You are on page 1of 13

Asia Pacific Education Review (2020) 21:197–209

https://doi.org/10.1007/s12564-019-09622-z

Benchmarking in Vietnam universities: teaching and research


and revenue efficiencies
Phung Phi Tran1 · Kuo‑Cheng Kuo2 · Wen‑Min Lu3   · Qian Long Kweh4

Received: 7 February 2019 / Revised: 1 October 2019 / Accepted: 10 December 2019 / Published online: 2 January 2020
© Education Research Institute, Seoul National University, Seoul, Korea 2020

Abstract
Assessing university resource allocation or misallocation is necessary to understand its impact on research and teaching
productivity. To achieve this insight, this study conducts a two-stage data envelopment analysis to judge the “teaching and
research efficiency” and “revenue efficiency” of 61 universities in Vietnam. We analyze what variables have the most effect
on the efficiencies of the universities through improving results. Moreover, this research applies scale elasticity to distin-
guish the benchmark performance leader among Vietnam universities. Results indicate that universities are more efficient
in terms of “revenue efficiency” than “teaching and research efficiency.” The scale elasticity decomposition findings show
that multidisciplinary universities perform better than those specializing in science and technology and social science and
economics. Finally, the competitive map to performance advancement strategies is exhibited. This map helps the university
management to enhance their efficiencies.

Keywords  Data envelopment analysis · Teaching and research efficiency · Revenue efficiency · Universities · Scale
elasticity

Introduction

Education plays a significant role in national, local, and indi-


vidual levels (Johnes et al. 2017). Forty-seven percent of the
* Wen‑Min Lu American population asserts that higher education is crucial
wenmin.lu@gmail.com to get ahead in life (Vistacollege 2019). Higher education
Phung Phi Tran assumes an essential contribution in national competitive-
tranthiphiphung@tdt.edu.vn ness (Tran and Villano 2017a). Moreover, education impacts
Kuo‑Cheng Kuo economic growth and financial remuneration, which should
kuochengkuo20@gmail.com be considered in the lifelong education premium (Chen
Qian Long Kweh et al. 2018). Universities play an important role on national
qlkweh@gmail.com development by providing excellent education to citizens
(Lu 2012). Evaluation of higher education, particularly
1
Department of International Business, Faculty of Business university performance, is a common practice for several
Administration, Ton Duc Thang University, 19 Nguyen Huu
Tho Street, Tan Phong Ward, District 7, Ho Chi Minh City, countries as a means of improving higher education quality
Vietnam (Martín 2006).
2
Department of International Trade, Chinese Culture Techniques that enable educational performance assess-
University, 55, Hwa‑Kang Road, Yang‑Ming‑San, ments should be utilized in the fields of teaching, research,
Taibei 11114, Taiwan and administration to improve current educational systems
3
Department of Financial Management, National Defense (Altamirano-Corro and Peniche-Vera 2014). However, the
University, No. 70, Sec. 2, Zhongyang North Rd., Beitou, execution of the current mechanism of university autonomy
Taibei 112, Taiwan continues to present shortcomings, decreasing the effec-
4
Faculty of Management, Canadian University Dubai, 1st tiveness of the autonomy of tertiary education institutions
Interchange, Sheikh Zayed Road, P.O. Box 117781, Dubai, (Hien 2017). The parallel existence of universities in piloting
United Arab Emirates

13
Vol.:(0123456789)

198 P. P. Tran et al.

autonomy and non-autonomy complicates the educational (Chen et al. 2009) for evaluating the operating universities’
management system. University managers should, there- performance. The first study to use this new method to eval-
fore, conduct detailed performance benchmarking analysis uate and benchmark Vietnam universities is that of Sahoo
of resource allocation and competitive advantage on opera- et al. (2014). The proposed method aims to sift through
tional styles in intense competition (Lu 2012). allocative inefficiency, which can emerge because of any
According to the competitiveness report of 138 selected sub-stage decision as to how much of intermediate factors
nations, Vietnam was ranked 60th in terms of higher educa- are applied and considered by managers in the higher edu-
tion and training (Schwab 2016). This ranking raised con- cation sector. Most of the real-life production technology
cerns not only as to whether Vietnam universities are opera- square measure are multi-stage in nature, and the sources of
tionally efficient but also whether dependent elements are accelerating returns are the sub-technologies (Sahoo et al.
impacting their work efficiencies (Tran and Villano 2017b). 2014). Therefore, estimating the size economies of univer-
Vietnam universities are more productive when the same sities is imperative not only for the network technology but
inputs lead to a greater output. In Vietnam, both public and also for the sub-technologies.
private universities are controlled under the same legal cir- The second contribution of this study is the use of
cumstances and financial management mechanisms. Accord- resource utilization inefficiency analysis to assist university
ing to Finance Newspaper, Vietnam’s annual budget for edu- managers not only to determine the network size economies
cation is approximately 20% of the total budget, which is but also to find their sources to enhance productivity fur-
equivalent to 5% of the GDP and even higher than those of ther. The intermediate stage supports the decomposition of
more economically developed countries such as Singapore technical efficiency (TE) and scale elasticity (SE) decom-
(3%), Hong Kong (3.5%), Thailand (3.8%), Malaysia (5.1%), position into specific SEs. This implies the need to analyze
and South Korea (5.2%) in 2011 (Nga 2017). Such statistics which variables have the most effect on the effectiveness of
reveal that the government’s spending plans on education universities through improvement results. In addition, this
support public universities, although the universities are study provides significant insights that can affect adminis-
admonished to depend entirely on the government’s regula- tration policies at the university level, such as proposing for
tions from the perspectives of expenses and revenues. managers to use a competitive map to boost efficiency. By
Hien (2010) looked into related investigations on the using a competitive map, universities can gain more inputs
research performance of 11 East and Southeast Asian coun- on dispensing their resources. The map is often regarded as
tries to identify the gaps among these nations in their levels a work device for operational improvement. An important
of economic development, and the findings were used as concern within the law enforcement agency analysis is to
basis to upgrade research effectiveness and reshape sci- create distinction within the model (Liu et al. 2009).
ence and higher education (HE) system in Vietnam. Using Third, to rank and acknowledge the truthfulness of role
data from 50 universities and 50 colleges, Tran and Vil- models within the education sector, we applied scale elas-
lano (2017a) approved a two-stage semi-parametric DEA to ticity in two-stage network DEA (Sahoo et al. 2014). Such
evaluate the efficiency of higher education institutes (HEIs) approach can distinguish the benchmark performance leader
and research the impacts of factors on their performance. in teaching and research and revenue stages. Based on the
Tran and Villano (2015) also integrated DEA-based dynamic benchmarking analysis, universities can gain significant
network model to test dynamic changes inefficiencies in insights that can help develop their core competitive advan-
116 Vietnam public colleges. Nonetheless, the identifica- tages and make the best possible outcomes. Moreover, ineffi-
tion performance of these methods is extensively dependent cient universities can learn from benchmark universities and
on black box wherein the focus of inspection is on inputs figure out better strategies. The paper also provides future
and outputs, and what goes outside the box is normally dis- reference for government finance supervision organizations
regarded. Thus, full classification of the aforementioned in selecting excellent universities for endowed manage-
methods is difficult using only the traditional DEA model. ment. In addition, the concluding remarks include policy
Furthermore, data are outdated compared to the current situ- implications.
ation because of new regulations arising from revolution. The rest of this paper is organized as follows. Section
To address these challenges, an effective detection method “Literature review” reviews related previous studies. To fur-
should be considered. ther distinguish the efficient universities, section “Research
This study thus considers three important objectives methodology” presents the research design and sample selec-
and makes corresponding contributions. The first objec- tion criteria. Section “Empirical analysis results” discusses
tive concerns estimating the performance of Vietnam uni- the empirical results. Finally, section “Concluding remarks”
versities through a two-stage structure including “teaching concludes with a summary of research contributions.
and research efficiency” and “revenue efficiency” by using
decomposing technical efficiency in two-stage network DEA

13
Benchmarking in Vietnam universities: teaching and research and revenue efficiencies 199

Literature review Application of DEA in universities

Overview of Vietnam universities In the past, various researchers embraced DEA to assess uni-
versity performance because it is a notable non-parametric
In 1975, after the war, Vietnam’s higher education followed method broadly assigned in operations research to assess the
the same subsidy model as that of other socialist countries efficiency of universities. In the extant literature on higher
(Chanh 2015). At this stage, the quality of higher education education, selecting between the input and output orienta-
is almost unchallenged. Students are considered elite and tion to examine and measure the efficiency of HEIs using the
carefully selected from the beginning with rigorous screen- DEA methodology depends on the specific research context.
ing rates. With the socio-economic reform in Vietnam since Some studies utilize the output orientation to maximize
1986, Vietnam’s higher education system has been imple- outputs, given the prevailing inputs (Ahn et al. 1988; Breu
menting many innovations (Nguyễn 2006). Since 2004, and Raab 1994; Johnes and Johnes 1995; Agasisti et al.
higher education has been aiming to train 400 students per 2012; Kuah and Wong 2013; Fu and Huang 2009; Fernando
10,000 people to catch up with various countries around the and Cabanda 2007). On the contrary, other researchers used
world. However, many graduates or graduate students exhib- the input orientation to minimize input resources and obtain
ited poor quality. As a result, the number of unemployed the target outputs. (Chu Ng and Li 2000; Korhonen et al.
graduates with bachelor’s and master’s degrees increased 2001; Avkiran 2001; Castano and Cabanda 2007; Duh et al.
to 26,000 from the second to fourth quarter of 2015 only 2014; Sav 2012; Tran and Villano 2017a). Higher education
(Quang 2016). not only applies multiple inputs to provide a diversity of
Based on Times Higher Education rankings in 2018, no outputs but also has multiple divisions as well.
Vietnam university was included in the top 350 universities Flegg et al. (2004) conducted DEA to inspect the spe-
in Asia due to weakness in various aspects such as research cialized productivity of 45 UK colleges. Turner (2005)
capacity, degree of influence on policies, curricula, teaching inspected the act of standardization colleagues utilizing an
staff, and facilities (Hoa 2018). The weakness of the Viet- “association table” approach. The examination contended
namese university system can also be attributed to lack of that using DEA to deal with benchmarking considers the
liberal education. Universities lack autonomy with regard to acknowledgment of assorted tasks, which is more suit-
management despite arguments that Vietnam must maintain able for UK colleges. Johnes (2006) used DEA to gauge
a different education system. Most importantly, the Vietnam- the effectiveness of financial management on UK colleges.
ese lack the spirit for serious learning to acquire knowledge Casu and Thanassoulis (2006) set up a DEA system to rec-
(Trung 2016). ognize works on prompting cost-productive focal managerial
In general, Vietnam’s higher education has achieved vital administrations in UK colleges from 1999 to 2000. Kao and
growth within the numbers of institutions, enrollments, Hung (2008) used DEA to survey the relative proficiency of
and research activities after 20 years of reform since 1997. the scholastic divisions at National Cheng Kung University
According to statistics of Vietnam Ministry of Education in Taiwan. Their findings revealed whether a department
and Training (MOET), the total number of lecturers and uses resources efficiently and if their efficiency breaks down
professors in 2016 increased by 4.6% compared with that in where change is required. Bayraktar et al. (2013) measured
the previous year. Despite the 0.48% increase in enrolling the efficiency of quality management applied on public and
students, the number of graduating students decreased by private universities in Turkey. They found that public uni-
13.68% due to several factors such as lack of quality devel- versities were particularly successful in controlling quality
opment, outdated quality assurance, and lack in quantity management practices for excellent teaching and research
and quality of teaching staff. The ratio of students to faculty performance. Thanassoulis et al. (2017) combined DEA
remains 24:1, that is, the ratio of doctoral students to lec- and analytical hierarchy process to suggest an integrated
turers increased from 20 to 22% (MOET 2017). Methods approach to higher education teaching judgment.
for teaching, learning, and assessing learning outcomes in In spite of the fruition of different studies exploring the
tertiary education are lagging, and the professional capacity managing performance of educational institutions, perfor-
of graduates do not meet occupational requirements, which mance assessment for Vietnam public and private univer-
imply the many inadequacies in the education system, such sities are rarely examined through a two-stage production
as equipping the students with creativity and life skills for structure. Utilizing the standard DEA using the black box
practical living. model does not consider the internal structure of HEIs that
connects multiple activities from varied divisions in the pro-
duction procedure. Moreover, the standard DEA neglects
assessment of divisional efficiency. Previous works never (1)
synchronously used decomposing TE in two-stage network

13

200 P. P. Tran et al.

Fig. 1  Two-stage production
processes for universities Full-time Teachers
Stage 1
Administration
Teaching and Research
Assistant Manpower
Efficiency
Facilities

Academic
Enrolling Graduated
Papers and
Students Students
Projects

Stage 2 Revenue from Tuition


Revenue Efficiency
Grants

DEA to assess the operating performance of universities, (2) • Administration support manpower (x2) is composed of
proposed decomposition of network SE approach to analyze administration staff such as educational administration,
what variables have the most effect on the effectiveness of student affairs, general affairs, accountant, librarians,
universities through improving results, and (3) applied SE in secretary, and security personnel.
two-stage network to rank and recognize the suitable models • Facilities (x3) is composed of a unit area, with a stand-
in the education sector. ard international square meter (m2) such as classrooms,
library, laboratory, and experimental zone.
• Academic papers and projects (z1) refer to the total
Research methodology amount of articles published in journals, conferences,
and projects, which have been used in measuring the
Two‑stage production processes for universities research capability of Vietnam universities.
• Enrolling students (z2) refers to the total number of stu-
To assess the performance of universities, this research dents who applied and entered courses in the present
proposes a two-stage network process to design two perfor- year.
mance models, namely “Teaching and Research Efficiency” • Graduated students (z3) refers to the total number of stu-
and “Revenue Efficiency.” As shown in Fig. 1, “Teaching dents who graduated in the present year.
and Research Efficiency” (Stage-1) estimates the university’s • Revenue from Tuition (y1) is composed of tuition and
ability to transform teaching and research inputs including minor revenue.
full-time teachers, administration assistant manpower, and • Grants (y2) includes the total of teaching and research-
facilities into three outputs including academic papers and subsidized grants and other subsidies.
projects, enrolling students, and graduated students. In other
words, the inputs: “full-time teachers, administration assis- Data collection
tant manpower, and facilities” may mean attractions and
ultimately production into “academic papers and projects, According to Statista, in 2016, there are 235 universities,
enrolling students, and graduated students.” “Revenue Effi- colleges, institutes, and university divisions in Vietnam. In
ciency” (Stage-2) gauges the university’s ability to improve this paper, we focus on benchmarking universities that are
returns from tuition and grants from a revenue perspective. with available data in their respective 2016 annual reports
Specifically, this stage examines whether a higher number and publicity reports. Having also checking related statistics
of students and research projects would mean the ultimate with the Vietnamese Ministry of Education, after which give
revenue and grants for the universities. The inputs/interme- us the maximum possible number of decision-making units
diates/outputs used in this research are defined as follows: (DMUs), we analyzed 61 universities (public and private)
in Vietnam. Each university is treated as a DMU in DEA
• Full-time teacher (x1) is determined as total professors analysis. Data are extracted from the 2016 annual report
(full-, associate-, and assistant professors), lecturers, and of each university and three-publicity report. Information
experts. on academic papers and projects is extracted from scien-
tific research department annual statistics report. Table 1

13
Benchmarking in Vietnam universities: teaching and research and revenue efficiencies 201

Table 1  Descriptive statistics and correlation coefficients


Variables Input variables Intermediate variables Output variables
x1 x2 x3 z1 z2 z3 y1 y2

Panel A: descriptive statistics


 Mean 558.6 289.1 55,357 193.5 2865.3 2505.6 1,599,922.6 116,143.3
 Min 85 21 6330 0 166 146 6736 72
 Max 2540 2056 6,437,000 2106 10,850 29,459 788,096 1,020,446
 Std. Dev 451.6 335.7 91,470.9 374 2346.7 3872.9 154,417.6 172,816.7
Panel B: Pearson correlation
 x1: full-time teacher (person) 1.000
 x2: administration assistant manpower (person) 0.799 1.000
 x3: Facilities ­(m2) 0.731 0.565 1.000
 z1: Academic papers and projects (number) 0.646 0.613 0.671 1.000
 z2: enrolling students (person) 0.727 0.645 0.495 0.472 1.000
 z3: graduated students (person) 0.620 0.809 0.313 0.323 0.740 1.000
 y1: revenue from tuition (VND$ million) 0.819 0.586 0.618 0.532 0.596 0.383 1.000
 y2: grants (VND$ million) 0.796 0.831 0.579 0.625 0.717 0.799 0.532 1.000

provides descriptive statistics and correlation analysis, each stage within an organization. This assessment leads to
indicating positive correlation coefficients in teaching and a black box system. The overall efficiency of an organiza-
research efficiency model between the chosen three input/ tion is not evaluated using solely a single stage but rather
intermediate/output. Moreover, Cooper (2001) stated that through multiple stages (Seiford and Zhu 1999; Kao and
when employing the DEA model, the number of DMUs Hwang 2008; Premachandra et al. 2012; Galagedera et al.
should be no less than triple of the total number of input/ 2016). In this regard, network DEA model can be devel-
intermediate/output factors investigated. In this research, oped to study a complete production process and the effects
the number of universities is 61, which is ten times higher of financial activities on the process within an organization
than the number of inputs and outputs considered. Therefore, (Färe and Grosskopf 2000). Liu and Lu (2012) extended the
the developed DEA model, which was executed through the single-stage network-based ranking method to examine the
General Algebraic Modeling System (GAMS) integrated rank of efficient DMUs, suggested a benchmarking unit for
development environment, possesses high construct valid- factors, and identified the advances of each efficient unit.
ity. In this table, we can recognize that number of gradu- The relational model designed by Kao and Hwang
ated students is higher than enrolling students are. Because (2008) can be employed to assess the performance of each
enrolling students firstly are students that are recruited in division in a two-stage production process and be further
the fall of previous year. Graduated students means students extended to network DEA (Kao 2009). However, the rela-
will graduate in the mentioned year. Second, Vietnam uni- tional model should only be applied under the acceptance of
versity system has three principal types of university such constant returns to scale (CRS). The Slacks Based Measure
as sciences and technology, social science and economics, (SBM) was first introduced to network DEA for evaluating
and multidisciplinary. The total study time of each type is DMUs performance (Tone and Tsutsui 2009). This measure
different (4–5 years). Third, some universities have in-ser- employs the probability of no proportional changes in the
vice education system and second degree system. It leads to input and output items in efficiency evaluations and consid-
uneven graduation. ers the importance of each division through the weighting
method, evaluating the overall efficiency and that of each
Measuring efficiency using a two‑stage network department. Chen et al. (2009) proposed the additive effi-
DEA model ciency decomposition approach to two-stage production
model; the overall efficiency is the weighted sum of each
DEA evaluates efficiency using a non-parametric produc- stage’s efficiency. This approach can be applied under the
tion function according to the concept of the frontier pro- assumption of CRS and variable returns to scale (VRS).
duction function proposed by Farrell (1957). However, the Cook et al. (2010) extend this approach to network DEA
conventional models of Charnes et al. (1978) and Banker and examine the weighted result of each division.
et al. (1984) neglect the complete process of production and To account for the allocative inefficiency (Sahoo et al.
therefore cannot explain the management information of 2014), the present study thus used two-stage network DEA

13

202 P. P. Tran et al.

(Chen et al. 2009) to evaluate the teaching and research, as indicates that the additive structure holding the multipliers
well as revenue efficiency. Related mathematical equations for the intermediate factors (produced as inputs and outputs)
are elucidated as follows: ( are the same. Using optimal multipliers from (1), we ( can
In the first stage, p outputs zdj , d = 1,(… , p are produced acquire input-oriented(TE of)the university in N N
) )
k TE VRS TEk
by n units of universities using m inputs xij , i = 1, … , m  ; in and the stages TES1 S1
and TES2 S1
as
) ( )
VRS TEk VRS TEk
stage two, the outputs formed in the first stage change into (∑s )/(∑m )
inputs, which
( are also named ) as intermediates, to produce
TENk =
r=1
u∗r yrk − 𝜔∗1 − 𝜔∗2
i=1
v∗i xik
s outputs yrj , r = 1, … , s  . vi is the multiplier for input i  ; (∑p )/(∑m )
TES1 (2)
∗ ∗ ∗
wd is the multiplier for intermediate measure d  ; ur is the k
= w
d=1 d
zdj − 𝜔 1
v
i=1 i
x ij
multiplier for output r. The technical efficiency (TE) of a net- (∑s )/(∑p )
TES2 = u∗r yrj − 𝜔∗2 w∗d zdj
work university can be estimated either in input-oriented or k r=1 d=1
output-oriented manner. We established the following input-
One can express TEN,t as the result of three terms. The
oriented VRS-based ) network DEA model for estimating the
k
first two terms defining the TEs in the stages ) VRS and
( SETE
S1
input TE TEN,t of the university k as
(
VRS
TEVRS . The third term exhibits an index Ik showing
S2
s
∑ whether the decision regarding the using of observed inter-
TENk = Max ur yrk − 𝜔1 − 𝜔2 (1) mediate factors as intermediate methods is optimal. The
proposed TE decomposition is shown below.
r=1

��s ����m �
TENk = u∗r yrk − 𝜔∗1 − 𝜔∗2 v∗i xik
r=1 i=1
∑p ∗ ∑s
�∑s �∑s �
w ∗
d=1 d dk
z − 𝜔 1
u ∗
r=1 r rk
y − 𝜔∗2 u∗ yt − 𝜔∗1 − 𝜔∗2
r=1 r rk
u∗ y − 𝜔∗2
r=1 r rk
= ∑m ∗ ⋅ ∑p ⋅ �∑p �∑p
w∗ z

v x w∗ z − 𝜔∗1 w∗ z
i=1 i ik d=1 d dk
� �∑p
d=1 d dk d=1 d dk (3)
1 − TE1S2 𝜔∗1

w∗ z
d=1 d dk
=TES1
k
⋅ TES2 k
⋅ k
� �∑p �
1 − 𝜔∗1 w∗ z
d=1 d dk
=TES1
k
⋅ TES2
k
⋅ IkSE

m
∑ IkSE is proxy for allocated inefficiency, in which case
vi xik = 1, (1.1) IkSE > (<)1 . Allocative inefficiency arises below the VRS
i=1
specification, yet vanishes under the CRS specification.
p m Therefore, one can illustrate that retaining the “same weight”
assumption on zdk as outputs and inputs as the VRS specifi-
∑ ∑
wd zdj − vi xij − 𝜔1 ≤ 0, (j = 1, … , n) (1.2)
d=1 i=1 cation is not ample to reject allocative inefficiency. Alloca-
tive inefficiency may be a broader conception that features
s
∑ p
∑ inefficiencies arising from any potential sub-optimal deci-
ur yrj − wd zdj − 𝜔2 ≤ 0, (j = 1, … , n) (1.3) sion as to how much zdk to turn out and consume in the
r=1 d=1 light of adjusting prices IkSE = 1 when TES2
k
= 1 meaning that
there is no allocative inefficiency in the use of observed zdk .
vi , wd , ur ≥ 0; 𝜔1 , 𝜔2 ∈ free. Note that optimal multipliers obtained from (1) may not be
distinctive, implying that TES1k and k do not seem unique.
TES2
Liang (2008) assumed the same weights on zd as an excel-
In line with Kao and Hwang (2008), assuming that TES1 k is
lent planning between the two stages based on the CRS spec-
more important, we first assumed the maximum value of
ification. Constraints (1.2) and (1.3) correspond to stages
k via
TES1
TES1VRS
and TES2VRS
 , respectively, whose particular inter-
cept multipliers are 𝜔1 and 𝜔2 , i.e., TENVRS is additive. This

13
Benchmarking in Vietnam universities: teaching and research and revenue efficiencies 203

Table 2  Technical efficiency decomposition and upper bound of scale elasticity estimates results
Total Research and teaching effi- Revenue efficiency
ciency
TE SE RTS TE1 RTS TE2 RTS
IRS DRS CRS IRS DRS CRS IRS DRS CRS

Mean
 Multidisciplinary 0.407 3.861 29 6 2 0.462 33 2 2 0.461 13 14 10
 Sciences and Technology 0.399 3.732 11 1 0 0.500 11 1 0 0.459 4 7 1
 Social Science and Economics 0.463 1.935 11 1 0 0.498 11 1 0 0.661 4 8 0
 Total 0.423 3.176 51 8 2 0.487 55 4 2 0.527 21 29 11

TE technical efficiency, SE scale efficiency, RTS returns to scale, IRS increasing return to scale, DRS decreasing return to scale, CRS constant
return to scale
∑p ( )
TES1
k = Max d=1
wd zdk − 𝜔1 Let its optimal solution vector be 𝛽k∗ , 𝜆∗j , 𝛾j∗ , z̃∗dk  . The
m
∑ inefficient university k demands to be planned onto the net-
vi xik = 1, work frontier by using the following formula:
i=1
s
(6)
( ∗
xik = 𝛽ik∗ xik − s−k , z̃∗dk , y∗rk = yrk + s+k ,
)

ur yrk − 𝜔1 − 𝜔2 = TENk
r=1
(4) where s−k and s+k are respectively input and output slacks
p m
∑ ∑ under (5). Following Baumol (1982), Sahoo et al. (2014)
wd zdj − vi xij − 𝜔1 ≤ 0, (j = 1, … , n)
d=1 i=1
defined the input-oriented Returns to Scale (RTS)1 of the
∑s p
∑ university k in TENk  , TES1
k  , and k .
TES2
ur yrj − wd zdj − 𝜔2 ≤ 0, (j = 1, … , n) m

∗ ∑m ∗
r=1 d=1
� 𝜕F N (⋅) �s 𝜕F N (⋅) 𝛽k i=1 vi xik
𝜀Nik ∗ ∗
, z̃dk , y∗rk
� �
xik ≡− xik yrk = ∑s
𝜕xik r=1 𝜕yrk u∗ y
vi , wd , ur ≥ 0; 𝜔1 , 𝜔2 ∈ free. i=1 r=1 r rk
𝛽h∗
To determine the returns to scale of network university = .
𝛽h∗ + 𝜔∗1 + 𝜔∗2
k , we applied model (5) to obtain the projection for the inef-
(7)
ficient university k . TENk uses 𝜆j and 𝛾j as intensity weights
Note that in (7), 1, due to (1.1) and
∑m
to form a linear combination of n universities. v∗ x =
i=1 i ik
𝜔2 . Based
on duality, the aim
∑s ∗ ∗ ∗ ∗
u
r=1 r
y rk = 𝛽 k
+ 𝜔 1
+
TENk = Min 𝛽h function values of (1) and (5) are the same, for instance
𝛽k∗ = r=1 u∗r yrk + 𝜔∗1 + 𝜔∗2 . Based on (7), the input-oriented
∑s
s.t.
n
∑ network is increasing returns to scale (IRS) (i.e., 𝜀Nik (⋅) > 1)
xij 𝜆j ≤ 𝛽k xik , (i = 1, … , m), if (𝜔∗2 + 𝜔∗2 ) < 0 in all optimal solutions, constant (CRS)
j=1 (i.e.,𝜀Nik (⋅) = 1 ) if (𝜔∗2 + 𝜔∗2 ) = 0 in all optimal solutions,
∑n
and decreasing (DRS) (i.e.,𝜀Nik (⋅) < 1 ) if (𝜔∗2 + 𝜔∗2 ) > 0 in all
zdj 𝜆j − z̃dk ≥ 0, (d = 1, … , p), optimal solutions. Following the same method, RTS in TES1
j=1 k
n
and TEkS2 can be determined.

𝜆j = 1, (sub − technology 1)
j=1 (5)
n

zdj 𝛾j − z̃dk ≤ 0, (d = 1, … , p),
j=1
n

yrj 𝛾j ≥ yrk , (r = 1, … , s),
j=1
n

𝛾j = 1, (sub − technology 2)
j=1
1
  RTS explains what happens as scale effects change when inputs are
𝛽h ≤ 1, 𝜆j , 𝛾j ≥ 0, z̃dk ∈ free. variable in a production function. Three possibilities exist: increasing,
constant, and decreasing RTS.

13

204 P. P. Tran et al.

Empirical analysis results These results reveal that some universities operated under
increasing returns of universities (IRS). The sources of IRS
Analyzing teaching and research and revenue remain although sub-stage for universities exhibits decreas-
efficiencies ing and/or constant returns. For instance, RTS exhibiting
IRS for An Giang University (AGU) is due to IRS in Stage-1
Table 2 demonstrates overall efficiency results and rankings even though CRS prevails in Stage 2. Similarly, RTS exhib-
of 61 Vietnam universities. The mean score of teaching iting IRS for VNU is precisely due to IRS in Stage 1 even
and research and revenue efficiencies are 0.487 and 0.527, though DRS prevails in Stage-2. Multidisciplinary universi-
respectively. The findings indicate that revenue efficiency ties generally exhibit efficiency in teaching and research yet
has higher values than teaching and research efficiency for show decreasing returns to scale in the revenue stage, sug-
these 61 universities. This outcome proposes that all pol- gesting that multidisciplinary universities have obtained the
icy staffs of these universities should prioritize improving managerial know-how to operate a university productively.
their teaching, increasing the number of staff and facilities, Nonetheless, they still have not reached their optimal level.
and improving research skills as proven by the number of Sciences and technology universities are technically inef-
research papers and projects, enhancement in recruitment ficient in terms of revenue but show increasing returns to
activities and operational efficiency, and improvement scale in the teaching and research stage, implying that these
in the quality and number of graduates. Our untabulated universities still have to work to achieve scale efficiency.
results show a total of 33 inefficient universities in teach- Generally, the number of publications, enrolling students,
ing and research stage and revenue stage. These universities and graduated students among universities should be viewed
should reallocate their input resources to decrease their input to earn revenue of scale in the education sector.
waste if they are to become more efficient in teaching and
research management and revenue management. For exam- Resource utilization inefficiency analysis
ple, the revenue-inefficient universities should reconsider
their amount of published papers and projects as well as the University management must evaluate the number of out-
quantity of enrolling and graduated students to improve their puts that can be increased and/or the magnitude of inputs
resource utilization. that can be preserved by inefficient teaching and research
We also explored how efficiency scores differ based on performance. This implies reduction in particular inputs for
principle types of a university. There are three types of a university to operate and improve efficiency in teaching
Vietnam universities in terms of disciplinary focus: sci- and research. Moreover, the growth in output can lead to
ences and technology, social science and economics, and decreased levels of resource inputs. Table 3 presents the
multidisciplinary. Table 2 uncovers that on average, multi- results of resource utilization inefficiency analysis for 56
disciplinary universities exhibited better efficiency than both inefficient teaching and research universities. For each ineffi-
sciences and technology and social science and economics. cient teaching and research university, we also distinguished
For instance, 29 multidisciplinary universities maintain an the inefficiency in terms of specific outputs and inputs. The
increasing return to scale, whereas sciences and technol- results suggest that the mean potential increase in annual
ogy and social science and economics universities (11 both) academic papers and projects represent more than 200%
exhibit increasing return to scale. The reason for this result improvement. The average potential increment of enrolling
is that the need for choosing multidisciplinary universities students is 32% improvement. As for graduate students, the
has been cultivated in Vietnam’s education sector recently. average potential increase is at 15%. The average potential
Moreover, we analyzed the results of returns to scale for reduction in the number of full-time teachers, number of
universities. From Table 2, 11 (in the revenue stage) and two staffs, and all facilities are also listed in Table 3. If the num-
(in the research and teaching stage) universities work at the ber of administration support manpower is better handled
constant returns to scale. Approximately 56.9% of Vietnam and the number of enrolling students is increased, resource
universities are encountering returns to scale with decrease utilization efficiency can be promoted in inefficient universi-
in revenue performance stage. Only 6.5% of the universities ties. We recommend that these universities utilize these two
indicate decreasing returns to scale in research and teach- areas to upgrade their performance.
ing stage. This result further reveals that universities are For deep understanding, insights into the resource utili-
confronting a highly competitive environment in revenue zation analysis need to be reviewed. Clearly, AGU should
stage and a low competitive environment in teaching and increase the number of academic papers and projects, enroll-
research stage. Twenty-one universities experience increase ing students, and graduate students by 7%, 49%, and 53%,
in returns to scale in the revenue stage and 55, in the research respectively. The AGU can achieve one of its goals in its
and teaching stage. current output level even if one of the following inputs is cut
off: 21% for full-time teachers, 12% for administration staff,

13
Benchmarking in Vietnam universities: teaching and research and revenue efficiencies 205

Table 3  Resource utilization University Input Output


inefficiency analysis
Full-time teacher Administration Facilities Academic Enrolling Graduated
(x1) assistant man- (x3) papers and students students
power projects (z2) (z3)
(x2) (z1)

AGU​ − 0.79 − 0.88 − 0.79 − 0.07 − 0.49 − 0.53


VNU 0.00 0.00 0.00 0.00 0.00 0.00
CTU​ − 0.82 − 0.82 − 0.84 − 0.81 − 0.63 − 0.60
VNUHCM 0.00 0.00 0.00 0.00 0.00 0.00
HUMG − 0.82 − 0.82 − 0.82 − 0.65 − 0.67 − 0.71
UDN − 0.77 − 0.77 − 0.89 − 0.75 0.88 2.05
HCMUT − 0.58 − 0.15 − 0.61 − 0.13 0.09 0.50
PTIT − 0.87 − 0.87 − 0.87 9.85 − 0.74 − 0.80
UTC​ − 0.85 − 0.85 − 0.85 − 0.41 − 0.85 − 0.84
VU − 0.78 − 0.78 − 0.87 − 0.45 − 0.58 − 0.47
QNU − 0.79 − 0.79 − 0.79 12.44 − 0.80 − 0.81
NTU − 0.81 − 0.81 − 0.96 0.73 − 0.76 − 0.68
VLU − 0.78 − 0.84 − 0.54 − 0.91 − 0.86 − 0.85
HCMIU − 0.46 − 0.84 − 0.59 − 0.92 − 0.50 0.27
NUCE − 0.82 − 0.82 − 0.82 0.16 − 0.33 − 0.46
UEH − 0.77 − 0.86 − 0.77 − 0.59 − 0.73 − 0.70
UIT − 0.28 − 0.75 − 0.22 − 0.80 − 0.66 − 0.22
UTEHY − 0.76 − 0.76 − 0.76 − 0.82 − 0.54 − 0.53
DTU − 0.84 − 0.91 − 0.84 − 0.97 − 0.82 − 0.76
HSU − 0.68 − 0.68 − 0.68 1.92 − 0.85 − 0.86
IUH − 0.88 − 0.88 − 0.90 − 0.13 − 0.76 − 0.69
UTE − 0.80 − 0.80 − 0.80 − 0.86 − 0.54 − 0.49
HUPH − 0.06 − 0.45 − 0.59 − 0.17 3.70 3.28
TTN − 0.67 − 0.70 − 0.67 − 0.91 − 0.42 − 0.45
NTTU​ − 0.90 − 0.96 − 0.88 0.00 0.04 1.00
HPU − 0.50 − 0.68 − 0.29 − 0.14 − 0.89 − 0.89
TUAF − 0.63 − 0.97 − 0.52 − 0.71 − 0.87 − 0.86
DLU − 0.73 − 0.73 − 0.73 − 0.54 − 0.80 − 0.73
HPU2 − 0.67 − 0.67 − 0.67 1.59 − 0.41 − 0.41
EPU − 0.71 − 0.71 − 0.71 5.31 0.30 0.52
TNUT − 0.70 − 0.70 − 0.70 − 0.49 − 0.88 − 0.89
DCMUSSH − 0.81 − 0.82 − 0.81 1.07 − 0.74 − 0.59
DTHU − 0.69 − 0.82 − 0.68 − 0.59 − 0.12 0.04
HUC − 0.33 − 0.68 − 0.33 3.51 − 0.48 − 0.35
SGU − 0.59 − 0.59 − 0.71 − 0.88 − 0.08 − 0.20
HCMUL − 0.61 − 0.61 − 0.69 0.55 − 0.68 − 0.69
HAUI − 0.80 − 0.75 − 0.75 − 0.49 − 0.35 − 0.33
HUP − 0.29 − 0.93 − 0.23 − 0.69 − 0.10 − 0.18
TLU − 0.50 − 0.50 − 0.81 0.12 0.00 0.12
HOU − 0.70 − 0.86 − 0.70 − 0.07 − 0.80 − 0.80
HVTC − 0.80 − 0.80 − 0.80 1.17 − 0.76 − 0.77
HUAF − 0.62 − 0.82 − 0.62 0.10 − 0.24 − 0.27
UEBVNU 0.00 0.00 0.00 0.00 0.00 0.00
HUFLIT − 0.68 − 0.68 − 0.82 − 0.88 − 0.70 − 0.66
NEU − 0.61 − 0.33 − 0.44 − 0.30 − 0.36 1.13
US − 0.82 − 0.82 − 0.83 − 0.93 − 0.50 − 0.50
HDU − 0.79 − 0.79 − 0.79 1.51 − 0.70 − 0.55

13

206 P. P. Tran et al.

Table 3  (continued) University Input Output


Full-time teacher Administration Facilities Academic Enrolling Graduated
(x1) assistant man- (x3) papers and students students
power projects (z2) (z3)
(x2) (z1)

COU − 0.72 − 0.65 − 0.82 0.31 − 0.87 − 0.86


FTU − 0.76 − 0.76 − 0.76 − 0.38 − 0.69 − 0.69
VNAM − 0.68 − 0.68 − 0.68 1.78 1.34 4.02
HUIFA 0.00 0.00 0.00 0.00 0.00 0.00
DAU − 0.57 − 0.57 − 0.57 2.28 2.17 3.26
PXU 0.00 0.00 0.00 0.00 0.00 0.00
PDU − 0.44 − 0.44 − 0.44 44.82 − 0.36 − 0.38
THP − 0.77 − 0.88 − 0.74 − 0.48 − 0.90 − 0.79
VNUHUS − 0.63 − 0.91 − 0.62 − 0.99 0.24 0.26
HPMU − 0.73 − 0.73 − 0.73 58.80 − 0.50 − 0.26
BVU − 0.37 − 0.04 − 0.30 − 0.82 − 0.65 − 0.62
QBU − 0.49 − 0.78 − 0.50 0.21 − 0.32 − 0.21
DNTU − 0.56 − 0.42 − 0.62 7.50 − 0.58 − 0.40
TDMU − 0.76 − 0.76 − 0.76 − 0.50 − 0.34 − 0.47
Mean − 0.62 − 0.67 − 0.63 2.22 − 0.32 − 0.15

Fig. 2  Competitive map of
Vietnam universities

or 21% for facilities. The findings present that decreasing the score in 2016, 61 universities in Vietnam can principally be
amount of slack for this university can lead to more efficient classified into six regions.
utilization of resources. Region A: Universities with the highest competitive-
ness and fastest growth include VNU, VNUHCM, HUIFA,
Competitive map for Vietnam universities UEBVNU, PXU, and HUPH. The universities have scores
of revenue efficiency and teaching and research efficiency
To distinguish the relative positions of 61 Vietnam universi- over 0.9. At present, these universities are working under the
ties and help enhance their performance, a competitive map perfect condition and are enhancing quickly, implying that
is produced by combining teaching and research efficiency they are on the right track, that is, to focus both on teaching
and revenue efficiency. In Fig. 2, the horizontal axis of the and researching skills to address internal weakness.
map presents teaching and research efficiency and the verti- Region B: Universities with low competitiveness in rev-
cal axis presents revenue efficiency. Considering the overall enue efficiency and fast growth in teaching and research

13
Benchmarking in Vietnam universities: teaching and research and revenue efficiencies 207

efficiency include BVU, VNUHUS, and UIT. These univer- or connection activities (Lu 2012). In the literature of edu-
sities have research and teaching efficiency score above 0.9 cational research, the idea of a two-stage model has not
and revenue efficiency score lower than 0.4. These results been discussed extensively. To evaluate university perfor-
imply efficient teaching and research capacity despite scor- mance, the present study builds a two-stage network struc-
ing averagely. These universities can still learn from bench- ture containing “Revenue Efficiency” and “Teaching and
mark universities and seek new breakthroughs to develop Research Efficiency” using the scale elasticity decomposi-
its revenue performance and maintain their competitive tion approach. Basing on the results of the scale elasticity
advantage. of two-stage network DEA, we ranked and identified the
Regions C and D: Universities with medium competi- leaders among the analyzed universities from teaching and
tiveness and fast growth include NEU, HUC, and HCMUT. research and revenue perspectives. From a holistic perspec-
These universities have research and teaching efficiency tive, this paper further suggests a direction for university
between 0.6 and 0.9 and revenue efficiency of less than 0.6. managers to advance their working efficiencies and promptly
In contrast with universities in region C, those in region D, withdraw strategies.
including HUP and HPU, have revenue efficiency lower than The results are as follows. First, universities showing
0.8 and research and teaching efficiency between 0.6 and 0.9. superiority in revenue efficiency contrasted to teaching and
At present, these universities have an average level of teach- research efficiency, suggesting that all politicians should
ing and research efficiency. This implies that the competitive concentrate on improving the quality of faculty teaching,
advantage of these universities expand gradually because increasing the number of administration staff, and restoring
they are upheld by sufficient resources. Furthermore, they facilities. Inefficient universities (54%) need to reallocate
concentrate only on the general prospect revenue. They have their input resources to reduce input waste and consequently
persistently learned from benchmark universities, learned become more efficient in teaching and research management
research and teaching strategies, and reduced the number of and revenue management. In relation to the link between effi-
facilities, teachers, and members of the faculty. ciency score and disciplinary focus of universities, including
Region E: Universities with low competitiveness in sciences and technology, social science and economics, and
teaching and research efficiency and fast growth in revenue multidisciplinary, we found that multidisciplinary universi-
efficiency include NTTU, VNAM, DAU, TLU, DNTU, and ties carry out better than the other two. Moreover, the results
PDU. These universities have revenue efficiency higher than of returns to scale for universities indicate that universities
0.9 and teaching and research efficiency of less than 0.6. are running into an immensely competitive environment in
These universities are required to have a bright prospect but revenue stage and an unremarkably competitive environ-
clearly lag behind the rest of the universities in terms of ment in teaching and research stage. The outcomes of our
overall research and teaching efficiency. Hence, it is recom- research can be supplementary for other research related to
mended that they enhance their internal administration to benchmarking university performance. Future research can
focus on long-term teaching and research strategies. incorporate the testing of performance by applying dynamic
Region F: Universities with low competitiveness and network DEA. Such an approach will provide a perspective
moderate pace of process include almost all universities. from the dynamic multidimensional university performance.
These universities have teaching and research efficiency We also assumed that the models and methods used in this
score and revenue efficiency score of less than 0.6 and 0.8, paper can be applied to other related research in different
respectively. These universities show low intensity and slow industries.
pace of progress and need to conquer the shortcomings by
learning from benchmarks, wisely enhancing internal staff
structures, and re-estimating interest of teaching to map out References
students’ beliefs and promote facilities. Gradually, they shall
be able to keep up with other universities and build up their Agasisti, T., Catalano, G., Landoni, P., & Verganti, R. (2012). Evalu-
ating the performance of academic departments: An analysis of
competitive advantages. research-related output efficiency. Research Evaluation, 21(1),
2–14.
Ahn, T., Charnes, A., & Cooper, W. W. (1988). Some statistical and
Concluding remarks DEA evaluations of relative efficiencies of public and private
institutions of higher learning. Socio-Economic Planning Sci-
ences, 22(6), 259–269.
Assessment of university performance is generally specified Altamirano-Corro, A., & Peniche-Vera, R. (2014). Measuring the insti-
in previous research, and DEA is commonly use. However, a tutional efficiency using DEA and AHP: The case of a Mexican
few imperative views have not yet been analyzed. From the university. Journal of Applied Research and Technology, 12(1),
63–71.
point of view of research methods, traditional DEA mod-
els are one-stage models and neglect intermediate factors

13

208 P. P. Tran et al.

Avkiran, N. K. (2001). Investigating technical and scale efficiencies of efficiency, and economic growth in the Asia-Pacific region (pp.
Australian universities through data envelopment analysis. Socio- 197–215). New York: Springer.
Economic Planning Sciences, 35(1), 57–80. Galagedera, D. U., Watson, J., Premachandra, I., & Chen, Y. (2016).
Banker, R. D., Charnes, A., & Cooper, W. W. (1984). Some models for Modeling leakage in two-stage DEA models: An application to
estimating technical and scale inefficiencies in data envelopment US mutual fund families. Omega, 61, 62–77.
analysis. Management Science, 30(9), 1078–1092. Hien, P. D. (2010). A comparative study of research capabilities of East
Baumol, W. J. (1982). Applied fairness theory and rationing policy. Asian countries and implications for Vietnam. Higher Education,
The American Economic Review, 72(4), 639–651. 60(6), 615–625.
Bayraktar, E., Tatoglu, E., & Zaim, S. (2013). Measuring the relative Hien, T. D. (2017). Renovating public higher education autonomy
efficiency of quality management practices in Turkish public and mechanism in the period of 2015–2017: Results and policy rec-
private universities. Journal of the Operational Research Society, ommendations. Retrieved from https​://tapch​itaic​hinh.vn/nghie​
64(12), 1810–1830. n-cuu--trao-doi/trao-doi-binh-luan/doi-moi-co-che-tu-chu-giao-
Breu, T. M., & Raab, R. L. (1994). Efficiency and perceived quality duc-dai-hoc-cong-lap-giaid​oan-20152​017-ket-qua-va-kien-nghi-
of the nation’s “top 25” National Universities and National Lib- chinh​-sach-13332​7.html.
eral Arts Colleges: An application of data envelopment analysis Hoa, M. N. (2018). Where is Vietnam University standing? Retrieved
to higher education. Socio-Economic Planning Sciences, 28(1), from https​: //tuoit​r e.vn/dh-viet-nam-dang-dung-o-dau-20180​
33–45. 21008​51425​07.html.
Castano, M. C. N., & Cabanda, E. C. (2007). Performance evaluation Johnes, J. (2006). Measuring teaching efficiency in higher education:
of the efficiency of Philippine Private Higher Educational Institu- An application of data envelopment analysis to economics gradu-
tions: application of frontier approaches. International Transac- ates from UK Universities 1993. European Journal of Operational
tions in Operational Research, 14(5), 431–444. Research, 174(1), 443–456.
Casu, B., & Thanassoulis, E. (2006). Evaluating cost efficiency in Johnes, J., & Johnes, G. (1995). Research funding and performance
central administrative services in UK universities. Omega, 34(5), in UK university departments of economics: A frontier analysis.
417–426. Economics of Education Review, 14(3), 301–314.
Chanh (2015). Education in South Vietnam 1954–1975 on the path Johnes, J., Portela, M., & Thanassoulis, E. (2017). Efficiency in
of construction and development. Retrieved from https​://khoav​ education. Journal of the Operational Research Society, 68(4),
anhoc ​ - ngonn ​ g u.edu.vn/home/index ​ . php?optio ​ n =com_conte​ 331–338.
nt&view=artic ​ l e&id=5227%3Agia ​ o -dc-min-nam-vit-nam- Kao, C. (2009). Efficiency decomposition in network data envelopment
1954-1975-tren-con-ng-xay-dng-va-phat-trin&catid​=115%3Agia​ analysis: A relational model. European Journal of Operational
o-dc&Itemi​d=189&lang=vi. Research, 192(3), 949–962.
Charnes, A., Cooper, W. W., & Rhodes, E. (1978). Measuring the effi- Kao, C., & Hung, H.-T. (2008). Efficiency analysis of university depart-
ciency of decision making units. European Journal of Operational ments: An empirical study. Omega, 36(4), 653–664.
Research, 2(6), 429–444. Kao, C., & Hwang, S.-N. (2008). Efficiency decomposition in two-
Chen, J., Zhang, Y., & Yin, Z. (2018). Education premium in the online stage data envelopment analysis: An application to non-life insur-
peer-to-peer lending marketplace: Evidence from the big data in ance companies in Taiwan. European Journal of Operational
China. The Singapore Economic Review, 63(01), 45–64. Research, 185(1), 418–429.
Chen, Y., Cook, W. D., Li, N., & Zhu, J. (2009). Additive efficiency Korhonen, P., Tainio, R., & Wallenius, J. (2001). Value efficiency
decomposition in two-stage DEA. European Journal of Opera- analysis of academic research. European Journal of Operational
tional Research, 196(3), 1170–1176. Research, 130(1), 121–132.
Chu Ng, Y., & Li, S. K. (2000). Measuring the research performance Kuah, C. T., & Wong, K. Y. (2013). Data Envelopment Analysis mod-
of Chinese higher education institutions: An application of data eling for measuring knowledge management performance in
envelopment analysis. Education Economics, 8(2), 139–156. Malaysian higher educational institutions. Information Develop-
Cook, W. D., Zhu, J., Bi, G., & Yang, F. (2010). Network DEA: Addi- ment, 29(3), 200–216.
tive efficiency decomposition. European Journal of Operational Liang, L., Wu, J., Cook, W. D., & Zhu, J. (2008). Alternative secondary
Research, 207(2), 1122–1129. goals in DEA cross-efficiency evaluation. International Journal
Cooper, W. W., Li, S., Seiford, L. M., Tone, K., Thrall, R. M., & Zhu, of Production Economics, 113(2), 1025–1030.
J. (2001). Sensitivity and stability analysis in DEA: Some recent Liu, J. S., & Lu, W. M. (2012). Network-based method for ranking of
developments. Journal of Productivity Analysis, 15(3), 217–246. efficient units in two-stage DEA models. Journal of the Opera-
Duh, R.-R., Chen, K.-T., Lin, R.-C., & Kuo, L.-C. (2014). Do internal tional Research Society, 63(8), 1153–1164.
controls improve operating efficiency of universities? Annals of Liu, J. S., Lu, W. M., Yang, C., & Chuang, M. (2009). A network-
Operations Research, 221(1), 173–195. based approach for increasing discrimination in data envelopment
Färe, R., & Grosskopf, S. (2000). Network DEA. Socio-Economic analysis. Journal of the Operational Research Society, 60(11),
Planning Sciences, 34(1), 35–49. 1502–1510.
Farrell, M. J. (1957). The measurement of productive efficiency. Jour- Lu, W.-M. (2012). Intellectual capital and university performance in
nal of the Royal Statistical Society Series A (General), 120(3), Taiwan. Economic Modelling, 29(4), 1081–1089.
253–290. Martín, E. (2006). Efficiency and Quality in the Current Higher Educa-
Fernando, B. I. S., & Cabanda, E. C. (2007). Measuring efficiency and tion Context in Europe: An application of the data envelopment
productive performance of colleges at the University of Santo analysis methodology to performance assessment of departments
Tomas: A nonparametric approach. International Transactions in within the University of Zaragoza. Quality in Higher Education,
Operational Research, 14(3), 217–229. 12(1), 57–79.
Flegg, A., Allen, D., Field, K., & Thurlow, T. (2004). Measuring the MOET (2017). Higher education statistics for the 2016–2017 school
efficiency of British universities: A multi-period data envelopment year. Retrieved from https​://www.moet.gov.vn/thong​-ke/Pages​/
analysis. Education Economics, 12(3), 231–249. thong​-ko-giao-duc-dai-hoc.aspx?ItemI​D=5137.
Fu, T.-T., & Huang, M.-Y. (2009). Performance ranking and manage- Nga, D. T. (2017). State investment in education and train-
ment efficiency in colleges of business: A study at the department ing: Current situation and some proposals. Retrieved
level in Taiwanese universities. In D. Lee (Ed.), Productivity, from https ​ : //t apch ​ i t aic ​ h inh.vn/nghie ​ n -cuu-trao-doi/

13
Benchmarking in Vietnam universities: teaching and research and revenue efficiencies 209

dau-tu-cua-nha-nuoc-cho-giao-duc-dao-tao-thuc-trang​-va-mot- Tone, K., & Tsutsui, M. (2009). Network DEA: A slacks-based meas-
so-de-xuat-13091​8.html. ure approach. European Journal of Operational Research, 197(1),
Nguyễn. (2006). Education in South Vietnam before 1975. Santa Ana, 243–252.
CA: Lê Văn Duyệt Foundation. Tran, C. D. T., & Villano, R. A. (2015). Measuring efficiency of
Premachandra, I., Zhu, J., Watson, J., & Galagedera, D. U. (2012). Vietnamese public colleges: An application of the DEA-based
Best-performing US mutual fund families from 1993 to 2008: Evi- dynamic network approach. International Transactions in Opera-
dence from a novel two-stage DEA model for efficiency decompo- tional Research., 25, 683–703.
sition. Journal of Banking & Finance, 36(12), 3302–3317. Tran, C. D. T., & Villano, R. A. (2017a). An empirical analysis of the
Quang (2016). 225,000 bachelors of unemployment and the cost of a performance of Vietnamese higher education institutions. Journal
sluggish education. Retrieved from https​://giaod​uc.net.vn/giao- of Further and Higher Education, 41(4), 530–544.
duc-24h/22500​0-cu-nhan-that-nghie​p-va-cai-gia-phai-tra-cua- Tran, C. D. T., & Villano, R. A. (2017b). Input Rigidities and Per-
mot-nen-giao-duc-i-ach-post1​65246​.gd. formance of Vietnamese Universities. Asian Economic Journal,
Sahoo, B. K., Zhu, J., Tone, K., & Klemen, B. M. (2014). Decom- 31(3), 253–273.
posing technical efficiency and scale elasticity in two-stage net- Trung, X. (2016). Mr. Tran Duc Canh pointed out the “spearhead” and
work DEA. European Journal of Operational Research, 233(3), “prison” sector in higher education. Retrieved from https​://giaod​
584–594. uc.net.vn/giao-duc-24h/ong-tran-duc-canh-chi-ra-nganh​-mui-
Sav, G. T. (2012). Four-stage DEA efficiency evaluations: Financial nhon-va-mui-tu-trong​-giao-duc-dai-hoc-post1​66809​.gd.
reforms in public university funding. International Journal of Turner, D. (2005). Benchmarking in universities: League tables revis-
Economics and Finance, 5(1), 24. ited. Oxford Review of Education, 31(3), 353–371.
Schwab, K. (2016). The Global Competitiveness Report 2016–2017. Vistacollege. (2019). The Importance of Higher Education in the 21st
Geneva: World Economic Forum. Century. Retrieved from https​://www.vista​colle​ge.edu/blog/resou​
Seiford, L. M., & Zhu, J. (1999). Profitability and marketability of rces/highe​r-educa​tion-in-the-21st-centu​ry/.
the top 55 US commercial banks. Management Science, 45(9),
1270–1288. Publisher’s Note Springer Nature remains neutral with regard to
Thanassoulis, E., Dey, P. K., Petridis, K., Goniadis, I., & Georgiou, jurisdictional claims in published maps and institutional affiliations.
A. C. (2017). Evaluating higher education teaching performance
using combined analytic hierarchy process and data envelopment
analysis. Journal of the Operational Research Society, 68(4),
431–445.

13

You might also like