The current issue and full text archive of this journal is available on Emerald Insight at:

https://www.emerald.com/insight/2515-8961.htm

Multicriteria
Multicriteria decision choices for decision
investment in innovative choices

upper-middle income countries

Marcela do Carmo Silva, Helder Gomes Costa and 321
Carlos Francisco Simões Gomes Received 15 February 2019
Department of Industrial Engineering, Revised 1 September 2019
Universidade Federal Fluminense, Niteroi, Brazil 27 November 2019
Accepted 15 January 2020

Abstract
Purpose – The purpose of this paper is to observe how to invest in upper-middle income countries via an
innovation perspective following global innovation index (GII) by multicriteria decision aid (MCDA)
approach, once MCDA was designed to support subjective decisions.
Design/methodology/approach – Pearson’s correlation was the milestone for understanding
innovation indicators at upper-middle income countries profiles. In a MCDA first step, the analytical
hierarchy process (AHP) was applied to obtain the criteria weight. In this step, the judgments or evaluations
inputted in AHP were collected from a sample composed by five experts in GII. After getting the criteria
weights compose to GII, Borda and Preference Ranking Organization Method for Enrichment Evaluations
(PROMÉTHÉE) methods were applied to obtain an MCDA-based GII. The inputs for this second step were:
the weights come from AHP output; and the countries performance came from GII data.
Findings – As a result, it was found out the upper-middle countries’ rank to invest and groups with
countries acting like “hubs” or “bridges” for economic sectors in near countries; when they are grouped
according to their maximum and minimum scores profiles, observing not only a particular region but also
similar profiles at diverse world areas.
Originality/value – Pearson-AHP-PROMÉTHÉE works as a supportive decision tool for several and
complex investment perspectives from criteria and alternatives analysis regarding innovation indicators for
upper-middle income countries. This combination also demonstrates grouping possibilities, aligning profiles
and not only ranking countries for investment and eliminating others but also grouping countries with similar
profiles via innovation indicators MCDA combined application.
Keywords MCDA, AHP, Global innovation index, Pearson,
World intellectual property organization, PROMÉTHÉE
Paper type Research paper

1. Introduction
The global innovation index (GII) is a yearly report where innovative countries are analyzed
and ranked considering their innovative profile. It is adopted by the World Intellectual
Property Organization (WIPO) for observing innovative tendencies regarding countries and
their perspectives via seven innovation indicators, where five are innovation input

© Marcela do Carmo Silva, Helder Gomes Costa and Carlos Francisco Simões Gomes. Published in
Innovation and Management Review. Published by Emerald Publishing Limited. This article is Innovation & Management
published under the Creative Commons Attribution (CC BY 4.0) license. Anyone may reproduce, Review
Vol. 17 No. 3, 2020
distribute, translate and create derivative works of this article (for both commercial and non- pp. 321-347
commercial purposes), subject to full attribution to the original publication and authors. The full Emerald Publishing Limited
2515-8961
terms of this license may be seen at http://creativecommons.org/licences/by/4.0/legalcode DOI 10.1108/INMR-02-2019-0016
INMR indicators and the other two are innovation output indicators. WIPO is a United Nations’
17,3 self-funding agency with 189 member states. It was established in 1967 as an institution in
Stockholm for promoting intellectual property global protection and innovative tendency
studies (Olwan, 2011).
On the one hand, GII is based on weighted sums and was not designed to work with
subjective features. On the other hand, multicriteria decision aid (MCDA) methods were
322 developed to work in situations where subjectivity is in the core of the problem. To
fulfil this gap, this paper proposes the use of MCDA methods to review the GII
calculation. Some of the research questions are: “Considering upper-middle income
countries’ scores in innovation indicators; which of them are better options for
investment? Is it possible to invest in some particular groups? Is it possible to align
Pearson’s correlation to analytical hierarchy process (AHP) to create weights for results
and analysis made in Preference Ranking Organization Method for Enrichment
Evaluations (PROMÉTHÉE)?
Section 2 observes GII 2016 innovation indicators and multicriteria methods used.
Section 3 describes the used multicriteria tools. Section 4 discusses results followed by final
considerations.

2. Background
2.1 Global innovation indicators
As reported by Cornell University, INSEAD, and WIPO (2016), the GII (edition 2016) is
composed of 82 indicators covering 16 composed indicators, 5 survey questions and 61 raw
data dimensions of innovative skills of the countries. Such indicators are analyzed through
the comparison among countries, consolidating their scores according to their pillars and
sub-pillars (Cornell University, INSEAD and WIPO, 2015).
As observed in Silva, Gavião, Gomes, and Lima (2017), a MCDA ranking method
(TOPSIS – Technique for Order of Preference by Similarity to Ideal Solution) was used
for observing Latin America and Caribbean countries’ innovative profiles in GII 2015.
In our paper, the challenge is to observe GII 2016’s upper-middle income innovative
countries by using other MCDA tools to analyze investments opportunities in diverse
world areas and upper-middle income countries. The results will be presented
according to decision-makers’ (DM) expectations taking into consideration the
possibility of these countries to be part of a group of investment receivers according to
their innovation indicators.
However, to analyze GII, it is important to be aligned with the six innovation policy
principles defined by GII 2015, but still present in GII 2016 in its restructured indicators, as
follows:
(1) Principle 1: Innovation policies should focus on maximizing innovation across all
industries in all economic, correlated and supporting sectors so that the global
production chain can develop technological innovation, being one of the pyramid’s
base principles.
(2) Principle 2: Innovation policies should support all types and innovation stages
because one of the national innovation policies errors is to define innovation
strategies on a microeconomic level, only focusing on the technological products
production, whilst innovation should extend throughout the whole production
chain to rethink the products mix that make up the production high value-added
sectors, composing the pyramid’s base as well.
 (3) Principle 3: To empower creativity and creative destruction to pursue innovation; Multicriteria
developing countries need to promote disruption in manufacturing to pursue decision
innovation. Innovative entrepreneurships will enable new players to take part in
new economic sectors, especially those who are well-positioned in terms of
choices
products and services. This is the last principle of the base of the pyramid.
(4) Principle 4: To keep the price of capital goods low – especially those related to
information, communications and technology – because without the new incoming 323
capital to invest in technology, the power to invest in innovation diminishes and
productivity stagnates, causing a drop in the global strategic competitiveness.
Monitoring quotas and tariffs, such as trade barriers, can therefore maintain the
capital cost low.
(5) Principle 5: To support the creation of technology transfers related to support
industries, where companies do not only need the access to innovation but also
the access to inter- and intra-sectoral technology transfers. In addition, the
digital infrastructure, skilled labor specialized in technological innovation and
technological knowledge must be aligned with the changes in various economic
sectors, starting with the fourth principle of the second layer of the pyramid.
(6) Principle 6: To develop a national innovation and production strategy with
companies that support such strategies through the creation of agencies dedicated
to fostering innovation, completing the innovation nationwide and reaching the top
of the pyramid (Cornell University, INSEAD, & WIPO, 2015, 2016).

The identification of the indicators starts with the definition of sub-pillars and pillars of
inputs and outputs. The first sub-pillar innovation input has five pillars, namely,
institutions (indicator 1 – I1), human capital and research (indicator 2 – I2), infrastructure
(indicator 3 – I3), market sophistication (indicator 4 – I4) and business sophistication
(indicator 5 – I5). The sub-pillar input defines some aspects related to the favorable
environment for innovation in the national economy. The second sub-pillar of the innovation
output has two pillars, which is a result of innovative activities results within a national
economic scenario. Even through there are only two pillars in this sub-pillar, they have the
same weight in total. They are defined as knowledge and technology outputs (indicator 6 –
I6) and creative outputs (indicator 7 – I7). Figure 1 demonstrates the seven innovation
indicators and their input and output sub-pillars (Cornell University, INSEAD, & WIPO,
2016; Silva et al., 2017).
The scores of the GII countries are represented by their average scores and
displayed from the bottom-up. It is important to observe how a hierarchical weighting
can be perceived from the bottom-up; being more recommendable because participants’
understanding of alternatives’ impact ranges can be better in the bottom-up than in the
top-down approach (Marttunen, Belton, & Lienert, 2018). That is the reason why the use
of Pearson’s correlation aligned with AHP-Borda and PROMETHÉE is important; the
use of these statistical resources helps DMs to make better decisions in terms of
investments in upper middle-income countries instead of only considering GII scores
singly.

2.2 Multicriteria
In the decision-making model, the following components are included criteria, weights and
grades (classification) that are given for each alternative, in each criterion. Assuming the
knowledge of the preferences of the DMs and the quality of the evaluation, it can be
INMR
17,3

324

Figure 1.
Seven pillars that
compose the GII 2016

admitted that one action is as good, better or worse than another, that is, to rank the
alternatives (Cardoso, Xavier, Gomes, and Adissi, 2009).
The decision analysis process usually considers a variety of alternatives, which must be
carefully evaluated so that the “best” decision can be made. The models based on
multicriteria decisions are recommended for problems that present several evaluation
criteria to consider (Dubois, 2003; Gomes, Costa, & Barros, 2017).
The DM has (Corrente, Figueira, Greco, & Slowiński, 2017; Fernández, Figueira, &
Navarro, 2019):

a set of alternatives A = {a1, a2, . . ., an}, A be a finite and stable set of potential
alternatives to be ranked;

a set of criteria C = {c1, c2, ., cj}, C be a set of criteria of performance with given
ordinal measurement scale, thus we have a total pre-ordering of each attribute c.
cj(a), a [ A, the best alternative on criterion cj;

a set of m decision-making agents (D1, D2, . . ., Dm) expressing their opinion on the
alternatives, through preference orderings;

a set of criteria weights W = {w(c1), w(c2), . . ., w(cj)}, in which each W preference is
an attribute weight vector satisfying w(c) > 0, for all c [ C.

The decision-making process largely consists of two phases (Dias, Antunes, Dantas, Castro,
& de Zamboni, 2018; Petrovic, Bojkovic, Stamenkovic, and Anic, 2018):
(1) construction of a decision-making problem and data preparation; and
(2) aggregation and exploitation.
Different aggregation methods are developed depending on which kind of data is prepared Multicriteria
in the first phase. The ranking method can be placed into two basic categories such as decision
cardinal and ordinal methods (Doumpos, & Figueira, 2018; Yu, Zhang, Liao, & Qi, 2018):
choices
(1) cardinal methods require DMs to express their degree of preference for one
alternative over another for each criterion; and
(2) ordinal methods require that only the rank order of the alternatives be known for
each criterion. 325
As the modeling presented herein is based on two MCDA methods, namely, AHP (Saaty,
1980; De Borda, 1781) and PROMÉTHÉE (Brans, Mareschal, & Vincke, 1984), a short
description about them is inserted as follows.
De Borda uses are usually for voting systems; however, it is an important multicriteria
aid method for understanding the differences of ordering alternatives, especially when the
criteria are weighted (Da Rocha, Da Silva, DE Barros, & Costa, 2016). On the other hand,
PROMÉTHÉE observes preferences and computes them in the software; illustrations
support the results to obtain a better perspective of all the preferences, for example, if a
cluster is formed by the alternatives, etc.
AHP gives the opportunity for understanding indicators and their pairwise importance
as criteria. This is a semi-qualitative method that involves a matrix-based pairwise
comparison among the different factors (Mishra and Chatterjee, 2018). The wide
applicability is because of its simplicity, easy of use and great flexibility. The AHP is a
decision-making tool used to deal with multicriteria evaluation (Chaudhary, Chhetri, Joshi,
Shrestha, & Kayastha, 2016; Dong, & Cooper, 2016).
Moreover, Pearson correlation shows strong and weak correlations among indicators,
which can be used to encourage investments based on the innovation level of upper-middle
income countries.
2.2.1 Analytical hierarchy process. AHP is based on the following principles:

problem definition;

structuring criteria in a tree or hierarchy composed of criteria and sub-criteria;

pairwise judgments;

prioritization based on the Eigen-value aggregation procedure; and

consistency analysis.

Emrouznejad and Marra (2017) elaborated a literature review and demonstrated that most
studies on AHP were published between 1979 and 2017, considering the large number of
works in the field (8,441 published pieces). They show that AHP has attracted the attention
of scholars in various fields because of its ability to provide support to different DMs, in
areas ranging from medical issues to computer science and environmental studies. They
identified that the limitation of the stand-alone AHP is the potential arbitrary judgment of
the DM, which can lead to inconsistencies. It is interesting to notice that, in their paper, there
is no identification of the multicriteria methodology, unlike we accomplished herein in
Figure 2.
The pairwise comparison is conducted in a simple way, by asking experts some
questions, such as how much a criterion is more important than another one. The collection
of the answers is based on the Saaty’ scale shown in Table 1.
AHP is mainly worthy because of its capacity of perceiving inconsistencies in judgments.
Similarly, the judgments made by the experts calculate the coherence of the experts through
INMR the consistency index (CI). Saaty (1980) suggests that if the consistency ratio (CR) is greater
17,3 than 0.1 (CR > 0.1), the judgments of the expert should be reviewed.
2.2.2 Borda. The Borda method was first proposed in De Borda (1781) as a procedure to
deal with incoherence in ranking preferences in electoral processes.
As reported in Costa (2014), the following steps are applied when De Borda method is
adopted:
326
Definition of DMs, judges and jury members.

Definition of elements or alternatives to be ranked.

Obtaining evaluations or established judgments by each DM for each
alternative.

Associating a rating score, order number to each alternative, considering individual
judgments from each judge.

Add order numbers obtaining a global number order for each alternative.

Obtaining a final alternatives’ score based on the global order numbers.

2.2.3 Preference Ranking Organization Method for Enrichment Evaluations. The

PROMÉTHÉE was first proposed by Brans et al. (1984) as method for ranking a
finite set of alternatives (Bouyssou, 1992; Bouyssou, Marchant, Pirlot, Perny,
Tsoukiàs, & Vincke, 2000). The PROMÉTHÉE method is important because it
involves concepts and parameters that have some physical or
economical interpretation that is easy for most DMs to understand (Sarrazin, & De
Smet, 2015).
It ranks the alternatives based on their “net flow,” i.e. the difference between how much
an alternative “a” is better than alternative “b” and how much an alternative “b” is better
than alternative “a” (Bogdanovic, Nikolic, & Ilic, 2012).
It is necessary to specify to each defined criteria a generalized preference function (Pj)
(Silva, & De Almeida-Filho, 2018), such as:

Figure 2.
Illustrating the
methodological
stepsSource: Own
elaboration.

Verbal judgment Numerical rating

Equally preferred 1
Moderately preferred 3
Very strongly preferred 5
Strongly preferred 7
Table 1. Extremely preferred 9
Pairwise comparison 2, 4, 6 and 8 are intermediate values
scale for AHP
preference Source: Saaty (1980)
 Pj : A x A ! ½0; 1 (1) Multicriteria
decision
Comparing

alternatives
 a1 e a2 belonging to A-set; it has: Pj ða1 ; a2 Þ ¼ P ðxÞ ¼
P f ða1 Þ  f ða2 Þ ; representing the preference level of a1 over a2 following criterion j;
choices
as:

P(x) = 0, a1 is not preferable regarding a2;

P(x) % 0, a1 is less preferable regarding a2; 327

P(x) % 1, a1 is strongly preferable regarding a2;

P(x) = 1, a1 is entirely preferable regarding a2.

When the criterion can be optimized, it is used x ¼ f ða1 Þ  f ða2 Þ for defining the preference
function. When the criterion is minimized, the preference function is x ¼ f ða2 Þ  f ða1 Þ.
The next step is to compute the aggregated preference indices p ða1 ; a2 Þ. It is computed
to each pair of alternatives, indicating that the percentage of the preference over alternative
a1 regarding alternative a2 is considered by attributing weights to each defined criteria:
X
n
p ða1 ; a2 Þ ¼ aj Pj ða1 ; a2 Þ (2)
j¼1

where:
X
n
aj ¼ 1 (3)
j¼1

0 # p ða1 ; a2 Þ # 1 8 a1 ; a2 2 A (4)

The next step is the definition of the positive and the negative outranking flows. The
positive flow indicates how an alternative a1 is outranking all the others belonging to the A-
set:

1 X n
Uþ ða1 Þ ¼ : p ða1 ; xÞ (5)
n  1 x2A

Thus, the higher the positive outflow of a1 (Uþ ða1 Þ), the better the alternative.
The negative outranking flow indicates how an alternative a 1is outranked by all the
others:

1 X n
U ða1 Þ ¼ : p ðx; a1 Þ (6)
n  1 x2A

The lower the negative flow of a1 (U ða1 Þ), the better the alternative.
The last step is to define the importance of the net outranking flow, after defining
positive and negative outranking flows. The objective is to define importance levels to each
alternative and structure following a decreasing order. The net outranking flow is the
balance between the positive and the negative outranking flows, as follows:
INMR Uða1 Þ ¼ Uþ ða1 Þ  U ða1 Þ (7)
17,3

where a1 is preferable over a2 ða1 Pa2 Þ whether : Uða1 Þ > Uða2 Þ;

where a1 is indifferent regarding a2 ða1 Ia2 Þ whether : Uða1 Þ ¼ Uða2 Þ; and

a1 Pa2 Uða1 Þ > Uða2 Þ

.
328 a1 Ia2 Uða1 Þ ¼ Uða2 Þ

2.3 Pearson’s correlation

The Pearson’s correlation coefficient (r) measures the linear association between two quantitative
variables. In the analysis of innovation indicators, the correlation coefficient measures the
pairwise association among indicators, considering the values of the set of indicators’ values
among themselves, as the variables analyzed. The value of r varies between 1 and 1, so it can
be positive or negative. A positive correlation between two innovation indicators indicates the
same tendency of either growth or decrease of these innovation indicators. Subsequently, a
negative correlation between two indicators indicates opposite tendencies of growth and decrease
of innovation values; in other words, as one of them tends to be a better reference in terms of
innovation indicator, the others tend to be less representative to be considered as innovation
indicators. These results depend on each country profile (Grácio & Oliveira, 2015).
Regardless of the direction of the correlation (positive or negative), the correlations can
also vary in terms of strength: from absence (equal to zero) to a very strong or even perfect
correlation (r value equal to 1 or 1). Considering xi the frequency of values of one indicator
X in relation to other indicators and yi the frequency of i values of one indicator Y with the
other innovation indicators, for i varying from 1 to n, with n equal to the number of
indicators in the GII 2016, the Pearson’s correlation coefficient (r) between X and Y is defined
by equation (1) (Galarça, Lima, Silveira, & Rufato, 2010; Grácio & Oliveira, 2015):
P P
P xi : yi
xi: yi 
r ¼ rffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
P 
n
P ffi (1)
P 2 ð xi Þ2 P 2 ð yi 2 Þ
xi  n : yi  n

The Pearson’s correlation coefficient results from the WIPO’s GII Audit 2016, which
demonstrated such as the GII 2015 Report the three highest Pearson’s correlation sub-pillars
that enabled their allocation in the seven innovation pillars. After other multivariate
statistical analysis procedures, the audit team adds the conceptual framework of
quantitative data and analysis to corroborate to the next GII Report to be published,
formulating an innovation ranking with innovation indicators along with their pillars and
sub-pillars (Cornell University, INSEAD and WIPO, 2015, 2016).

3. Methodology
As it can see in Figure 2, the methodology is structured in three phases, namely,

(1) using AHP to define criteria weights;

(2) ranking the alternatives;
(3) comparing the results.
3.1 Using analytical hierarchy process to define criteria weights Multicriteria
The AHP was used to obtain the weight of each one of the criteria of GII 2016, after aligning decision
it with Pearson’s correlation. The alignment was made by the global average of Pearson’s
correlation innovation indicators used by WIPO compared to the AHP standard scale.
choices
According to Macharis, Springael, De Brucker, and Verbeke (2004), this procedure provides
an opportunity to apply AHP-Borda as a weight to PROMÉTHÉE.
The judgments were made during meetings by experts in economic development,
innovation (based on the understanding of GII reports published by WIPO), operational
329
research, applied mathematics and computer science.

3.2 Ranking the alternatives

This phase is divided into two. In the first, the De Borda method is applied to rank the
countries, having as input the weights from AHP and the countries’ evaluations under GII
criteria. In the second one, the PROMÉTHÉE is adopted to obtain the ranking of the
countries. The inputs of this step are the weights from AHP and the countries’ performance
according to the GII. As one can notice, in this phase the weights generated through AHP
were used as an input to be used in the MCDA method – as first proposed by Da Rocha, Da
Silva, DE Barros, & Costa, 2016.

3.3 Comparing the results

From Pearson’s correlation initial analysis, it is possible to understand the importance of the
innovation indicators for upper-middle income countries (i.e. understanding their pertinence
for these peculiar countries). After aligning Pearson’s ranges to AHP Saaty’ scale, AHP was
computed and combined with De Borda ranking system. Such AHP combinations made the
Operational Research team execute quantitative procedures by using PROMÉTHÉE,
observing an outranking of results that indicated potential investments groups of countries
presenting a “hub” profile.
This paper aims to discover if these innovation indicators can be considered good
parameters for understanding the profile of upper-middle income countries. Pearson’s
correlation was the basis for understanding how upper-middle income countries act,
considering the innovation indicators of these countries’ profiles.
Pearson’s correlation was identified as the milestone for this paper because it was
necessary to understand how innovation indicators were representative for upper-middle
income countries. After carrying out Pearson’s correlation analysis in these indicators
aligned with the calculated and validated AHP and considering the representativeness of the
indicators according to the DMs of the operational research team, it was necessary to
provide a brief understanding on MCDA; more specifically, how Borda was applied in the
AHP results.
Hence, after achieving weights via AHP, these weights were used in an ordering-ranking
with Borda for a pilot project to make investors understand the MCDA concepts and how
AHP weights were used in PROMÉTHÉE; whose results were computed via GAIA
software. The use of the software GAIA was necessary herein because it is PROMÉTHÉE’s
seminal software. The use of GAIA provides the original perspective of the decision-making
process. Thus, the exploratory nature might show relevant variables necessary to
understand innovation indicators in the environment of national and international strategy
and competitiveness (Martins, Lima, & Costa, 2015). In this paper, inspired by Da Rocha
et al. (2016), the De Borda method combined with AHP was used as the sum of the
alternatives; the AHP results were used as weights that resulted in the global net flow.
INMR PROMÉTHÉE II results provide a ranking that can be used to solve the decision-making
17,3 problem according to the alternatives considered herein. The software GAIA was used to
compute all the data. From these last multicriteria applications, the decision for investment
was explored jointly with investors.

3.4 Multicriteria decision aid computing procedures

330 This paper observes via MCDA tools if innovation indicators might be considered by
investors for investment possibilities in upper-middle income countries according to the GII
2016 data. Pearson’s correlation was the basis for starting an AHP application regarding
innovation indicators.
Pearson’s correlation practiced by GII 2016 had innovation indicators audited with the
same system from 2015: all the sub-pillars with the highest Pearson correlation coefficients
were bracketed inside the innovation indicators; thus, all 21 sub-pillars were defined in each
of one 7 innovation indicators, as shown in Table 2 (Cornell University, INSEAD, & WIPO,
2016).
Initially, Pearson’s correlation shows a concentration between 0.4 and 0.7. If this
concentration has not been observed, the calculations of the MCDA stages would not have
been pertinent in terms of undertaking investments. Pearson’s correlation shows a statistical
correlation regarding the strength of each pair of variables observed whilst AHP observes
the importance. Some analysts understand the strength brought by Pearson’s correlation as
an opportunity for aligning a statistical data from GII 2016 to a MCDA tool such as AHP,
whose importance serves as several other multicriteria analyses when used with other
multicriteria aid tools.
Saaty’s scale was used for analysts who observed usual concentration of odd numbers in
the scale, which could not indicate a good alignment between Pearson’s correlation and AHP
scale; contrary to a fairly common practice among multicriteria aid tools when AHP is
applied with odd numbers (Franek, & Kresta, 2014). The use with Saaty’s scale of
importance was tested with 8 and 9; however, such importance numbers showed
incoherence when applied to Pearson’s correlation because no indicators have an absolute
dominance over their pairs; the dominance indicated is the most adequate high scale level
for aligning Pearson’s correlation organized in ranges according to Table 3.
From this perspective, analysts decided to range Pearson’s correlation for observing the
behavior of indicators inside these ranges when they were aligned with AHP’s consistency
considering the matrix size. Intervals from 0 to 0.99 were used, being the first interval from 0
to 0.39 because such interval did not appear at GII 2016 Pearson’s correlation, turning
Saaty’s comparison scale into an important resource, represented by the number 1. The
following ranges were established in 0.09 ranges by each interval and were equalized to each
AHP scale within the comparison scale shown in Table 3.
This alignment allows to start AHP computations according to Table 4, where the AHP
scale was inserted aligned with the range from the coefficient average of Pearson’s
correlation.
Table 5 shows AHP results with each innovation indicators and their weight.
In the third stage, after observing both importance and strength among GII 2016
indicators, it was necessary to show DMs a perspective for investment in upper-middle
income countries, defined by analysts as a pilot project for the decision-making process
considering multicriteria investments perspectives. With such innovation indicators
representative percentage calculated after alignment, the next stage was the better
understanding for sorting middle-income countries according to these innovation indicators
 Indicator 1 Indicator 2 – Indicator 4 – Indicator 5 – Indicator 6 –
– human capital Indicator 3 – market business knowledge and Indicator 7 –
Sub-pillar institutions and research infrastructure sophistication sophistication technology outputs creative outputs

Input
Political environment 0.94 0.77 0.84 0.7 0.74 0.69 0.8
Regulatory environment 0.92 0.63 0.67 0.58 0.65 0.6 0.67
Business environment 0.9 0.71 0.75 0.68 0.62 0.68 0.71
Education 0.52 0.75 0.54 0.4 0.48 0.51 0.52
Tertiary education 0.65 0.79 0.75 0.57 0.46 0.52 0.57
Research and development (R&D) 0.69 0.89 0.77 0.78 0.81 0.84 0.74
Information and communication
technologies (ICTs) 0.77 0.83 0.94 0.7 0.64 0.69 0.76
General infrastructure 0.62 0.62 0.74 0.55 0.58 0.56 0.54
Ecological sustainability 0.62 0.6 0.74 0.52 0.51 0.52 0.63
Credit 0.65 0.6 0.58 0.85 0.59 0.52 0.57
Investment 0.48 0.5 0.42 0.76 0.52 0.49 0.4
Trade, competition and market scale 0.51 0.66 0.71 0.71 0.5 0.64 0.6
Knowledge workers 0.63 0.8 0.68 0.67 0.85 0.73 0.67
Innovation linkages 0.53 0.4 0.42 0.43 0.72 0.49 0.49
Knowledge absorption 0.6 0.56 0.56 0.54 0.82 0.71 0.59
Output
Knowledge creation 0.63 0.79 0.63 0.66 0.75 0.88 0.77
Knowledge impact 0.51 0.51 0.56 0.44 0.5 0.73 0.59
Knowledge diffusion 0.52 0.54 0.51 0.52 0.64 0.73 0.5
Intangible assets 0.62 0.61 0.67 0.54 0.56 0.65 0.89
Creative goods and services 0.67 0.63 0.64 0.54 0.61 0.69 0.84
Online creativity 0.81 0.78 0.77 0.67 0.76 0.77 0.88
Source: Cornell University, INSEAD and WIPO (2016)

each sub-pillar inside

Table 2.

innovation indicators
Pearson correlation
331
choices
decision
Multicriteria

coefficient results for

INMR using other MCDA information. To identify the level of comprehension, the AHP-Borda
17,3 method was used.
The analysts decided to show the pilot project using the AHP-Borda to support
innovation investment tendencies, for considering this tool simple for being understood by
anyone not involved in MCDA analysis. Analysts made two different procedures using
Borda: a simple Borda demonstration using the row sum from the seven innovation
332 indicators of upper-middle income countries and also the AHP results used as weights for
calculating Borda via weights. The different results are shown in Table 6, where it is
possible to observe that only three positions (positions: 7, 24 and 32) in the rank do not
change using these two different ways of calculating Borda.
After achieving weights via AHP, these weights were used in an ordering-ranking with
Borda in a pilot project to make investors understand the MCDA concepts. After
understanding the analysis, AHP weights were used in PROMÉTHÉE, whose results were
computed via GAIA software. Thus, exploratory nature analyses might show relevant
variables to understand innovation indicators in the environment of national and
international strategy and competitiveness (Martins et al., 2015).
When analysts explain the method for choosing innovative upper-middle income
countries to invest in, it was possible to understand tendencies and outranking changings
using weights or not. As the qualitative data, according to GII 2016, was aligned with the
quantitative result of the pilot project, analysts keep on calculating the MCDA using AHP-
PROMÉTHÉE at GAIA software. With DMs understanding the supportive tools,
considering the global and complete analysis with the upper-middle income countries, we
used AHP-PROMÉTHÉE II to conclude the analysis in which some GII 2016 countries show
possibilities for investment to promote the development of innovation.

4. Results and discussion

After analysts explained the method for choosing innovative upper-middle income countries
to invest in, it was possible to understand the tendencies and rankings using weights or not.
As the qualitative data, according to GII 2016, was aligned with the quantitative result of the
pilot project, analysts keep on calculating the MCDA using AHP-PROMÉTHÉE and GAIA.
The AHP results were used as weights in PROMÉTHÉE II. The GAIA software
computation revealed a different investment scenario for upper-middle income countries
from the GII 2016 ranking: Malaysia becomes the first country to be chosen for investment,
followed by Montenegro, Bulgaria, Romania and China.
Figure 3 show an adaptation from the GAIA software illustration of net flow (Phi) result,
the positive preference flow result (Phiþ) and the negative preference flow result (Phi).

Pearson’s range AHP scale compared to

organized from GII’s audit Pearson’s correlation Saaty’s comparison scale Saaty’s scale of importance

[0, 0.39] 1 Equal importance 1

[0.40, 0.49] 2 Weak dominance 3
[0.50, 0.59] 3 Strong dominance 5
[0.60, 0.69] 4 Demonstrated dominance 7
[0.70, 0.79] 5 Absolute dominance 9
[0.80, 0.89] 6 (Intermediate values) (2, 4, 6, 8)
Table 3. [0.90, 0.99] 7
Aligning Pearson’s
range to AHP Source: The authors
 Indicator 1 Indicator 2 – Indicator 4 – Indicator 5 – Indicator 6 –
– human capital Indicator 3 – market business knowledge and Indicator 7 –
Description institutions and research infrastructure sophistication sophistication technology outputs creative outputs

Indicator 1 – institutions 1 5 5 4 4 4 5
Indicator 2 – human capital and
research 0.2 1 4 3 3 4 4
Indicator 3 – infrastructure 0.2 0.25 1 4 3 4 4
Indicator 4 – market sophistication 0.25 0.333333 0.25 1 3 3 3
Indicator 5 – business sophistication 0.25 0.333333 0.33333333 0.33333333 1 4 3
Indicator 6 – knowledge and
technology outputs 0.33333333 0.25 0.25 0.33333333 0.25 1 4
Indicator 7 – creative outputs 0.33333333 0.25 0.25 0.33333333 0.33333333 0.25 1
Source: The authors

computations
AHP initial
Table 4.
333
choices
decision
Multicriteria
INMR Malaysia shows the distance from the other upper-middle income countries having 0.8737 in
17,3 preference for investments considering the innovative profile of innovation indicators
weights used in PROMÉTHÉE II calculus.
Figure 4 shows an adaptation from a GAIA image with a net flow graph result. It is
possible to observe clusters being formed as a sequence of investments in the same region;
for example, it is possible to start an investment in Montenegro, followed by Bulgaria and
334 Romania, with the same potential investment because they are developing industries in their
region as a cluster. Observing the results presented by the graph, it is possible to understand
the investment cycle in upper-middle income countries because there is an investment
tendency rebounding in these economies and inside the integrated chain from each region
where these countries belong to.
Figure 5 has an interesting perspective of strong and weak innovation indicators for each
selected country. Despite being an individual perspective stemming from PROMÉTHÉE I in
each innovation indicator, it is interesting to analyze it because it indicates deeply in which
investment scenarios it is possible to invest in to obtain profits in the short-term, considering
the economic sectors involved in each countries and investments that only provide financial
returns in the long-term.
The use of all these MCDA tools was important to encourage DMs to invest in countries
not only taking into consideration the GII 2016 ranking of the most innovative countries by
area, region or economic profile. These MCDA tools combined showed how the qualitative
data available at GII 2016 is important to be analyzed from a quantitative perspective for
multicriteria decision supportive tools because there are possibilities to see beyond the
ranking report, the global innovation tendency and how countries will project their
economies using their strong and weak innovation indicators.
It is possible to observe investment opportunities organizing the countries in subsets
according to the following analysis results from AHP-BORDA and AHP- PROMÉTHÉE.
Table 7 shows the GII’s 2016 upper-middle income countries’ ranking organized in the initial
analysis. The three first subsets present a bold minimum number and the fourth subset
presents a bold maximum number to highlight borderline numbers
Table 8 compares ranking results of Borda row sum, AHP-Borda and AHP-
PROMÉTHÉE II where only Paraguay at position 32 presents the same result for all the
three MCDAs adopted.
Considering the information presented in Table 8, the results of the Pearson’s correlation
and AHP-Borda result in a Borda’s row sum of 0.87. The results of the Pearson’s correlation
that stem from AHP-Borda and AHP-PROMÉTHÉE II was 0.98. The result of the Pearson’s
correlation and Borda’s row sum with AHP-PROMÉTHÉE II was 0.88. These correlations
demonstrate a strong consistency of results.

Item description Weight (%)

Indicator 1 – institutions 35.75

Indicator 2 – human capital and research 19.63
Indicator 3 – infrastructure 15.42
Indicator 4 – market sophistication 10.30
Indicator 5 – business sophistication 8.42
Table 5. Indicator 6 – knowledge and technology outputs 6.36
Innovation Indicator 7 – creative outputs 4.12
indicators’ AHP
results Source: The authors
 Global order Global order number Final rank Final rank by
Multicriteria
UM countries number (row sum) (AHP weights) by row sum AHP weights decision
choices
China 361.7 52.7027 1 2
Malaysia 329.5 54.2216 2 1
Bulgaria 301.6 49.1042 3 5
Turkey 283.8 44.7568 6 15
Costa Rica 288.5 48.0115 4 8 335
Romania 283.7 48.3001 7 7
Montenegro 287.8 49.2535 5 4
Thailand 275 43.5735 10 19
Mauritius 280.6 51.2203 9 3
South Africa 281.7 48.3881 8 6
Mongolia 267.6 44.8256 13 12
TFYR of Macedonia 268.5 47.0867 11 9
Mexico 265.7 44.7746 14 13
Colombia 267.9 44.5585 12 16
Serbia 257.8 45.3774 17 11
Panama 255 42.5575 18 22
Brazil 260.9 43.1900 15 20
Lebanon 244.2 39.6076 21 29
Peru 259.4 44.0146 16 17
Kazakhstan 251.3 45.4998 19 10
Dominican Republic 235.7 38.6272 25 30
Tunisia 236.6 42.0417 24 24
Iran, Islamic Rep. 229.2 37.5875 29 31
Belarus 247.6 43.7283 20 18
Jordan 228.1 40.8777 30 26
Azerbaijan 233.6 40.0873 27 28
Bosnia and Herzegovina 240 42.7529 22 21
Jamaica 229.7 41.4919 28 25
Botswana 238.5 44.7571 23 14
Albania 235.1 42.4701 26 23
Namibia 222.9 40.5823 31 27
Paraguay 217 35.6831 32 32
Ecuador 210.2 34.3115 33 34
Algeria 196.3 34.3877 34 33 Table 6.
Borda results using
Source: The authors row sum and AHP

Although China, according to GII 2016, is the first upper-middle income country, the
infrastructure innovation indicator presents a weak profile in the short-term qualitative and
quantitative data analysis because the increase in steel inventory from 2016 to 2017 shows
that the processing industries perspective for developing new production chain links
become less preferable for investing than ICT sectors, which can be observed in other
innovative indicators in recent years.
Similarly, Malaysia presents an ICT economic sector stronger than other countries
geographically and by economic profile according to the GII 2016 selection of upper-middle
income countries. As part of the Asian tigers as a country with a strong electronic
processing industry, Malaysia’s innovation indicators presented three stronger scores than
other upper-middle income countries and the strongest innovation indicators scores in
comparison to other Asian countries. Regarding the ICT economic sector profile, with a
cyclical demand and offer movement for new production process and new adjustments in
INMR
17,3

336

Figure 3.
Adaptation from
GAIA with net flow
results

the production chain for competitiveness, Malaysia becomes the first destination of
investments when observing the countries selected.
Table 9 presents a proposal for investors, considering the integration between
production chain and transnationality involving economic sectors in different countries and
continents. Considering the profiles of the countries, the basic competitive industrial
structure and the supply flows in several production chains, it is possible to observe four
groups in which countries present a similar behavior, whose integrated investment
development of different regions and areas promote competitiveness according to the
innovation indicators observed in these countries. It is interesting to analyze these groups as
an opportunity for technological frontiers.
Table 10 shows the minimum score’s validation for each country obtained in each
indicator by subset. By analyzing these scores, it was possible to observe, for instance, that
Thailand – despite having a Group 1 profile – is allocated in Group 2; the same way, even
though Paraguay presents a Group 3 profile, it was allocated in Group 4.
In the light of the above-mentioned information, groups where re-organized according to
Table 11 observing the new subsets for investments of upper-middle economies.
After the profile identification, it is possible to notice that Group 1 has four
countries showing a peripheral industry behavior in their regions; Group 2 has
opportunities for investments in countries with integrated production chain in extractive
 Multicriteria
decision
choices

337

Figure 4.
Adaptation from
GAIA screen graphs
INMR
17,3

338

Figure 5.
Walking weights

industries (Oil & Gas) and manufacturing industry belonging to a hub in their regions for
supplying added value industries. Group 3 presents opportunities for investing in these
countries where they have production sectors or economic sectors where the investment
leverages also other integrated sectors.
Regarding the quantitative perspective, using Borda’s row sum, AHP-Borda and AHP-
PROMÉTHÉE II shown in Table 8, it is also interesting for investors to observe that despite the
use of three different MCDA applications, it was possible to envision investments in these
countries, i.e. four groups with similar country behaviors. When changes in the ranking results
become concentrated in some countries creating an “investment belt,” it indicates another
opportunity for DMs to define their investments preferences more accurately.
Therefore, achieving a threshold from the minimum score at each of the three first
groups and the maximum for the last group becomes an opportunity for raising upper-
middle countries’ investment standard; it means that for each country there is a growth goal
according to the minimum innovation indicator belonging to the next group level. Table 10
indicates such thresholds moving within the groups.

5. Conclusion and final considerations

The question proposed by this paper was answered with the combined tools MCDA, AHP-
PROMÉTHÉE II giving opportunity for investing in upper-middle income countries. These
countries were analyzed for their innovation indicators; being corroborated by short- and
long-term economic prospects.
After meeting the purpose of this research, it was possible to identify minimum values
for scoring countries in subsets, whose values may become goals for reaching by the
countries raising their economic development sectors.
The result brought up by the hierarchical model used in the two models of MCDA
considered the methodology and adoption of criteria and sub criteria that presented some
technological potential. These methodological concepts were determinant for proposing
technological arrangements compatible with AHP and PROMÉTHÉE II multicriteria
decision support aid models. The results obtained are compatible with the authors’ proposal
 Indicator 6
Indicator 2 Indicator 4 Indicator 5 knowledge and Indicator 7
Indicator 1 human capital Indicator 3 market business technology creative
Upper-middle countries institutions and research infrastructure sophistication sophistication outputs outputs

Montenegro 69.2 35.8 45.8 43.8 36 23.4 33.8

South Africa 69.1 33.1 37.4 58.7 32.2 24.7 26.5
Costa Rica 66.7 32.1 47.4 38.1 40.4 26.8 37
Turkey 54.6 39.2 43.6 47.7 27.6 29.1 42
Romania 69 30.2 48.9 41.8 30.2 31 32.6
Bulgaria 67.8 32.1 46.7 43.7 36.2 32.1 43
Malaysia 70.9 43.3 49.2 55 41.8 33.4 35.9
China 55.2 48.1 52 56.6 53.8 53.3 42.7
Belarus 56 42.6 43.6 39.1 28.7 28.1 9.5
Kazakhstan 66.5 31.4 46.8 38.6 25.4 21.2 21.4
Brazil 55.3 32.5 44.9 43.9 37 23.7 23.6
Serbia 65.7 32.8 43.7 34.3 28.1 27.7 25.5
Peru 60.4 27.5 45 50 32.9 16.4 27.2
Colombia 58.2 27.9 52.5 49.4 30.8 21.2 27.9
Mauritius 80.5 30 43.1 49.5 25.6 22.9 29
Panama 59.6 22 47 42.3 30.7 24.2 29.2
Mexico 60.5 33.7 42.8 45.7 29.8 23.3 29.9
Thailand 54.7 30.7 42.8 51.4 35.3 29 31.1
TFYR of Macedonia 71.4 33.4 34.5 40.7 31.5 21.9 35.1
Mongolia 64.8 26.9 38.3 49.9 27.8 23.4 36.5
Namibia 64.7 21 36.2 40 21.3 8.8 30.9
Jamaica 65.3 24.2 31.2 40.3 28.8 16.3 23.6
Albania 62.3 23.6 43.6 51.2 21.9 17.2 15.3
Azerbaijan 56.4 22.9 40.8 52.1 19.7 17.6 24.1
Bosnia and Herzegovina 58.7 41 33 42.8 27.1 17.8 19.6
Botswana 68.3 32.6 37 42.8 23.9 18.6 15.3
Dominican Republic 53.6 19 38.9 45.5 32.1 19.1 27.5
Tunisia 58.3 38 41.6 29 23.7 19.9 26.1
Jordan 62.6 25.4 38.5 32 21.5 21.7 26.4
(continued)

Table 7.

middle income
rank for upper-
GII’s 2016 WIPO’s
339
choices
decision
Multicriteria

economies
 17,3

340
INMR

Table 7.
Indicator 6
Indicator 2 Indicator 4 Indicator 5 knowledge and Indicator 7
Indicator 1 human capital Indicator 3 market business technology creative
Upper-middle countries institutions and research infrastructure sophistication sophistication outputs outputs

Lebanon 52.1 29.8 37.5 37.9 31.7 22.4 32.8

Iran, Islamic Republic 45.9 36.9 36.7 36.2 22.8 24 26.7
Paraguay 47.9 24.4 34.1 42.3 25 11.8 31.5
Algeria 45.7 28.2 37.2 31.7 21.2 17.7 14.6
Ecuador 44.6 21.4 38.7 40.7 24.2 13.2 27.4

Source: Adapted from Cornell University, INSEAD and WIPO (2016)

UM countries Borda row sum AHP-Borda AHP-PROMETHEE II
Multicriteria
decision
China 1 2 5 choices
Malaysia 2 1 1
Bulgaria 3 5 3
Turkey 6 15 13
Costa Rica 4 8 7
Romania 7 7 4 341
Montenegro 5 4 2
Thailand 10 19 18
Mauritius 9 3 6
South Africa 8 6 8
Mongolia 13 12 16
TFYR of Macedonia 11 9 9
Mexico 14 13 11
Colombia 12 16 15
Serbia 17 11 10
Panama 18 22 21
Brazil 15 20 19
Lebanon 21 29 29
Peru 16 17 17
Kazakhstan 19 10 12
Dominican Republic 25 30 31
Tunisia 24 24 23
Iran, Islamic Republic 29 31 30
Belarus 20 18 20
Jordan 30 26 26
Azerbaijan 27 28 28
Bosnia and Herzegovina 22 21 22
Jamaica 28 25 25
Botswana 23 14 14
Albania 26 23 24
Namibia 31 27 27 Table 8.
Paraguay 32 32 32
Ecuador 33 34 33
Rankings
Algeria 34 33 34 comparison with
different MCDAs
Source: The authors combination

of offering to DMs a complete ranking of alternatives for contributing with the decision-
making process improvement in terms of choices related to technological investments
alternatives.
This paper presents quantitative observations based on which upper-middle
income country DMs can invest considering strong and weak innovation indicators
according to their perspectives, supported by the qualitative observation obtained
from GII 2016.
When first analyzing the ranking of innovation indicators, the MCDA provided a
better support to investment decision-making because all the innovation indicators
were considered in the results in a deep way when Pearson’s correlation coefficient was
the highest one in the sub-pillars of each innovation indicator and PROMÉTHÉE II
completes the ranking. AHP weights are used in PROMÉTHÉE II computations
justifying the rank according to the weights. A pilot project for investors proposed the
MCDA using AHP-Borda.
INMR AHP- PROMÉTHÉE
17,3 Borda row sum AHP-Borda II Groups

China Malaysia Malaysia Group 1: China, Malaysia, Bulgaria,

Malaysia China Montenegro Costa Rica, Montenegro, Romania
Bulgaria Mauritius Bulgaria and South Africa
Costa Rica Montenegro Romania
342 Montenegro Bulgaria China
Turkey South Africa Mauritius
Romania Romania Costa Rica
South Africa Costa Rica South Africa
Mauritius TFYR of Macedonia TFYR of Macedonia Group 2: Mauritius, Thailand,
Thailand Kazakhstan Serbia Macedonia, Colombia, Mongolia,
TFYR of Macedonia Serbia Mexico Mexico, Brazil, Peru, Serbia,
Colombia Mongolia Kazakhstan Kazakhstan and Belarus
Mongolia Mexico Turkey
Mexico Botswana Botswana
Brazil Turkey Colombia
Peru Colombia Mongolia
Serbia Peru Peru
Panama Belarus Thailand
Kazakhstan Thailand Brazil
Belarus Brazil Belarus
Lebanon Bosnia and Panama Group 3: Lebanon, Bosnia and
Herzegovina Herzegovina, Botswana, Tunisia,
Bosnia and Panama Bosnia and Dominican Republic, Albania,
Herzegovina Herzegovina Azerbaijan, Jamaica, Iran Islamic
Botswana Albania Tunisia Rep., Jordan and Namibia
Tunisia Tunisia Albania
Dominican Republic Jamaica Jamaica
Albania Jordan Jordan
Azerbaijan Namibia Namibia
Jamaica Azerbaijan Azerbaijan
Iran, Islamic Lebanon Lebanon
Republic.
Jordan Dominican Republic Iran, Islamic Rep.
Table 9. Namibia Iran, Islamic Rep. Dominican Republic
Paraguay Paraguay Paraguay Group 4: Paraguay, Ecuador and
Groups of countries Ecuador Algeria Ecuador Algeria
for investing Algeria Ecuador Algeria
according to their
similar profiles Source: The authors

It is important to demonstrate herein that when investing in one single economic sector of a
country, there is a subsequent investment in another country, which boosts transnational
economy and investment. The three MCDA tools used herein provide the possibility of
investing according to investors’ preference in four different countries groups correlated due
to their integrated supply chain in some economic sectors.
All these MCDA methods used in this paper are important to observe how the
investment decision can change when criteria are deeply computed and associated with
weights from other multicriteria tools, increasing the investment understanding and
the country profile, the cluster association among neighboring partners inserted or not
in the alternatives analyzed.
 Indicator 1 Indicator 2 – Indicator 4 – Indicator 5 – Indicator 6 –
– Human capital Indicator 3 – market business knowledge and Indicator 7 –
UM countries institutions and research infrastructure sophistication sophistication technology outputs creative outputs

Group 1 – minimum 54.6 30.2 37.4 38.1 27.6 23.4 26.5

Group 2 – minimum 54.7 22 34.5 34.3 25.4 16.4 9.5
Group 3 – minimum 45.9 19 31.2 29 19.7 8.8 9.5

Source: Adapted from Cornell University, INSEAD, & WIPO (2016)

Scores’ validation
343
choices
decision
Multicriteria

analysis
Table 10.
INMR Group Countries’ subset
17,3
Group 1 China, Malaysia, Bulgaria, Costa Rica, Montenegro,
Romania, South Africa and Thailand
Group 2 Mauritius, Macedonia, Colombia, Mongolia,
Mexico, Brazil, Peru, Serbia, Kazakhstan and
Belarus
344 Group 3 Lebanon, Bosnia and Herzegovina, Botswana,
Tunisia, Dominican Republic, Albania, Azerbaijan,
Jamaica, Iran Islamic Rep., Jordan, Namibia and
Table 11. Paraguay
Re-organized subset Group 4 Ecuador and Algeria
of countries for
investments Source: The authors

References
Bogdanovic, D., Nikolic, D., & Ilic, I. (2012). Mining method selection by integrated AHP and
PROMETHEE method. Anais da Academia Brasileira de Ciências, 84, 219–233. https://doi.org/
10.1590/S0001-37652012005000013.
Bouyssou, D. (1992). Ranking methods based on valued preference relations: a characterization of the
net flow method. European Journal of Operational Research, 60, 61–67, https://doi.org/10.1016/
0377-2217(92)90333-5.
Bouyssou, D., Marchant, T., Pirlot, M., Perny, P., TsoukiàS, A., & Vincke, P. (2000). Evaluation and
decision models: a critical perspective, New York, NY: Kluwer Academic Publishers.
Brans, J.P. & Mareschal, B. (1990). The PROMETHEE methods for MCDM; the PROMCALC, GAIA
and BANK ADVISER software. In Readings in multiple criteria decision aid, Ed. C. Bana e
Costa, Springer-Verlag.
Brans, J. P., Mareschal, B., & Vincke, P. (1984). Promethee: a new family of outranking methods in
multicriteria analysis, Washington, DC: North-Holland.
Cardoso, R. S., Xavier, L. C., Gomes, C. F. S., & Adissi, P. J. (2009). Uso de SAD no apoio à decisão na
destinação de resíduos plásticos e gestão de materiais. Pesquisa Operacional, 29, 67–95, https://
doi.org/10.1590/S0101-74382009000100004.
Chaudhary, P., Chhetri, S. K., Joshi, K. M., Shrestha, B. M., & Kayastha, P. (2016). Application of an
analytic hierarchy process (AHP) in the GIS interface for suitable fire site selection: a case study
from Kathmandu metropolitan city, Nepal. Socio-Economic Planning Sciences, 53, 60–71,
https://doi.org/10.1016/j.seps.2015.10.001.
Cornell University, INSEAD and WIPO. (2015). The global innovation index 2015: Effective innovation
policies for development, Ithaca, Fontainebleau and Geneva: Cornell University, INSEAD and
WIPO.
Cornell University, INSEAD and WIPO. (2016). The global innovation index 2016: Winning with
global innovation, Ithaca, Fontainebleau and Geneva: Cornell University, INSEAD and
WIPO.
Corrente, S., Figueira, J. R., Greco, S., & Slowiński, R. (2017). A robust ranking method extending
ELECTRE III to hierarchy of interacting criteria, imprecise weights and stochastic analysis.
Omega, 73, 1–17. https://doi.org/10.1016/j.omega.2016.11.008.
Costa, H. G. (2014). Sistemas de votação pelo método de Borda. Relatorios de Pesquisa em Engenharia
de Produção, 14, 1–10.
Da Rocha, P. M., Da Silva, G. B., DE Barros, A. P., & Costa, H. G. (2016). Analysis of the operational
performance of Brazilian airport terminals: a multicriteria approach with De Borda-AHP
 integration. Journal of Air Transport Management, 51, 19–26, https://doi.org/10.1016/j. Multicriteria
jairtraman.2015.11.003.
decision
De Borda, J.-C. (1781). Mémoire sur les élections au scrutiny, Paris: L’Académie des Sciences.
choices
Dias, L. C., Antunes, C. H., Dantas, G., Castro, N., & de Zamboni, L. (2018). A multi-criteria
approach to sort and rank policies based on Delphi qualitative assessments and ELECTRE
TRI: the case of smart grids in Brazil. Omega, 76, 100–111. https://doi.org/10.1016/j.
omega.2017.04.004.
Dong, Q., & Cooper, O. (2016). A peer-to-peer dynamic adaptive consensus reaching model for the group
345
AHP decision making. European Journal of Operational Research, 250, 521–530, https://doi.org/
10.1016/j.ejor.2015.09.016.
Doumpos, M., & Figueira, J. R. (2018). A multicriteria outranking approach for modeling corporate
credit ratings: an application of the electre tri-nC method. Omega, 82, 166–180, https://doi.org/
10.1016/j.omega.2018.01.003.
Dubois, D. (2003). Book review: Evaluation and decision models: a critical perspective. In D. Bouyssou,
T. Marchant, M. Pirlot, P. Perny, A. Tsoukias, & P. Vincke, (Eds). Fuzzy sets and systems,
Boston/London/Dordrecht: Kluwer Academic Publishers, 139, 469–472.
Emrouznejad, A., & Marra, M. (2017). The state of the art development of AHP (1979-2017): a literature
review with a social network analysis. International Journal of Production Research, 55,
6653–6675, https://doi.org/10.1080/00207543.2017.1334976.
Fernández, E., Figueira, J. R., & Navarro, J. (2019). An interval extension of the outranking approach
and its application to multiple-criteria ordinal classification. Omega, 84, 189–198. https://doi.
org/10.1016/j.omega.2018.05.003.
Franek, J., & Kresta, A. (2014). “Judgment scales and consistency measure in AHP”, Enterprise and the
Competitive Environment 2014 conference, ECE 2014, 6–7 March 2014, Brno, Czech Republic In:
Procedia Procedia Economics and Finance, Vol.12, pp. 164 – 173. DOI:10.1016/S2212-5671(14)
00332-3.
Galarça, S. P., Lima, C. S. M., Silveira, G., & Rufato, A. R. (2010). Pearson correlation and path analysis
identifying variables for the characterization of Pyrus communis L. rootstock. Ciência e
Agrotecnologia, 34, 860–869, https://doi.org/10.1590/S1413-70542010000400010.
Gomes, C. F. S., Costa, H. G., & Barros, A. P. (2017). Sensibility analysis of MCDA using prospective in
Brazilian energy sector. Journal of Modelling in Management, 12, 475–497, https://doi.org/
10.1108/JM2-01-2016-0005.
Grácio, M. C. C., & Oliveira, E. F. T. (2015). Indicadores de proximidades em análise de cocitação de
autores: um estudo comparativo entre coeficiente de Correlação de Pearson e Cosseno de Salton.
Informação & Sociedade: Estudos, 25, 105–116.
Macharis, C., Springael, J., De Brucker, K., & Verbeke, A. (2004). PROMETHEE and AHP: the design of
operational synergies in multicriteria analysis. Strengthening PROMETHEE with ideas of AHP.
European Journal of Operational Research, 153, 307–317, https://doi.org/10.1016/S0377-2217(03)
00153-X.
Martins, E. F., Lima, G. B. A., & Costa, H. G. (2015). Análise estratégica multicritério socio-econômico-
ambiental como ferramenta de apoio à decisão em uma empresa de distribuição de energia.
Sistemas & Gestão, 10, 96–106, https://doi.org/10.7177/sg.2015.V10.N1.A8.
Marttunen, M., Belton, V., & Lienert, J. (2018). Are objectives hierarchy related biases observed in
practice? a meta-analysis of environmental and energy applications of multi-criteria decision
analysis. European Journal of Operational Research, 265, 178–194, https://doi.org/10.1016/j.
ejor.2017.02.038.
Mishra, M., & Chatterjee, S. (2018). Application of analytical hierarchy process (AHP) algorithm to
income insecurity susceptibility mapping – a study in the district of purulia. Socio-Economic
Planning Sciences”. , 62, 56–74, https://doi.org/10.1016/j.seps.2017.07.002.
INMR Olwan, R. M. (2011). “Intellectual property and development: Theory and practice”, Thesis of Doctor of
Philosophy Queensland University of Technology, working paper, Brisbane, Retrieved from
17,3 http://eprints.qut.edu.au/54839/1/Rami_Olwan_Thesis.pdf.
Petrovic, M., Bojkovic, N., Stamenkovic, M., & Anic, I. (2018). Supporting performance appraisal in
ELECTRE based stepwise benchmarking model. Omega, 78, 237–251. https://doi.org/10.1016/j.
omega.2017.07.002.
Saaty, T. L. (1980). Analytic hierarchy process, New York, NY: McGraw-Hill.
346
Sarrazin, R., & De Smet, Y. (2015). Applying multicriteria decision analysis to design safe road projects.
European Journal of Transport and Infrastructure Research, 15, 613–634.
Silva, L. G. D. O., & De Almeida-Filho, A. T. (2018). A new PROMETHEE-based approach applied
within a framework for conflict analysis in evidence theory integrating three conflict measures.
Expert Systems with Applications, 113, 223–232, https://doi.org/10.1016/j.eswa.2018.07.002.
Silva, M. C., Gavião, L. O., Gomes, C. F. S., & Lima, G. B. A. (2017). A proposal for the application of
multicriteria analysis to rank countries according to innovation using the indicators provided by
the WIPO. Rai Revista de Administração e Inovação, 14, 188–198, https://doi.org/10.1016/j.
rai.2017.05.003.
Yu, X., Zhang, S., Liao, X., & Qi, X. (2018). ELECTRE methods in prioritized MCDM environment.
Information Sciences, 424, 301–316. https://doi.org/10.1016/j.ins.2017.09.061.
Appendix Multicriteria
decision
choices
Abbreviation Description

AHP Analytical hierarchy process

CI Consistency index
CR Consistency ratio
347
DM Decision-makers
GII Global innovation index
ICT Information and communication technologies
MCDA Multicriteria decision aid
PROMETHEE Preference Ranking Organization Method for Enrichment Evaluation
TOPSIS Technique for Order of Preference by Similarity to Ideal Solution
UM Upper-middle income countries
WIPO World Intellectual Property Organization
Table A1.
Source: The authors List of abbreviations

About the authors

Marcela do Carmo Silva is a Post-PhD Student at Production Engineering (UFF, 2020). Master in
Management Systems (Fluminense Federal University – UFF, 2015). Graduations in Economic
Sciences (2003) and Production Engineering (2011) by the State University of Rio de Janeiro; MBA in
Project Management (UCP, 2013) and MBA in Organizations and Strategy (UFF, 2015). Professional
experience: Financial, Treasury, Accounts Payable, Accounts Receivable, Controllership, Cost
Management and Revenue from Simple Companies, Presumed Profit and Real Profit, Formulation of
kpis and parameterization of goals. Projects: Consulting, Project Planning, Process Mapping (BPMN),
Process Development (TO-BE), Risk Analysis, Systems Design and Project Management. Marcela do
Carmo Silva is the corrosponding author and can be contacted at: professoramarceladocarmo@gmail.
com
Helder Gomes Costa is a Industrial Engineering Graduate Program Professor and Coordinator at
Fluminense Federal University. Graduation in Mechanical Engineering from Universidade Federal
Fluminense (1987), Master’s at Mechanical Engineering from Pontifícia Universidade Catolica do Rio
de Janeiro (1991) and Doctorate at Mechanical Engineering from Pontifícia Universidade Catolica do
Rio de Janeiro (1994). He is currently full professor at Universidade Federal Fluminense acting on the
following subjects: multicriteria decision-making and performance evaluation.
Carlos Francisco Simões Gomes is a adjunct Professor in Industrial Engineering Department at
Fluminense Federal University. Master’s at Production Engineering from Universidade Federal
Fluminense (1995) and PhD at Production Engineering from Universidade Federal do Rio de Janeiro
(1999). Professional experience in Administration, acting on the following subjects: multicriteria,
operacional research, logistics, decision and management.

For instructions on how to order reprints of this article, please visit our website:
www.emeraldgrouppublishing.com/licensing/reprints.htm
Or contact us for further details: [email protected]