Markets vs Management: What Drives Profitability? TORIo Jaime A Roquebert; Robert L. Phillips; Peter A Westfall Strategic Management Journal, Vol. 17, No8(Oct, 1996), 653-664 Stable url: http://links.jstor.org/sici?sici=0143-2095%28199610%2917%03a8%03c653%3amvmw%27p903e2.0.c0%3b2-e Strategic Management Journal is currently published by John wiley sons Your use of the JSTOR archive indicates your acceptance of JSTOR's Terms and Conditions of Use, available at http://www.jstor.org/about/terms.htmlJstOr'sTermsandConditionsofUseprovidesinpartthatunlessyou have obtained prior permission, you may not download an entire issue of a journal or multiple copies of articles, and you may use content in the JSTOR archive only for your personal, non-commercial use Please contact the publisher regarding any further use of this work. Publisher contact information may be obtained at http://wwwjstor.org/journals/jwiley.html Each copy of any part of a jSTOR transmission must contain the same copyright notice that appears on the screen or printed page of such transmission jStOR is an independent not-for-profit organization dedicated to creating and preserving a digital archive of scholarly journals. For more information regarding JSTOR, please contact support@jstor. org http://www」]stor.org Wed nov204:19:11200
Strategic Management Journal, VoL. 17, 653-664(1996) MARKETS VS MANAGEMENT: WHAT DRIVES' PROFITABILITY? JAIME A. ROQUEBERT ROBERT L. PHILLIPS and PETER A WEStFALL College of Business Administration, Texas Tech University, Lubbock, Texas, U.S.A This study addresses the issue of the relative degree of variance in ROA accounted for b ndustry, corporate, and sBU effects while controlling for the business cycle and the interaction between the business cycle and industry. Two key articles, Schme (1991), are discussed in detail Research results on a recent data base(COMPUSTAT), using variance components analysis (VARCOMP) are presented that not only confirm most of the Rumelt(1991) findings, but also suggest the existence of a corporate effect, heretofore The debate between researchers in industrial The paper has three sections that parallel the organization economics (IOE) and the field of above purposes. First, we briefly review the theo strategic management (SM)concerning the ques- retical background, followed by an extensive tion of the principal source of profits(markets or comparison of the two head-to-head articles on organizational behavior) has been going on for the topic. Next, we present the results of our more than 50 years. Perhaps the term 'debate' is empirical analysis. Finally, we suggest a possible somewhat strong in that there has not been so reconciliation of differing findings and extend the much of a debate as a case of neglect. The key theoretical relationship between corporate diversi question of the principal source of profits appears fication and corporate effects on sBu profitabil- to have had very little influence on the further ity development of theoretical orientations in both the IOE and the SM fields, each acting as if its view were dominant-industry for IOE, manage THEORETICAL BACKGROUND ment for SM. Only recently have researchers from both fields gone head-to-head so to speak Ed Mason, the 'fatherof IOE, argued in the The present research attempts to build on the late 1930s that there was a rather deterministic head-to-head'research by(1) contrasting and association between market structure and profita- comparing key methods and findings on the topic, bility(Mason, 1939 ). The logic of the argument (2)investigating the issue further by pitting the rested on the premise that structural character IOE model against the quintessential SM model, istics of the industry or market, typically oper using a relatively recent, broad sBu panel span- ationalized through summary measures such as ning 7 years, and (3)exploring possible reasons concentration (Lenz, 1981), placed constraints on or any divergence with previous research find- the conduct (or strategies)firms could pursue The constrained conduct, in turn. led to differen- tial performance among firms according to the ROA: ind 1939). within the IOE field, firms in an industry were CCC0143-2095/96/080653-12 Received 4 April 7994 o 1996 by John Wiley Sons, Ltd Final revision received 5 December 1995
654 J. A. Roquebert, R L. Phillips and P. A. Westfall thought to be alike in all strategically important ation-specific'( Porter, 1991: 97). It was respects except for scale, and therefore, the focus implicitly assumed that managers'perceptions or unit of analysis was the industry(Weiss, 1971; and choices largely accounted for the variance in Rumelt, Schendel, and Teece, 1994). The early companies' performance. The emphasis on taking industrial organization economic work was also a general manager's perspective led to a largely concerned with capturing complexity: conse- process-oriented, as opposed to a content-oriented quently, explanations were based upon detailed stream of research( Porter, 1981) and extensive industry studies( Porter, 1991) Investigation of the evolution of organizational It should be pointed out that Mason was not theory models suggests a somewhat in-between the only theorist commenting on the subject. Even view of the world. As in industrial organization at the time, there were theorists arguing what we economics, performance models in organizational might now suggest belongs to the genre of the theory have been largely deterministic(White and resource-based theory of the firm. Nourse and Hammermesh, 1981). However, until recently the Drury(1938)suggested that influences specific theory and measures of environmental(industry) to the firm determined performance; i.e., manage- influence have been distinct from those employed ment basically determined firm advantages and in industrial organization economics. Further, and firms were not simply at the mercy of industry more importantly, the 'determinism' was not only factors. Over 50 years have passed since Mason's from the influence of the external environment ( 1939)proposition with apparently little dampen- but also from the organization structure itself. In ing of the industry elan. Montgomery and Porter essence, the organizing principle was that the (1991: xiv-xv)affirm the importance of indus- structure of the organization and its fit to the environment determine the relative degree of role industry conditions play in the performance h here were major shifts during the decades of Present research continues to affirm the importan of individual firms Seeking to explain perform 1970s and 1980s in the strategic management ance differences across firms recent studies have and industrial organization economics fields, with repeatedly shown that average industry profitabil- respect to the unit of analysis, the reevaluation ity is, by far, the most significant predictor of of assumptions, and the relative focus across firm performance. It is far more important than fields. The reasons for shifting theoretical orien- it is now uncontestable that industry analysi tations appear to have been the inability should play a vital role in strategy formation classical economics to explain intraindustry ( Note: Montgomery and Porter, 1991, used a inductive nature of case studies, and perhaps a single article to support the above quotation- healthy'cross-fertilization between fields Schmalensee, 1985, which will be discussed in Masons deterministic approach implied that detail later ) comparisons across industries would be valid In contrast to the basic theoretical orientation while Nourse's hypothesis supported the notion of IOE, the strategy field seems to have germi- of intraindustry heterogeneity(Bass, Cattin, and nated in the early 1960s out of the case study Wittink, 1977: 194 ) Further, the feedback loops tradition at Harvard(Teece, 1990). Learned of cause-effect-cause relationships regarding al.(1969)and Andrews(1971)constituted for environment, conduct, and performance would the most part the organizing framework that render purely deterministic theories unreconcil Ba 1991). In the classic strategy model, a firms competi- Recently, a 'new'stream of research has tive advantage was gained from a combination of reemerged by the name of resource-based the- external and internal factors, known as oppor- ories, which propose that firm idiosyncrasies in tunities and strengths applied against threats and the accumulation of unique and inimitable weaknesses. The early literature was developed resources create sustained competitive advantage around broad principles, reflecting an orientation (Barney, 1991; Collis, 1991; Conner, 1991; Lipp- toward prescriptions for practitioners and the 'rec- man and rumelt, 1982; Rumelt, 1984). It should ognition, indeed the preoccupation, with the fact be kept in mind that the seeds of this notion that competition was complex and highly situ- were present at the birth of the field (e. g, Nourse
Markets vs Managemen 655 and Drury, 1938). This rapidly developing body answer the question what is the relative impact of literature argues that the forgotten half of of industry factors vs. firm effects on financial Andrews model (the marshaling of internal performance? resources to develop distinctive competencies) Of all the past research, however, two studies needs to be brought back into the core of strategy dominate the literature. First is the classic article ( Bartlett and Ghoshal, 1991: 8-9). A clear impli- by Schmalensee (1985), cited earlier by cation of resource-based theory is the obvious Montgomery and Porter (1991) in support of their choice of unit of analysis; namely, that it requires emphasis on industry factors. Schmalensee(1985) a focus on the individual firm investigated estimates of the relative importance G A highly important second implication relates of firm(corporate ), market(industry ), and market the methodologies suitable for organization share on business unit profitability. He used the studies. Since idiosyncratic attributes rep Federal Trade Commission(FTC) line of business source of competitiveness, a reliance on detailed (LB )data for 1975 and both regression analysis ase studies, or at least in-depth analysis, appears and an analysis technique that was then unique to to be required. It do st. at least at the ioe lite first glance, that resource-based theory should (See Appendix 1 for further discussion) supplant the traditional and well-established IOE Schmalensee (1985)concluded his regression frameworks, but rather that it supplements them. and variance component analyses with four prop What we have seen then, in essence, is the ositions, two of which are relevant to the present strategic management field coming almost full work. The relevant propositions state that circle in terms of what early policy researchers (1) firm effects do not exist, and(2)industry identified as the basis for good strategy formu- effects exist and are important, accounting for at lation'( Collis, 1991: 65) least 75 percent of the variance of industry rates The deterministic vs. choice orientation of the of return on assets(Schmalensee, 1985: 349) effectiveness research does not account solely for The second major work to address the relative differences in units of analysis(as shown by part variance question was that of Rumelt(1991), of the organization theory literature), but rather who explored the variance accountability issue of each approach brings with it different perspec- industry vs. other factors. Because Rumelt(1991) tives, biases, and methodologies. As Rumelt was concerned with Schmalensee's(1985)use of (1974: 560)points out: a single year's data, he extended Schmalensee's he mismatch arises because polic (1985) methodology to cross-sectional and time researchers and economists have been interested series data by using 4 years'worth of data, and substantially different phenomena. The central included a term in his equation not only for oncerns of business policy are the observed corporate management( what Schmalensee, 1985 heterogeneity of firms and firm's choice of called firm effects ), but for SBU level manage- oduct-market commitments. By contrast, the ment also basic phenomena of interest in neoclassical theory is the functioning of the price system under norms of decentralized decision making COMPARING SCHMALENSEE (1985) It should also be added that economists tend to TO RUMELT(1991) see firms as players in a multifactor economic game, and their interest is in the game and its Schmalensee(1985) used the Ftc lB data for outcomes, rather than in the particular play or the year 1975. SBU returns on assets (ROAs performance of individual firms'(Nelson, 1991: were decomposed by an equation in the form 61) rom the foregoing discussion, it is apparent ri=H+a;+B;+ ySy+ei that there are opposing assumptions about the strength of environmental forces(choice or where ri is the rate of return of firm is operations determinism) seen in different literatures. Second, in industry j(SBU ROA); S is the market share; each set of assumptions implies justification of a are firm effects, p are industry effects, u and fferent units of analysis. In light of these two y are constants; and e are disturbances arguments, the strategy field has recently tried to Schmalensee (1985: 343-344)stated that
656 J. A Roquebert, R. L. Phillips and P. A. Westfall whereas none of the coefficients in the specified by year interactions, o represent the SBU effects, equation could be given a'defensible structural and E represent error. Variances associated with interpretationthe model taken as a whole could particular effects address" the relative merits of at least the extreme 02, and o2 versions of the classical, revisionist, and mana- Since Schmalensee (1985)used only a single genial positions. The relative magnitudes of the years worth of data, if we were to translate variables would suggest support or lack of support Schmalensee's model using Rumelt's specification for each (t= l year)we would have Schmalensee (1985: 344)tested for the nce of industry effects (nonidentical B), corpor- rikl =u+a+Bk+ y1+8n+i E(4) ate effects (nonidentical a), and share effects (nonzero y), using ordinary least squares and =(μ+Y1)+(a1+8a)+Bk+(中+E) the usual F-statistics. The results indicated that =H+a+βk+∈ corporate effects simply do not exist ( Schmalensee, 1985: 346). In contrast, tests for As pointed out by Rumelt(1991), it is clear that the existence of industry and share effects were persistent effects(oi and di) are not estimable significant at least at the level of 0.045. Schma- when only 1 year of data is considered; time- lensee (1985)reformulated his basic model leav- series data is needed to isolate their effects out the term representing corporate effects, In Schmalensee's (1985)model, industr he as(see Equation 1), and applied the effects'are the combined effects of industry and riance component analysis. His results allocated industry-time interaction in Rumelt's (1991) less than 1 percent of the variance to share model. It can also be seen that Schmalensee's effects, 19.5 percent to industry effects, and the error effects' are the combined effects of error remaining variance(80%)to error and the persistent SBU effects of Rumelt's model It should also be noted that Schmalensee(1985: The corporate effects are the same in both mod- 345)addressed the issue of whether or not it is els. It should be noted that Schmalensee con- defensible to work with industry-level data, i. e, sidered a market share variable in the variance examining the average SBu Roa by industry. component analysis; however, its contribution to y so doing, his basic equation then became variance explained was negligible (less than 1 percent), and will not be included in the remain R;=u+B +terms in as, Ss, and es(2)der of the discussion. Assuming that all effects are independent, the variances are related as fol where R, is the average SBU ROA for industry lows(anything with a prime refers to Schmalen- j. He then suggested that industry- level analysis see‘ s model):2=σa+ os and o2=σd+σ2 would to be sensible if estimates of the The theoretical corporate variance, o3 is identical variance of B(industry effects)are large relative in both models to the cross-section variance of the R, Thus, Schmalensee's(1985) industry effects (Schmalensee, 1985: 345 ). It was through the (about 20 percent; from Table 1, p. 348)corre- latter analysis that he found that the industry sponds to the sum of Rumelt's(1991)industry effects in Equation 2 accounted for over 75 per- year interaction, plus "industry variance compo- cent of the variance nents(about 16 percent), as given in Rumelt's As indicated above, Rumelt (1991) used the (1991) Table 3, p 178.(Note: Rumelt, 1991 ame data as Schmalensee but added to it the drew two samples, A and B. Sample a closely data for the years 1974, 1976, and 1977. Rumelt followed Schmalensee's, 1985, selection pro- (1991) then postulated the following model cedure and, among other things, excluded all BUs with sales less than 1 percent of the FtC r=+Q1+阝k+Y:+8n+中读+∈(3) industry total sales for a given year. Sample B was an expanded Sample A through the addition ere rikt represents the sBu rOa for a given of smaller SBUs) SBU of corporation k in industry i in year t; a Schmalensee's estimate of o2(error)is about are industry effects, B are the corporate effects, 80, whereas Rumelt's estimate of o2+o2 y represent the year effects, 8 are the industry (combined error and SBU effects )is 83 percent-
arkets vs Management 657 ather a remarkable agreement (see Table 4 for ent subsets of the data, or by considering differing a comparison of findings ) The agreement numbers of years in a longitudinal study. Com- gests, of course, had Schmalensee(1985) pounding this difficulty is the fact that there is more than I year's worth of data and, had he no unique measure for industry aver luded an SBU term in his equation, he might ages. We have described two possibilities: the well have arrived at a similar value for the time-series vs. cross-sectional methods amount of sBU variance accounted for as Rumelt Further, by analyzing industry averages, one (1991)found. (The comparison is clearly pointed implicitly suggests that the appropriate estimate out in Rumelt's, 1991, Table 4, p. 179) of the industry effect is the average of all uni Let us now focus our attention on the major within the industry. Such an implication leads to differences between Schmalensee (1985)and an erroneous conclusion. If the random effects Rumelt(1991) on their respective findings con- model is considered valid, then the industry aver cerning industry averages. Recall from Equation ages are inefficient estimates of industry effects 2 above, the equation for examining the average The appropriate estimates are the best linear profitability of industries. The percentage unbiased predictors (or BLUPs, see Searle, accounted for by industry effects was estimated Casella, and McCulloch, 1992) to be approximately 75 percent by Schmalensee Thus, to avoid the difficulties associated with (1985), but only 48 percent for Rumelt (Table choice of data base, choice of variance measure 5, 1991: 11). There are at least three possible for industry average, and choice of estimation of reasons for the discrepancy. ndustry effects themselves, we recommend the First, as indicated above, the industry term of decomposition of the actual Roa variance(the Schmalensee's (1985) model encompasses both rikr), rather than the variance of the industry the persistent and nonpersistent industry effects, averages, as the primary vehicle for assessing the i.e., includes the industry effects and industry by relative importance of factors. Such a decompo- year interactions of Rumelt's (1991)model, sition then may be used for determining decompo- Second, two sets of data were used which or time-series), or for other estimates (e.g yielded a different number of observations per BLUPs), if desired industry. The variance decomposition procedure is very sensitive to such differe Third, the two theorists calculated the industry THE CURRENT RESEARCH averages in a different way. Schmalensee(1985) computes the average return for all corporations We wanted not only to compare and contrast the in the given year, then considers a cross-sectional Schmalensee/Rumelt differences, but also to enter variance. Rumelt(1991)calculates the average the debate using a more up-to-date data base over all corporations and years, then considers while employing a variance components analysis variance of the time-series averages similar to Schmalensee (1985) and Rumelt It appeared that some of the difference might (1991). Thus, our mathematical model was as fol be due to the differing methods of average calcu- lows lations; thus, we examined Rumelt's(1991) find- gs on the decomposition of variance of industry r面k=+α1+阝k+Y:+δn+中+∈ Appendix 1 for a detailed, mathematical dis- where rikt represents the SBu ROa for a giver average calculations ). In conclusion, the discrep- are industry effects, B are the corporate effects, ancies between the 75 percent(Schmalensee) and y represent the year effects, 8 are the industry the 50 percent(Rumelt) may be the result of the by year interactions, represent the SBU effects, overall precision with which the industry effects and E represent error. are estimated. This precision may be more related Since we were conducting a longitudinal study, to a particular data base used than it is to the we expected to obtain results similar to rumelt actual ROA values-one can obtain drastically i.e., a high SBU effect, a low industry effect different variance decompositions through differ- and probably a zero or near zero corporate effect
658 J. A Roquebert, R. L Phillips and p A. Westfall since both Schmalensee's(1985)and Rumelt's The Frc data were designed to obtain data from (1991)corporate effects were trivial, at best only the corporations with the greatest market Data b cessful,corporations. On the other hand, there vantage in looking at a sample In the present effort, we selected the COMPU- not only better represents businesses in general, STAT@ Business Information Industry Segment but is much more recent. Despite the above Data as the population from which to sample. advantages, it must be acknowledged that the In addition to the financial information at the Ftc data were collected under regulatory force consolidated corporate level, COMPUSTAT com- whereas the COMPUSTAT data are taken from piles information required by the Securities and data that companies provide in their respective Exchange Commission (SEC) and the Financial statements. ccounting Standards Board (FASB) Statement Number 14. Corporations are required to provide nformation on their principal lines of business Sample selection anmd statistical procedure (industry segments). In the data base, each cor- In order to parallel Schmalensee (1985)and poration is required to report information on seg- Rumelt (1991) we restricted our sample to 4 ments which account for 10 percent or more of digit SIc codes connected with manufacturing consolidated sales, operating profits, or assets. (codes 2000-3999). Second, we only selected Before proceeding, it may be well to compare those corporations with two or more SBUs. Third, the comPustat data to the ftc data used by we eliminated all SBU ROAS greater than four both Schmalensee(1985) and Rumelt (1991). standard deviations from the mean in order that Table 1 presents key aspects of the two data extreme values did not have undue influence on ases. As can be seen from Table 1. there are the anal important differences between the two data bases. Of the four estimation procedures in the The compustat data base is SAS/STATO package for the variance comp nents analysis(PROC VARCOMP), we selected o more recent(1985-91)VS. FTC (1974-77); the maximum likelihood procedure because of o broader in scope(746 manufacturing SICs vs. certain advantages. One benefit is that the asymp 260 for FTC); totic covariance matrix of the parameter estimates for a greater time period(7 years of data Vs. is readily available as a by-product of the esti for FTC) o larger (almost 3000 corporations involve manufacturing vs. fewer than 500; over Y 6000 onal elements of the covariance matrix are stan- dard errors of the parameter estimates which can manufacturing lines of business vs. fewer than in turn, be used to create asymptotically valid 4000 for the Ftc data) confidence limits on the parameters. In other o less restrictive(includes smaller corporations- words, we can estimate the significance levels of average SBUs per corporation of 1.95 vs 9.17 the resulting z-values. We must, however, add a for the ftc data ). caveat. The validity of the asymptotic approxi- mation requires that the true underlying parameter To be sure, both data bases have their advantages. be nonzero That the use of these standard errors Table 1. Comparison of the COMPUSTAT and FTc zero is, strictly speaking, not valid for hypothes data bases testing. Nevertheless, it is useful to supply the standard errors along with the estimates to pro Average of vide a measure of accuracy. Further, the order FTC sample COMPUSTAT of magnitude of the parameter(expressed as a Industry categories 942 percentage of the total variance accounted for) Corporations 6873 an be used as an indicator of the likelihood that Lines of business (LOB) 4, 193 13.398 the underlying true value is nonzero--that is, the LOB rporation 917 195 greater the parameter, the more likely the basi is met
Markets vs Management 659 a second benefit is that the maximum likeli- As can be seen from table 3. our results hood estimation scheme is an iterative procedure produced a high degree of agreement with the which incorporates the relative magnitudes of the Rumelt(1991) results of the Ftc data base in estimates at each step of the iteration. The alter- percent variance accounted for by both industry native is to use single-step(noniterated) quadratic and SBU effects. The results of the MIVQUEO estimates which do not account for the relative analysis provided the same relative order of vari- magnitudes of the variance components, and such ance accounted for, but as expected due to estimates lose efficiency relative to the iterated reasons discussed earlier, produced smaller values estimates, particularly in large samples(see Searle for the parameters and a much larger value for et al, 1992, for further discussion) the error term(industry =4.6%0; corporate=7.0%, Due to computational constraints with the SBU=32.6%; industry by year interaction maximum likelihood method, we employed a ran- 2.9%0, and error= 52.6%). We suggest that the dom sampling procedure, without replacement. relative size of the parameters of the MIVQUEO Ten samples of 100-150 corporations were ana- results tend to support the maximum likelihood lyzed. Each sample, on the average, contained parameter estimates 100-150 corporations in over 160 industries. The However, there is a significant divergence 10 samples resulted in 16,596 different sbu between the present results and both the schma ROAs (all that met the selection criteria in the lensee (1985) and Rumelt (1991) results data base) being analyzed. The average number Specifically, we estimate a nontrivial corporate of SBUs per corporation was 4.01. Table 2 pre- effect, i.e., 18 percent of variance for us vs. less sents the descriptive statistics for each sample. than 2 percent for Rumelt(1991)and near zero In order to further investigate the robustness of for Schmalensee (1985). See Table 4 for a com he procedure, we used the MIVQUEO estimates, parison of Schmalensee's (1985), Rumelt's which are more easily computed(but less efficient (1991), and current results. (Note: we compare than the maximum likelihood estimates), allowing our results to Rumelt's, 1991, Sample B as that us to analyze the entire data set with a single run. was the sample that was less restrictive with respect to the size of sBU market share and thus, more comparable to our data) Table 3 shows the results for each includ- ing the parameter estimates for the various effects, TOWARD A POSSIBLE the percentage of total variance parameter RECONCILIATION represents, the value of the dia asymptotic covariance matrix, and the estimated It should be acknowledged that Schmalensee significance (1985)recognized that 80 percent of the variance Table 2. Descriptive statistics for each sample Sample Average S.D. Minimum Maximum Levels ROA ROA ROA Industry Cor SBU Sample 1 1, 752 l134 22.53-138.2413239 437 Sample 2 1, 773 0521973-1376014938 266 45l Sample 3 1,711 11 20.3 131.87 13559 432 1096 21.79 -13847 128,35 Sample51,60310.36 -134.38 145.16 6909039 387 415 Sample 6 1, 512 20.90 31.22 1464 403 Sample 7 1, 699 1003 22.27-144.68121.74 129920.55-147.0614241 Sample 9 1,721 11.4 13868 143.10 Sample 10 1, 568 9.29 23.48-143.22144.6 403 16,59611.00991 -147.0614938 237.5 105.1 4214
660 J. A Roquebert, R. L. Phillips and P. A. Westfall Table 3. VARCOMP results for each sample ndustry Error Total Industry Corporation Variance component estimate 2.510.392 n= 1752 Percent of total Estimate variance (diagonal 94045 Significance level 3.34 0072107 omponent estmate 100.34 92.121552008520.55116.76485.81 n=1773 42%240%100.0% riance(diagonal) 92860 7095600.350.84373150.75 5860933,3616.39 Sample 3 Variance 468,11 (diagonal) 39260 嚣盤醫 7399281990.8800020197573.05 n=1457 129%49,2%0.2%0.0%35.2%100.0% ariance(diagonal) 876.132096.78 0.00 Sample 5 Varance onent estmate 125.701.5617.851984760898 449.24 286739123.80 mple 6 Variance onent estimate 10560 145.9 13 488.11 =1512 Percent of tot stimate variance(diagonal 431.20 920.705577808510243 ample 7 Variance 173.9 173.00 2003162805 ce(diagonal 65284 232 317255.5244.788213544.77 Estimate variance(diagonal) 380.20 263,3 15859613.51 n=1721 Percent of tot 嚣嚣 354%100.0% 7.12657.22 Percent of tot 42.0% 36.1%100.0% Estimate variance(diagonal) 1496.55 507.75 1.70 Significance level 7.110.83 23.50 ercent of total 10.1% 179%371%04%2.3%32.2%100.0% average perce 10.2% 179%37.1%0.5%2.3%32.0%1000% in the rOAs of sBUs was in his analysis unre- Also, it is interesting to note that rumelt con- lated to industry, share, or corporate (firm) cluded his section on corporate effects with the effects. Thus, it should come as no surprise that observation that it was surprising to find only a include an vanishingly small additional term in the equation (SBU effects ), suggested that both his result and that of Schma some of the previously unexplained 80 percent lensee remains a puzzle and deserves further would be accounted for investigation'(1991: 182)
Markets vs Management 661 Table 4. Comparison of results (percent of variance perform sensitivity analysis on the analytical accounted for) results by successively increasing the required Rumelt Current number of SBUs per corporation to be included Variable Schmalensee in the analysis. We went from a minimum of two SBUs per corporation to three, then four, Industry x Year n. a 2.3 etc. We were able to increase the SBu require- Market share n.a. ment through six. After a minimum of six SBUs 1.64 17.9 per corporation, our statistical power trailed off precipitously. However, the minimum of six 44.79 32.0 SBUS pe resulted 100.0 99.5 ber of SBUs per corporation of 7.82. Figure 1 presents the results of our diversification sensi- tivity analysis together with a power function terpolation Three possibilities immediately come to mind As can be seen from Figure l, there is a strong in order to explain the difference in the corporate relationship between the number of SBUs per parameter between the present research and the corporation and the percentage of SBu rOa results of both Rumelt (1991)and Schmalensee variance accounted for by the corporate effect. (1985 ) First, time must be a consideration. Per- What is particularly puzzling is our result which haps the world shaped up a bit differently in the suggests the corporate effect is inversely related late 1980s and early 1990s vs. the mid-1970s. to diversification is exactly the opposite of the Second, the difference in methods, although result obtained by Kessides(1987) minor, might have something to do with it. On the other hand, if we use a power function Finally, we must consider the differences in the to model the diversification sensitivity results and data bases. Recall that the Ftc data base con- then plot the average number of SBUs per corpor tained a much higher number of SBUs per corpor- ation in Rumelt's(1991)B' sample(the larger ation (Table 1). Thus, diversification of the cor- sample), we obtain a value for the corporate oration may have something to do with the effect of about 9 pe ercent(see the vertical axis differing result of Figure 1). Although we have not completely There is not much we can immediately do with converged with RumeIt's(1991) results of 1.64 respect to the first speculation-the world-has- percent, we have narrowed the gap from about changed hypothesis. And, the differences in 16 percent (2% vs. 18%)to about 7 percent methods, in our judgment, should not have made (2% vS 9%o). In the variance component analysis that much of a difference in outcomes (recall we used both an iterative and noniterative method But, we could investigate the third speculation- R2·0.849 diversification of corporation. Rumelt(1991), cit ing a working paper from Kessides(see the 3 16 citation on Kessides, 1987), reported that Schma lensee's data was reanalyzed in which all corpora tions which were active in fewer than three indus were excluded. The results produced a " statistically significant corporate effect(Rumelt, 1991: 170). The implication is, of course, the greater the diversification the greater the corporate a effect since Schmalensee (1985), who inclu the less diversified corporations, found no sig- nificant corporate effect The average number of lines of business per corporation for Rumelt's(1991) Sample B was Averag· Number of sBUs p· r Corporation 6.07. Our average number of lines of business Figure 1. Size of corporate effect vs degree of diver per corporation was 4.01. Thus, we began to sification
