Serving Multiple Constituencies in the Business School: MBA ...

In Press - Academy of Management Journal

Serving Multiple Constituencies in the Business School: MBA Program vs. Research Performance

James S. Trieschmann Terry College of Business

University of Georgia, Athens GA 30602

Alan R. Dennis Kelley School of Business

Indiana University, Bloomington IN 47405 e-mail: [email protected]

Gregory B. Northcraft College of Commerce and Business Administration

University of Illinois, Champaign IL 66801

Albert W. Niemi Cox Graduate School of Management

Southern Methodist University, Dallas TX 75275

May 17, 2000

2

Serving Multiple Constituencies in the Business School:

MBA-Program vs. Research Performance1

ABSTRACT

Business schools strive to meet two goals: knowledge exploration through research and

knowledge exploitation through instruction. Our results indicate that research performance

(pages published in leading journals) and MBA program performance (from the popular press)

are different. Research performance was improved by the number of faculty, the proportion of

full professors, the proportion of assistant professors, and editorships. MBA program

performance was improved by the budget per faculty member, and the proportion of full

professors.

1. We would like to thank Srini Reddy, Dale Goodhue, and Bob Vandenberg all of The

University of Georgia for statistical advice, and the editor and anonymous reviewers for helpful

comments on previous drafts. This article was prepared while Alan Dennis was on faculty at

The University of Georgia.

3

March and Sutton (1997) have noted that, “Explaining variation in performance or

effectiveness is…one of the more enduring themes in the study of organizations.” However,

studying organization-level performance is problematic because of the multiple constituencies

which organizations serve (e.g., Zammuto, 1984; Tsui, 1990). The existence of these multiple-

constituencies means that, “It is not clear that organizational purpose can be portrayed as unitary

or that the multiple purposes of an organization are reliably consistent” (March & Sutton, 1997).

The failure of measures of organizational performance to reflect an organization’s multiple

constituencies may lead the organization to treat the satisfaction of one of its constituencies as

the primary goal and the satisfaction of others as pathology, rather than maintaining a healthy

tension between them (March & Sutton, 1997). From a resource dependence perspective

(Pfeffer & Salancik, 1978), this could lead to alienation of important constituencies and eventual

demise of the organization.

The purpose of this paper is to explore multiple constituency problems in measuring the

organizational performance of business schools. The study described in this paper examines (a)

whether diverse measures of business school performance – those apparently reflecting different

constituencies – agree, and (b) what factors are likely to influence them..

CONCEPTUAL BACKGROUND AND HYPOTHESES

Business schools typically acknowledge at least two constituencies: 1) students and

business practitioners and 2) academics (e.g., Stahl, Leap, & Wei, 1988; D’Aveni, 1996). Within

business schools, distinct sub-cultures – what March & Sutton (1997) refer to as the “soldiers of

organizational performance and the priests of research purity” – have developed to serve these

two constituencies. These two business school sub-cultures also map onto March’s (1991) more

general distinction between exploration and exploitation in organizations. The priests of

research purity are the explorers who create deep knowledge about organizations; their

constituency is composed of like-minded academics. The soldiers of organizational performance

are the exploiters who disseminate (and apply) that knowledge in the pursuit of improved

organizational practice; their constituency is composed of students and practitioners who

4

embrace the resulting recommendations. The interdependence of these functions means that

organizational success may depend upon striking a balance between the two.

In the case of business schools, emphasizing either exploration (academic constituency)

or exploitation (student and business constituency) may prove adaptive in the short run, but

harmful in the long run. Pure exploration without an eye toward exploitation eventually forgets

the point of exploration; pure exploitation without an eye toward exploration eventually loses its

foundation for innovation and adaptation (March, 1991). Thus, it seems critical that accepted

measures of performance capture the goals of multiple constituencies. The risk when they do not

is perverse learning – that organizational members will “strive to meet measured performance

objectives to the detriment of objectives that are not measured” (Meyer & Gupta, 1994).

Over the past few years, the popular press routinely has published rankings of business

schools (e.g., BusinessWeek, 1998a; U.S. News and World Report, 1999). These rankings have

become increasingly important to business schools. The status conveyed by such visible

measures of organizational performance allows organizations to “cream” inputs (e.g.,

recruitment of students and faculty, support of alumni and donors) and “stream” outputs (e.g.,

placement of graduates) (D’Aveni, 1996). University presidents, vice presidents, and deans have

become quite aware of the rankings and (if positive) include them in their speeches and program

evaluations (e.g., see Southern Methodist University, 1997).

The best-known business school rankings (e.g., BusinessWeek, 1998a; U.S. News &

World Report, 1999) evaluate business school performance from the viewpoint of the MBA

program. In contrast, articles in academic journals have assessed business school performance

from the perspective of research productivity -- the quantity of research published in various

academic and professional journals. Unfortunately, in contrast to the popular press rankings of

business school, these studies have evaluated organizational research performance in a single

journal or a group of journals within a given academic discipline (e.g., Reeve & Hutchinson,

1995; Stahl, et al., 1988; Zivney, T.L. & Bertin, 1992). Since 1988 (Niemi, 1988), no surveys of

research have presented a broad, business-school-wide assessment of research performance

5

comparable to the popular press evaluations of MBA programs. This paper reports a measure of

business school research performance which reflects performance across major disciplines.

The risk of focusing on a single measure (e.g., MBA program performance) is that

current popular press business school rankings may overemphasize the student/business

constituency and the exploitation role, and thereby dysfunctionally focus attention and effort on

exploitation to the exclusion of complementary exploration. Thus, a primary purpose of this

paper is to assess whether current, exploitation-focused measures of organizational performance

(i.e., teaching) are different than exploration-focused measures of performance (i.e., research).

As a starting point, it is hypothesized that:

H1: Exploitation-focused measures of business school performance (i.e., MBA-program

rankings) and exploration-focused measures of business school performance (i.e., research

rankings) will reflect distinct competencies.

Tsui (1990) suggests that organizations should be more likely to simultaneously satisfy

their multiple constituencies when the environment is more munificent. If exploration and

exploitation are distinct competencies, then schools with greater financial resources should be

better able to address their multiple constituencies. In contrast, schools with lesser financial

resources may be more likely to focus their efforts on satisfying one constituency to the

detriment of the other. This argument should apply to the tension between external and internal

constituencies, and also to the tension among internal constituencies (e.g., departments).

Therefore, for schools with lesser financial resources, we would expect greater differences

between the performance measures targeting exploration and those targeting exploitation, and

greater differences among those measures for different disciplines. It is hypothesized:

H2a: Research performance and MBA program performance will be improved by

greater financial resources.

H2b: There will be smaller differences between measures of exploration (research) and

exploitation (MBA program) performance for schools with greater financial resources.

H2c: There will be smaller differences between measures of exploration performance

6

(research) among disciplines within a school for schools with greater financial resources.

Resource-dependence theory (Pfeffer & Salancik, 1978) suggests that attention paid to a

constituency should reflect the organization’s reliance on that constituency for resources. This

suggests that private schools – which are more dependent on students and businesses for

resources – should devote more time and effort to this constituency through exploitation of

knowledge (i.e., as reflected by MBA rankings). Because public institutions have some minimum

level of guaranteed state funding for undergraduate programs, they can devote less attention to

satisfying the student/business constituency via MBA programs, and therefore instead may

emphasize their academic constituency through exploration of knowledge (i.e., research). Thus:

H3: Privately-funded schools will have higher exploitation performance (MBA program),

while publicly-funded schools will have higher exploration performance (research).

If exploration and exploitation are truly distinct goals for business schools, they should be

influenced by different factors. If the satisfaction of multiple constituencies is key to survival,

such differences would have important implications for resource allocations. A critical problem

for organizations is that satisfying multiple constituencies can require a diverse portfolio of skills

and responses (Tsui, 1990), some of which enhance the organization’s reputation in the eyes of

one constituency while simultaneously diminishing it in the eyes of its other constituencies

(D’Aveni, 1996). This problem is a particular concern when those skills and responses compete

for scarce resources in the organization.

Knowledge exploration and exploitation are labor intensive activities that draw on the

same scarce resources (i.e., the faculty). There are practical limits to the number of journal

articles that any one faculty member can produce, so as the number of faculty increase, the

potential to produce more research should also increase. Because of the differential emphasis in

many business schools on publishing (exploration) for junior faculty and instruction/service

(exploitation) for senior faculty, faculty composition in terms of the proportion of assistant,

associate, and full professors might also be expected to influence research and MBA-program

performance measures. A large proportion of assistant professors, whose primary measure of

7

individual performance for promotion is exploration, is more likely to result in more

publications. Conversely, a large proportion of full professors, who have reached a terminal rank

and may be drawn more deeply into exploitation activities such as instruction and service

(including executive education and consulting), is less likely to result in more publications. Such

a switch in emphasis to exploitation is also likely to improve the quality of MBA instruction. A

close involvement in the publishing process through in-house journal editorships may also be

linked to research performance, either because editorships are awarded based on prior research

prowess or because the presence of an in-house editor is likely to encourage local authors to

submit their best work to the editor's journal (Stahl, et al., 1988). Therefore:

H4a: Research performance will be positively related to the number of journal

editorships.

H4b: Research performance will be improved as the proportion of assistant professors

increases, but impaired as the proportion of full professors increases.

H4c: MBA program performance will be improved as the proportion of full professors

increases.

METHODS

Research Performance Measure

This paper reports a measure of business school research performance which is based on

articles published in leading business research journals. Because the quality of published

research is debatable (e.g., Byrne, 1990), as is the quality of the journals in which the research is

published (e.g., Glick et al., 1997), an important concern is the extent to which the research

performance measure is significantly influenced by the inclusion or exclusion of certain journals.

The research performance measure examined the number of pages published in 20 "top

tier" business research journals. We focused on the 13 year time period (from 1986 to 1998)

starting immediately after the last published cross-discipline ranking study (Niemi, 1988). A

three-step approach was used to identify the journals included in the study. First, we used the

AACSB (1998) definition of business school disciplines to identify a set of 13 disciplines, which

8

we integrated into a set of 8 discipline groups: accounting, finance, management, management

information systems, management science, marketing, production/operations management, and a

combination of insurance/international business/real estate. Second, we examined 7 to 13

articles in each discipline group that ranked journals or listed "top" journals in that discipline,

resulting in a set of candidate journals for each. Finally, we selected the specific journals for

inclusion in the study. We attempted to select journals so that there was the same relative

number of journals in each discipline group as there were faculty (i.e., more journals from larger

groups), so that there was at least one journal from every discipline, and so that the selected

journals were ranked highly by both journal ranking articles and citation impact factors (ISI,

1995; 1996; 1997; Starbuck, 1997). The journals in each discipline group are shown in Table 1.

The journal selection process is described on the Web at:

www.kelley.indiana.edu/ardennis/rankings.

[ insert Table 1 about here ]

For our measure of research performance, we chose the page count approach used in

previous studies (Henry & Burch, 1974; Niemi, 1988). We counted the number of pages

published by each university, adjusted for the relative page size of each journal, using Academy

of Management Journal as the standard. The conversion was accomplished by calculating the

mean number of characters per line based on a randomly selected 20-line section of text in each

journal for each year (i.e., 12 samples per journal) multiplied by the average number of lines per

page; this number was then divided by the average for AMJ to produce a conversion index for

each journal. As an aside, we note that this measures the performance of a school at one point in

time, not the performance of the current faculty; faculty compositions change, and we make no

attempt to trace faculty movements and credit past publications to their new school.

For articles written by professors from different universities, we followed the approach of

Niemi (1988) which gives equal credit to all schools regardless of the number of authors on the

paper or number of authors from each school (a modified version of the “normal count” method

of Lindsey, 1980). We made no attempt to exclude publications by faculty from outside the

9

business school; our measure is of research published in major business research journals,

regardless of whether the author(s) are from within the business school or not (see Niemi, 1988).

To avoid overemphasizing some disciplines and underemphasizing others, we

standardized the number of pages across disciplines so that the number of pages published within

each discipline was weighted by the number of faculty in that discipline. We first divided the

number of pages published in each journal by the total number of pages published in that

discipline to convert the page counts into percentages of pages within discipline, and then

multiplied by the number of faculty in that discipline as reported by AACSB (1998) (see Table 1

for the number of faculty). The results were then summed for each school.

To assess the stability of the standardized, weighted pages published over time, we

performed a Cronbach alpha analysis, treating each year as separate measure. The alpha was .92

indicating reliability over time. To assess the extent to which the measure is significantly

influenced by the specific journals included in the sample, we calculated Cronbach alpha (.97)

for the entire set of journals (i.e., treating each journal as a separate item). We then omitted each

of the individual journals in turn and recalculated the alpha for the remaining journals. In all

cases, the overall alpha remained .97, suggesting that the overall measure of research

effectiveness is quite robust with respect to which top journals are included. We also performed

a Cronbach alpha analysis on the journals within each discipline group (accounting, marketing,

etc.) except for the production/ operations management group (which had only one journal).

Alphas were acceptably strong (i.e., above .80) for all but the insurance/international

business/real estate group (.67) and the management information systems group (.73).

MBA Program Performance Measure

We used the U. S. News and World Report MBA program evaluations because they are

published annually and rank more MBA programs (the top 50). U.S. News reports a continuous

score (and a ranking based on the score) which is composed of an opinion survey of business

school deans and graduate coordinators (25% weighting), an opinion survey of corporate

recruiters (15%), the median salary of MBA graduates (14%), proportion of MBAs employed

10

upon graduation (7%), proportion employed after three months (12.25%), the ratio of recruiting

firms to graduates (1.75%), GMAT test scores (16.25%), undergraduate grade point average

(7.5%), and the proportion of applicants accepted (1.25%).

Our statistical analyses focus on the MBA rankings from 1995-1999. Over this five year

time period, 49 business schools were ranked in the top fifty in at least four of these five years.

We chose these 49 schools for our analyses. To assess the stability of the MBA program score

over time, we performed a Cronbach alpha analysis, treating each year in our analyses (1995

through 1999) as a separate observation; an MBA score of 10% less than the lowest score in that

year was assigned for any “missing” data (i.e., those schools not ranked in the top 50 for a given

year). The Cronbach alpha was .98, indicating high reliability over time.

Potential Factors Affecting Performance

We measured in-house editorships as the number of years (or part years) during 1993-

1997 that any school in the sample was home to an editor or associate editor for any of the

sampled journals (we did not include editorial board memberships, or if a journal had more than

three associate editors, the associate editorships). We used this time period because of the time

lag between the editorial acceptance of an article and its publication (Stahl, et al., 1988). This

was standardized and weighted in the same manner as the number of pages published.

Following Stahl, et al., (1988), we contacted AACSB to obtain the number of tenure-

track research faculty in each rank at each business school for 1996 (AACSB, 1997a). This

number excludes visiting faculty, non-tenure-track faculty (e.g., lecturers, executive professors),

faculty holding administrative appointments (e.g., deans) and emeritus faculty. It is important to

note that faculty size data for some schools include economics departments and other

departments such as business law, public policy and so on, while in other cases it does not; thus,

results based on the faculty size data need to be interpreted with caution. BusinessWeek (1998b)

provided budget data for 47 of the 49 schools (BusinessWeek obtained the budget data by

surveying deans and MBA program directors in 1997, requesting 1996-97 data). Because total

budget was significantly related to the number of faculty (r=.52, p=.001), we used budget per

11

faculty member as our measure of financial resources.

Analyses

Our analyses compare MBA program performance and business research performance.

While research performance data (standardized, weighted pages published and the resulting

rankings) are available for all schools (because one can see the extent to which any given school

has published in any time period), MBA performance data (scores and rankings) are only

available for the top 50 schools in any given year. Therefore, our analyses focus only on the last

five years, and only on the 49 top MBA schools whose performance has been almost consistently

rated over this time period. Because MBA program data are only available on the those schools

that were ranked in the top 50 for a given year, this analysis suffers from reduced variability in

the measures because we are forced to sample based on the dependent measure (see Hannan &

Carroll, 1995).

We conducted two separate analyses to test H1: 1) on the rank data; 2) on the underlying

continuous data that produced the rank data. To test the rank data, which are the primary popular

press measure, we first calculated an average MBA rank for each school over the 1995-1999

time period (using a value of 55 for any year in which a school was not ranked in the top 50) and

then converted the average into a rank from 1 to 49. We then calculated a research performance

ranking for each of the schools by calculating the total number of standardized, weighted pages

published from 1994-1998 and converting this into a rank. We choose this time period to

correspond with the 1995-1999 time period for the MBA performance because the MBA

performance data published in each year are gathered based on performance in the preceding

year (i.e., MBA performance data published in 1999 are based on performance in 1998). We

used the Wilcoxon matched pairs signed rank test (Siegel, 1956) which does not assume that the

distribution is normal or that the data are interval scaled to test whether the two sets of ranks

were significantly different.

We then examined the underlying continuous measures. We calculated the average MBA

program score for each school, using a value of 10% more than the lowest score in that year for

12

any school that was not rated. One might be tempted to examine the correlation between MBA

program score and research pages published, but our question is not whether the two measures

are related, but whether they are the same. Instead, we computed Z-scores for the average MBA

score and for the standardized, weighted pages published (both using just the 49 schools in our

data set), and then calculated the absolute difference between the two Z-scores for each school.

We used a t-test to test whether the mean absolute Z-score was significantly different from zero.

We used a LISREL model to test the effects of different factors on MBA and research

performance (H2a, H3, H4a, H4b, H4c), using only the 47 schools for which we had budget data.

We used the average MBA score (not rank) and the number of standardized, weighted pages

published as dependent variables, and budget per faculty member, total number of faculty,

proportion of assistant and full professors, weighted years of editorships and the public-private

status of the school as independent variables.

To test H2b, we first computed Z-scores for the average MBA score and Z-scores for the

standardized, weighted pages published by each of the 47 schools for which we had budget data.

We then calculated the absolute difference between the research Z-score and the MBA Z-score

for each school and used a linear regression to see if there was a relationship between the

absolute difference and the budget per faculty member (including the same five other factors

used in the LISREL model above as covariates).

To test H2c, we first computed Z-scores for the standardized, weighted of pages

published within each discipline by each of the 47 schools for which we had budget data. We

then calculated the sum of the absolute differences between a school’s average discipline Z-score

and each of its eight discipline-specific Z-scores and performed a linear regression to see if there

was a relationship between the absolute difference and the budget per faculty member (including

the other five factors used in the LISREL model above as covariates).

RESULTS

The top 50 universities in research performance over the entire 13-year period are listed

in Table 2, along with their mean MBA program performance from 1995-1999. (Research

13

rankings for the Top 100 schools are on the Web at: www.kelley.indiana.edu/ardennis/rankings).

Over the 13-year time period, 431 universities published 213,158 pages in these 20 top tier

journals (adjusted for the average size of an AMJ page and crediting multi-authored articles to

all schools). The top ten ranked schools accounted for 25.5% of the total pages in these journals,

the top 25 schools accounted for 48.0%, the top 50 for 69.5%, and the top 100 for 87.2%.

[ insert Table 2 about here ]

MBA-Program vs. Research Rankings

As shown in Table 2, 19 of the top 50 research schools have not been consistently ranked

in the top 50 MBA programs. The Wilcoxon test on the rank data found significant differences

between the research rankings and MBA rankings (t(48)=2.32, p<.05). The t-test on the mean

absolute difference between the two underlying continuous measures of performance found

significant differences (t(48)=8.44, p<.001), although the two are significantly correlated (r=.57,

p<.001). We conclude that MBA-program performance measures and research performance

measures are distinct; H1 is supported.

Factors Affecting Performance

The results of the LISREL analysis on the factors affecting performance is presented in

Figure 1. While the goodness of fit and χ2 statistics show that the model fits the data, we should

caution readers that it is built using only a small sample and thus may suffer from the usual

problems of small sample sizes (some insignificant effects may have been found to be significant

with a larger sample). We are also sampling based on a dependent variable (MBA performance)

which may limit the observed variability in the sample, also reducing power. Although there is a

significant correlation between the MBA score and the research pages published, the LISREL

model shows that neither MBA score nor pages published affects the other measure, after

accounting for the effects of the other factors. In short, any relationship between the two

measures is due to the effects of and relationships among underlying factors, not to direct effects.

[ insert Figure 1 about here ]

The number of research pages published is positively affected by the total number of

14

faculty, the proportion of full professors, the proportion of assistant professors and the number of

years of editorships. The MBA score is positively affected by the budget per faculty member,

and the proportion of full professors. Thus H2a is partially supported, H3 is not supported, H4a

is supported, H4b is partially supported, and H4c is supported.

Factors Affecting Variability in Performance

H2b, which argued that the difference between research performance and MBA

performance would be reduced for schools with greater financial resources, was not supported.

The linear regression found that the absolute difference between the measures for each school

was affected only by the number of faculty (t(40)=2.07, p<.05); as the number of faculty

increased, the absolute difference increased. We repeated the analysis, but this time using the

simple difference between the Z-scores, not the absolute difference, and found that as the number

of faculty increased, the difference favored research performance (t(40)=4.66, p<.001) but as the

budget per faculty member increased, the difference favored MBA performance (t(40)=2.51,

p<.05). In short, larger schools tended to emphasize research performance, while richer schools

tended to emphasize MBA performance; or schools that excel in research tend to acquire more

faculty, while those which excel in MBA education tend to gather more financial resources per

faculty member.

H2c, which argued that there would be fewer performance differences among disciplines

within schools with greater financial resources, was not supported. The linear regression found

that the total absolute difference between the discipline research performance measures for each

school increased as the number of faculty increased (t(40)=3.27, p<.01) and as the budget per

faculty member increased (t(40)=2.49, p<.05). That is, there was a significant relationship

between financial resources and the variability in performance across disciplines, but it was the

opposite of what we hypothesized; richer schools displayed greater differences among disciplines.

Table 3, which presents the within discipline rankings for each of the top fifty research

schools, shows that three of the top 10 research schools (Columbia, Chicago, and Northwestern)

have no publications in the selected MIS and POM journals. These three schools are also among

15

the top five schools in budget per faculty member. We performed a exploratory factor analysis

treating each discipline as a separate item, and found the disciplines to load onto two factors, one

for MIS and POM (with correlations of .60 and .78, respectively), and one for the other six

disciplines (with correlations ranging from .68 to .89).

[ insert Table 3 about here ]

DISCUSSION

First and foremost, the results indicate that the popular press measure of MBA program

performance is significantly different from our measure of business research performance. MBA

program rankings were linked to budget per faculty member, and to a lesser extent to the

proportion of full professors. It is not surprising that financial resources affect MBA performance

suggesting perhaps that this performance enables schools to acquire the financial resources

needed to reinforce the MBA program (D’Aveni, 1996). Some resources such as specialized

placement offices and staff, classroom facilities, alumni relations, and MBA program marketing

are uniquely associated with the MBA program, and schools with larger budgets can invest in

those resources to improve their MBA performance without affecting research performance.

Larger schools tend to produce more research because faculty are the production engine

for research, and there are limits as to how much research any one faculty member can

consistently produce. We had argued that a greater proportion of full professors might be

associated with poorer performance (due to a shift in their attention away from research), but the

data suggest the opposite: more full professors are linked to better research performance. This

suggests that those faculty who continue to publish after tenure and are promoted to full

professor also tend to continue publishing after promotion to full professor, either because of the

added incentives of professorships and chairs, or because the promotion and tenure system has

successfully identified those faculty who are productive researchers. One might also argue that

the presence of more full professors may provide additional support and encouragement so that

assistant and associate professor publish more.

Our results indicate that research performance is not consistent across disciplines within a

16

school. With the exception of Michigan, all of the top 10 schools have at least one discipline

(usually three) in which they are not ranked in the top 20. Implicitly or explicitly, most schools

tend to focus their research strength in a few disciplines, rather than spread efforts more evenly

across disciplines. Contrary to our hypothesis, this focus was most clear among schools with

greater financial resources: richer schools tended to have a greater focus. We speculate that

increased financial resources enable deans to shift resources to build a strength in a few areas,

and/or that focusing on a few areas creates a salient institutional identify that attracts

endowments and faculty to those areas, both of which are self-reinforcing.

Understanding how schools develop specific foci is an important issue for future

research. Perhaps different sub-cultures that place different values on exploration and

exploitation emerge within schools and constrain efforts (March & Sutton, 1997). Perhaps the

proximity of schools to constituencies may affect their focus (e.g., schools who serve Wall Street

may emphasize finance, while schools who serve manufacturing emphasize production). As an

aside, we note that there were significant correlations between budget per faculty and research

performance in accounting and finance -- two disciplines with the highest mean salary --

suggesting that increased wealth enabled a focus on "expensive" disciplines. Our exploratory

factor analysis suggests that performance in MIS and POM may be different from performance

in the more "established" disciplines. It is unclear whether this may reflect explicit strategic

decision making, a slower acceptance of on MIS and POM in general, or a slower acceptance of

the need for change to include these disciplines at the most successful schools.

More research is clearly warranted to understand how some universities have managed to

significantly improve their research performance (e.g., Texas, Harvard, Penn State, Vanderbilt),

when many of the factors affecting performance are hard to manage. While deans can affect

performance by increasing the number of faculty, they have less control over the proportion of

full professors or editorships. We speculate that internal policies that focus on serving the

research constituency are important in motivating faculty. Future research should also explore

whether schools with high rankings across both MBA program and business research measures

17

accomplish this via constituency specialization among faculty (e.g., teachers vs. researchers), or

by making it possible for individual faculty to address multiple constituencies more effectively,

or by focusing within constituencies (e.g., many private schools do not have large undergraduate

programs as do public universities; the role of Ph.D. programs may also warrant future research).

Finally, about 70% of the research in the leading business journals (by our measure) was

produced by the top 50 research schools. In other words, the production of business research is a

highly concentrated "industry," with these 50 schools forming the core of what the Economist

(1997) calls the "global knowledge factory" for business research. If these 50 schools were to

focus unduly on knowledge exploitation rather than knowledge exploration, the global

production of business research would likely be reduced. The value of business research is

debatable, but in the past, business research has transformed markets (e.g., the Black-Scholes

option-pricing model that created the derivatives industry (AACSB, 1997b)).

Our study has an important implication for business school deans and university

presidents. The finding that measures of research performance and MBA-program performance

are different, suggests that focusing only on one of these measures (e.g., popular press MBA

program rankings) risks “perverse learning” and an imbalance in the tension between exploration

and exploitation, possibly threatening a school’s long-term viability -- and the viability of the

global knowledge factory. In contrast, a “balanced scorecard” (Kaplan & Norton, 1992)

approach to assessing organizational performance uses multiple indicators of performance in

order to ensure a healthy balance among multiple-constituency demands. In the short-run, re-

emphasizing the multiple-constituency nature of business school performance in this way might

decrease a school’s performance on some dimensions (e.g., MBA rankings) as efforts are

redirected toward neglected constituencies. However, such short-term losses may be minimized

by identifying ways in which resources can be applied to serve both constituencies.

The tension in serving multiple constituencies has a long history in the business school

(AACSB, 1997b). For much of 20th century, most business schools focused on undergraduate

instruction and very practical applied research. The Carnegie report (Pierson, 1959), which was

18

highly critical of this exploitation focus, provoked a sharp turn to a focus on research,

particularly "scientific" research. The Porter-McKibbin report (1988), which was highly critical

of theoretical research, coupled with the rise of popular press MBA program rankings, has, we

believe, driven the focus back to MBA instruction at the possible expense of research (cf.,

AACSB, 1997b). Our results show that over the past five years, schools with greater financial

resources have tended to emphasize MBA program performance over research performance.

While the popular press has highlighted one facet of performance in business schools, it

may have done so at the expense of a healthy tension between exploitation and the critical

exploration role also performed by these schools. Published rankings create status, and as noted

by D’Aveni (1996), status has self-reinforcing effects on organizational performance. However,

if the measures of performance ignore or overlook critical constituencies and dimensions of

performance (e.g., research performance), then those self-reinforcing influences on performance

will be similarly misdirected. We believe that a more balanced measure that considers both MBA

program and research performance -- and understands the healthy tension between them -- is

needed to help ensure that the lessons learned from published rankings are productive rather than

perverse.

19

TABLE 1

Journals Studied (and Number of Faculty) in Each Discipline Area

Accounting (3,336a)

• Accounting Review

• J of Accounting & Economics

• J of Accounting Research

Managementc (3,457)

• Administrative Sciences Quarterly

• Academy of Management Journal

• Academy of Management Review

• Strategic Management Journal

Finance (2,409)

• J of Finance

• J of Financial Economics

Management Information Systems (1,627)

• Information Systems Research

• MIS Quarterly

Insurance, International Business &

Real Estate (584)

• J of Risk and Insurance

• J of International Business Studies

• Real Estate Economicsb

Marketing (2,432)

• J of Consumer Research

• J of Marketing

• J of Marketing Research

Management Science (752)

• Management Science

• Operations Research

Production/Operations Mgmt (877)

• J of Operations Management

a) Numbers in parentheses are the total number of faculty in these areas from AACSB (1998).

b) Prior to 1995, this was entitled Journal of the American Real Estate and Urban Economics

Association.

c) Includes management, strategy, human resource management, and organizational behavior.

20

TABLE 2

Business School Research Rankings

Research Rank Research

Rank

1986-98

University

Percent of

Weighted

Pages 1986-89 1990-93 1994-98

Average

MBA Rank

1995-99

1 Pennsylvania 3.71% 1 1 1 2

2 Michigan 3.03% 4 2 2 10

3 Stanford 2.59% 2 3 7 1

4 New York University 2.55% 5 4 5 14

5 Chicago 2.45% 7 5 3 6

6 Columbia 2.37% 3 10 6 7

7 Minnesota 2.33% 6 6 10 29

8↑ Texas - Austin 2.29% 14 7 4 19

9↑ Harvard 2.12% 20 8 8 3

10 Northwestern 2.05% 9 9 9 5

11 Washington - Seattle 1.72% 8 18 14 49*

12 Southern California 1.72% 18 15 11 27

13 Massachusetts Inst Tech 1.69% 13 11 18 4

14 Illinois 1.68% 12 14 13 40*

15↓ California - Berkeley 1.64% 10 12 19 12

16↓ California - Los Angeles 1.55% 11 13 22 13

17↑ Cornell 1.54% 24 20 12 18

18↓ Ohio State 1.45% 16 17 26 24

19 Duke 1.43% 15 19 20 9

20↑ Wisconsin - Madison 1.42% 26 22 17 *

21 Indiana 1.39% 22 21 21 20

22↓ Arizona 1.38% 17 27 27 38

23↑ UNC - Chapel Hill 1.34% 29 30 15 17

24 Carnegie Mellon 1.30% 28 23 24 15

21

25↑ Penn State 1.28% 36 26 16 35

26 Texas A & M 1.27% 23 29 25 45

27↓ Florida 1.25% 19 25 31 43*

28↑ Arizona State 1.25% 30 24 23 34

29↓ Rochester 1.22% 25 16 36 22

30 South Carolina 1.19% 27 32 30

31↓ Iowa 1.08% 21 39 32

32 Purdue 1.05% 33 33 29 21

33↑ Georgia 0.98% 43 37 28 41*

34↑ Rutgers 0.95% 46 28 34

35 Michigan State 0.89% 31 48 33 32

36 Pittsburgh 0.86% 32 41 38 46*

37 Southern Methodist 0.84% 38 45 35 *

38 Colorado - Boulder 0.83% 47 31 43

39↑ California - Irvine 0.75% 82 34 37

40↓ Cincinnati 0.68% 35 38 54

41↑ Georgia State 0.68% 64 52 41

42↑ Notre Dame 0.67% 55 50 40 *

43↓ Dartmouth 0.67% 39 35 56 8

44↑ Maryland - College Park 0.65% 50 58 42 25

45 Washington - St. Louis 0.64% 37 77 39 25

46↓ Utah 0.63% 34 43 64

47↓ Louisiana State 0.62% 40 40 57

48 Houston 0.62% 56 36 60

49↑ Vanderbilt 0.61% 65 47 44 30

50↓ Florida State 0.59% 50 56 58

* Not consistently ranked in top 50 by U.S. News & World Report; numeric rankings calculated for those missing

only one year (with missing year counted as a rank of 55). A blank indicates not ranked in top 50.

↑ and ↓ show schools whose research rankings rose or fell by 8 or more positions over the time period.

22

TABLE 3

Within-Discipline Research Rankings

Rankings within Disciplines Rank

1986-

1997

University Acc Fin Ins, IB

& RE

Mgt MIS Mgt

Sci

Mkt POM

1 Pennsylvania 1 2 1 3 34 3 1 38

2 Michigan 2 5 19 2 17 8 12 15

3 Stanford 4 19 81 1 35 4 14 55

4 New York University 20 3 10 4 9 26 6 56

5 Chicago 3 1 117 41 21 8

6 Columbia 10 8 38 5 2 5

7 Minnesota 22 34 17 7 1 24 9 1

8 Texas - Austin 13 15 3 20 3 5 3 41

9 Harvard 16 4 49 8 7 18 25 26

10 Northwestern 7 9 156 6 11 10

11 Washington - Seattle 5 16 50 15 16 22 44 39

12 Southern California 14 11 17 24 29 13 25

13 Massachusetts Inst Tech 37 14 52 27 2 1 32 107

14 Illinois 21 18 7 9 28 32 17 16

15 California - Berkeley 11 20 13 11 112 12 24 111

16 California - Los Angeles 23 7 27 24 20 15 18 89

17 Cornell 12 13 61 13 142 17 37 73

18 Ohio State 26 6 21 36 52 14 30 6

19 Duke 19 10 78 37 7 11 23

20 Wisconsin - Madison 33 25 5 25 107 25 4 11

21 Indiana 34 17 12 18 30 43 16 8

22 Arizona 18 28 230 33 10 36 7 97

23 UNC - Chapel Hill 8 38 11 22 45 19 62 9

24 Carnegie Mellon 25 31 19 5 6 73 34

23

25 Penn State 30 32 20 14 37 31 23 30

26 Texas A & M 50 62 39 10 25 58 19 7

27 Florida 17 29 24 91 126 39 2 32

28 Arizona State 28 24 60 21 69 51 15 12

29 Rochester 6 12 79 198 29 23 87 44

30 South Carolina 74 80 2 26 12 59 21 3

31 Iowa 9 22 84 54 129 40 34 103

32 Purdue 40 23 53 12 64 9 67

33 Georgia 31 39 6 66 6 52 49 28

34 Rutgers 46 42 8 23 62 20 20

35 Michigan State 35 61 23 30 128 43 2

36 Pittsburgh 47 48 87 44 8 35 27

37 Southern Methodist 48 27 46 43 13 82 36

38 Colorado - Boulder 24 73 92 38 19 46 35 107

39 California - Irvine 97 41 48 28 14 38 38

40 Cincinnati 100 185 15 40 55 76 28 4

41 Georgia State 68 66 9 58 4 110 55 98

42 Notre Dame 36 26 101 69 176 42 31

43 Dartmouth 72 49 45 16 55 51 70

44 Maryland - College Park 67 40 70 31 18 34 60

45 Washington - St. Louis 15 58 65 95 53 19

46 Utah 73 30 71 87 93 22 14

47 Louisiana State 66 35 16 34 116 61 39 79

48 Houston 52 53 51 68 23 63 48 22

49 Vanderbilt 60 33 235 63 105 48 31 27

50 Florida State 38 90 34 72 15 62 85 36

A blank indicates that the university has published no research in this discipline's selected journals.

24

FIGURE 1

LISREL Results for Factors Affecting the Rankings

ResearchPages

MBAScore

.12.29

Full Prof %

.37*

.25*

TotalFaculty

Budget per Capita

Privatevs.

Public

Asst Prof %

Editors

.27*

.24**

.54***

.09

-.12

.05

.55***

.00

Note:Darker lines indicate significant relationships* p < .05 ** p < .01*** p<.001

Squared Multiple Correlations: Research Pages: .71; MBA Score: .66

χ32=0.30, p=0.96

Root Mean Square Residual (RMR) = 0.0045

Goodness of Fit Index (GFI)= 1.00

Adjusted Goodness of Fit Index (AGFI)= 0.98

25

References

AACSB. 1997a. Response to Information Request on Faculty Composition. American

Association of Collegiate Schools of Business.

AACSB. 1997b. “Research Relevance Remains a Challenge for Business Schools,” AACSB

Newsline, 27(3): 1-5.

AACSB. 1998. Faculty Positions Filled, Vacant, and Planned Growth. Newsline 28, 3, 27.

Armstrong, J. S. & Sperry, T. 1994. The Ombudsman: Business School Prestige -- Research

versus Teaching. Interfaces, 24: 2, 13-43.

BusinessWeek, 1998a. "The Best B-Schools," October, 19, 86-94

BusinessWeek, 1998b. "1998/99 B-School Budget" accessed 10/25/99,

http://www.businessweek.com/bschools/faqsnfigs/index.htm

Byrne, J. A., 1990. Is research in the ivory tower ‘fuzzy, irrelevant, pretentious’?,

D’Aveni, R.A. 1996. A Multiple-constituency, Status-based Approach to Interorganizational

Mobility of Faculty and Input-output competition among top business schools.

Organization Science, 7, 166-189.

Economist. 1997. Inside the Knowledge Factory, October 4.

Glick, W., McKelvey, W., Cooper, M., Huber, G., & Zmud, W. 1997. Survey Feedback from

Surveys of Journal Reputations, Report of the 1997 INFORMS Committee Review of

Organization Science.

Hannan, M. T. & Carroll, G. R. (eds.) (1995) Organizations in industry : strategy, structure, and

selection, New York, Oxford University Press.

Henry, W. R. & Burch, E. E. 1974. Institutional Contributions to Scholarly Journals of Business.

Journal of Business, 47, 56-66.

Hollman, K. W., Murrey, J. H., Jr., & Homaifar, G. 1991. The structure and disciplinary

boundaries of insurance: A citational analysis of JRI articles, The Journal of Risk and

Insurance 58, 714-721.

26

Kaplan, R.S. & Norton, D.P. (1992) The Balanced Scorecard – Measures that Drive

Performance. Harvard Business Review, January/February, 71-79.

Kirkpatrick, S. A. & Locke, E. A. 1992. The Development of Measures of Faculty Scholarship.

Group & Organizational Management, 17, 5-23.

Lindsey, D. 1980. Production and Citation Measure in the Sociology of Science: The Problem of

Multiple Authorships. Social Studies of Science. 10: 145-162.

Mansfield, E. 1987. Statistics for Business and Economics, 3rd edition, W.W. Norton &

Company, New York.

March J.G. 1991. Exploration and exploitation in organizational learning. Organization

Science. 2, 71-87.

March, J.G. 1996. Continuity and change in theories of organizational action. Administrative

Sciences Quarterly, 41, 278-287.

March, J.G. & Sutton, R. I. 1997. Organizational performance as a dependent variable.

Organization Science, 8, 698-706.

McDowell, J. M. & Amacher, R. C. 1986. Economic Value of an In-House Editorship. Public

Choice, 48, 101-112.

Meyer, M.W., & Gupta, V. 1994 The performance paradox. In B.M. Staw & L.L. Cummings

(Eds.), Research in Organization Behavior, 16, 309-369.

Northcraft, G. B., Neale, M. A., & Huber, V. L. 1988. Decision bias and social influence in

human resource decision making. In K. Rowland & G. Ferris eds. Research in Personnel

and Human Resource Management. Greenwich, CT: JAI Press, 157-189.

Niemi, A. W., 1987. Institutional contributions to the leading finance journals, 1975 through

1986: A note, The Journal of Finance 42, 1389-1397.

Niemi, A. W., 1988. Research productivity of American business schools, 1975-85, Review of

Business and Economic Research, Spring 1988, 1-17.

Pierson, F.C. 1959. The Education of American Businessmen: A Study of University-College

Programs in Business Administration, The Carnegie Series in American Education, New

27

York: McGraw-Hill

Porter, L. & McKibbin, L. 1988. Management Education and Development: Drift or Thrust into

the 21st Century? New York: McGraw-Hill.

Reeve, R. C. & Hutchinson, P. J. 1988. The Contribution of Non-U.S. Institutions to Academic

Accounting Journals. Abacus 24, 1 90-94.

Siegal, S. Nonparametric Statistics for the Behavioral Sciences. New York: McGraw Hill.

Southern Methodist University 1997. Edwin L. Cox School of Business Annual Report, Dallas,

Texas.

Stahl, M. J., Leap, T. L. & Wei, Z. Z. 1988. Publication in leading management journals as a

measure of institutional research productivity, Academy of Management Journal 31, 707-

720.

Starbuck, W. 1997. Estimates of the 1997 Impact Ratings. Unpublished Web document.

http://www.stern.nyu.edu/~wstarbuc/cites.html and

http://www.stern.nyu.edu/~wstarbuc/whatmean.html

Tsui, A.S. (1990) A Multiple-constituency Model of Effectiveness: An Empirical Examination

at the Human Resource Subunit Level. Administrative Sciences Quarterly, 35, 458-483.

U.S. News and World Report, (1999) “Schools of Business,” America’s Best Graduate Schools,

special issue, March 29, 86-87.

Zivney, T. L. & Bertin, W. J. 1992. Publish or perish: What the competition is really doing. The

Journal of Finance 47, 295-329.

28

JAMES S. TRIESCHMANN holds the Dudley L. Moore, Jr. Chair of Insurance in the Terry College of Business at the University of Georgia and has served as Associate Dean for Faculty and Research. ALAN R. DENNIS holds John T. Chambers Chair of Information Systems in the Kelley School of Business at Indiana University. His current research interests include the use of computer technologies to support group brainstorming, decision making, and collaborative work. GREGORY B. NORTHCRAFT is the Harry J. Gray Professor of Executive Leadership in the Department of Business Administration, and Labor and Industrial Relations, at the University of Illinois. His major research interests include collaboration in teams, conflict management, managerial decision making, and employee motivation and job design, particularly in high-technology manufacturing settings. He is currently Editor of the Academy of Management Journal. ALBERT W. NIEMI, JR. is Dean and holds the Tolleson Chair in Business Leadership in the Cox Graduate School of Management at Southern Methodist University.