Journal of Cleaner Production 224 (2019) 811e822

Contents lists available at ScienceDirect

Journal of Cleaner Production

journal homepage: www.elsevier.com/locate/jclepro

Optimal human resource planning with P-graph for universities

undergoing transition
Kathleen B. Aviso a, *, Anthony S.F. Chiu b, e, Feorillo P.A. Demeterio III c,
Rochelle Irene G. Lucas d, Ming-Lang Tseng e, f, Raymond R. Tan a
a
Chemical Engineering Department, Gokongwei College of Engineering, De La Salle University, Manila, Philippines
b
Industrial Engineering Department, Gokongwei College of Engineering, De La Salle University, Manila, Philippines
c
Filipino Department, College of Liberal Arts, De La Salle University, Manila, Philippines
d
Department of English and Applied Linguistics, Brother Andrew Gonzalez FSC College of Education, De La Salle University, Manila, Philippines
e
Institute of Innovation and Circular Economy, Asia University, Taichung, Taiwan
f
Department of Medical Research, China Medical University, Taiwan

a r t i c l e i n f o a b s t r a c t

Article history: This study deals with the problem of human resources planning due to the additional workforce re-
Received 7 January 2019 quirements of research activities. In practice, planning of human resource expansion is seldom done in a
Received in revised form systematic manner, thus leading to sub-optimal results. In this work, a P-graph model is developed as a
22 February 2019
decision support tool to aid in planning expansion of staffing levels for Higher Education Institutions.
Accepted 19 March 2019
Available online 20 March 2019
Many higher education institutions in the developing world are undergoing the transition from being
teaching-intensive to becoming increasingly research-oriented. This shift is widely recognized as an
important adjustment to the need for universities to play a greater role in creating knowledge capital,
Keywords:
Higher education institution (HEI)
which is an essential component to fuel the next phase of growth of developing countries. Research
Human resource (HR) planning universities are an essential resource for facilitating eco-innovation in industry. New targets that involve
Input-output model increased research output tend to put strain on institutional processes and resources that were previ-
Optimization ously configured to meet the demands of teaching-intensive organizations. The model formulation is
P-graph based on an input-output framework that reflects interdependencies among different employee cate-
gories. The use of the model is illustrated with a representative case study of a typical Higher Education
Institution in the Philippines.
© 2019 Elsevier Ltd. All rights reserved.

1. Introduction transfer and industry engagement success rate (Sengupta and Ray,
2017). Similar correlations are found at national scales between
Higher education institutions (HEIs) play an important role in research output and technology innovation (Wong and Wang,
promoting sustainability through their core functions of research, 2015). Knowledge transfer can then drive eco-innovation that will
instruction and outreach. Current international trends show provide new solutions to environmental or regulatory issues faced
increased emphasis on sustainability in university culture (Adams by industry (Pham et al., 2019). For example, new technologies for
et al., 2018), organizational structure (Soini et al., 2018) and infor- recycling of waste electronic products will be essential to conserve
mation systems (Goni et al., 2017). Integration of sustainability in finite mineral resources and to operationalize Circular Economy
the institutional mission helps shape student attitudes (Dagiliu
te_ (CE) policies (Amato and Beolchini, 2018). Such trends are well
et al., 2018), and potentially paves the way for sustainable entre- established in mature research-intensive HEIs in developed coun-
preneurship that translate research output into commercial green tries, but will also be needed in the developing world in order to
technologies (Fichter and Tiemann, 2018). In universities, a strong enable emerging economies to develop along sustainable growth
research foundation has been found to be linked to knowledge trajectories (Lee et al., 2018). Such gains can be made possible
through the knowledge-based (scientific or technological) outputs
(Hamdoun et al., 2018) as well as social influence (i.e., through
education and public engagement) (Lin et al., 2016). Thus, there is a
* Corresponding author.
E-mail address: kathleen.aviso@dlsu.edu.ph (K.B. Aviso).
need for HEIs in developing countries to take on a greater role in

https://doi.org/10.1016/j.jclepro.2019.03.213
0959-6526/© 2019 Elsevier Ltd. All rights reserved.
812 K.B. Aviso et al. / Journal of Cleaner Production 224 (2019) 811e822

building knowledge capital through research on sustainable 2019). Thus, effective decision support tools for HR management
technologies. in HEIs need to have a strong structural component that accounts
For HEIs, becoming research-intensive is desirable not only for for workflow linkages. Input-output (IO) models present a prom-
the prestige that such status brings, but more so for the capacity ising approach. IO models were first developed for economic sys-
that it gives to contribute to the socio-economic and cultural de- tems analysis (Leontief, 1936). This framework is discussed in great
velopments of their home countries through the continuous gen- detail, along with many extensions, by Miller and Blair (2009); it
eration of knowledge and innovations, as well as formation of more can be readily implemented in spreadsheets or equation-based
critical, creative and productive citizens. Most of the leading modelling software (Tan et al., 2018a). Correa (2002) proposed
research-intensive universities at present were either established the application of IO models to the analysis of large-scale social
as research universities right at the outset, or were teaching- systems. In subsequent works, the IO approach was used for HR
intensive universities that were re-designed as research univer- planning problems in organizations (Correa and Craft, 1999). In this
sities before the onset of the current age of mass and universal approach, the fundamental problem was to determine the correct
higher education. However, in developing countries, which more staffing levels for different employee categories or office units,
often than not came out from long periods of colonization, the given organizational workload. The IO framework is able to account
establishment of research-intensive universities also involved the for workload resulting from the demands of both external and in-
re-designing of selected teaching-intensive universities into ternal customers. This methodology has been applied to deter-
research universities, but already during the era of mass and uni- mining staffing levels in libraries (Correa and Correa, 1996),
versal higher education (Altbach and Salmi, 2011). This means that municipal offices (Correa and Guajardo, 2001) and hospitals
such selected teaching-intensive universities were already (Correa and Parker, 2005), among others. For organizational sys-
burdened with huge numbers of academic staff and students, tems with zero degrees of freedom, unique solutions for correct
making their transition into research universities more challenging, staffing levels can be determined. For systems with positive de-
requiring more financial resources and political will power, and grees of freedom, optimization models based on the IO matrix can
more importantly sharper human resource planning. Sloppy or be formulated. For example, a fuzzy optimization model was
non-existent human resource (HR) planning will only result to developed by Aviso et al. (2018) to determine optimal personnel
dysfunctional combinations of teachers, researchers, administra- reassignment during a temporary crisis in an understaffed organi-
tors, students and support staff. This situation may be reflective of zation. A subsequent work used the P-graph framework to solve a
what is happening in countries such as the Philippines, where the similar problem (Aviso et al., 2017); this approach results from the
few universities that are venturing into such re-designing had capability of representing the optimization of IO systems as a
concentrated their efforts in increasing the numbers of their Process Network Synthesis (PNS) problem (Aviso et al., 2015).
research faculty members without giving due attention on whether In this work, a P-graph model for HR planning in HEIs in tran-
they had the proportionate numbers of research-capable students sition towards becoming more research-intensive is developed. The
and support staff. Thus, HEIs that are currently re-designed, or will common practical response in such institutions is increased
be re-designed, into research-intensive universities will benefit recruitment of faculty researchers. However, such transitions also
from a planning tool that can peer into their current HR configu- require augmentation of staffing levels in administrative support
ration and project at any given development phase their optimal offices in order to manage the increase in overhead workload for
combinations of teachers, researchers, administrators, students and managing research projects. The IO framework provides a means of
support staff relative to their targets and financial considerations. quantifying interactions among staff categories in order to account
It has been found that globalization and internationalization in for both direct and indirect workload changes; the use of P-graph
higher education institutions is accomplished through the then allows optimal (and near-optimal) HR plans to be developed.
achievement of academic excellence, innovation and research The methodology developed in this work is intended to provide a
productivity. Nonetheless, many institutions fail to realize the static snapshot of ideal staffing levels for short- or medium-term
importance of HR management for the creation of a knowledge projections, and does not account for dynamic effects accounting
economy (Chatterja and Kiran, 2017). The core function of human for factors such as turnover, retirement and international mobility.
resource management is to iterate organizational strategy into The rest of this paper is organized as follows. The next section gives
human resource prerogatives. Hence, human resource plays a ma- an overview of the P-graph framework as it applies to HR problems.
jor role in the fulfilment of organizational vision and aspiration A formal problem statement is then given. Then, a representative
(NOE et al., 2003). The main function of HR planning in an HEI is to case study is solved to illustrate the P-graph approach. The alter-
ensure that the staffing complement is sufficient for the institu- native staffing schemes generated are discussed from a practical
tional plans and policies. Effective HR planning in HEIs needs to human resource management perspective. Finally, conclusions and
consider both direct and indirect staffing requirements (Yousif and prospective research directions are given.
Shaout, 2018). In an HEI undergoing transition to a research uni-
versity, an extensive review of the functions of its staffing levels is 2. P-graph framework for HR planning
imperative to address the growing targets. Rigorous analysis of the
nature of the work assigned to the institution's personnel is P-graph is a mathematical framework for solving PNS problems,
essential. Furthermore, HR management should also consider the originally in the area of chemical plant design, but more recently for
different processes and management strategies to contribute to the a broad range of problems characterized by a common structure
growth of the organization (Kucharcíkov a et al., 2015). (Tan et al., 2018b). It is based on five axioms that are common to all
Strategic HR management is imperative in the attainment of an PNS problems (Friedler et al., 1992a). Efficient algorithms for
institution's competitive edge (Allui and Sahni, 2016). There is generating combinatorially feasible networks have been developed
relatively sparse literature on the use of systematic decision sup- using the information implicit in the five axioms (Friedler et al.,
port tools to aid in human resource planning in HEIs (Macke and 1992b). P-graph also allows the determination of the maximal
Genari, 2019). Early examples of methods used include goal pro- structure, which differs from superstructures used in many Process
gramming (Feuer, 1985) and simplified population balance models Integration (PI) problems because it is generated with mathemat-
(Wei, 1998). Optimal organizational structure and task allocation is ical rigor that eliminates the risk of human error (Friedler et al.,
critical for fostering an innovative work culture (Fonseca et al., 1993).
 K.B. Aviso et al. / Journal of Cleaner Production 224 (2019) 811e822 813

The P-graph framework consists of three component algorithms

Maximal structure generation (MSG) is the rigorous, automated

generation of a structure that is the union of all combinatorially
feasible network structures.

Solution structure generation (SSG) is the rigorous elucidation
of all combinatorially feasible network structures.

Accelerated branch-and-bound (ABB) is the efficient determi-
nation of an optimal network based on a predefined objective
function. The computational efficiency of ABB can be attributed
to the use of information inherent in PNS problems to exclude
redundant solutions from the search space.

The P-graph framework can also generate the n-best solutions of

a PNS problem, a capability that is particularly useful for practical
problems. In practice, solutions that are not mathematically
optimal in the strict sense may only be marginally inferior to the
Fig. 1b. P-graph representation for interactions between staff categories in an HEI.
true optimum, and should be examined closely and considered for
implementation. The P-graph approach has been applied to prob-
lems similar in structure to the original PNS problem for plant
design; a review paper by Lam (2013) discusses such applications,
with an emphasis on supply chain optimization. Geographical and INPUT OUTPUT
topical trends in P-graph applications were discussed by Klemes STAFF CATEGORY
STREAMS STREAMS
and Varbanov (2015). Varbanov et al. (2017) examined further
trends and potential directions in a subsequent paper. Notable
applications of P-graph for organizational problems include opti-
mization of business process workflow (Tick, 2007) and workforce Fig. 2a. Block diagram for a staff category.
reallocation (Aviso et al., 2017).
The use of P-graph in the context of HR planning is the direct
such as the administrators and the faculty; administrators super-
result of its similarity to IO models as established by Aviso et al.
vise faculty and the support staff; etc. Each type of employee, for
(2015). The equivalent mathematical model for the P-graph
example a faculty member, can also be viewed in isolation, in block
framework is discussed in the Appendix. Interactions among
diagram form (Fig. 2a) or P-graph form (Fig. 2b). The input streams
different staff categories in an HEI can be represented as an IO block
represent demands that the average faculty member (or any staff
diagram (Fig. 1a) or alternatively as a P-graph (Fig. 1b). In this
category) places on other employees, as measured in person-hours.
modelling framework, all work is measured in terms of person-
Such demands on other employees' time will be indicated by
hours, which are assumed to be proportional to actual output
negative numerical values, as they signify “consumption” of other
(e.g., number of research papers published) based on average staff
people's productive output. The output streams represent different
productivity levels. The streams can represent typical interactions,
types of work output of the average faculty member (or any staff
or internal transactions, that signify typical demands that each type
category), also measured in person-hours. Work output is indicated
of employee makes on colleagues. For example, support staff pro-
by positive numerical values.
vide various types of administrative support to other employees

ADMINISTRATION SUPERVISION
(SUPERVISION)

FACULTY RESEARCH
(RESEARCH)

SUPPORT SUPPORT STAFF

(SUPPORT)

Fig. 1a. Block diagram for interactions between staff categories in an HEI.
814 K.B. Aviso et al. / Journal of Cleaner Production 224 (2019) 811e822

Current efforts by the national government via the Department of

Science and Technology and the Commission on Higher Education
to build a knowledge-based economy has led to initiatives to in-
crease research intensity in these HEIs. Hence, the transition from a
teaching-intensive to a research-intensive institution is actually
becoming a trend in as far as this handful leading Philippine HEIs
and the numerous Philippine state universities and colleges are
concerned. Bernardo (2003) suggested Philippine HEIs need to
consider two things to achieve such a transition. Firstly, HEIs should
invest in developing their research capabilities through training
their faculty members, providing research infrastructure, hiring
more research support staff and formulating workable research
policies. Secondly, these institutions should be able to identify
research areas where collaborations with external partners can be
further explored. As such, Philippine HEIs must examine their
research development programs to be able to develop their own
specific niches to contribute to the globally placed knowledge
production system that is currently taking place.
For the typical HEI, there are seven staff categories, as given in
Table 1. Note that faculty members are further classified as pure
teaching, pure research and combined research/teaching.
Staff categories are modelled as operating units in P-graph. The
P-graph models used in this work are all available upon request via
communication to the corresponding author, and can be run using
the software P-graph Studio hosted by the University of Pannonia
(P-graph, 2019). Based on typical job descriptions, six types of
interaction between the different staff categories are identified,
namely, general supervision/management, research mentoring,
research, teaching (including consultation), knowledge transfer
Fig. 2b. P-graph representation of a staff category.
and support. All workflows are measured in person-hours per
week, with the number of person-hours being proportional to the
actual output demands (e.g., the average number of person-hours
3. Formal problem statement needed to produce one scientific article as research output can be
estimated empirically). The interactions between staff classifica-
In the proposed methodology, the problem addressed via the P- tions can be viewed as “transaction costs” of working within an
graph model can be formally stated as: organization, and are modelled as material streams in P-graph.
Examples include routine tasks such as meetings, sending or

Given a set of homogeneous categories of HEI staff, within which reading e-mails, and other similar activities. The interactions are
work assignments are assumed to be fully interchangeable, and summarized in Table 2 where negative entries indicate inputs or
for which average productivity and salary level is also fixed; need while positive entries indicate output or productivity of a staff

Given a set of workflows or tasks, which are associated with staff category. The equivalent P-graph model is shown in Fig. 3 which
categories as inputs or outputs; can be better understood with the accompanying nomenclature

Given that each staff category has a constant ratio of workflow found in Table 3. As initially discussed, the staff categories which
inputs and outputs, based on the employee's job description and are indicated as column headings in Table 2 are represented as
interactions with co-workers; rectangular boxes (or process units) in Fig. 3, while the nature of

Given the current staffing levels and workflow outputs; interaction or function which can be found in the rows of Table 2

Given that the HEI has a predefined set of net output for each of are represented by circular nodes in Fig. 3. Fictitious nodes to
the workflows that arise from strategic targets; represent the actual count of the workforce are also indicated, this

The objective is to determine the staffing level in each category includes the nodes labelled as FAC to represent faculty who can
to deliver the required net output at minimum cost. The provide different types of functions, SS to represent support ser-
resulting recommended staffing levels are intended to be used vices and ADMIN to represent human resources who function
as benchmarks or targets for HR management. entirely as administrators. For example, a faculty who is entirely
involved in teaching (e.g. 2nd column in Table 2) can provide 36
In addition to the generic assumptions used here, additional person-hours/week of teaching service. However, the teaching
case-specific assumptions can also be added, as will be illustrated in faculty will require 2 person-hours/week of supervision and 12
the next section. person-hours/week of support services. This is represented in the
process unit labelled FAC_T where an outgoing stream labelled 36
u/y is directed towards the blue colored node labelled TEACH to
4. Case study on a typical HEI in the Philippines represent the 36 person-hours/week of teaching provided by the
teaching faculty. Similarly, input streams labelled as 2 u/y comes
The Philippine higher education system consists predominantly from the node labelled as SUPERVISION and 12 u/y from the node
of teaching-oriented universities that were historically established labelled as SUPPORT to represent the 2 person-hours/week and 12
to educate a professional workforce for a largely local job market. person-hours/week requirements respectively. To generate Fig. 3,
For example, only eight Philippine HEIs appear among the 500 the same procedure is done for all the remaining staff categories in
institutions listed in the 2019 QS Asian University Rankings. Table 2. The properties of the nodes and process units can be
 K.B. Aviso et al. / Journal of Cleaner Production 224 (2019) 811e822 815

Table 1
Summary of HEI staff classifications and functions.

Human Resource Classification Functions

Faculty General academic staff with teaching, research and extension functions; the relative proportion of these functions can give rise to further
subcategories.
Teaching assistant Contractual staff engaged to assist in teaching-related functions.
Research staff Contractual staff engaged to assist in research-related functions. This category includes postdocs and research associates, but excludes
graduate students.
General support staff Staff employed for various organizational support functions (e.g., finance, procurement, legal, etc.)
Faculty with administrative Administrators appointed to part-time management functions, while still retaining partial faculty duties.
functions
Pure administrators Administrators appointed to full-time management functions.

Table 2
HEI staff interaction matrix in person-hours/week.

Faculty Faculty Faculty (Teaching/ Teaching Research Support Administrator/ Pure

(Teaching) (Research) Research) Assistant Staff Staff Faculty Administrator

General 2 2 2 2 2 4 15 40
supervision
Research 0 6 3 0 4 0 2 0
mentoring
Research 0 24 12 0 36 0 8 0
Teaching 36 0 18 36 0 0 12 0
Knowledge 0 6 3 0 2 0 2 0
Transfer
Support 12 12 12 8 10 40 30 60

Fig. 3. Equivalent P-graph representation of Table 2.

adjusted in the P-graph studio software to indicate minimum and 4.1. Baseline scenario
maximum limits as defined by the problem specifications.
The average weekly salary for each staff category is reflected in Using SSG, it is possible to identify 86 different structures,
Table 4. Note that the approximate exchange rate is PhP60 ¼ V1. indicating that different combinations of these staff categories can
Additional assumptions used here are as follows: exist in the HEI. Due to space constraints, these potential structures
are not shown here. The baseline scenario assumes the following:

Legal restrictions force HEIs to retain their personnel.

The retained personnel can be trained to gain improved skills to
the minimum weekly total output for teaching is equivalent to
allow them to assume new tasks pertaining to research and 18,000 person-hours
knowledge transfer.
816 K.B. Aviso et al. / Journal of Cleaner Production 224 (2019) 811e822

Table 3
Nomenclature for P-graph representation of case study.

P-graph Representation Label Meaning

ADMIN Human resource hired as administrators

FAC Human resource hired as faculty
SS Human resource hired as support staff
RES_MENT Research mentoring function
SUPERVISION Supervisory function
SUPPORT Support function
KNOW_TRAN Knowledge transfer function
RESEARCH Research function
TEACH Teaching function
AD_FAC Faculty with administrative, teaching and research duties
FAC_R Faculty with only research duties
FAC_T Faculty with only teaching duties
FAC_TR Faculty with both research and teaching duties
GSS General Support Staff
PURE_AD Human resource with purely administrative duties
RS Research staff
TA Teaching assistant

Table 4
Baseline salary and number of personnel.

Faculty Faculty Faculty (Teaching/ Teaching Research Support Administrator/ Pure Total
(Teaching) (Research) Research) Assistant Staff Staff Faculty Administrator

Number of personnel 417 0 79 0 0 248 132 0 876

Salary per person (thousand 27 36 33 12 12 10.8 33 45 N/A
PhP/week)
Total Salary (M PhP/week) 11.250 0 2.594 0 0 2.674 4.358 0 20.877

the minimum weekly total output for research is 1,500 person- Suppose that the HEI seeks to increase research output by 300%
hours over a planning horizon of five years, under the assumption that

the minimum weekly total output for knowledge transfer is 500 such an increase will require a proportionate level of increase in
person-hours research person-hours. The new requirements and assumptions are

there are no teaching assistants and no faculty that have purely as follows:
research duties

the minimum weekly total output for teaching is 18,000 person-
The minimum weekly total output corresponds to a university hours
with an output profile of 90% teaching, 7.5% research and 2.5%
the minimum weekly total output for research is 6,000 person-
knowledge transfer. Upon optimization, the baseline scenario re- hours
sults in an academic staff requirement of 628 faculty members, of
the minimum weekly total output for knowledge transfer is 500
which 417 have only teaching duties, 79 have combined teaching person-hours
and research, 132 faculty administrators. Also, 248 support staff are
the total number of faculty and support staff employed should
needed as indicated in Table 4. This structure generates 18,000 be more than the baseline scenario
person-hours of teaching, 2,000 person-hours of research and 500
a maximum of 500 teaching assistants may be hired while an
person-hours of knowledge transfer. The cost of personnel salaries unlimited number of research faculty can be employed
(including employee benefits) is PhP 20.877 M per week (V 347,956
per week), or PhP 1,085 M per annum (V 18,093,712 per annum). The HR problem that faces management is to determine the
The last row in Table 4 summarizes the total personnel cost asso- most cost-effective means of achieving this target, by identifying
ciated with each staff category. The detailed allocation of workflow possible ideal staffing schemes and their associated costs in salaries
for the baseline scenario is summarized in Table 5 and illustrated in and employee benefits. It is possible to use the capabilities of P-
P-graph form in Fig. 4. graph to generate structurally distinct HR plans, each of which

Table 5
Baseline scenario total workflows (in person-hours/week).

Faculty Faculty Faculty (Teaching/ Teaching Research Support Administrator/ Pure Final
(Teaching) (Research) Research) Assistant Staff Staff Faculty Administrator Output

General 833 0 157 0 0 991 1,981 0 0

supervision
Research 0 0 236 0 0 0 264 0 500
mentoring
Research 0 0 943 0 0 0 1,057 0 2,000
Teaching 15,000 0 1,415 0 0 0 1,585 0 18,000
Knowledge 0 0 236 0 0 0 264 0 500
Transfer
Support 5,000 0 943 0 0 9,906 3,962 0 0
 K.B. Aviso et al. / Journal of Cleaner Production 224 (2019) 811e822 817

Fig. 4. HR structure for baseline scenario.

represents an option to be considered for implementation. These failure to do so can result in process bottlenecks within the HEI.
alternatives can be evaluated based on cost, as well as non- This new structure meets the desired teaching and research out-
quantitative aspects that cannot be directly integrated in P-graph. puts, and results in 692 person-hours for knowledge transfer. Total
Key features of the four HR scenarios generated using the P-graph cost of personnel salaries increases to PhP 20.61 M per week
model are described in Table 6. Each scenario is then discussed in (V376,785 per week), or 8.29% increase from the baseline.
detail in the following subsections. In all the succeeding scenarios, While this is a cost-optimal solution, it can be seen that the
the HEI retains its current employees due to legal restrictions. A segregation of teaching and research functions can have serious
summary of the different scenarios is shown in Table 7. repercussions on the quality of education, as it effectively hinders
the bulk of the student population from learning from the top re-
4.2. Scenario A: cost-optimal staffing plan searchers of the HEIs. Furthermore, excessive reliance on research
staff (who generally have lower salaries than faculty) may threaten
The cost-optimal network can be generated simply by directly the long-term sustainability of the research growth of the institu-
minimizing total salaries and is illustrated in Fig. 5. This solution tion. Thus, the capability of P-graph to generate alternative struc-
requires a slight shift in workforce structure such that there are tures becomes valuable for HR planning. In this case, 23 additional
now teaching assistants and research staff. For this scenario only 1 near-optimal structures are generated which provide options for
additional teaching assistant is necessary. There should be 448 the kind of HR structure an HEI will prefer.
faculty dedicated to purely teaching duties, and 25 research faculty
without teaching duties. There is also a requirement for 115 4.3. Scenario B: use of teaching assistants
research staff plus 155 faculty administrators. This new structure
maintains the faculty count at 628. There is now a need to employ The near-optimal structure shown in Fig. 6 demonstrates a HR
research staff, teaching assistants and identify faculty dedicated to profile which employs teaching assistants, but does not have
just conducting research. The general support staff requirement has research staff. This network was identified from the 23 near-
increased to 287, which is a 15.7% increase from the baseline. The optimal structures generated using the assumptions in Scenario
latter increase is difficult to deduce intuitively, but in practice, A. However, it is possible to restrict the optimization model to this

Table 6
Key Features of four HR scenarios.

Scenarios Features

A Extensive use of research and staff personnel to assist faculty

B Extensive use of teaching assistants to free up faculty
C No dedicated teaching assistant or research personnel
D All faculty have multiple functions to perform (e.g. teaching and research)
818 K.B. Aviso et al. / Journal of Cleaner Production 224 (2019) 811e822

Table 7
Summary of scenarios.

Number of Number of Number of Faculty Number of Number of Number of Number of Number of Pure Cost (in M
Faculty Faculty (Teaching/Research) Teaching Research Staff Support Staff Administrator/ Administrator PhP/week)
(Teaching) (Research) Assistant Faculty

Baseline 417 0 79 0 0 248 132 0 20.877

Scenario A 448 25 0 1 115 287 155 0 20.607
Scenario B 268 195 0 177 0 298 165 0 25.041
Scenario C 500 250 0 0 0 331 0 71 29.250
Scenario D 0 0 628 186 0 337 0 74 29.947

Fig. 5. HR structure for Scenario A.

particular network topology where the maximum research staff 4.4. Scenario C: expansion of faculty talent pool
and faculty with dual duties of research and teaching are set to zero.
The total faculty population is still 628. Of this total, 268 faculty This scenario relies on aggressive expansion of the faculty talent
(43%) are assigned to perform purely teaching roles and 195 (31%) pool through recruitment, and does not consider hiring teaching
to purely research roles. The remaining 165 (26%) function as fac- assistants and research staff to handle the increased requirements
ulty administrators. The faculty pool is now supported by 177 new for research productivity. Similar to the structure in Scenario B, this
teaching assistants, who can absorb the teaching load previously network was identified from the 23 near-optimal structures
assigned to research faculty. This structure results in a total cost of generated using the assumptions in Scenario A, the network is
PhP 25.04 M per week (V417,358 per week), and meets the output shown in Fig. 7. As a result, the total number of faculty needed to
requirements of 18,000 person-hours of teaching, 6,000 person- meet the minimum requirement for research is now 750, which is a
hours for research and 1,500 person-hours for knowledge trans- 19.4% increase from the baseline scenario. Furthermore, an addi-
fer. The cost of salaries is 19.95% higher than the baseline scenario. tional 71 personnel have to be hired to function as pure adminis-
This scheme offers the prospect of stronger long-term research, as trators. The main feature of this structure is having the personnel
it relies on research faculty to increase output levels. It can be work only on dedicated roles, where 500 faculty (67%) are involved
argued that this HR plan is still subject to a similar weakness as purely on teaching and 250 faculty (33%) are focused purely on
Scenario A, in that research knowledge will have limited capacity to research. A total of 331 support staff are needed to support these
trickle down into the curriculum. activities, which is a 33.5% increase from the baseline. This network
 K.B. Aviso et al. / Journal of Cleaner Production 224 (2019) 811e822 819

Fig. 6. HR structure for Scenario B.

Fig. 7. HR structure for Scenario C.

results in a total cost of PhP 29.25 M per week (V487,500 per week) proportions. The network was also identified from the 23 near-
with 18,000 person-hours of teaching, 6,000 person-hours for optimal structures obtained using the assumptions of Scenario A
research and 1,500 person-hours for knowledge transfer. The total and the network is shown in Fig. 8. This scenario maintains the
cost of staff salaries is 40.10% higher than in the baseline scenario. faculty population at 628, as in the baseline scenario. However, they
are now expected to be engaged in both teaching and research.
Furthermore, an additional 74 personnel should be hired to act as
4.5. Scenario D: integrated research and teaching
pure administrators, along with 186 teaching assistants. The latter
are needed to take up the teaching load so that faculty can spend
For this scenario, the faculty members are expected to perform
more work hours on research. Furthermore, the number of support
multiple functions (i.e. both teaching and research) in various
820 K.B. Aviso et al. / Journal of Cleaner Production 224 (2019) 811e822

Fig. 8. HR structure for Scenario D.

staff should be increased to 337, which is a 36% increase from the (Salmi, 2009). Of these options, upgrading an existing HEI is typi-
baseline. As a result, this HR structure has the total cost of PhP cally the least costly path, but is also subject to management con-
29.95 M per week (V499,120 per week), and can provide 18,000 straints such as institutional culture and inertia. Institutional
man-hours for teaching, 7,536 man-hours for research and 1,884 reforms need to be managed to ensure acceptance by the university
man-hours for knowledge transfer. This scenario thus exceeds the personnel and stakeholders (e.g., students and alumni), and should
minimum man-hour requirement for both research and knowledge also be adequately supported by rigorous projections. The model
transfer, but total salaries increase by 43.44%. developed here is intended to provide support for both strategic
Qualitative analysis of intangible aspects of this HR plan sug- decision-making and communication.
gests that it may be the most sustainable one among the four The absence of systematic HR planning in many HEIs means that
scenarios. First, reliance on faculty to do research (including the increasing number of faculty researchers, with their increasing
research mentoring) ensures continuity of research, through the capacity to undertake research, is not accompanied with propor-
training of early career researchers (i.e., junior faculty) by their tionate increases in the number of research staff and support staff.
seniors (Carpintero, 2015). Development of the talent pool in this Thus, the capacity of support units (e.g., finance, legal or procure-
manner, while more costly in the short-term, will ensure the long- ment offices) can become a bottleneck during a period of rapid
term sustainability of gains made in research output. Training of growth in research intensity. Such bottlenecks can result in friction
junior faculty will ensure that the HEI staffing becomes robust in among co-workers and demoralization among researchers, and
the face of disruptions due to turnover or retirement, which needs may cause compliance issues for the institution (e.g., project delays
to be considered as part of HR dynamics (Wei, 1998). Integration of and cost overruns). Hence, the computer-aided capability to
teaching and research functions, even in varying proportions, en- generate a menu of HR planning scenarios could benefit not only
sures that knowledge generation within the HEI remains strongly HEIs in the Philippines, but also universities with a similar macro-
linked to the students’ educational experience. In particular, con- environment that are undergoing a similar transition from
tact with active researchers will thus ensure that the HEI produces teaching-intensive to research-intensive institutions. In practice,
graduates who understand the process of innovation and knowl- university leaders need to be able to map out medium-term plans
edge generation. Such additional knowledge ultimately translates projecting directions over a period of approximately 3e5 years.
into better career prospects and mindsets as educated citizens. Such a time frame is sufficient to allow for policy directions to
trickle down through the management levels, so that decisions for
staff recruitment, training, and reclassification can be done with
5. Implications for HR management in HEIs adequate lead time. The P-graph model can generate alternative HR
plans to be considered for implementation, and the final plan
HEIs play a critical role in establishing an ecosystem that enables selected serves to provide benchmark or targets for staff recruit-
innovation-drive development through generation of both human ment and reassignment to suit strategic goals of the HEI.
and knowledge capital. These influences are reflected, for example, The model as presented here assumes that general staffing
in the Global Innovation Index (Cornell University, Institut categories cut across the entire institution. The resulting staffing
Europe en d'Administration des Affaires, and World Intellectual
plans are thus high-level benchmarks for use at the level of central
Property Organization, 2018) and other similar country-level as- university administration. However, in principle, the same model-
sessments. Given the importance of HEIs in national development, ling framework can be applied at a higher level of resolution to
governments typically stimulate upgrading or merging of in- reflect individual departments or offices. The current model has
stitutions, or else create entirely new ones based on strategic needs
 K.B. Aviso et al. / Journal of Cleaner Production 224 (2019) 811e822 821

some limitations that can be dealt with in future variants. For

example, each generated HR plan represents a static benchmark, min cT x (A-1)
but a multi-period variant spanning multiple decades may be
Subject to:
desired in some cases. Variations in employee age and productivity
are also not reflected. Other aspects not directly incorporated in the Ax ¼ y (A-2)
model can arise in practical situations. For instance, in many HEIs in
developing countries, university lecturers may lack an appropriate
yL  y  yU (A-3)
doctorate degree; thus, completion of such degrees by young staff
can be considered within a time frame that is sufficient for Ph.D.
studies. Such training results in improvement in productivity which diagðbÞxL  x  diagðbÞxU (A-4)
will result in changes in the model coefficients over the course of
time. The impact of internationalization through short- or long- bj 2f0; 1g cj (A-5)
term employment of foreign staff also needs to be integrated, as
this is an essential aspect of modern higher education. The objective (Eq. A-1) is to minimize the total cost incurred
resulting from the salary of the staff. This is a product of the salary
per person per week which is contained in the salary vector ðcT Þ
6. Conclusions
and the total number of personnel per staff category as indicated in
the total workforce vector ðxÞ. As each type of staff performs their
A P-graph model was developed for optimal human resource
tasks, their activities will include routine work and interacting with
planning in HEIs in transition from being primarily teaching-
other staff classifications which can be represented by the inter-
intensive towards becoming more research-oriented. Such transi-
action matrix, A. Depending on the quantity and type of available
tions are needed as HEIs need to play a role in developing both
workforce, the entire organization thus generates the net output of
human and knowledge capital for achieving sustainability in
work as indicated by the net output vector, y, (Eq. A-2). The net
developing countries. Thus, strengthening the capacity of HEIs can
output of work should meet the targets set by the organization as
play an important role in forming sustainable development paths of
defined by the lower ðyL Þ and upper ðyU Þ demand vector targets
emerging economies through scientific research and knowledge
(Eq. A-3). Furthermore, the total number of staff per category
transfer to stimulate eco-innovation in industry. The model
should be within the defined lower ðxL Þ and upper ðxU Þ workforce
developed here takes into account the interdependencies among
limits (Eq. A-4) where b is a vector of binary variables (Eq. A-5)
general staff categories, so that appropriate staffing levels can be
which indicates whether a certain workforce category j is present
identified to support institutional targets. A case study represen-
ðbj ¼ 1Þ or not ðbj ¼ 0Þ.
tative of a typical HEI in the Philippines was solved to illustrate the
model's capabilities. Implications for cost-effective human resource
management were also discussed. The findings show that there are References
different strategies which can be implemented to achieve targets
Adams, R., Martin, S., Boom, K., 2018. University culture and sustainability:
for increased research productivity. These strategies can look at designing and implementing an enabling framework. J. Clean. Prod. 171,
employing different HR structures such that top management can 434e445.
identify the required human and financial resources. An examina- Allui, A., Sahni, J., 2016. Strategic human resource management in higher education
institutions: empirical evidence from Saudi. Procedia Soc. Behav. Sci. 235,
tion of the different options will require an assessment of man- 361e371.
agement priorities since selecting the kind of HR structure can Altbach, P.G., Salmi, J., 2011. Introduction. In: Altbach, P.G., Salmi, Jamil (Eds.), The
potentially influence institutional culture and sustainability. Road to Academic Excellence: the Making of World-Class Research Universities.
World Bank, Washington, DC, pp. 1e9.
Future work can focus on developing and applying different Amato, A., Beolchini, F., 2018. End of life liquid crystal displays recycling: a patent
model variants. For example, it is possible to develop a multi-period review. J. Environ. Manag. 225, 1e9.
variant to account for annual adjustments in strategic targets of an Aviso, K.B., Cayamanda, C.D., Solis, F.D.B., Danga, A.M.R., Promentilla, M.A.B.,
Santos, J.R., Tan, R.R., 2015. P-graph approach for GDP-optimal allocation of
organization. Robust model formulations can also be developed to resources, commodities and capital in economic systems under climate change-
allow for uncertainties in staff performance levels; this approach induced crisis conditions. J. Clean. Prod. 92, 308e317.
can also be coupled with Monte Carlo simulation so that alternative Aviso, K.B., Cayamanda, C.D., Mayol, A.P., Yu, K.D.S., 2017. Optimizing human
resource allocation in organizations during crisis conditions: a P-graph
staffing strategies can be simulated in silico. Temporal aspects to
approach. Process Integr. Optim. Sustain. 1, 59e68.
account for staff training, turnover and attrition can also be incor- Aviso, K.B., Mayol, A.P., Promentilla, M.A.B., Santos, J.R., Tan, R.R., Ubando, A.T.,
porated, along with other widely recognized indicators of HEI Yu, K.D.S., 2018. Allocating human resources in organizations operating under
crisis conditions: a fuzzy input-output optimization modeling framework.
quality (e.g., international mobility and research income). It is also
Resour. Conserv. Recycl. 128, 250e258.
important for the model to be field-tested for actual applications in Bernardo, A.B.I., 2003. International higher education: models, conditions and is-
developing countries. sues. In: Tullao, T.S. (Ed.), Education and Globalization. Philippine Institute for
Development Studies, Manila, pp. 213e272.
Carpintero, S., 2015. Qualities that mentors in the university setting should have.
Acknowledgement Procedia Soc. Behav. Sci. 197, 255e258.
Chatterjia, N., Kiran, R., 2017. Role of human and relational capital of universities as
underpinnings of a knowledge economy: a structural modelling perspective
This work was supported by the Philippine Commission on from north Indian universities. Int. J. Educ. Dev. 56, 52e61.
Higher Education (CHED) via the Philippine Higher Education Cornell University, Institut Europe en d'Administration des Affaires, and World In-
Research Network (PHERNet) Sustainability Studies Program at De tellectual Property Organization, 2018. The Global Innovation Index 2018:
Energizing the World with Innovation. Fointainebleu, and Geneva, Ithaca.
La Salle University. We are also grateful to the P-graph Studio Correa, H., 2002. An input-output operationalization of societal systems. J. Socio
software team based at the University of Pannonia, Hungary. Econ. 31, 115e123.
Correa, H., Correa, V., 1996. An application of input-output analysis to the admin-
istration of a library. Libr. Inf. Sci. Res. 18, 343e356.
Appendix Correa, H., Craft, J., 1999. Input-output analysis for organizational human resources
management. Omega 27, 87e99.
Correa, H., Guajardo, S.A., 2001. An application of input-output analysis to city's
The P-graph model for staffing allocation can be formulated into municipal government. Soc. Econ. Plann. Sci. 35, 83e108.
the equivalent mixed integer linear program as follows: Correa, H., Parker, B.R., 2005. An application of organizational input-output analysis
822 K.B. Aviso et al. / Journal of Cleaner Production 224 (2019) 811e822

to hospital management. Soc. Econ. Plann. Sci. 39, 307e333. in technological and vocational higher education: balanced scorecard measures
Dagiliu
te,
_ R., Liobikiene,_ G., Minelgaite, _ A., 2018. Sustainability at universities: with uncertainty. J. Clean. Prod. 120, 1e12.
students' perceptions from green and non-green universities. J. Clean. Prod. 181, Macke, J., Genari, D., 2019. Systematic literature review on sustainable human
473e482. resource management. J. Clean. Prod. https://doi.org/10.1016/j.jcle-
Feuer, M.J., 1985. Organizational decline, extended work life and implications for pro.2018.10.091 (in press).
faculty planning. Soc. Econ. Plann. Sci. 3, 213e221. Miller, R.E., Blair, P.D., 2009. Input-output Analysis: Foundations and Extensions,
Fichter, K., Tiemann, I., 2018. Factors influencing university support for sustainable second ed. Cambridge University Press, Cambridge, UK.
entrepreneurship: insights from explorative case studies. J. Clean. Prod. 175, Noe, R.A., Hollenbeck, J.R., Gerhart, B., Wright, P.M., 2003. Human Resources Man-
512e524. agement: Gaining a Competitive Advantage, fourth ed. McGraw-Hill, New York.
Fonseca, T., de Faria, P., Lima, F., 2019. Human capital and innovation: the impor- P-graph, 2019. www.p-graph.com. (Accessed 4 February 2019).
tance of the optimal organizational task structure. Res. Pol. https://doi.org/ Pham, D.D.T., Paille, P., Halilem, N., 2019. Systematic review on environmental
10.1016/j.respol.2018.10.010 (in press). innovativeness: a knowledge-based resource view. J. Clean. Prod. 211,
Friedler, F., Tarjan, K., Huang, Y.W., Fan, L.T., 1992a. Graph-theoretic approach to 1088e1099.
process synthesis: axioms and theorems. Chem. Eng. Sci. 47, 1973e1988. Salmi, J., 2009. The Challenge of Establishing World-Class Universities. World Bank,
Friedler, F., Tarjan, K., Huang, Y.W., Fan, L.T., 1992b. Combinatorial algorithms for Washington, DC.
process synthesis. Comput. Chem. Eng. 16, 313e320. Sengupta, A., Ray, A.S., 2017. University research and knowledge transfer: a dynamic
Friedler, F., Tarjan, K., Huang, Y.W., Fan, L.T., 1993. Graph-theoretic approach to view of ambidexterity in British universities. Res. Pol. 46, 881e897.
process synthesis: polynomial algorithm for maximal structure generation. Soini, K., Jurgilevich, A., Pietika €inen, J., Korhonen-Kurki, K., 2018. Universities
Comput. Chem. Eng. 17, 929e942. responding to the call for sustainability: a typology of sustainability centres.
Goni, F.A., Chofreh, A.G., Mukhtar, M., Sahran, S., Shukor, S.A., Klemes, J.J., 2017. J. Clean. Prod. 170, 1423e1432.
Strategic alignment between sustainability and information systems: a case Tan, R.R., Aviso, K.B., Promentilla, M.A.B., Yu, K.D.S., Santos, J.R., 2018a. Input-Output
analysis in Malaysian public Higher Education Institutions. J. Clean. Prod. 168, Models for Sustainable Industrial Systems: Implementation Using LINGO.
263e270. Springer, Singapore.
Hamdoun, M., Jabbour, C.J.C., Othman, H.B., 2018. Knowledge transfer and organi- Tan, R.R., Aviso, K.B., Klemes, J.J., Lam, H.L., Varbanov, P.S., Friedler, F., 2018b. To-
zational innovation: impacts of quality and environmental management. wards generalized process networks: prospective new research frontiers for the
J. Clean. Prod. 193, 759e770. P-graph framework. Chem. Eng. Trans. 70, 91e96.
Klemes, J.J., Varbanov, P.S., 2015. Spreading the message: P-Graph enhancements: Tick, J., 2007. P-Graph-based workflow modelling. Acta Polytech. Hung. 4, 75e88.
implementations and applications. Chem. Eng. Trans. 45, 1333e1338. Varbanov, P.S., Friedler, F., Klemes, J.J., 2017. Process network design and optimi-
Kucharcíkova , A., Tokarcíkova , E., Blaskov
a, M., 2015. Human capital management e sation using P-graph: the success, the challenges and potential roadmap. Chem.
aspect of the human capital efficiency in university education. Procedia Soc. Eng. Trans. 61, 1549e1554.
Behav. Sci. 177, 48e60. Wei, J., 1998. A steady-state planning model for faculty balance. Ind. Eng. Chem. Res.
Lam, H.L., 2013. Extended P-graph applications in supply chain and process network 37, 2078e2080.
synthesis. Curr. Opin. Chem. Eng. 2, 475e486. Wong, C.-Y., Wang, L., 2015. Trajectories of science and technology and their co-
Lee, C.T., Lim, J.S., Fan, Y.V., Liu, X., Fujiwara, T., Klemes, J.J., 2018. J. Clean. Prod. 176, evolution in BRICS: insights from publication and patent analysis. J. Inf. 9 (1),
726e735. 90e101.
Leontief, W., 1936. Quantitative input and output relations in the economic system Yousif, M.A., Shaout, A., 2018. Fuzzy logic computational model for performance
of the United States. Rev. Econ. Stat. 18, 105e125. evaluation of Sudanese Universities and academic staff. J. King Saud Univ.
Lin, M.H., Hu, J.Y., Tseng, M.L., Chiu, A.S.F., Lin, C.Y., 2016. Sustainable development Comput. Inf. Sci. 30, 80e119.