Also in recent years, some researchers began to use DEA-based Malmquist
index to evaluate the change of university research productivity. Hu and Liang
( 2007 ) took 25 merger universities as sample, and calculated their research pro-
ductivity change over 1999–2002. They found that technical change was the prime
resource of research efficiency improvement, and the scale effect of merger uni-
versities was not substantial at all. Luo ( 2009 ) examined the research efficiency and
productivity status of higher education system in 29 provinces from a provincial
level, and found that the overall efficiency decreased across China from eastern,
central, to western, and the allocative efficiency of most provinces was on a
descending trend.
2.5 Comments on Empirical Literature and Quantitative Approach...
Approach
2.5.1 Comments on Empirical Literature...................
By combing the relevant empirical studies on university research efficiency and
productivity, wefind the following questions and points needed to be improved.
First, most of existing literature is focused on exploiting technical efficiency and
scale efficiency of university research production, while only a few examining cost
efficiency and allocative efficiency of research production. If there could be analysis
targeting the same subject from multiple perspectives, the existing literature would
provide us with more abundant and meaningfulfindings.
Second, most of existing literature employs DEA to exploit the status of effi-
ciency and productivity of university research activities. Due to the misuse of this
method, researchers can’t examine the criterion-related validity, which is hard to
ensure the validity of conclusions.
Third, in most literature, there lacks a theoretical framework for constructing the
input-output indicators system on research efficiency evaluation. Currently, the
selection of indicator is heavily relied on the availability of data and referring to
similar studies. Thus, the lack of theoretical framework makes it hard for different
researchers to obtain consistentfindings in their own studies, and it would make the
researchers be trouble to interpret of the evaluation outcomes. Furthermore, when
existing studies consider research output, their indicator system tends to measure
the output quality based on bibliometric database. For example, a regular method is
to use the articles published on academic journals in SCI or SSCI, the impact factor
of the journal, the citations of a paper, to measure the research quality of targeted
unit. There are certain limitations in using this kind of indicators to measure
research output quality, which is especially true in the evaluation on university as
whole, because there exists heterogeneity in publication and citation indicators
among different disciplines, and taking these indicators into the econometric models
or DEA, SFA estimations would result into serious estimation bias. Therefore, in
22 2 Evaluation on University Research Efficiency and Productivity...