Strategies and more Exercises Construction in PDF only on Docsity!
WORKING
PAPERS
AN EMPIRICAL ANALYSIS OF THE BOSTON CONSULTING GROUP'S PORTFOLIO MODEL
Malcolm B. Coate
WORKING PAPER NO. 71
August 1982
FfC Bureau of Ecooomics working papers are preliminary materials circulated to stimulate discussion and critical comment All data contained in them are in the public domain. This includes information obtained by the Commissioo which has become part of public record. The analyses and conclusions of the authors and do not necessarily reflect the views of other members of the Bureau of Economics, other Commission staff, or the Commission itself. set forth are those Upon request, single copies or the paper will be provided. References in publications to FfC Bureau of Economics working papers by FfC economists (other than acknowledgement by a writer that he has access to such unpublished materials) should be cleared with the author to protect the tentative character of these papers.
BUREAU OF ECONOMICS
FEDERAL TRADE COMMISSION
WASHINGTON, DC 20580
AN EMPIRI CAL ANALYSIS O F T HE
B OS T ON C ONSUL TING GR OUP'S P OR T F OLI O M ODEL
Malcolm B. Coate June 30, 198 2
the^ The^ analysesauthor and^ and do^ conclusions not necessarily^ set^ forthreflect^ in^ thisthe viewspaper ofare other^ those^ of members of the Bureau of Economics, other Co m mission staff, or the Com mission.
T HE BOS T ON C ONSUL TING GR OUP'S P OR T F OLI O M ODEL
The B CG literature states, "the real measure of management's success is the increase in the present value of future cash payoff" [4]. This implies that the firm should follow a discounted-cash-flow maximization rule. Thus, the goal of the B CG m odel is to maximize the long-run profitability (present value) of the fir m's business units. The investment strategies of each unit are the basic control variables used to optimize the return on the portfolio of business units. These strategies are based on t he position of the various units in the portfolio matrix. There are four fundamental concepts that form the foundation of the B CG matrix approach to planning.l^ They are: (1) the separability of the firm into independent business units, (2) the limitation on corporate resources, (3) the existence of the experience curve, and (4) the importance of the industry growth rate [7). These ideas are used to structure a planning matrix and define optimal strategies. The business-unit concept implies that the firm C il define strategies for each unit without having to consider their effect on the other business units. This allows the fir m to subdivide itself into a meaningful portfolio of businesses. The actual definition of each unit should attempt to capture all the possible synergies while still maintaining a significant subdivision of the firm into component parts.
(^1) The B CG also assumes that price will decline with costs in the long run and seems to ignore the risk of a strategy [71 •
The li mitation on corporate resources requires the firm to m ake all its strategic decisions simultaneously, because invest m ent in one unit reduces the funds available to the other units of the firm. The B CG expects the firm to try to build a balanced p ortfolio of business units and use cash from its mature busi nesses to fund the investment in its growth business. This c oncept implies that the fir m must pass over some attractive investment projects because the necessary investment resources are n ot available. The experience curve implies a negative relationship exists between costs and cumulative output. This sug gests that relative m arket share will have a positive effect on profitability, b ecause the business with the largest equilibrium share (and thus the most cumulative output} must have the largest margin. (^2) The large margin will allow the high-share units to generate cash that can be reinvested in the business or transferred to other units. This relationship between share and profitability is the most i mportant concept in the portfolio model because it serves to identify the most profitable type of business unit. The B CG portfolio model also defines a relationship between t he industry growth rate and the investment in each unit. The B CG observes the firm must invest heavily in high-growth industries to
(^2) Other explanations can be offered for the relative-market share/profitability valid even if the experience relationship, curve sodoes the notportfolio exist [16]model • can be
3 ü
are used to divide the business units the fir m allocate the
into different cells. The B CG model sug gests that the cash generated by each business, to maximize long-run profitability of its portfolio subject to the required balance in cash generated and used. The desired with
movement in the portfolio matrix is illustrated in figure 1, the B CG's title for each type of unit in the appropriate box. Hedley notes "the first goal should be to m aintain position in the 'c ash cows' but to guard against the frequent temptation to reinvest in them excessively" [1 5, p. 11]. Next, the firm should invest to preserve the market.share of the " stars." Any surplus funds are invested in the best "question marks" to acquire additional market share. The firm invests in "stars" and selected "question marks" in the hope that they will beco me "cash cows" when their market growth rate decreases. The " do g " and the re mainder of the "question mark" businesses are managed to generate·cash or are divested fro m the portfolio, because the B CG model sug gests further investment will never yield a future cash return [3]. Thus, the B CG model advises the fir m to use the excess funds fro m the "cash cow " units to finance invest ment in high-growth units, with the goal of creating new "cash cows" when the industry growth rate slows. This "ability of the diversified company to redirect its cash flow internally is
to
extremely imp ortant" [4]. rt· allows the corporation to continue grow and earn profits as its individual business units move through the stages of the product 1ife cycle.
L' .....
v 7 ......
Figure l
'Ihe BCG Portfolio Matrix
HIGH
MarketGrowth (^) lO%
HIG!
STAR QUESTICN MARK
CASH CC!fl rxx;
Relative Daninance
unit is independent of the rest of fir m. This hypothesis contra d icts the relatedness theory of diversification, a theory that advises a firm to diversify into areas similar to its main lines o f business. Rumelt found that firms with related diversification strategies were significantly more profitable than the average f ir m [19]. This suggests that the firm can take advantage of so me operational synergies between individual business units. Carter used the Herfindahl numbers-equivalent index for firms with a centralized organizational structure as a proxy for relatedness [ 6]. He found that these firms are slightly more profitable than analogous fir ms with a multidivisional (n oncentralized) structure, b ut the difference was not significant. Thus, the existing weak evidence suggests that the independent-business-unit hypothesis s hould be rejected. But it may be possible to construct another test of the hypothesis. A quantitative measure of the operational synergy between business units requires so me definition of relatedness. The Standard Industûial Classification (S I C ) code can be used to de fine business units in the same two-digit S I C industry as related. Sc herer [20, p. 60] notes that the S I C-code system emp hasizes similarities in the production process, so most of the potential for manufacturing synergy should be captured. But the measure will not necessarily incorporate marketing relationships. The actual variable must also give so me consideration to the share of a fir m's sales in a S I C group, to proxy the potential size of the
synergy. The approach, used to measure synergy in a group, c o mp utes the product of the num ber of possible synergistic business units in a group and the square of the percentage share of a firm's sales in an industry classification. This variable gives more weight to industry groups where the related opp ortunities or the level of the firm's sales offers a chance for significant synergies. 20 REL = (^) i=lE (n (^) i -^1 ) (Si ) 2 where ni = the nu mber of the fir m's business units in the i'th two-digit S I C indu stry si This variable is equal to zero if the fir m does not operate more than one business in each S I C industry and is large for a diversified fir m that operates a number of business units in a single S I C industry group. The B CG hypothesis of minimal opera tional synergy between units cannot be rejected if the variable has an insignificant coefficient in the profitability model. The Limitation on Investment Funds The limitation on investment funds implies that internal financing is valuable to the firm, since it allows the fir m to generate additional funds not available in the capital market. Thus, internal financing may make it easier for a fir m to under take some new profitable projects that require large capital
= the total sales of the fir m's business units in the i'th industry group, divided by total fir m sales.
Experience
The diversification variable can range from zero (for a fir m in a single market) to approximately one (for a well-diversified fir m ). The financial-synergy hypothesis implies the diversification variable should have a positive sign in the profitability model. The Curve The experience curve implies that the profitability of a business is proportional to its relative share of the market. This relative-market-share/profitability hypothesis has some initial support from a few B CG fir m and industry case studies [ 9]. Also, an aggregated for m of the relationship has substantial econo metric support. (^) Relative market share had a significant positive effect on profitability in a Federal Tr ade Co m mission (F T C ) study [24] and in a later study by !mel and Helmberger [ 17] .s Both papers controlled for industry concentration but only had data for fir ms in agricultural processing industries. A num ber of other studies have found that absolute market share is a significant deter minant of profitability at either the fir m [11, 22] or business-unit level [18]. This relationship could be caused by the omission of relative market share fro m the model. Finally, various P IMS studies found that relative and absolute market share were related to profitability [12, 21]. The existing studies support offer some support for the B CG úelationship between relative market share and profitability. But
concentration^ Relative^ share ratio.^ was^ defined^ by^ dividing^ market^ share^ by^ the
the industrial-organization literature either uses a narrow set of industries [17, 24] or absolute instead of relative market share [11 , 18 , 22]. Also, most of the studies use fir m data instead of business-unit data. The PIMS stud y uses a sample of 1, 000 b usiness units but suffers fro m a number of econometric diffi c ulties [20 , 21]. Thus, another study would be useful to add to the piecemeal support of the relative -mark et-share/profitability h ypotpesis. The relative market share of a business is defined as the ratio of the unit's share to the industry concentration level. 6 Then the overall fir m share measure is the weigh ted average of each unit's share. It can be calcul ated as RMS = (^) jr Sj RMSj
where Sj = the share of the fir m' s sales in the j'th four-digit SI C industry RMS·J = the relative market share of the fir m in its j'th business. This variable will be relatively large for fir ms that tend to do minate their industries and small for firms that hold a marginal p osition in each mark et. The share variable should have a
(^6) ' The B CG measure of relative share (the share of the firm dividedbecause byit thewould share not ofapproxi its largestmate the co relative mp etitor) cost is notposition used, of the leading firm in a regression model.
units of each fir m. Fir m size is included to investigate the residual effect of absolute size on profitability. It is measured by the inverse of the logarithm of assets (net of investments in unconsolidated subsidiaries) [13] • Shep herd [22] n otes that the effect of size on profitability is indeterminate because large size may be associated with higher costs, in addition to higher revenues. The advertising-to-sales ratio and t he research-and-development-to-sales ratio are included to test the effect of these variables on profitability. Sc herer [20, p. 38 8] observes that an oligopolistic market structure may lead to either investment in product-differentiation advertising and increased profits or overinvestment in advertising and reduced profits. An analogous argument could be made for the research variable. Thus, the sign of these variables is theoretically indeter minate. Co manor and Wilson [ 8] have found a positive advertising effect, and Scherer notes that this finding "has been replicated using diverse profitability measures and fir m or industry samples" [20, p. 286]. Thus, a positive sign is expected for advertising. !mel and Hei mberger [17] reported that research and development intensity had a significant positive effect on pro fitability. Therefore, a positive sign should also be expected for the research variable. The capital-to-sales ratio is includ ed in the model, to account for interindustry variations in capital intensity, when a return-on-sales variable is used to measure pro fitability. The ratio should have a positive sign, since the firm's return on sales does not consider the capital stock. A
few industry variables are also included in the study. Concentration should increase the profitability of every fir m in t he industry, if it acts as a proxy for shared market power. Therefore, a positive sign is expected. The percentage increase in industry sales is included to allow growth to affect industry profits. The sign of the growth variable is indeter minate. If growth is a proxy for undercapacity, high growth should be linked to high profitability. But the portfolio theory sug gests that the i nvestment necessary in high-growth industries could reduce the m easured short-r un profitability of these businesses. Thus, a negative relationship could be found. In conclusion, the basic B CG hypotheses will be supported if we find the following:
- The relatedness variable is insignificant;
- The diversification variable is significant;
- The relative-market-share variable is significant.
E S TIM A TI ON O F T HE M ODEL The regression model requires a complicated data set, to test the three hypotheses. The construction of the file required merging the 1978 Economic Information System (E IS) data, 1976- Comp ustat financial infor mation, and 1977 census data on four digit SI C industries. Each data set contributed to the calcula tion of the variables used in the regression model. The EIS tape de fined the firm's market share and the percentage of its sales in a given industry. The Comp ustat file provided profit, sales,
Two profitability measures--the return on sales and the return on equity--were used as dependent variables.8 Since the dependent variable is defined at the fir m level, a weighting scheme is required to calculate the business-unit-based explana tory variables. Carter notes that the weights should be consistent with the profit measure used in the model [6]. The return-on-sales variable implies the fir m can use a simple sales- share weighting system. But the return-on-equity variable requires the weights to be based on the equity of the unit. Individual equity measures for each business unit are not avail able, so we used the two-digit industry capital/output ratio to estimate a capital-share measure from the sales/share data.^9 This weighting system approximates the equity share of each business. Thus, the two dependent variables have slightly different sets of independent variables. The profitability model was estimated with ordinary-least squares ( OLS), and the results for the two dependent variables are presented in table 1. A graphical analysis of the error terms indicated that heteroscedastøcity was a problem, so a generalized least-squares (GLS) for mulation was comp utedùlO (^) These
interest^8 The^ return-on-sales expense, while^ measurethe return^ incorporates on equity^ both only^ profits includes^ and profits. (^9) The necessary data were taken from tables A- 1 and A-2 of [26] • (^10) Following an earlier F T C study, the fourth root of assets was used as the correction factor for the data [24] •
Table 1.--Regression Results for OLS Model Returnon E quity REL (^) (-. 5 0)-.0009 ( .0 003• 4) DIV • 0 007* ( 3 • 5 )
RMS .10 2 5 *
.o 54* (3.1) l/Log 10 A .38 0 7* (2.8 )
AD/S ( 2.1)• 7267* (3.0) .376*
RD/S (2.2)• 778 0 * (3. 5)^ .446*
C4 (-.0 3)-.00001 (-1.0) -.
G ( 2.4).0 3467* ( 3 • • 0 186*4)
K/S (10. 5).0726*
Constant (^) (-1.7)-.1068 (-3.3)^ -.0^910 * S MSE .018^ .0 0 5 R^2 .2136 .5 147
to the^ SMSE mean^ is ofthe the^ ratio dependent^ of^ the variable.standard^ error^ of^ the^ regression *Statistically significant at the a = .0 5 level.
Returnon Sales
