2012. Vol.3, No.2, 224-227
Published Online April 2012 in SciRes (http://www.SciRP.org/journal/ce) http://dx.doi.org/10.4236/ce.2012.32035
Copyright © 2012 SciRes.
The Relationship between Institutional Efficiency and
Instructional Quality in Higher Education
University of Central Oklahoma, Edmond, USA
Received March 4th, 2012; revised April 2nd, 2012; accepted April 12th, 2012
Wage setting methodologies for university faculty may be merit/market based or administered. Failure to
exploit the fact that faculty productivity depends on abilities and wages results in inefficient use of uni-
versity budgets. If such inefficiencies exist it suggests suboptimal productivity of the existing faculty and
the inability to attract new qualified faculty. As motivation for this analysis, a simple model of university
faculty “output” maximization is presented. Efficient budget allocation requires that faculty compensation
be structured so that marginal productivities are equated across faculty. This paper examines and com-
pares the efficiency of several regional universities in the US, identified as “peers”, employing the Data
Envelopment Analysis (DEA) estimation method. The results suggest the existence of inefficiencies and
more notably, that the homogeneity assumption regarding the peers is questionable.
Keywords: Salary; Efficiency; DEA; Higher Education; Education Administration
Introduction and Related Literature
The topic of university faculty salaries has been addressed
frequently through research with varying focal points of interest.
Cohn (1973) develops a multiple regression model to analyze
the factors that might have an effect on faculty salaries. Al-
though the empirical results of this study are dated and of lim-
ited applicability to the current economic situation of university
professors and their pay, many of the model’s independent
variables that are identified still have explanatory power today.
These include the type of institution, whether it is public or
private, geographic location, quality measures, institution size,
and state per capita income. The variable that Cohn finds to
have the greatest effect on faculty salaries is that of institution
Tuckman and Tuckman (1976) also use regression methods
“to analyze the determinants of salary structure at American
universities”. The explanatory variables of their model empha-
size the rate and amount that faculty publish, faculty personal
characteristics such as gender and age, the university’s geo-
graphic location, and the area of faculty expertise. Two of the
more definitive outcomes are that publishing and research raise
salaries compared to teaching and that academic field correlates
to a significant variation in faculty salaries.
Hoenack (1982) uses a theoretical approach to analyze how
prices affect the choices made by faculty, students, legislators,
and others involved in higher education with regard to the effi-
cient use of resources. He concludes that inefficiency results
from the existing prices, and he suggests how efficiency could
be improved through changes in prices.
Alexander (2001) analyzes the impact that the growing dis-
parity between faculty salaries at private universities compared
to those at public universities is having on the ability of public
institutions to attract and retain top-notch faculty. He describes
the potential for the development of two separate higher educa-
tion systems, one private and the other public. As private uni-
versities are better able to compete financially, there exists a
possible “brain drain” as higher quality faculty migrate toward
better salaries. The differential between faculty salaries at pub-
lic and private universities is reported by Smallwood (2006)
from a survey conducted by the College and University Profes-
sional Association (CUPA) for Human Resources. This survey
reveals that salaries for 2005-2006 increased by 3.7 percent at
private universities compared to only 3.1 percent at public uni-
In a series of articles contained in the ASHE-ERIC Higher
Education Report (2001) several aspects of faculty compensa-
tion systems are addressed. The first article describes how the
quality of an institution is related to how well the institution
achieves its self-stated mission. Its mission is heavily depend-
ent on the faculty, which in turn is strongly affected by the
faculty compensation system. Faculty compensation is deter-
mined by a variety of factors, both external and internal. Nota-
bly, one of the internal factors cited is “market pay in the disci-
pline”. The second article describes the three main types of
faculty compensation systems: 1) the contract salary system or
merit pay; 2) the single salary system; and 3) nontraditional
faculty compensation systems. The third article describes the
commonly used arguments in favor of either merit pay or the
single salary system. The fourth article describes the advan-
tages and disadvantages of the three compensation systems
mentioned above. Interestingly, one of the primary disadvan-
tages of the single salary compensation system is listed as “a
lack of efficiency in the use of human resources”.
With regard to the methods that have been developed and
utilized to measure technical and/or allocative inefficiency, the
literature is replete with a history of scholarly contributions.
Some of the more notable and most frequently cited of these
include; Farrell (1957), Aigner et al., (1977), Kumbhakar and
Wang (2006) among others.
This paper first describes a simple microeconomic model of
faculty production and budget efficiency. It then presents some
preliminary efficiency results for a select group of Finance
Departments at peer universities.
This model is based on the discussion in Rassouli-Currier
and Currier (2008). For a given faculty member, let y denote
“productivity”, i.e., number of publications, and let x = (x1, ···,
xn) denote a vector of university expenditures on productivity
enhancement. The vector x could include expenditures on
wages and salaries, technology support, library facilities, con-
ference travel support etc. The individual faculty member’s
output depends on effort e and the vector x, as summarized by
the “production function”
,yfxe This production func-
tion is an increasing function of e and xi, i = 1, ···, n. In addition,
the faculty member’s utility is
,Uxy e reflecting the “dis-
utility of effort”. We assume that a (small) increase in any xi
will increase the marginal benefit of an increase in y and reduce
the additional effort necessary to achieve this increase in y.
Given the vector x, the faculty member selects effort (and cor-
responding output) to:
Subject to ,
This is illustrated in Figure 1 for the case of n = 1. Utility
increases as we move downward and to the right in the diagram.
The indifference curves are concave, reflecting the assumption
that a faculty member’s willingness to exert additional effort to
increase productivity is highest when effort is low.
For any vector x, individual utility maximization implies so-
yyx. Assuming appropriate dif-
ferentiability, we have 0
yx, i = 1, ···, n. Thus, a ceteris
paribus increase in any xi will increase faculty productivity
(output). Suppose now that n = 1, there are m individual faculty
members and that the university has overall resources (e.g.,
salary budget) of $B. Then the university budget constraint is
. Under an administered salary program, it is es-
sentially the case that j
, j = 1, ···, m. Thus, differences
in rank not withstanding, each faculty member receives the
same fraction of the total budget allocation. Faculty member j’s
output will be j
with aggregate faculty output
Alternatively, suppose that salaries are set in such a way as to
maximize total faculty output, given the university’s overall
budget constraint. In this case, the university administration
selects individual salaries to solve:
yielding the vector of optimal salaries
corresponding faculty output levels Since
salaries are chosen to maximize total faculty output, it must be
the case that **
This is illus-
trated in Figure 2 where it is shown that selection of individual
salary levels on the basis of the optimality criterion permits the
university to operate on the “efficient” production frontier E
whereas any other non-optimal salary assignment rule, meas-
ured by the number of faculty, percentage faculty holding PhD
or equivalent and percentage of the tenured/on tenure-track
faculty, forces the university below the efficient frontier.
The Data Set
The main source of data was the Finance Departments of 27
universities and colleges that were identified as the peers to
University of Central Oklahoma in 2006 (the choice of Finance
Department was arbitrary). However, the data collecting phase
turned out to be extremely challenging. Universities’ published
data on faculty salary and their contribution to the department is
either aggregated at the college/university level or not well
documented. In addition the CUPA website was not particularly
helpful. Ultimately, data was collected on several variables of
interest for 17 out of the 27 institutions. The list of variables
and their summary statistics are presented in Table 1.
DEA Estimation Method
The basic idea of the DEA approach is to view universities as
Decision Making Units (DMUs) with multiple inputs and out-
puts. DEA is a non-parametric productivity analysis model that,
unlike its parametric counter-part, allows for multiple inputs
and outputs to be considered simultaneously. In addition, it
Efficient production frontier.
Copyright © 2012 SciRes. 225
Summary statistics for the variables (n = 17).
Variable Mean SD Minimum Maximum Sum
RA/AB 42.00 40.16 2.00 134.00 714.00
OPA 32.41 28.83 1.00 96.00 551.00
TEACH 8.82 3.81 4.00 17.00 150.00
PHD 0.80 0.14 0.53 1.00 13.53
TT 0.78 0.21 0.25 1.00 13.21
Where: Y1 = RA/AB is the number of refereed articles/authored books; Y2 = OPA
is other professional activities such as published articles of merit, working papers,
professional presentations etc; X1 = TEACH is the number of teaching staff; X2 =
PHD is the percentage of the faculty holding Ph. Ds or equivalent; X3 = TT is the
percentage of the faculty that is tenured or is on tenure track.
does not require any restrictive assumptions regarding the func-
tional form of the model.
On the down side, DEA assumes that all DMUs have the
same deterministic (as opposed to stochastic) production fron-
tier and that any deviation from the frontier is due to ineffi-
ciency, which may not be realistic. However, it is a reasonably
powerful diagnostic tool that can be used to measure the effi-
ciency of a set of homogenous DMUs individually (e.g., peer
universities) relative to the most efficient unit. Studying the
reasons for any possible inefficiencies and finding remedies to
eliminate them is the responsibility of the unit’s decision maker
Following Coelli et al., (1998) in the DEA method, the tech-
nical efficiency is identified as a proportional increase in the
output vector with a given input vector. Therefore, the out-
put-oriented measure of technical efficiency (in case of a pro-
duction function) is the solution to the following constant re-
turns to scale (CRS) DEA linear programming problem:
where is a scalar, and yi and xi are column vectors of out-
puts and inputs respectively for the ith university. λ is an N × 1
vector of constants. The variable Y is an M × N output matrix
and X is a K × N input matrix, and the proportional increase in
outputs that could be achieved by the ith university, holding
inputs constant, is , (1) with
1 1 the univer-
sity’s efficiency score, which is between 0 and 1 (For a com-
plete explanation of DEA and its advantages and disadvantages
see Coelli et al., 1998)
The empirical model for the DEA estimation is defined as
Output Inputf where Y1, Y2 are considered the outputs
and X1, X2 and X3 are inputs. Due to difficulties obtaining dis-
aggregated salary data, X1, X2 and X3 are proxies for the budget.
The justification here is that the higher the faculty salary
(budget), the better the universities ability to hire more faculty
in general and have a larger body of teaching staff, a higher
percentage of faculty holding PhDs and being tenured or on
tenure track. The DEA efficiency estimation for each institution,
under the assumption of variable return to scale (VRS), was
computed using DEAP 2.1 software developed by T. J. Coelli.
VRS was chosen due to the rather restrictive nature of CRS.
The list of the universities and their efficiency scores are not
included in the paper to preserve both the privacy of the institu-
tion and because the scores in and of themselves are not the
focus of this research.
The summary statistic of the efficiency scores suggests a
mean score of 0.74 with a standard error equal to 0.08. The
95% confidence interval for the mean has a 0.18 margin of
error. There is a wide range of efficiency from 0.023 to 1. The
scores are negatively skewed (skewness = –0.99), as expected
from the efficiency scores, suggesting the existence of ineffi-
ciency in at least some of the universities under study. Based on
the theoretical model here, these inefficiencies stem from the
existence of inefficient allocations of the total budgets.
Qualifications and Concluding Remarks
Historically, universities with larger proportions of budget
allocated to teaching staff (which generally translates to a larger
number of faculty) are assumed to be more productive i.e., to
have higher efficiency. These universities generally have a
higher percentage of faculty holding a Doctorate, tenured or on
tenure track. Some examples among the sample considered here
are University of Colorado-Denver and the University of Texas-
San Antonio. These universities have larger number of faculty
with well above average number of refereed publications. This
observation does not necessarily hold for all universities.
This study attempts to get a preliminary idea of the charac-
teristics of a “peer group” identified by UCO in 2006. The pre-
liminary nature of this research suggests several shortcomings
such as those stemming from the deterministic nature of DEA
and the inability to perform statistical testing etc. However, the
most important shortcoming of results stems from the enor-
mous difficulty of the obtaining disaggregated salary data from
universities. Despite all of this, the study can be considered a
first step and a baseline for future studies. As one potential
example, one can argue that perhaps the criteria that identify
universities as peers should be reexamined since our prelimi-
nary results seem to suggest that the existence of a strong ho-
mogeneity among these universities is highly questionable. It is
noteworthy that since 2007, UCO has adopted a new set of
peers. This change, at least implicitly, could be a validation for
the results of this paper.
Aigner, D. J., Lovell, C. A. K., & Schmidt, P. (1977). Formulation and
estimation of stochastic frontier production function models. Journal
of Econometrics, 6, 21-38. doi:10.1016/0304-4076(77)90052-5
Alexander, F. K. (2001). The silent crisis: The relative fiscal capacity of
public universities to compete for faculty. The Review of Higher
Education, 24, 113-129.
Coelli, T. J., Rao, D. S. P., & Battese, G. E. (1998). An introduction to
efficiency and productivity analysis. Norwell: Kluwer Academic
Cohn, E. (1973). Factors affecting variations in faculty salaries and
compensation in institutions of higher education. The Journal of
Higher Education, 44, 124-136. doi:10.2307/1980571
Faculty Compensation Systems Used in Higher Education (2001).
ASHE-ERIC Higher Ed uca ti on Report 28, 15-24.
Copyright © 2012 SciRes.
Copyright © 2012 SciRes. 227
Hoenack, S. A. (1982). Pricing and efficiency in higher education. The
Journal of Higher Education, 53, 403-418. doi:10.2307/1981606
Kumbhakar, Subal, C., & Wang, H.-J. (2006). Estimation of technical
and allocative inefficiency: A primal system approach. Journal of
Econometrics, 134, 419-440. doi:10.1016/j.jeconom.2005.07.001
Rassouli-Currier, S., & Currier, K. (2008). Inefficiency in higher: An
empirical application to peer regional universities. Southwestern
Economic Review: Research Notes.
Smallwood, S. (2006). Faculty salaries rise by 3.4%; Law professors
still earn the most. Chronicle of Higher Edu c ation, 52, A12.
Talluri, S. (2000). Data envelopment analysis: Models and extensions.
Decision Line, 31, 8-11.
Tuckman, B. H., & Tuckman, H. P. (1976). The structure of salaries at
american universities. The Journa l of Higher Education, 47, 51-64.