Determinants of Firm Performance: The Relative Importance of economic and Organizational Factors TORIo Gary S. Hansen; Birger Wernerfelt Strategic Management Journal, Vol. 10, No 5(Sep. -Oct., 1989), 399-411 Stable url: http://links.jstor.org/sici?sici=0143-2095%02819890 10%2910%3A5%3C399%3 ADOFPTR%3E20CO%3B2-A trategic 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 STOR 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 nov204:08:03200
Strategic Management Journal, Vol. 10, 399-411(1989) DETERMINANTS OF FIRM PERFORMANCE: THE RELATIVE IMPORTANCE OF ECONOMIC AND ORGANIZATIONAL FACTORS GARY S HANSEN Graduate School of Management, University of Washington, Seattle, Washington U. SA BIRGER WERNERFELT Alfred P. Sloan School of Management, Massachusetts Institute of Technology, Cambridge, Massachusetts, U.S.A We decompose the inter firm variance in profit rates into economic and orga e model from each paradig we find that bol actors are of firm performance. Further findings are the ects are d that organizational factors explain about twice 如 INTRODUCTION Theory or empirical evidence of linl performance abound within each paradigm, but In the business policy literature there are two surprisingly little has been done to integrate the major streams of research on the determinants two and evaluate the relative effect of each on of firm performance. One is based primarily firm profitability. Notable exceptions are recent importance of external market factors in deter- kani (1988), Miller (1986), White(1986), White mining firm success. The other line of research and Hammermesh(1981), and Lenz(1981), who builds on the behavioral and sociological paradigm discussed and/or evaluated a limited number of and sees organizational factors and their fit with contingent relationships between economic and the environment as the major determinants of administrative factors. No work has been done success. Within this school of thought, little direct however, to assess the relative importance of attention is given to the firms competitive these two sets of explanatory factor position. Similarly, economics traditionally has In h an integrated disregarded factors internal to the firm. examination of firm profitability. Utilizing a d behavioral data base I The following statement, from Buzzell and Gale(1987), is we construct and test three models of firm typical. 'Our treatment of strategy is also confined to performance, first an example from an economic In contrast, some other elements of strategies cannot be perspective, second an example from an organi- dimensions that can be measured in reasonably clear terms readily measured, or perhaps measured at all. There has zational perspective and the third an integration differences unquestionably affect performance. But we know economic and organizational componen to its ms of their cultures, and such the inter-firm variance in prc of no way to measure the key policies, management Before describing the study we would like to or personality factors that shape corporate cultures. make two things clear. First, we do not propose that would be equally difficult to quantify or even to synthesize all economic and organizational theories of firm performance. We have taken an 0143-2095/89060399-13506 Received 17 September 1987 C 1989 by John Wiley Sons, Ltd Revised /7 January 1989
400 G. S. Hansen and B. Wernerfelt example from each class of models. We will argue considered in the literature, We divide these into that these examples are somewhat representative the three classes mentioned above. Each is but they are, nonetheless, only examples. Readers discussed in turn simplistic may well expect a more complex ample to capture more of the variance in firm Industry variables profits. This problem of selecting a representative A long tradition, most often associated with Bain model seems particularly acute for the organi -(1956) is concerned with identifying properties e because of its numerous of industries contributing to above-average pre theories and levels of analysis found in the itability. A large set of variables (growth, literature, but that model actually does better in concentration, capital intensity, advertising inten our sample. Second, we use accounting rates of sity, etc. ) have performed differently in different return as our measure of performance. Within studies, but the overall importance of these the economic tradition these have been the factors is beyond dispute(Ravenscraft, 1983). In subject of some debate(Bentson, 1985), but they a study such as ours, where interest is focused are still commonly used, and arguments have on the importance of industry per se, rather than been raised in their defense(Long and Ravens- on characteristics of more or less attractive craft, 1984; Jacobson, 1987). The choice of profit industries, the effect of industry can be captured rates is less obvious vis-a-vis the organizational by the average industry profits. A recent study literature. While profits have been used within by Schmalensee (1985) shows that differences that tradition, so has a large number of other between industries as measured by average concepts of performance(e. g. satisfaction, sur- industry return on assets account for almost vival, etc. ) If the organizational model had all the explained variance in business unit erformed less well, this would have been a performance serious problem In the next two sections we present economic and organizational models. We then Variables relating the firm to its competitors describe our data and give the results, ending The key member of this class is relative market with a discussion of the implications of our share, a variable which has been widely used in findings strategy and is emphasized PIMS(PIMS, 1977; Buzzell and Gale, 1987) Originally perceived as the source of market ECONOMIC MODEL OF FIRM power(Shepherd, 1972) market share and more PERFORMANCE specifically relative market share as viewed for this study serves as a proxy for some firm-specific Industrial organization economics has proven relative competitive advantage resulting from extremely useful to researchers of strategy content learning effects and other firm specific assets in providing a basic theoretical perspective on(Karnani, 1984) the infiuence of market structure on firm strategy nd performance. While there is a range of inants of firm-level Firm variables profitability include: (1)characteristics of the We complete our model with firm size. This industry in which the firm competes;(2) the most often interpreted as a source of organi firms position relative to its competitors; and(3) zational costs(Shepherd, 1972), or X-inefficiencies the quality or quantity of the firm's resources. (Leibenstein, 1976). From a strategy perspective Scherer (1980: Ch. 9) surveyed many of the we note that size also may be an indicator of pecific models of both industry- and firm-level diversification, which by and large has been performance, and Porter's review(1981)describes found to affect performance negatively(Rumelt he influence of the IO paradigm on business 1982; Porter, 1987; Wernerfelt and Montgomery Our economic model, while only an exampl Overall, the typical economic model of firm includes several of the explanatory variables performance explains from 15 to 40 percent of
Determinants of Firm Performance the variance in profit rates across firms. Apart be over-differentiated in one area and under from random effects. measurement errors, and differentiated in another, but on the whole be o forth, one can suggest at least three expla- just about right? In contrast, firm performance nations for the remaining variance. First, there is an aggregate phenomenon may be important economic variables, the extent One research stream which has attempted to of which cannot be measured (e. g. assets that capture the multidimensional aspect of these are specific to an industry or a trading partner). significant organizational ph Second, the true' model may be such that of structure, motivation, group dynamics, job intervening economic variables differ from case enrichment, decision-making, leadership, goal to case, making aggregate analysis difficult. Third, setting and planning, etc.is that of organi- with very few exceptions (e.g. Armour and zational climate. Long a prominent concept Teece, 1978), organizational factors are not within the organizational sciences, organizational considered in this literature climate was originally defined as follows The concept of climate provides a useful bridge ORGANIZATIONAL MODEL OF FIRM between theories of individual motivation and PERFORMANCE behavior. on one hand. and organizational theories, on the other. Organizational climate, as defined here, refers to the perceived subjective Perhaps even more than their economist counter- effects of the formal system, the informal parts, organizational researchers have developed tyle of the managers and other important a wide variety of models of performance. While environmental factors on the attitudes, beliefs, are rich in the breadth and depth of their studies 1900,)ular oa dons of the people who work the organization behavior and theory literatures of organization structures, systems, and people, the variety of conjectures and empirically tested And more recently as models makes aggregation difficult. For example just determining the appropriate construct of the perceived properties or characteristics found performance or effectivene in the work el vironment that result from ranging from employee satisfaction to shareholder actions taken consciously or unconsciously by wealth(Cameron, 1986; Goodman and Pennings an organization and that presumably affect ubsequent behavior(Steers and Lee, 1983: 82) 1977; Steers, 1975). In broad terms this stream of research suggests that managers can influence Just as geographic regions have different the behavior of their employees(and thus the climates' as a result of the immediate interaction performance of the organization) by taking into of t account factors such as the formal and informal rain/snow to make them favorable or unfavorable structure, the planning, reward, control and climates for living, so can a firm have as the information systems, their skills and personalities, interaction of its facilities, structures, systems and the relation of these to the environment That and people a favorable or unfavorable work is, managers influence organizational outcomes by climate establishing context, and that context is the 196Os. and still a major result of a complex set of psychological, sociologi- concept today, climate uniquely refers to a broad cal. and physical interactions class of organizational and perceptual variables The difficulty in working with such multifaceted that reflect individual-organizational interactions developing, collecting and aggregating appro- Steers and Lee. 1983: Field and Abelson.1982 priate measures(Bonoma, 1985; Bower, 1982). James and Jones, 1979; Schneider,1975;Litwin Many constructs within the literature are ditficult and Stringer, 1968). It is important because it to measure and those which are relatively easier provides a conceptual link between analysis at to capture are otten at the micro (individual) the organizational level and at the employee level. For example, can we say that a firm on vel, precisely the requirements of this stud the whole is bureaucratic just because it has Unlike objective measures of organization everal levels to its hierarchy? Can a firr ructures such as 'M-form' or systems such
402 G.S. Hansen and B. Wernerfelt capital budgeting policies, climate as measured climate across departments in the by employee response to questionnaires reflects zation, the departmental effects are much weaker the individual's per ceptions of that employe 41-42) organizational effects(Drexler, 1977 han ti bout the effect or presence or nature of certain organizational phenomena. Climate is not tructure-size,production processes, arrange Denison(1982), using the same climate instru ments, or number of levels. Structure may ment with substantially more firms, also demon influence human behavior, but it is not necessary strated that climate measures were more appro- to examine human behavior to describe an priate at the organizational level rather than at organization's structure. Additionally, the same the group or individual levels. Glick's(1985) structures in different organizations may produce review of the psychological and organizational very different climates (Springer and Gable, climate literature and the empirical corrections 980)as structure is only one of the many he makes to Drexler's work leads him to conclude factors that significantly infuence the worker's Thus, the concrete conclusion is that Drexler's Numerous studies have demonstrated how reliable measures of organizational clima.a- perceptions of his or her work environment aggregated perceptual measures are indee changes in organizational structures, systems and Given constraints on data access it was not practices have altered climate measures and possible to duplicate the above tests for this hence individual performance(Pritchard and study; nor was it deemed necessary, given these Karasick, 1973; Litwin and Stringer, 1968; For- prior tests of the same instrument shand and Gilmer, 1964). Lawler et al.(1974) Of course, there are many competing theories studied 117 research laboratories and demon- and concepts of firm-level performance and no strated that both organizational structure(span of single construct has emerged in the literature control, size, levels) and organizational processes We will interpret a positive association between (performance reviews, budgeting, collaboration) overall firm climate and profitability as support re more closely associated with climate meas- for one theory of organizational determinants of ures than with performance(both subjective and performance. It is possible to interpret the climate objective) measures, and that organizational scores in light of competing theories. That is climate was directly linked to performance. Other high climate scores may indicate that the key more clinical efforts have shown linkages between contingencies are satisfied, or that corporate managerial practices and attributes or dimensions culture is appropriate to the environment, etc of organization climate and firm performance (Denison, 1984). If one subscribes to such an ditional climate model of firm performance from our models are biased towards zero. We To empirically validate that climate was indeed use the climate data because they have some a firm-level construct, Drexler(1977)examined significant history in the literature 1256 work groups representing 6996 individuals many elements of organizational phenomena,are 21 organizations to test the strength of the appropriate for analysis at the firm level,are organizational climate construct at the organi- largely infuenced by managerial actions and are zational level rather than at a departmental available for a reasonable number of representa or some sub-organizational level. His findings tive firms strongly support the use of our measures of organizational climate for firm or organizational analvSIS DATA AND MEASURES The results reported in this study should encour- The sample includes 60 Fortune 1000 firms ge those researchers who consider organi representing both dominant and lesser members zational climate to be an organizational attribute A large share of the variance in measures of of the ese firms togetl limate that describe organization-wide cor comprise over 300 lines of business as determined at the four-digit SIC level. While the sample is While there are differences in organization not large, it is clearly representative of major
Determinants of Firm Performance 403 ORGANIZATIONAL FACTORS (structure, systems, size, history FACTORS (sociological, political, FACTORS economic, technological) skills, personalities, age) ORGANIZATIONAL CLIMATE Decision Making Practices Communication Flow Leadershi rocesses INDIVIDUAL BEHAVIOR ORGanIzation PERFORMANCE Figure 1. A traditional model of organizational climate corporations in the United States. For a more assets as reported by Compustat. Because the detailed discussion of the sample and its character- availability of organizational survey data, dis- istics see Hansen(1987) cussed later, dictated the sample, not all the data are from a single year. In order to adjust profit for annual effects such as inflation, the risk-free Performance measure rate as determined by the T-bill ratewas The measure for firm-level performance(FIRM) subtracted from the Roa for each firm as was selected as the 5-year average return on prescribed by Shepherd(1970: 50-51). This multi-
404 G.S. Hansen and B. Wernerfelt year average seemed appropriate given the long- zations (soo) instrument described in Taylor term strategic and structural variables in this and Bowers(1972). Tested and developed by the study, and is consistent with previous strategy Institute for Social Research at the university 1988: Rumelt, 1982; of Michigan, this questionnaire many Christensen and Montgomery, 1981) dimensions of organizational factors including characteristics of communication flow, emphasis Economic variables on human resources, decision-making practices organization of work, job design, and goal The economic variables are at the four-digit SIC emphasis. It has already been noted that this level and come from Trinet/EIS, the FTc specific operationalization of organizational cli- Line of Business Data, and the Census of mate is accepted as appropriate for organizational Manufacturers. Firm financial information was level studies (Glick, 1985; Mossholder and obtained through Compustat tapes Bedeian, 1983; Denison, 1982; Drexler, 1977) Industry profitability(IND), is defined as the Briefly, the Soo operationalizes climate in a sales weighted average return on assets across somewhat prescriptive manner. That is, it assumes the firm's lines of business. It would be desirable that the presence of work groups with clear, to take our sample firms' own effect out of consistent and high individual, group, and organi average industry profits, but we were not able zational standards and goals, linked through to do this consistently. As mentioned above, effective communications utilizing participatory IND was selected to summarize the effects of decision-making techniques, is evidence of good all industry-level variables such as growth rate, management. Employees who feel properl oncentration, barriers, etc. Given this, inclusion rewarded with pay and recognition, and who of an additional industry-level variable would have leaders/managers who train, help, listen and lead to a misspecification of the model are experts in their tasks, are more productive To indicate firm competitive position, we use Finally, workers who are members of work the relative market share as calculated by the groups that have standards and are mutually sales weighted ratio of the firms market share supportive lead to better performance. For divided by the four-firm concentration ratio in this study we use the results of over 50,000 its four-digit SIC industry. This ratio is similar questionnaires administered to 60 publicly traded to the relative market share as developed by non-regulated Fortune 1000 firms. The majority PIMS(1977). We of course expect the sign of of the firms(47)came from the Soo data and this variable(RELMs) to be positive the rest from a similar instrument used by the At the firm level we use firm size (SIZE), Forum Corporation(1974).(A Chow test allowed defined as the natural logarithm of total assets. us to pool the two after testing for homogeneity This should measure inefficiencies resulting from in construct and content)2 size or diversification and we expect a negative Our data consist of averages per firm. In a sign(as in Shepherd, 1972). A common problem few cases only one of many divisions was with this measure is that the size variable enters surveyed, but in most cases unweighted averages in the denominator of FIRM. Accordingly, among multiple divisions are involved. 3 Because measurement errors in this variable will generate of the large sample size our data are better than a negative bias in its coefficient and increase the those obtained from 'key informant'methods amount of variance it nts for in the but it is beyond dispute that the data contait egression. While the magnitudes of these effects large amounts of noise are difficult to assess, their salience is limited All the major dimensions of the SOo question given that our results attribute relatively low naire are intended to measure different constructs variance to economic facto limate variables to have a noticeable amount of Organizational variables collinearity. Thus, only two of the variables were As mentioned above, it is very difficult to get 2 This test wa Pindyck and Rubinfeld (1981 121-123)and was conducted at the 5 percent level good data on organizational factors. Our measure 3 Unfortunately we have not been able to get evidence on of climate is derived from the Survey of Organi- the amount of variation across divisions of given firms
Determinants of Firm Performance 405 selcted for our models. In particular, we use (a) Table 1. Description of variables Emphasis on Human Resources(HRM EMPH) which measures the employee's perception of how concerned the organization is with his Standard welfare, work conditions, etc, and(b) Emphasis varable cases value Mean deviation on Goal Accomplishment(GOAL. EMPH)which FIRM 60-0.1000.2370.0470.065 measures the employee's perception of relative ind 600.0100.2250.1120.046 emphasis on achieving aggressive goals or objec- RELMS 600.0230.8200.2080.181 tives. We chose these variables for four reasons, SIZE 603.1489.8507.2771.520 One, they are well grounded in major streams HRM. EMPE 2.4874.1143.0920.354 of research. HRM EMPH comes directly out of GOAL. EMPH 2.8014.4103.5010.383 the human relations school of thought(Barney 986: McGregor, 1960; Roethlisberger and Dick son,1947). The other variable, GOAL. EMPH can be associated with Barnards(1968) organi- RESULTS zation purpose and is more representative of the scientific management school of research, Table 3 shows the results of three models of firm specifically the work on goal theory by Locke performance-the economic, the organizational (1978). The strength of goal theory research in and the integrated. For each variable the b both laboratory and field settings, and the size coefficient, its significance level based upon its of its impact, is significant and well established t-statistic, and the standardized b weights are in the literature. Second, the two variables reported. In addition, the computed F-ratio for represent the logical tension between attention the regression, its probability (p-level)and the to an employees needs and task accomplishment. unadjusted(r2)and adjusted(R )R2'values are This, too, has been an item of provided e literature(Blake and Mouton, 1964). Third these variables were the least correlated among Economic model the climate dimensions, And fourth these two (in addition to a few others) were available from The least-squares estimation for the economic both the soo and the Forum instrument, model was significant at the 0.009 level and the enabling maximum sample size. While our selec- signs of all the coefficients were in the expected tion may appear to be somewhat arbitrary, it direction. Somewhat surprising was both the should be noted that our results are very similar insignificance of the relative market share variable for other pairs of climate variables and the relatively low R2 of 0. 141. A few recent studies suggest that high absolute or relati Data description share may not be as closely associated with firm profits as argued in the BCG(1972) framework Tables 1 and 2 provide the descriptive statistics Jacobson and Aaker, 1985; Rumelt and Wensley, for the variables and the correlation matrix. 1981)and that even low market share firms may It can be seen from the correlation matrix indeed be just as profitable given certain favorable hat the climate measures--HRM EMPH and industry- and firm-specific conditions(Woo, 1981 GOAL. EMPH--exhibit the strongest corre- Hammermesh, Anderson and Harris, 1978) Itions with firm-level profit. The sample was Nevertheless, the economic model as a whole is tested for multicollinearity and for heteroscedas- successful in explaining firm profit performance ticity using the Goldfield-Quandt tests(Bass, Cattin and wittink, 1978). This revealed no Organizational model The organizational model is also highly significant For each firm, many employees answered many que with its coefficients in the expected positive pertaining to each of these two variables. The values we e.g.,Taylor and Bowers(1972)for a copy of the orstions are the average scores across individuals and Se direction. While the HRM.EMPh variable is lestionnaire highly significant(p-level less than 0.000)the and further details apparent interrelation between it and the
406 G. S. Hansen and B. Wernerfelt Table 2. Correlation matrix of variables in models FIRMTT IND丌 RELMS SIZE HRM EMPH GOAL. EMPH 1.000 IND丌 0.293 1.000 RELMS 0.119 0.075 SIZE 0.259 0.07 0.37 1.000 HRM. EMPH 0.600 0.110 -0.162 1.000 GOAL, EMPH 0.438 0.053 0.021 0.064 0.545 1.000 GOAL. EMPH variable keeps the latter insignifi- Variance decomposition cant. When run as two models with the variables kept separate, the GOAL. EMPH variable is To decompose the interfirm variance in profit positive and significant at the 0.005 level. The rates we will start with the combined model and organizational model alone explains substantially use f-tests to see if there are significant differences more of the profit variance than the economic in the amount of explained variance as we drop model alone. Given the extremely large number either group of variables from the complete of surveys used to formulate the measures, and model. Figure 2 starts at the bottom with the the long history of importance in the management of the F-test between it and the two submodels literature of motivating employees and goal Finally, the two submodels are tested from the theory(Locke, 1978), the results are not overly null model. This method is presented by Kmenta surprising to organization theorists (although those from an economic perspective may find (1971)and utilized by Schmalensee(1985) them noteworthy) Both economic and organizational factors are highly bethe at either level the organizational factors explain Integrated model more variance than the economic factors To The third, or integrated, model of firm perfor- formally analyze this, Table 4 gives the incremen- mance is also highly significant. The signs of the tal contributions to R2 for the economic and coefficients are in the expected direction, and it organizational models(Theil, 1971). Three thing explains even more of the firm's performance are important. First, both the models explain ubstantial amounts of firm profitability. Second with an R2 of 0.457. The RELMS variable is the organizational factors account for about twice now significant and with the expected sign as much variance as the economic factors.And indicating the importance of the firm's market third, the models are approximately orthogonal chare relative to its major competitors. However the GOAL. EMPH variable remains insignificant. Suggesting that these are indeed two independent It is worth noting that the R2 of this me only slightly smaller than the sum of the R2 values of the two partial models. So our specific economic and organizational factors appear to CONCLUSIONS AND IMPLICATIONS be roughly independent contributors to perfo We integrated two sample models of firm mance performance, one from the economic paradigm and one from the organizational paradigm. The results confirm the nodel, that climate is the effect of numerous environmental of both sets of factors in explaining perfo ormance We are not arguing that these underl However, the results also indicate that organi dog businesses or weak competiti variance in firm profit rates as economic factors Regarding the mod ality. There are
Determinants of Firm Performance 0∞cm寸s E卫卫一5 含军