Analysis of the Sustainable Development Indicators in the OECD Countries

Abstract

Sustainable development is a key task for governments that should end poverty, ensure prosperity, create better conditions for health, education or social needs. The set of indicators to be monitored for evaluation of successes or failures of the sustainable development varies by intergovernmental organizations like OECD or EU. To discover the status and dynamics of variables which are part of the sustainable development goals of the OECD countries is the main aim of the presented analysis. To measure the convergence of socio-economic indicators the coefficient of variation was used. The Pearson’s correlations coefficient and regression analysis were applied to detect the linear relationship between a pair of variables. The OECD countries were compared not only by using univariate statistical methods but also by applying a multivariate approach. The cluster analysis and principal component analysis were used for a set of indicators to monitor the countries from a wider perspective. The analyzed indicators GDP per capita or real change in GDP per capita belong to variables of economic activity. Variables of life expectancy at birth, standardized death rates for noncommunicable diseases belong to indicators of health. Altogether fifteen selected indicators were used for a multivariate analysis of OECD countries in two periods of time.

Full text

sustainability
Article
Analysis of the Sustainable Development Indicators
in the OECD Countries
Silvia Megyesiova * and Vanda Lieskovska
Faculty of Business Economics with seat in Košice, Tajovského, University of Economics, Bratislava, 13,
04130 Košice, Slovakia; vanda.lieskovska@euke.sk
*Correspondence: silvia.megyesiova@euke.sk; Tel.: +421-55-722-3228
Received: 29 October 2018; Accepted: 28 November 2018; Published: 3 December 2018


Abstract:
Sustainable development is a key task for governments that should end poverty, ensure
prosperity, create better conditions for health, education or social needs. The set of indicators
to be monitored for evaluation of successes or failures of the sustainable development varies by
intergovernmental organizations like OECD or EU. To discover the status and dynamics of variables
which are part of the sustainable development goals of the OECD countries is the main aim of
the presented analysis. To measure the convergence of socio-economic indicators the coefficient of
variation was used. The Pearson’s correlations coefficient and regression analysis were applied to
detect the linear relationship between a pair of variables. The OECD countries were compared not
only by using univariate statistical methods but also by applying a multivariate approach. The cluster
analysis and principal component analysis were used for a set of indicators to monitor the countries
from a wider perspective. The analyzed indicators GDP per capita or real change in GDP per capita
belong to variables of economic activity. Variables of life expectancy at birth, standardized death rates
for noncommunicable diseases belong to indicators of health. Altogether fifteen selected indicators
were used for a multivariate analysis of OECD countries in two periods of time.
Keywords:
expenditure on health per capita; gross domestic product; life expectancy; death rates;
convergence; cluster analysis; sustainability; OECD countries
1. Introduction
The sustainable development belongs to the main goals and policies of societies all over the world.
The 2030 Agenda for Sustainable Development that was adopted by the international community in
September 2015 represents a set of goals to end poverty, protect the planet and ensure prosperity for
all [1,2].
For the United Nations and the countries, it took several decades to reach this new and very
ambitious Agenda 2030. In 1992 at the Earth Summit, Agenda 21 was adopted. Agenda 21, the Rio
Declaration on Environment and Development, is a comprehensive plan of action to be taken in
every area in which human impacts on the environment. It is a global partnership for sustainable
development to improve human lives and protect the environment [
3
]. In the same year the
Commission on Sustainable Development (SD) was established to monitor, report and ensure the
implementation of the agreements of the Rio Declaration.
The new Millennium brings new pretentious aims for the UN countries. The aims were
summarized in the Millennium Declaration that the Member States adopted in the year 2000.
The eight
Millennium Development Goals include targets for: Reduction of extreme poverty and
hunger, achievements of universal primary education, promotion of gender equality, reduction
of children mortality, improvement of maternal health, halting the spread of HIV/AIDS, malaria
and other diseases, ensuring environmental sustainability and securing global partnership for
Sustainability 2018,10, 4554; doi:10.3390/su10124554 www.mdpi.com/journal/sustainability

Sustainability 2018,10, 4554 2 of 22
development by 2015 [
4
]. Ten years after the Rio Earth Summit the World Summit on SD held
in Johannesburg in 2002 reaffirmed the commitments to the reduction of poverty and environment
protection.
The Johannesburg
Summit 2002 brought attention and action toward improving people’s
lives, protecting natural resources in a world with a growing population [5].
The UN Member States decided to launch a process to develop a set of Sustainable Development
Goals (SDGs) in Rio in 2012, during the Conference on SD (Rio+20 conference). In Rio the Member
States renewed the commitment to SD and guaranteed promotion of an economically, socially and
environmentally sustainable future for the planet and for the present and future generations in the
Resolution: “The future we want” [
6
]. In 2013 an Open Working Group was set up by the General
Assembly with an ambitious objective to develop a proposal on the SDGs. The SDGs were built upon
the Millennium Goals. The negotiation process on the post-2015 development agenda culminated
in the adoption of the 2030 Agenda for Sustainable Development with 17 SDGs and 169 associated
targets at the United Nations SD Summit in September 2015 in New York [1,2].
The implementation of the 2030 Agenda brings benefits for all, for today’s generation and for
future generations, for planet and prosperity, it seeks to strengthen peace and freedom. The benefits
from the performance of the SDGs over the next fifteen years should be notable for [1]:
•People through ending poverty and hunger;
•
Planet by protecting the planet from degradation, sustainable management of its natural resources
and by taking urgent actions on climate change;
•Prosperity by ensuring of prosperous and fulfilling lives for all human beings;
•Peace by determining to foster peaceful, just and inclusive societies free from fear and violence;
•
Partnership by finding the means required to implement the Agenda through Global Partnership
for SD.
The core of the 2030 Agenda are the seventeen Sustainable Development Goals (SDGs) and
associated targets to be achieved over the next 15 years. The SDGs are the most salient points
for understanding and achieving environmental and human development ambitions up to the
year 2030 [
7
], they recognize that ending poverty must be associate with strategies for sustainable
economic growth, for including wide range of social needs, like health, education, social protection,
job opportunities
, protection of environment and tackling climate change [
2
,
3
].The SDGs are not legally
binding, but the governments are expected to take ownership and establish national frameworks for
the achievement of the following seventeen goals [8]:
1. No poverty;
2. Zero hunger;
3. Good health and well-being;
4. Quality education;
5. Gender equality;
6. Clean water and sanitation;
7. Affordable and clean energy;
8. Decent work and economic growth;
9. Industry, innovation and infrastructure;
10.
Reduced inequalities;
11.
Sustainable cities and communities;
12.
Responsible consumption and production;
13.
Climate action;
14.
Life bellow water;
15.
Life on land;
16.
Peace, justice and strong institutions;
17.
Partnership for the goals.

Sustainability 2018,10, 4554 3 of 22
As the SDGs are universally applicable to all countries the EU, OECD and other international
organizations are committed to being the frontrunners in implement the SDGs into their
policies [9,10]
.
OECD and also the EU developed a special indicator set to monitor the SDGs in the OECD
framework [
11
] and the EU framework [
12
–
14
]. OECD will support countries for identification
of the current stand in relation to the SDGs and it will propose sustainable pathways based on
evidence [
15
]. Among the strategies of the OECD toward the 2030 Agenda is the improving policy
coherence through a variety of projects and initiatives. Promoting investment in SD will be the next
ambitious strategy, new resources should be established for ensuring long term development and
improvement of investment conditions. The next initiative should ensure the planet’s sustainability
and create a balance between socio-economic progress and the ecosystems [15,16].
The SDGs feature in 10 European Commission’s priorities [
17
,
18
]: Jobs, growth and investment;
digital single market; energy union and climate; internal market; a deeper and fairer economic
and monetary union; a balanced and progressive trade policy to harness globalization; justice and
fundamental rights; migration; a stronger global actor; democratic change. The strategies, instruments
and actions contributing to the SDGs within the EU are in details presented in the Communication from
the Commission: European action for sustainability [
19
]. The implementation and progress toward the
SDGs are in the interest not only of the United Nations or other intergovernmental organizations, but it
is in the interest of researchers all over the globe. Some very interesting research was conducted in the
field of academic accounting and its role in furthering achievement of the SDGs through enhanced
understanding, critiquing and advancing of accounting policy, practice and theorizing [
7
], while some
other researchers analyzed the differences between the traditionally discrete domains of financial
reporting and sustainability reporting [
20
], they discussed the adoption of corporate governance,
environmental and social practices in order to react to unexpected shocks, while preserving business
sustainability [
21
]. The last but not the least, was the original research on evaluation of the quality
of non-financial information in two selected EU Member States before the implementation of the EU
Non-financial reporting Directive that should increase the quality of sustainability reporting of the
companies [
22
]. Further research on implementation and achievements in the field of the 2030 Agenda
is very needed and perspective.
The main aim of the presented article and analysis is to discover the status and developments
of the selected indicators in the OECD countries. The indicators are part of the SDGs. Using suitable
statistical techniques (one-dimensional or multivariate approaches of analysis) it was possible to follow
the changes of a solo indicator or to look at more than one variable and so to describe the status of the
OECD countries.
2. Materials and Methods
The socio-economic indicators used for analyses of the status and development of the 35 OECD
countries were downloaded from the OECD.Stat database [
23
]. The database is freely accessible and
includes data and metadata for all OECD countries and also for some non-member economies.
For analysis of a solo indicator the univariate statistical approach was chosen [
24
,
25
]. A variable
was characterized by its average level, minimum, maximum, standard deviation, median, first and
third quartile. The analysis focused on the variability of the analyzed indicators. To compare the
changes of variability in selected indicators the range is not always suitable because it measures the
variability only through the maximal and the minimal values of an indicator. Much more suitable for
comparison of a convergence in case of variability is a relative measure of variability, for example the
coefficient of variation. The coefficient of variation (CV) belongs to the so-called sigma convergence
coefficients [
26
,
27
] used to measure the convergence process of selected socio-economic variables.
Convergence is an often-discussed issue not only on the country level but also on the regional
levels [28–30]
. To detect the linear relationship between a pair of indicators the correlation and
regression analysis were used [24,25].

Sustainability 2018,10, 4554 4 of 22
Besides a univariate statistical analytical approach, the contribution is devoted to the application
of multidimensional methods in assessing the state of OECD countries. Among the multivariate
statistical method, the cluster analysis and the principal component analysis were applied to a set of
selected socio-economic indicators. The indicators used for multivariate or univariate analysis are part
of the SDGs of the OECD countries. Cluster analysis is a very useful tool that tries to identify structures
within the selected dataset. For grouping objects of similar kind into categories different algorithms
and methods are used. These methods are often called a segmentation analysis that organizes selected
data into meaningful structures, these methods do not make any distinction between independent or
dependent variables [
31
,
32
]. Cluster analysis (CA) tries to sort different objects (in our case different
OECD countries). CA identifies homogenous groups of objects in a way, that the objects are similar to
one another within the same cluster and dissimilar to the objects in other clusters. It means that we can
expect that the degree of association between two objects is maximal if they belong to the same group
and minimal otherwise [
31
–
33
]. Cluster analysis is a very useful technique that helps to organize
observed date or cases into two or more homogenous groups and an advantage of this analysis is that
it doesn’t require any prior knowledge of which object belongs to which clusters. Different measures
have been used to measure the distance for different data types and several different hierarchical
or non-hierarchical methods are used to determine which clusters should be joined at each stage,
for example nearest neighbor method, furthest neighbor method, average linkage method, centroid
method, Ward’s method, k-means clustering method [34,35].
For cluster analysis 35 OECD Member States were used as objects and a few socio-economic
variables as indicators. The data for analysis comes from the OECD.Stat database [
23
]. The problem of
data collinearity is a usual problem of multivariate statistical analysis. It is necessary to think about
the best way how to solve the problem if the selected variables will be strongly correlated with each
other. For this reason, a principal component analysis should be taken into consideration. Principal
component analysis (PCA) belongs to a group of techniques that creates a smaller number of linear
combinations of analyzed variables. The reduced “new” variables should account for and explain
most of the variance in correlation matrix pattern [
36
,
37
]. PCA is a dimensionality reduction method.
PCA helps the determine a minimum number of factors that will account for the maximum variance
in the dataset in use. It means, that principal component analysis is a dimension-reduction tool that
can be used to reduce a large set of variables to a smaller set of uncorrelated variables called principal
components. PCA creates the same number of components as is the number of analyzed original
variables but usually only a few of them are used for the next analysis, for example in our study for the
cluster analysis technique. The first principal component accounts for the highest variability in the data
and each subsequent component account for as much of the remaining variability as possible [
37
–
39
].
For the next analysis it is useful to take only the first meigenvectors which explain a predetermined
threshold of the total variability, for example the first meigenvectors should in common explain at
least 80% of the total variability of the original dataset. The next way is to check the scree plot of
the PCA
, which plots the variance explained by each of the components. The number of components
to be used in the next analysis can be assessed from the scree plot by the inflection point of the
principal components.
The principal component analysis is a very useful analytical technique for the reduction of the
original dataset dimensions and for the creation of uncorrelated “new” dataset for the subsequent
multivariate analysis.
3. Results
Gross domestic product is a measure for the economic activity, it measures the value of the total
final output of goods and services produced by an economy within a certain period of time [
40
].
It is
also used as a proxy for the development of a country’s material living standards. GDP per capita
is calculated as the ratio of GDP to the average population, it can be expressed in purchasing power
parities (PPP), which represents a common currency that eliminates the differences in price an enables

Sustainability 2018,10, 4554 5 of 22
a meaningful comparison of the GDP [
40
]. The limitations of the GDP are, that it does not monitor the
environmental or social effects of economic activity [41].
GDP is in the interest of the Agenda 2030. The change in real GDP per capita belong to the Goal
8—Promote sustained, inclusive and sustainable economic growth, full and productive employment
and decent work for all, while GDP per capita in PPP is a part of the EU indicators of the Goal
10—Reduce inequality within and among countries [12,13].
Not only the economic activity is in the focus of OECD countries. Health is in point of view of the
population in each country. Health is the main aim of the Goal 3—Ensure healthy lives and promote
well-being for all at all ages [
1
,
2
,
8
]. In the interest of the Goal 3 is the life expectancies at birth and the
death rates of population. The selected indicators analyzed in the presented paper are a part of The
Agenda 2030 for Sustainable Development.
3.1. Gross Domestic Product and Current Expenditure on Health per Capita
In OECD countries the living standard increased steadily. The GDP per capita in PPP current
prices jumped from an average level of USD 23,616.5 to USD 42,429.2 (see Table 1). In 2000 only five
Member States had the GDP per capita lower than $10,000, namely Latvia ($8013), Estonia ($9385),
Turkey ($9426), Chile ($9524) and Mexico ($9974). On the other hand, in the same year the per capita
GDP was higher than $30,000 in Ireland, the Netherland, Switzerland, the United States, Norway and
in Luxembourg ($55,221). The average GDP per capita increased to approximately $42,429
in 2016
.
Mexico had in 2016 according to the GDP per capita figures ($18,583) the lowest living standard
between the OECD Member States. Extremely high GDP per capita was typical for Switzerland
($62,898), Ireland ($72,772) and again Luxembourg ($105,768). Although the standard deviation of the
GDP per capita increased from 2000
till 2016
, the convergence of countries from the GDP per capita
point of view should be measured using a relative rate of variability. The coefficient of variation (CV) is
a suitable relative measure of a convergence process of living standard across countries [
42
,
43
].
The CV
decreased from a level of 43% in 2000 to 38.8% in 2016. The decreasing relative variability is a good
signal of a convergence process of the GDP per capita of the OECD population.
Table 1. GDP per capita in OECD countries (purchasing power parities (PPP), current prices).
Sustainability 2018, 10, x FOR PEER REVIEW 5 of 23
and decent work for all, while GDP per capita in PPP is a part of the EU indicators of the Goal 10—
Reduce inequality within and among countries [12,13].
Not only the economic activity is in the focus of OECD countries. Health is in point of view of
the population in each country. Health is the main aim of the Goal 3—Ensure healthy lives and
promote well-being for all at all ages [1,2,8]. In the interest of the Goal 3 is the life expectancies at
birth and the death rates of population. The selected indicators analyzed in the presented paper are
a part of The Agenda 2030 for Sustainable Development.
3.1. Gross Domestic Product and Current Expenditure on Health per Capita
In OECD countries the living standard increased steadily. The GDP per capita in PPP current
prices jumped from an average level of USD 23,616.5 to USD 42,429.2 (see Table 1). In 2000 only five
Member States had the GDP per capita lower than $10,000, namely Latvia ($8013), Estonia ($9385),
Turkey ($9426), Chile ($9524) and Mexico ($9974). On the other hand, in the same year the per capita
GDP was higher than $30,000 in Ireland, the Netherland, Switzerland, the United States, Norway and
in Luxembourg ($55,221). The average GDP per capita increased to approximately $42,429 in 2016.
Mexico had in 2016 according to the GDP per capita figures ($18,583) the lowest living standard
between the OECD Member States. Extremely high GDP per capita was typical for Switzerland
($62,898), Ireland ($72,772) and again Luxembourg ($105,768). Although the standard deviation of
the GDP per capita increased from 2000 till 2016, the convergence of countries from the GDP per
capita point of view should be measured using a relative rate of variability. The coefficient of
variation (CV) is a suitable relative measure of a convergence process of living standard across
countries [42,43]. The CV decreased from a level of 43% in 2000 to 38.8% in 2016. The decreasing
relative variability is a good signal of a convergence process of the GDP per capita of the OECD
population.
In OECD countries the relative measure of variability of per capita expenditure on health (in
PPP, current prices) is higher compared to the CV of GDP/capita. In 2000 the CV of current
expenditures on health per capita was as high as 55%. Also, in this case the variability began to decline
after 2000 and the CV reached 50.7% in 2016. The average amount spent on health per capita increased
from $1788 in 2000 to $3997 in 2016 (see Table 2). The health expenditure (HE) per capita in 2000 was
lower than USD 600 altogether in five countries, namely in Turkey ($425), Latvia ($437), Mexico
($484), Estonia ($486) and Poland ($564). In the same year, very high expenditure on health per capita
was typical for countries which faced also a very high level of GDP per capita. So, for example in
Switzerland the expenditure on health per capita stood at $3332, in Luxembourg $3405 and the United
States $4559.
Table 1. GDP per capita in OECD countries (purchasing power parities (PPP), current prices).
Source: OECD.Stat database [23]—own calculations based on OECD database.
Source: OECD.Stat database [23]—own calculations based on OECD database.
In OECD countries the relative measure of variability of per capita expenditure on health (in PPP,
current prices) is higher compared to the CV of GDP/capita. In 2000 the CV of current expenditures on
health per capita was as high as 55%. Also, in this case the variability began to decline after 2000 and
the CV reached 50.7% in 2016. The average amount spent on health per capita increased from $1788 in
2000 to $3997 in 2016 (see Table 2). The health expenditure (HE) per capita in 2000 was lower than

Sustainability 2018,10, 4554 6 of 22
USD 600 altogether in five countries, namely in Turkey ($425), Latvia ($437), Mexico ($484), Estonia
($486) and Poland ($564). In the same year, very high expenditure on health per capita was typical for
countries which faced also a very high level of GDP per capita. So, for example in Switzerland the
expenditure on health per capita stood at $3332, in Luxembourg $3405 and the United States $4559.
The ranking of the countries with the lowest or the highest HE per capita did not change a lot
till 2016. Again, the highest expenses on health were reached in Luxembourg ($7463), Switzerland
($7919) and the United States. In the USA the expenditure on health per capita reached almost $10,000.
Compared with other OECD countries it is an extremely high level of HE per capita. On the opposite
side are the countries with the lowest expenses on health per capita in 2016, like Mexico ($1080),
Turkey ($1088), Latvia ($1466), Poland ($1798).
Table 2. Current expenditure on health per capita in OECD countries (PPP, current prices).
Sustainability 2018, 10, x FOR PEER REVIEW 6 of 23
The ranking of the countries with the lowest or the highest HE per capita did not change a lot
till 2016. Again, the highest expenses on health were reached in Luxembourg ($7463), Switzerland
($7919) and the United States. In the USA the expenditure on health per capita reached almost
$10,000. Compared with other OECD countries it is an extremely high level of HE per capita. On the
opposite side are the countries with the lowest expenses on health per capita in 2016, like Mexico
($1080), Turkey ($1088), Latvia ($1466), Poland ($1798).
Table 2. Current expenditure on health per capita in OECD countries (PPP, current prices).
Source: OECD.Stat database [23]—own calculations based on OECD database.
According to the finding that countries have a similar ranking position by both variables (GDP
per capita and health expenditures per capita) makes it possible to thing about a linear relationship
between these characteristics. In Figure 1 the association of these variables is presented. In 2016 the
Pearson’s correlation coefficient between both variables was very high and positive (rxy = 0.83). It
means that in countries with high GDP per capita we can expect a high level of expenditure per capita
and vice versa. From the lowest values of both characteristics till a level of about $5600 in the case of
HE and about $60,000 in the case of GDP per capita the countries tightly copy the linear regression
line, but from these levels the deviation of the countries from the theoretical regression line is rifer,
in 2016 it was the case of Ireland, Luxembourg, Switzerland and the United States. For Luxembourg,
we would according to the regression line expect due to a very high GDP per capita level a higher
per capita expenditure on health, and in the case of the USA, a lower health expenditure is expected
according to the GDP figures. The high correlation coefficient and also the high determination index
(R2 = 0.69) between both variables indicate a strong linear association between expenditure on health
and GDP per capita in OECD countries.
Source: OECD.Stat database [23]—own calculations based on OECD database.
According to the finding that countries have a similar ranking position by both variables (GDP
per capita and health expenditures per capita) makes it possible to thing about a linear relationship
between these characteristics. In Figure 1the association of these variables is presented. In 2016
the Pearson’s correlation coefficient between both variables was very high and positive (r
xy
= 0.83).
It means
that in countries with high GDP per capita we can expect a high level of expenditure per capita
and vice versa. From the lowest values of both characteristics till a level of about $5600 in the case of
HE and about $60,000 in the case of GDP per capita the countries tightly copy the linear regression line,
but from these levels the deviation of the countries from the theoretical regression line is rifer, in 2016 it
was the case of Ireland, Luxembourg, Switzerland and the United States.
For Luxembourg
, we would
according to the regression line expect due to a very high GDP per capita level a higher per capita
expenditure on health, and in the case of the USA, a lower health expenditure is expected according to
the GDP figures. The high correlation coefficient and also the high determination index (R
2
= 0.69)
between both variables indicate a strong linear association between expenditure on health and GDP
per capita in OECD countries.
The next assumption was that the GDP per capita, as a very rough estimate of the overall country’s
productivity, will grow more strongly than the expenditure on health per capita. The expectation that
productivity generates the growth of expenditure on health has not been confirmed. Almost in all
OECD countries the increase of real GDP per capita was lower than the real change in HE per capita
between 2000 and 2016 (see Figure 2). Most notable was the difference between both characteristics in
Korea and Chile. In Korea the real GDP per capita increased in the selected time span by 69.7%, but the
real expenditure on health per capita grow by 222.9% and so the difference between both growth rates
hit 153.3 percentage points (p.p.). In Chile the gap was as high as 108.7 p.p. In six OECD countries the

Sustainability 2018,10, 4554 7 of 22
difference between growth rates was higher than 50 p.p. but lower than 100 p.p., namely in United
Kingdom (74 p.p.), Sweden (62.6 p.p.), Japan (58.4 p.p.), the Netherlands (55.2 p.p.), Slovakia (55 p.p.)
and Estonia (51.4 p.p.). In only two countries the real GDP per capita growth trespassed the real
health expenses per capita changes. In Iceland the GDP per capita increased by 32% while the health
expenditures per capita increased more moderately by 27.5%, in Turkey the changes reached 76.4%
and 68.2% respectively. Even in countries most affected by the economic crisis the health expenses per
capita grows moderately. For example, in Greece where the real GDP per capita between 2000 and
2016 declined by 2.7% the health expenses increased by 9.7%. In Italy the decline of GDP per capita
was more significant, it dropped by 5.3%, but the real health expenses per capita changed oppositely
and grow by 11.6%.
Sustainability 2018, 10, x FOR PEER REVIEW 7 of 23
Figure 1. Association of expenditure on health per capita and GDP per capita in OECD countries in
2016 (per capita in PPP, USD, current prices, 2016). Source: OECD.Stat database [23]—own
calculations based on OECD database. Country codes: AUS—Australia, AUT—Austria, BEL—
Belgium, CAN—Canada, CHL—Chile, CZE—Czech Republic, DNK—Denmark, EST—Estonia,
FIN—Finland, FRA—France, DEU—Germany, GRC—Greece, HUN—Hungary, ISL—Iceland, IRL—
Ireland, IST—Israel, ITA—Italy, JPN—Japan, SO—Korea, LVA—Latvia, LUX—Luxembourg, MEX—
Mexico, NLD—Netherlands, NZL—New Zealand, NOR—Norway, POL—Poland, PRT—Portugal,
SVK—Slovakia, SVN—Slovenia, ESP—Spain, SWE—Sweden, CHE—Switzerland, TUR—Turkey,
GBR—United Kingdom, USA—United States.
The next assumption was that the GDP per capita, as a very rough estimate of the overall
country’s productivity, will grow more strongly than the expenditure on health per capita. The
expectation that productivity generates the growth of expenditure on health has not been confirmed.
Almost in all OECD countries the increase of real GDP per capita was lower than the real change in
HE per capita between 2000 and 2016 (see Figure 2). Most notable was the difference between both
characteristics in Korea and Chile. In Korea the real GDP per capita increased in the selected time
span by 69.7%, but the real expenditure on health per capita grow by 222.9% and so the difference
between both growth rates hit 153.3 percentage points (p.p.). In Chile the gap was as high as 108.7
p.p. In six OECD countries the difference between growth rates was higher than 50 p.p. but lower
than 100 p.p., namely in United Kingdom (74 p.p.), Sweden (62.6 p.p.), Japan (58.4 p.p.), the
Netherlands (55.2 p.p.), Slovakia (55 p.p.) and Estonia (51.4 p.p.). In only two countries the real GDP
per capita growth trespassed the real health expenses per capita changes. In Iceland the GDP per
capita increased by 32% while the health expenditures per capita increased more moderately by
27.5%, in Turkey the changes reached 76.4% and 68.2% respectively. Even in countries most affected
by the economic crisis the health expenses per capita grows moderately. For example, in Greece
where the real GDP per capita between 2000 and 2016 declined by 2.7% the health expenses increased
by 9.7%. In Italy the decline of GDP per capita was more significant, it dropped by 5.3%, but the real
health expenses per capita changed oppositely and grow by 11.6%.
y = 6.74x + 15488
R² = 0.69
10 000
20 000
30 000
40 000
50 000
60 000
70 000
80 000
90 000
100 000
110 000
120 000
GDP per capita
Health expenditures per capita
LUX
MEX
TUR LVA
CHL
IRL
USA
CHE
NOR
ISL
SVK
CZE
GRC
SVN
NZL
FIN
FRA
BEL DEU
NLD
ITA
Figure 1.
Association of expenditure on health per capita and GDP per capita in OECD countries
in 2016 (per capita in PPP, USD, current prices, 2016). Source: OECD.Stat database [
23
]—own
calculations based on OECD database. Country codes: AUS—Australia, AUT—Austria, BEL—Belgium,
CAN—Canada, CHL—Chile, CZE—Czech Republic, DNK—Denmark, EST—Estonia, FIN—Finland,
FRA—France, DEU—Germany, GRC—Greece, HUN—Hungary, ISL—Iceland, IRL—Ireland,
IST—Israel, ITA—Italy, JPN—Japan, SO—Korea, LVA—Latvia, LUX—Luxembourg, MEX—Mexico,
NLD—Netherlands, NZL—New Zealand, NOR—Norway, POL—Poland, PRT—Portugal, SVK—Slovakia,
SVN—Slovenia, ESP—Spain, SWE—Sweden, CHE—Switzerland, TUR—Turkey, GBR—United Kingdom,
USA—United States.
Strong increases of the health expenditures are visible also through an increase of the current health
expenditures as a percent of GDP (see Figure 3). In the time span of selected years (
2000 and 2016
) the
health expenditures as a share of GDP increased most significantly in the USA by 4.7 p.p. The USA
had reached the highest shares of the health expenses between the OECD countries with 12.5% in 2000
and 17.2% in 2016. On the other hand, the smallest share was achieved in Turkey with a starting level
of 4.6% in 2000 and with an ending ratio of 4.3% in 2016. The range between maximum and minimum
increased from 8.5 p.p. in 2000 to 12.9 p.p. in 2016. While in 2000 only in one country (USA) the total
health expenditures as a percent of GDP was higher than 10%, in 2016 it was a reality in 12 countries:
Canada, Denmark, Austria, Belgium, Norway, the Netherlands, Japan, France, Sweden, Germany,
Switzerland and USA. The threshold of 10% was overpassed in most developed OECD countries with
a high living standard, high levels of GDP per capita, high life expectancies at birth.

Sustainability 2018,10, 4554 8 of 22
Sustainability 2018, 10, x FOR PEER REVIEW 8 of 23
Figure 2. Real change in Expenditure on health per capita and GDP per capita in OECD countries
between 2000 and 2016 in %. Source: OECD.Stat database [23]—own calculations based on OECD
database.
Strong increases of the health expenditures are visible also through an increase of the current
health expenditures as a percent of GDP (see Figure 3). In the time span of selected years (2000 and
2016) the health expenditures as a share of GDP increased most significantly in the USA by 4.7 p.p.
The USA had reached the highest shares of the health expenses between the OECD countries with
12.5% in 2000 and 17.2% in 2016. On the other hand, the smallest share was achieved in Turkey with
a starting level of 4.6% in 2000 and with an ending ratio of 4.3% in 2016. The range between maximum
and minimum increased from 8.5 p.p. in 2000 to 12.9 p.p. in 2016. While in 2000 only in one country
(USA) the total health expenditures as a percent of GDP was higher than 10%, in 2016 it was a reality
in 12 countries: Canada, Denmark, Austria, Belgium, Norway, the Netherlands, Japan, France,
Sweden, Germany, Switzerland and USA. The threshold of 10% was overpassed in most developed
OECD countries with a high living standard, high levels of GDP per capita, high life expectancies at
birth.
y = 0.47x -0.1478
R² = 0.58
- 10.0
0.0
10.0
20.0
30.0
40.0
50.0
60.0
70.0
80.0
90.0
100.0
110.0
120.0
0.0 20.0 40.0 60.0 80.0 100.0 120.0 140.0 160.0 180.0 200.0 220.0
real change in % -GDP per capita, 2016 -2000
real change in % -expanditure on health per capita, 2016 -2000
ITA
GRC
PRT FRA DNK
AUT
ISL
ISR
SVN
TUR
HUN
NZL SWE
GBR
JPN
NLD
CZE
AUS
IRL
POL EST
SVK
LVA
CHL
SO
USA
NOR
0.0 2.0 4.0 6.0 8.0 10.0 12.0 14.0 16.0 18.0
TUR
LVA
MEX
LUX
POL
EST
SVK
CZE
ISR
HUN
SO
IRL
GRC
CHL
SVN
ISL
ITA
PRT
ESP
NZL
FIN
AUS
GBR
CAN
DNK
AUT
BEL
NOR
NLD
JPN
FRA
SWE
DEU
CHE
USA
2016 2000
Figure 2.
Real change in Expenditure on health per capita and GDP per capita in OECD countries between
2000 and 2016 in %. Source: OECD.Stat database [23]—own calculations based on OECD database.
Sustainability 2018, 10, x FOR PEER REVIEW 8 of 23
Figure 2. Real change in Expenditure on health per capita and GDP per capita in OECD countries
between 2000 and 2016 in %. Source: OECD.Stat database [23]—own calculations based on OECD
database.
Strong increases of the health expenditures are visible also through an increase of the current
health expenditures as a percent of GDP (see Figure 3). In the time span of selected years (2000 and
2016) the health expenditures as a share of GDP increased most significantly in the USA by 4.7 p.p.
The USA had reached the highest shares of the health expenses between the OECD countries with
12.5% in 2000 and 17.2% in 2016. On the other hand, the smallest share was achieved in Turkey with
a starting level of 4.6% in 2000 and with an ending ratio of 4.3% in 2016. The range between maximum
and minimum increased from 8.5 p.p. in 2000 to 12.9 p.p. in 2016. While in 2000 only in one country
(USA) the total health expenditures as a percent of GDP was higher than 10%, in 2016 it was a reality
in 12 countries: Canada, Denmark, Austria, Belgium, Norway, the Netherlands, Japan, France,
Sweden, Germany, Switzerland and USA. The threshold of 10% was overpassed in most developed
OECD countries with a high living standard, high levels of GDP per capita, high life expectancies at
birth.
y = 0.47x -0.1478
R² = 0.58
- 10.0
0.0
10.0
20.0
30.0
40.0
50.0
60.0
70.0
80.0
90.0
100.0
110.0
120.0
0.0 20.0 40.0 60.0 80.0 100.0 120.0 140.0 160.0 180.0 200.0 220.0
real change in % -GDP per capita, 2016 -2000
real change in % -expanditure on health per capita, 2016 -2000
ITA
GRC
PRT FRA DNK
AUT
ISL
ISR
SVN
TUR
HUN
NZL SWE
GBR
JPN
NLD
CZE
AUS
IRL
POL EST
SVK
LVA
CHL
SO
USA
NOR
0.0 2.0 4.0 6.0 8.0 10.0 12.0 14.0 16.0 18.0
TUR
LVA
MEX
LUX
POL
EST
SVK
CZE
ISR
HUN
SO
IRL
GRC
CHL
SVN
ISL
ITA
PRT
ESP
NZL
FIN
AUS
GBR
CAN
DNK
AUT
BEL
NOR
NLD
JPN
FRA
SWE
DEU
CHE
USA
2016 2000
Figure 3.
Current expenditure on health as a share of GDP in 2016 (in %). Source: OECD.Stat
database [23]—own calculations based on OECD database.
Strong real growth of HE should positively affect the health outcome of the population in OECD
countries. To discover the expected positive changes in health the following characteristics were
chosen: Life expectancy (at birth and at the age 65), standardized death rates of noncommunicable
diseases. Moreover, these variables were included in the multivariate analysis.
3.2. Life Expectancy at Birth and at Age 65
Life expectancy (LE) at birth refers to the mean number of years a new-born child can expect to
live assuming that current mortality levels remain throughout his or her life constant [
40
]. LE at birth

Sustainability 2018,10, 4554 9 of 22
belongs to the indicators of EU SDG indicator set for Goal 3 [
12
,
13
]. Live expectancies are very often
used for international comparison of living standards, economic development in selected countries,
they are used as an indicator of public health condition and status, they can explain the relation
between the LE and pollution in selected areas [
44
,
45
]. Live expectancy (LE) is also calculated for
specific ages and characterize the average number of years that a person at that age can be expected
to live. To measure the success in declining mortality usually LE at a certain age is a very useful toll
how to identify the progress of improved health care, changing living standards. For this purpose,
the age
65 is a boundary that is generally accepted. LE at age 65 is the average number of years still to
be lived by a person who has reached the age 65, if subjected throughout the rest of his or her life to
the age-specific probabilities of dying [40].
Life expectancy at birth increased progressively in the OECD countries. The average LE for
females was 73 years in 1970 and jumped to 83.2 in 2015. The LE for males was lower by 6.3 years
compared to the life expectancies women in 1970 and stood at 66.7 years. The gap of LE at birth
between both sexes is closing. In 2015 the difference between the average levels of LE between men and
women was 5.2 years, the LE for men increased to 77.9 in 2015 (see Table 3). The correlation between
LE for males and females is strong a positive. The box plot presentation of the women’s LE at birth in
Figure 4identified four extreme values of LE for Turkey (56.3 years), Mexico (63.2), Chile (65.4) and
Korea (65.8) in 1970. It is necessary to mention that in this year the data for Canada, Israel, Italy and
Latvia were not available. Moreover, in the next decades one country had extremely low LE for women
at birth, but positively can be considered the elimination of the extreme values. In 2015 no extreme LE
for females was discovered. Not only women’s LE at birth were, in some countries, extremely low,
but the situation was, in some patterns, similar also for men. In 1970 altogether four countries had
extremely low LE at birth for men: Turkey (52 years), Mexico (58.5), Korea (58.7), Chile (59.1). It is
again necessary to mention that Canada, Israel, Italy and Latvia did not have any available data of
LE for this period. In 2015 still in one country the LE at birth for men was extremely low compared
to other OECD member states (Latvia; 69.7 years). While the minimum of LE at birth for females
was typical for Turkey and Mexico (2015), the lowest LE at birth for men was only at the beginning
reached in Turkey, later on the lowest values were typical for the former communist countries, Estonia
and Latvia.
Table 3. Selected characteristics of life expectancy (LE) at birth in OECD countries.
Sustainability 2018, 10, x FOR PEER REVIEW 10 of 23
Table 3. Selected characteristics of life expectancy (LE) at birth in OECD countries.
Source: OECD.Stat database [23]—own calculations based on OECD database.
Figure 4. Box-plots of life expectancies at birth in OECD countries. Source: OECD.Stat database [23]—
own calculations based on OECD database.
A very positive development was discovered for the LE at age 65. While the average LE at birth
for women increased by about 14% and for man by 16.8%, the LE at age 65 increased more rapidly.
Women in OECD countries at age 65 could expect to live another 21.1 years in 2015 compared with
15.6 years in 1970, which consists of an overall increase by about 35%. Men’s LE at 65 in the same
time span jumped by more than 40% from 12.7 years in 1970 to 17.9 years in 2015. One of the reasons
for increasing spending on health per capita can be the increasing life expectancy, especially the LE
at higher ages. The aging of the population is nowadays an issue that is often discussed from the
perspective of the economic aspects of aging, the health care system, the expenditures for health care,
and the sociological or psychological problems of the elderly population. Extreme values of LE at age
65 were not detected (see Figure 5) as often as in the case of the LE at birth. The lowest levels were
typical for Turkey and Hungary in the case of women and in Korea or some former communist
countries like the Czech Republic, Estonia, Latvia for men (see Table 4). On the other hand, high LE
at age 65 was achieved for women in very developed countries like Sweden, Iceland, France, Japan
and for men in Greece, Iceland, Switzerland and Australia.
Source: OECD.Stat database [23]—own calculations based on OECD database.

Sustainability 2018,10, 4554 10 of 22
Sustainability 2018, 10, x FOR PEER REVIEW 10 of 23
Table 3. Selected characteristics of life expectancy (LE) at birth in OECD countries.
Source: OECD.Stat database [23]—own calculations based on OECD database.
Figure 4. Box-plots of life expectancies at birth in OECD countries. Source: OECD.Stat database [23]—
own calculations based on OECD database.
A very positive development was discovered for the LE at age 65. While the average LE at birth
for women increased by about 14% and for man by 16.8%, the LE at age 65 increased more rapidly.
Women in OECD countries at age 65 could expect to live another 21.1 years in 2015 compared with
15.6 years in 1970, which consists of an overall increase by about 35%. Men’s LE at 65 in the same
time span jumped by more than 40% from 12.7 years in 1970 to 17.9 years in 2015. One of the reasons
for increasing spending on health per capita can be the increasing life expectancy, especially the LE
at higher ages. The aging of the population is nowadays an issue that is often discussed from the
perspective of the economic aspects of aging, the health care system, the expenditures for health care,
and the sociological or psychological problems of the elderly population. Extreme values of LE at age
65 were not detected (see Figure 5) as often as in the case of the LE at birth. The lowest levels were
typical for Turkey and Hungary in the case of women and in Korea or some former communist
countries like the Czech Republic, Estonia, Latvia for men (see Table 4). On the other hand, high LE
at age 65 was achieved for women in very developed countries like Sweden, Iceland, France, Japan
and for men in Greece, Iceland, Switzerland and Australia.
Figure 4.
Box-plots of life expectancies at birth in OECD countries. Source: OECD.Stat
database [23]—own calculations based on OECD database.
A very positive development was discovered for the LE at age 65. While the average LE at birth
for women increased by about 14% and for man by 16.8%, the LE at age 65 increased more rapidly.
Women in OECD countries at age 65 could expect to live another 21.1 years in 2015 compared with
15.6 years in 1970, which consists of an overall increase by about 35%. Men’s LE at 65 in the same
time span jumped by more than 40% from 12.7 years in 1970 to 17.9 years in 2015. One of the reasons
for increasing spending on health per capita can be the increasing life expectancy, especially the LE
at higher ages. The aging of the population is nowadays an issue that is often discussed from the
perspective of the economic aspects of aging, the health care system, the expenditures for health care,
and the sociological or psychological problems of the elderly population. Extreme values of LE at
age 65 were not detected (see Figure 5) as often as in the case of the LE at birth. The lowest levels
were typical for Turkey and Hungary in the case of women and in Korea or some former communist
countries like the Czech Republic, Estonia, Latvia for men (see Table 4). On the other hand, high LE at
age 65 was achieved for women in very developed countries like Sweden, Iceland, France, Japan and
for men in Greece, Iceland, Switzerland and Australia.
Table 4. Selected characteristics of LE at age 65 in OECD countries.
Sustainability 2018, 10, x FOR PEER REVIEW 11 of 23
Figure 5. Box-plots of life expectancies at age 65 in OECD countries. Source: OECD.Stat database
[23]—own calculations based on OECD database.
Table 4. Selected characteristics of LE at age 65 in OECD countries.
Source: OECD.Stat database [23]—own calculations based on OECD database.
3.3. Clusters of OECD Countries
Multivariate analysis of the EU countries allows to include in the analysis not only a solo
indicator but to select a few variables and to make an analysis with a group of the selected indicators.
The first important issue is to appoint the set of variables. The researchers should pay enough
attention to the selection procedure, as the outcome of the multivariate analysis will depend on the
originally analyzed dataset. The main aim of a multivariate analysis was the creation of relatively
good isolated groups of OECD countries, where the countries, characterized as objects, within a
specific cluster are similar to each other and are dissimilar to the objects in different clusters. The
multivariate analysis was carried out by professional statistical software SAS Enterprise Guide 6.1.
For analytical purpose, some socio-economic indicators were chosen, some of the indicators are part
of the new sustainability effort of the EU or the OECD [12,13,46]. The variables are taken for the latest
available period, some of the indicators are available already for the year 2016, but some of them only
for years 2013 or 2015. Some variables are identified separately for males and/or females and some
variables are commonly used for the CA classified as total (standardized death rates). The commonly
used indicators are also the expenditures for health care per inhabitant in PPS and the ratio of the
total health care expenditures to GDP in percent, between some indicators we expect a strong linear
association [47].
The OECD countries were grouped into clusters based on their similarity in two periods of time.
For classification altogether 15 characteristics were used: Some indicators of economic situation in
Source: OECD.Stat database [23]—own calculations based on OECD database.

Sustainability 2018,10, 4554 11 of 22
Sustainability 2018, 10, x FOR PEER REVIEW 11 of 23
Figure 5. Box-plots of life expectancies at age 65 in OECD countries. Source: OECD.Stat database
[23]—own calculations based on OECD database.
Table 4. Selected characteristics of LE at age 65 in OECD countries.
Source: OECD.Stat database [23]—own calculations based on OECD database.
3.3. Clusters of OECD Countries
Multivariate analysis of the EU countries allows to include in the analysis not only a solo
indicator but to select a few variables and to make an analysis with a group of the selected indicators.
The first important issue is to appoint the set of variables. The researchers should pay enough
attention to the selection procedure, as the outcome of the multivariate analysis will depend on the
originally analyzed dataset. The main aim of a multivariate analysis was the creation of relatively
good isolated groups of OECD countries, where the countries, characterized as objects, within a
specific cluster are similar to each other and are dissimilar to the objects in different clusters. The
multivariate analysis was carried out by professional statistical software SAS Enterprise Guide 6.1.
For analytical purpose, some socio-economic indicators were chosen, some of the indicators are part
of the new sustainability effort of the EU or the OECD [12,13,46]. The variables are taken for the latest
available period, some of the indicators are available already for the year 2016, but some of them only
for years 2013 or 2015. Some variables are identified separately for males and/or females and some
variables are commonly used for the CA classified as total (standardized death rates). The commonly
used indicators are also the expenditures for health care per inhabitant in PPS and the ratio of the
total health care expenditures to GDP in percent, between some indicators we expect a strong linear
association [47].
The OECD countries were grouped into clusters based on their similarity in two periods of time.
For classification altogether 15 characteristics were used: Some indicators of economic situation in
Figure 5.
Box-plots of life expectancies at age 65 in OECD countries. Source: OECD.Stat
database [23]—own calculations based on OECD database.
3.3. Clusters of OECD Countries
Multivariate analysis of the EU countries allows to include in the analysis not only a solo indicator
but to select a few variables and to make an analysis with a group of the selected indicators.
The first
important issue is to appoint the set of variables. The researchers should pay enough attention to
the selection procedure, as the outcome of the multivariate analysis will depend on the originally
analyzed dataset. The main aim of a multivariate analysis was the creation of relatively good isolated
groups of OECD countries, where the countries, characterized as objects, within a specific cluster are
similar to each other and are dissimilar to the objects in different clusters. The multivariate analysis
was carried out by professional statistical software SAS Enterprise Guide 6.1. For analytical purpose,
some socio-economic indicators were chosen,
some of
the indicators are part of the new sustainability
effort of the EU or the OECD [
12
,
13
,
46
]. The variables are taken for the latest available period, some of
the indicators are available already for the year 2016, but some of them only for years 2013 or 2015.
Some variables are identified separately for males and/or females and some variables are commonly
used for the CA classified as total (standardized death rates). The commonly used indicators are also
the expenditures for health care per inhabitant in PPS and the ratio of the total health care expenditures
to GDP in percent, between some indicators we expect a strong linear association [47].
The OECD countries were grouped into clusters based on their similarity in two periods of time.
For classification altogether 15 characteristics were used: Some indicators of economic situation in
the countries, indicators of health expenditures, life expectancy at birth and standardized death rates
of noncommunicable diseases. In the case of strong correlations between the selected indicators the
most important principal components were used to carry out the CA instead of the original strongly
correlated dataset. The principal components are linearly independent and so, they are suitable as
input data for the cluster analysis method. Only the “most important” principle components were
chosen to form the mutually isolated clusters of OECD countries; the optimal number of clusters was
determined, using more ways of deciding [48–50].
The list of selected indicators used for the cluster analysis is as follows:
x
1GDP per capita, in PPP, current prices,
x
2Real change in GDP per capita in PPP,
x
3Current expenditure on health per capita (all functions), in PPP, current prices,
x
4Real change in Health expenditure per capita in PPP,
x
5Health expenditures as % of GDP,
x
6
Government expenditure on health (compulsory schemes) as % of the current expenditure
on health,
x
7Household out-of-pocket payments on health as % of current expenditure on health,

Sustainability 2018,10, 4554 12 of 22
x
8Life expectancy at birth—females,
x
9Life expectancy at birth—males,
x
10 Standardized death rates—malignant neoplasms (MN),
x
11 Standardized death rates—diabetes mellitus (DM),
x
12 Standardized death rates—mental and behavioral disorders (MBD),
x
13 Standardized death rates—diseases of the nervous system (NS),
x
14 Standardized death rates—diseases of the circulatory system (CS),
x
15 Standardized death rates—diseases of the respiratory system (RS).
3.3.1. Cluster Analysis—First Period
In the first period of cluster analysis the variables x
1
, x
3
, x
5
–x
15
represents the data from the
beginning of the analyzed period, for year the 2000. To include the dynamics of GDP per capita and
health expenditures per capita, the indicators x
2
and x
4
indicate the real change in both indicators
between 2000 and 2008. Due to a strong correlation between some pairs of the variables presented in
Table 5the PCA was used for calculation of uncorrelated principals.
Table 5. Pearson’s correlation coefficients (first period).
Sustainability 2018, 10, x FOR PEER REVIEW 12 of 23
the countries, indicators of health expenditures, life expectancy at birth and standardized death rates
of noncommunicable diseases. In the case of strong correlations between the selected indicators the
most important principal components were used to carry out the CA instead of the original strongly
correlated dataset. The principal components are linearly independent and so, they are suitable as
input data for the cluster analysis method. Only the “most important” principle components were
chosen to form the mutually isolated clusters of OECD countries; the optimal number of clusters was
determined, using more ways of deciding [48–50].
The list of selected indicators used for the cluster analysis is as follows:
x
1
GDP per capita, in PPP, current prices,
x
2
Real change in GDP per capita in PPP,
x
3
Current expenditure on health per capita (all functions), in PPP, current
prices,
x
4
Real change in Health expenditure per capita in PPP,
x
5
Health expenditures as % of GDP,
x
6
Government expenditure on health (compulsory schemes) as % of the
current expenditure on health,
x
7
Household out-of-pocket payments on health as % of current expenditure
on health,
x
8
Life expectancy at birth—females,
x
9
Life expectancy at birth—males,
x
10
Standardized death rates—malignant neoplasms (MN),
x
11
Standardized death rates—diabetes mellitus (DM),
x
12
Standardized death rates—mental and behavioral disorders (MBD),
x
13
Standardized death rates—diseases of the nervous system (NS),
x
14
Standardized death rates—diseases of the circulatory system (CS),
x
15
Standardized death rates—diseases of the respiratory system (RS).
3.3.1. Cluster Analysis—First Period
In the first period of cluster analysis the variables x1, x3, x5–x15 represents the data from the
beginning of the analyzed period, for year the 2000. To include the dynamics of GDP per capita and
health expenditures per capita, the indicators x2 and x4 indicate the real change in both indicators
between 2000 and 2008. Due to a strong correlation between some pairs of the variables presented in
Table 5 the PCA was used for calculation of uncorrelated principals.
Table 5. Pearson’s correlation coefficients (first period)
Source: OECD.Stat database [23]—own calculations based on OECD database.
For CA only the first five most important principals were used, these five principal components
explain about 84% (see Figure 6) of the variation of the original dataset.
Variables
x
1
x
2
x
3
x
4
x
5
x
6
x
7
x
8
x
9
x
10
x
11
x
12
x
13
x
14
x
15
x
1
1
x
2
-0.610 1
x
3
0.897 -0.614 1
x
4
-0.527 0.808 -0.621 1
x
5
0.520 -0.514 0.824 -0. 649 1
x
6
0.262 -0.091 0.106 -0.118 0.011 1
x
7
-0.489 0.293 -0.474 0.256 -0.395 -0.768 1
x
8
0.644 -0.608 0.612 -0.545 0.522 0.279 - 0.336 1
x
9
0.696 -0.800 0.667 -0.647 0. 534 0.222 -0. 344 0.879 1
x
10
0.037 0.207 0.027 0.187 0. 096 0.427 -0.416 0.042 -0.133 1
x
11
-0.294 -0.200 -0. 267 -0.130 -0.231 -0.530 0.501 -0.350 - 0.137 -0.478 1
x
12
0.414 -0.308 0.320 -0.173 0.097 -0.044 -0. 085 0.284 0.385 -0.220 -0.048 1
x
13
0.532 -0.422 0.623 -0.457 0.532 -0.017 -0.282 0.282 0.445 -0.074 -0.105 0.560 1
x
14
-0.430 0.752 -0.390 0. 557 -0.236 0.163 0.024 -0.474 -0.697 0.596 -0.329 -0.488 -0.402 1
x
15
0.118 -0.312 0.005 0.107 -0.111 -0.053 -0.083 0.055 0.229 -0.053 0.127 0. 184 0.039 -0.339 1
Source: OECD.Stat database [23]—own calculations based on OECD database.
For CA only the first five most important principals were used, these five principal components
explain about 84% (see Figure 6) of the variation of the original dataset.
Sustainability 2018, 10, x FOR PEER REVIEW 13 of 23
Figure 6. Scree plot of principal component analysis—first period. Source: OECD.Stat database [23]—
own calculations based on OECD database.
In the first period of CA according to the cluster tree (Figure 7) the 35 OECD countries were split
into five relatively isolated clusters. The created clusters separated 15 most developed countries into
one larger cluster and another smaller cluster with eight countries. On the other hand, the less
developed OECD countries were classified into a cluster with four objects, the East-European (former
communist countries) were separated into a cluster with six objects. One cluster content only two
countries (Ireland and the United Kingdom), these countries are similar not only due to the analyzed
indicators but are also geographically very closely located.
Figure 7. Tree of cluster analysis (dendrogram)—first period. Source: OECD.Stat database [23]—own
calculations based on OECD database.
The main features of the clusters are presented by the cluster centroids in Table 6.
Cluster 1 (15 countries—Australia, Italy, New Zealand, Spain, Sweden, Japan, Belgium, the
Netherlands, Denmark, Austria, Germany, Greece, Slovenia, Israel, Portugal).
Countries in these clusters are “middle”, meaning they include some characteristics, like GDP
or expenditures on health per capita, standardized death rates (SDR) of malignant neoplasms, of
diseases of the NS, CS and RS. The real change in GDP per capita was the smallest between 2000 and
2008 (14.2%) while the expenditures on health per capita jumped more intensively in the same time
span (by 30.7%). Health expenditure as the share of GDP was quite high in these countries and high
was also the contribution of their governments on current health expenditures. The LE for women
and men was as high as in the countries in Cluster 3, but one very significant difference can be visible
Figure 6.
Scree plot of principal component analysis—first period. Source: OECD.Stat database [
23
]—own
calculations based on OECD database.

Sustainability 2018,10, 4554 13 of 22
In the first period of CA according to the cluster tree (Figure 7) the 35 OECD countries were
split into five relatively isolated clusters. The created clusters separated 15 most developed countries
into one larger cluster and another smaller cluster with eight countries. On the other hand, the less
developed OECD countries were classified into a cluster with four objects, the East-European (former
communist countries) were separated into a cluster with six objects. One cluster content only two
countries (Ireland and the United Kingdom), these countries are similar not only due to the analyzed
indicators but are also geographically very closely located.
Sustainability 2018, 10, x FOR PEER REVIEW 13 of 23
Figure 6. Scree plot of principal component analysis—first period. Source: OECD.Stat database [23]—
own calculations based on OECD database.
In the first period of CA according to the cluster tree (Figure 7) the 35 OECD countries were split
into five relatively isolated clusters. The created clusters separated 15 most developed countries into
one larger cluster and another smaller cluster with eight countries. On the other hand, the less
developed OECD countries were classified into a cluster with four objects, the East-European (former
communist countries) were separated into a cluster with six objects. One cluster content only two
countries (Ireland and the United Kingdom), these countries are similar not only due to the analyzed
indicators but are also geographically very closely located.
Figure 7. Tree of cluster analysis (dendrogram)—first period. Source: OECD.Stat database [23]—own
calculations based on OECD database.
The main features of the clusters are presented by the cluster centroids in Table 6.
Cluster 1 (15 countries—Australia, Italy, New Zealand, Spain, Sweden, Japan, Belgium, the
Netherlands, Denmark, Austria, Germany, Greece, Slovenia, Israel, Portugal).
Countries in these clusters are “middle”, meaning they include some characteristics, like GDP
or expenditures on health per capita, standardized death rates (SDR) of malignant neoplasms, of
diseases of the NS, CS and RS. The real change in GDP per capita was the smallest between 2000 and
2008 (14.2%) while the expenditures on health per capita jumped more intensively in the same time
span (by 30.7%). Health expenditure as the share of GDP was quite high in these countries and high
was also the contribution of their governments on current health expenditures. The LE for women
and men was as high as in the countries in Cluster 3, but one very significant difference can be visible
Figure 7. Tree of cluster analysis (dendrogram)—first period. Source: OECD.Stat database [23]—own
calculations based on OECD database.
The main features of the clusters are presented by the cluster centroids in Table 6.
Cluster 1 (15 countries—Australia, Italy, New Zealand, Spain, Sweden, Japan, Belgium,
the Netherlands, Denmark, Austria, Germany, Greece, Slovenia, Israel, Portugal).
Countries in these clusters are “middle”, meaning they include some characteristics, like GDP or
expenditures on health per capita, standardized death rates (SDR) of malignant neoplasms, of diseases
of the NS, CS and RS. The real change in GDP per capita was the smallest between 2000 and 2008
(14.2%) while the expenditures on health per capita jumped more intensively in the same time span
(by 30.7%). Health expenditure as the share of GDP was quite high in these countries and high was
also the contribution of their governments on current health expenditures. The LE for women and
men was as high as in the countries in Cluster 3, but one very significant difference can be visible
between Clusters 1 and 3. The very high life expectancies at birth were achieved with much lower
current expenditures on health per capita. The countries in this cluster can be positively rated due to
low levels of SDR of mental and behavioral disorders, but negatively evaluated through very high
SDR for diabetes mellitus.
Cluster 2 (two countries—Ireland, United Kingdom).
In this cluster only two countries were placed, both are geographically very closely located, and
both are island lying. These countries have a higher GDP per capita compared to Cluster 1 with a
higher increase in this indicator (16.3%). At the beginning of the analyzed period Ireland and GBR
were in the “middle” of clusters based on health expenditures per capita, which were combined
with a bit very high increase of these expenditures by nearly 64%. These strong increase of current
expenditure on health per capita compared with the moderate change in GDP per capita caused the

Sustainability 2018,10, 4554 14 of 22
highest gap between both real change indicators which stood at 47.3 p.p. Governments’ expenditure
on health as % of total health expenditures was the highest with a level of 78.4%. Positively can be
rated the countries in this cluster due to a very low SDR for diabetes mellitus, but the standardized
death rates for other illnesses were quite high. Cluster 2 had the second worst position based on
SDR for malignant neoplasms, diseases of the nervous system and of the circulatory system. But the
worst is their position from the point of view of SDR related to the diseases of the respiratory system.
Household out-of-pocket payments on health (11.8%) was the lowest one among all of the clusters.
Cluster 3 (eight countries—Canada, Iceland, France, Luxembourg, Norway, Switzerland, Finland,
United States).
These countries had the highest average level of GDP per capita and health expenditures per
capita. The health expenditures (HE) per capita increased more rapidly compared to the changes of
GDP/capita. Due to the very high current expenditure on health per capita the indicator of HE as
% of GDP (8.6%) was the highest among all of the clusters. Government expenditure on health as %
of the total HE was only 70.5% especially due to a very low rate in the USA (44.2%). Specific for the
countries in Cluster 3 was the highest LE for females and males. Positively can be rated the cluster for
lower levels of SDR for malignant neoplasms, diseases of the CS or RS. On the other hand, the very
high LE in these eight countries can be the main reason for the highest SDR for mental and behavioral
disorders and highest levels of SDR for diseases of the nervous system.
Custer 4 (four countries—Chile, South Korea, Turkey, Mexico).
The emerging and the least developed countries of the OECD were joined together into Cluster 4.
Typical for these countries are very low levels of GDP per capita combined with the lowest current
health expenses on health. The growth rate of health expenditures per capita reached an extremely
high level of 61.2% and so the gap between the real changes of indicators x
2
and x
4
was as high
as 33.6%. These countries had the lowest health expenditures share of GDP (4.9%), the lowest
government expenditure on health as % of health expenses (53%) and surprisingly the highest
household out-of-pocket payments on health (41.9%). The high household out-of-pocket payments
in less developed countries is a risky condition for the overall health of the population, a danger of
public health issues because the health expenses can be financed only by richer persons while the
poorer population can be at risk of health and health care. According to the SDR data for malignant
neoplasms and diseases of the circulatory system these countries achieved the lowest average levels
among all five clusters. Very negatively should be rated the highest SDR for diabetes mellitus that
averaged at 62 deaths per 100,000 population, with an extreme for Mexico at 121 deaths per 100,000.
Not only the SDR for diabetes were high but also for diseases of the RS was the second highest between
the analyzed clusters.
Cluster 5 (six countries—Czech Republic, Slovakia, Estonia, Poland, Hungary, Latvia).
The former communist countries created the last Cluster 5. These countries are similar not only
due to their former communist regimes, due to their common entry into the European Union in 2004,
due to their territory closeness, but also due to the chosen indicators. On average for the countries is
typical the lowest GDP per capita level and second lower HE per capita. On the other side they reached
the highest real GDP growth that averaged at 54.6% and the highest real change in the expenditures on
health (75.8%). The health expenditures as % of GDP stood at 5.6%, but due to a strong increase of HE
per capita the percentage can grow in the future. As a surprise of these East European countries is a
quite high share of household out-of-pocket payments on health expenditure, that was with 24.6% the
second largest. This cluster achieved the lowest levels of LE for women (77.1) and men (68.0). The low
LE can be a result of an extremely high SDR for diseases of the CS (more than 770 deaths per 100,000)
and the highest SDR for malignant neoplasms. Good results can be seen in the lowest SDR for mental
and behavioral disorders, diseases of the nervous system and RS. These countries should focus on
better national screening programs for early detection of initial cancer stages that could be treated with
more successful results.

Sustainability 2018,10, 4554 15 of 22
Table 6. Cluster centroids of OECD countries—first period.
Sustainability 2018, 10, x FOR PEER REVIEW 15 of 23
strong increase of HE per capita the percentage can grow in the future. As a surprise of these East
European countries is a quite high share of household out-of-pocket payments on health expenditure,
that was with 24.6% the second largest. This cluster achieved the lowest levels of LE for women (77.1)
and men (68.0). The low LE can be a result of an extremely high SDR for diseases of the CS (more
than 770 deaths per 100,000) and the highest SDR for malignant neoplasms. Good results can be seen
in the lowest SDR for mental and behavioral disorders, diseases of the nervous system and RS. These
countries should focus on better national screening programs for early detection of initial cancer
stages that could be treated with more successful results.
Table 6. Cluster centroids of OECD countries—first period
Source: OECD.Stat database [23]—own calculations based on OECD database.
3.3.2. Cluster Analysis—Second Period
For the second period of cluster analysis the variables used to create the homogenous clusters
have been taken from the most recent period according to the availability of the concrete indicator.
Variables x1, x3, x5, x6 represents the data from the end of the analyzed period, from the year 2016.
Unfortunately, not all of the characteristics were available for most actual period, so for example the
life expectancy for females (x8) and also males (x9) cover the year 2015, variables of mortality x10–x15
are calculated for 2013 and the real indices of GDP per capita and health expenditures per capita
(indicators x2 and x4) represent the real change in both indicators between 2009 and 2016. For cluster
analysis six first most important principal components were used which explained more than 85%
(see Figure 8) of the variation of the original dataset. The PCA was used due to a strong association
between some pairs of the selected variables, the correlation coefficients can be found in Table 7.
In the second period of CA the 35 OECD countries were split into six relatively isolated clusters
(see the cluster tree—Figure 9). The created clusters separated 17 most developed countries into one
large cluster. One cluster consists of Switzerland and the United Stated, the countries with a very
high percentage of HE to GDP. Cluster three is created by countries with a very low real change in
GDP. Clusters 4 and 6 consists of the less developed countries of the OECD. In the second period of
CA also one cluster with only one object was created, Mexico was the only country that is not similar
to any OECD countries according to used indicators for the analysis.
Clusters of OCED countries cluster 1 cluste r 2 clust er 3 clus ter 4 clus ter 5
Number of countries 15 2846
GDP pe r capita
(PPP, curr. p.) in 2000
25 359 28 098 34 454 11 752 11 227
Real growth rate, GDP per capita,
(2000 - 2008)
1.142 1.163 1.146 1. 275 1.546
HExpend. per capita
(PPP, curr . P.) in 2000
1 976 1 673 2 944 558 636
Real growth rate, HExpend. per
capita, (2000 - 2008)
1.307 1.636 1.273 1. 612 1.758
HExpend. (% of GDP) in 2000 7.8 6.0 8.6 4. 9 5.6
HExpend. (government
compulsory of total, %) in 2000
73.6 78.4 70.5 53. 0 74.2
HExpend., Out of pocket
pam ynets (% of total ) in 2000
19.9 11.8 18.4 41. 9 24.6
LE_females (2000) 81.4 79.8 81.5 77. 2 77.1
LE_males (2000) 75.6 74.8 75.7 71. 4 68.0
SDR _MN (2000) 231.8 257.7 226.0 169. 9 278.4
SDR_DM ( 2000) 23.6 13.1 17.7 62. 1 15.9
SDR_MBD ( 2000) 17.8 21.4 31.5 23.0 4.3
SDR_NS (2000) 19.2 22.7 32.8 16. 1 13.8
SDR_CS (2000) 384.7 435.1 353.1 259. 6 772.6
SDR_RS (2000) 82. 9 167.0 78.2 100.1 58.0
Source: OECD.Stat database [23]—own calculations based on OECD database.
3.3.2. Cluster Analysis—Second Period
For the second period of cluster analysis the variables used to create the homogenous clusters
have been taken from the most recent period according to the availability of the concrete indicator.
Variables x
1
, x
3
, x
5
, x
6
represents the data from the end of the analyzed period, from the year 2016.
Unfortunately, not all of the characteristics were available for most actual period, so for example the
life expectancy for females (x8) and also males (x9) cover the year 2015, variables of mortality x10–x15
are calculated for 2013 and the real indices of GDP per capita and health expenditures per capita
(indicators x2and x4) represent the real change in both indicators between 2009 and 2016. For cluster
analysis six first most important principal components were used which explained more than 85%
(see Figure 8) of the variation of the original dataset. The PCA was used due to a strong association
between some pairs of the selected variables, the correlation coefficients can be found in Table 7.
Sustainability 2018, 10, x FOR PEER REVIEW 16 of 23
Table 7. Pearson’s correlation coefficients (second period).
Source: OECD.Stat database [23]—own calculations based on OECD database.
Figure 8. Scree plot of principal component analysis—second period. Source: OECD.Stat database
[23]—own calculations based on OECD database.
Figure 9. Tree of Cluster analysis (dendrogram)—second period. Source: OECD.Stat database [23]—
own calculations based on OECD database.
The main features of the clusters are presented by the cluster centroids in Table 8.
Variables
x
1
x
2
x
3
x
4
x
5
x
6
x
7
x
8
x
9
x
10
x
11
x
12
x
13
x
14
x
15
x
11
x
2-0.072 1
x
30.829 -0.259 1
x
4-0.125 0.457 -0.019 1
x
50.353 -0.451 0.797 0.066 1
x
60.327 -0.040 0.173 -0.095 0.012 1
x
7-0.528 0.050 -0.555 0.131 -0.435 -0.723 1
x
80.432 -0.417 0.414 -0.042 0.380 0. 315 -0.370 1
x
90.558 -0.428 0.565 -0.103 0.480 0. 361 -0.478 0.864 1
x
10 -0.087 0.113 -0.145 -0.077 -0. 101 0.247 -0. 176 -0.238 -0.322 1
x
11 -0.339 0.163 -0.341 0.004 -0.313 -0.429 0.448 -0.541 -0. 378 -0.519 1
x
12 0.586 -0.191 0.728 0.084 0.622 0.206 -0. 418 0.187 0.455 0. 079 -0.292 1
x
13 0.310 -0.067 0.339 -0.044 0.270 0. 162 -0.292 0. 249 0.379 -0.258 - 0.228 0.391 1
x
14 -0.405 0.389 -0.486 0.011 -0.485 -0.063 0.296 -0.698 -0.855 0.575 0.012 -0.406 -0.312 1
x
15 -0.083 0.199 -0.040 0.006 0.019 - 0.059 -0.028 -0.188 0.032 -0.108 0.272 0. 084 -0.149 -0.289 1
Figure 8.
Scree plot of principal component analysis—second period. Source: OECD.Stat
database [23]—own calculations based on OECD database.
In the second period of CA the 35 OECD countries were split into six relatively isolated clusters
(see the cluster tree—Figure 9). The created clusters separated 17 most developed countries into one

Sustainability 2018,10, 4554 16 of 22
large cluster. One cluster consists of Switzerland and the United Stated, the countries with a very high
percentage of HE to GDP. Cluster three is created by countries with a very low real change in GDP.
Clusters 4 and 6 consists of the less developed countries of the OECD. In the second period of CA also
one cluster with only one object was created, Mexico was the only country that is not similar to any
OECD countries according to used indicators for the analysis.
Table 7. Pearson’s correlation coefficients (second period).
Sustainability 2018, 10, x FOR PEER REVIEW 16 of 23
Table 7. Pearson’s correlation coefficients (second period).
Source: OECD.Stat database [23]—own calculations based on OECD database.
Figure 8. Scree plot of principal component analysis—second period. Source: OECD.Stat database
[23]—own calculations based on OECD database.
Figure 9. Tree of Cluster analysis (dendrogram)—second period. Source: OECD.Stat database [23]—
own calculations based on OECD database.
The main features of the clusters are presented by the cluster centroids in Table 8.
Variables
x
1
x
2
x
3
x
4
x
5
x
6
x
7
x
8
x
9
x
10
x
11
x
12
x
13
x
14
x
15
x
11
x
2-0.072 1
x
30.829 -0.259 1
x
4-0.125 0.457 -0.019 1
x
50.353 -0.451 0.797 0.066 1
x
60.327 -0.040 0.173 -0.095 0.012 1
x
7-0.528 0.050 -0.555 0.131 -0.435 -0.723 1
x
80.432 -0.417 0.414 -0.042 0.380 0. 315 -0.370 1
x
90.558 -0.428 0.565 -0.103 0.480 0. 361 -0.478 0.864 1
x
10 -0.087 0.113 -0.145 -0.077 -0. 101 0.247 -0. 176 -0.238 -0.322 1
x
11 -0.339 0.163 -0.341 0.004 -0.313 -0.429 0.448 -0.541 -0. 378 -0.519 1
x
12 0.586 -0.191 0.728 0.084 0.622 0.206 -0. 418 0.187 0.455 0. 079 -0.292 1
x
13 0.310 -0.067 0.339 -0.044 0.270 0. 162 -0.292 0. 249 0.379 -0.258 - 0.228 0.391 1
x
14 -0.405 0.389 -0.486 0.011 -0.485 -0.063 0.296 -0.698 -0.855 0.575 0.012 -0.406 -0.312 1
x
15 -0.083 0.199 -0.040 0.006 0.019 - 0.059 -0.028 -0.188 0.032 -0.108 0.272 0. 084 -0.149 -0.289 1
Source: OECD.Stat database [23]—own calculations based on OECD database.
Sustainability 2018, 10, x FOR PEER REVIEW 16 of 23
Table 7. Pearson’s correlation coefficients (second period).
Source: OECD.Stat database [23]—own calculations based on OECD database.
Figure 8. Scree plot of principal component analysis—second period. Source: OECD.Stat database
[23]—own calculations based on OECD database.
Figure 9. Tree of Cluster analysis (dendrogram)—second period. Source: OECD.Stat database [23]—
own calculations based on OECD database.
The main features of the clusters are presented by the cluster centroids in Table 8.
Variables
x
1
x
2
x
3
x
4
x
5
x
6
x
7
x
8
x
9
x
10
x
11
x
12
x
13
x
14
x
15
x
11
x
2-0.072 1
x
30.829 -0.259 1
x
4-0.125 0. 457 -0.019 1
x
50.353 -0.451 0.797 0.066 1
x
60.327 -0.040 0.173 -0.095 0. 012 1
x
7-0.528 0. 050 -0.555 0.131 -0. 435 -0.723 1
x
80.432 -0.417 0.414 -0.042 0. 380 0.315 -0.370 1
x
90.558 -0.428 0.565 -0.103 0. 480 0.361 -0.478 0.864 1
x
10 -0.087 0. 113 -0.145 -0.077 -0.101 0.247 -0.176 -0.238 -0. 322 1
x
11 -0.339 0. 163 -0.341 0.004 -0. 313 -0.429 0.448 -0. 541 -0.378 -0.519 1
x
12 0.586 -0.191 0.728 0.084 0.622 0.206 -0.418 0.187 0.455 0.079 -0.292 1
x
13 0.310 -0.067 0.339 -0.044 0. 270 0.162 -0.292 0.249 0.379 - 0.258 -0.228 0.391 1
x
14 -0.405 0. 389 -0.486 0.011 -0. 485 -0.063 0.296 -0. 698 -0.855 0.575 0.012 -0.406 -0.312 1
x
15 -0.083 0. 199 -0.040 0.006 0.019 -0.059 -0.028 - 0.188 0.032 -0.108 0.272 0. 084 -0.149 -0.289 1
Figure 9.
Tree of Cluster analysis (dendrogram)—second period. Source: OECD.Stat database [
23
]—own
calculations based on OECD database.
The main features of the clusters are presented by the cluster centroids in Table 8.
Cluster 1 (17 countries—Australia, Austria, France, Sweden, Belgium, Canada, Germany,
New Zealand
, the Netherland, Norway, Japan, Denmark, United Kingdom, Ireland Finland,
Iceland Luxembourg).
The first cluster joined together countries with a very high living standard. The average GDP
per capita and HE per capita are the second highest between the clusters with a real growth between
2009 and 2016 by about 10% in the case of GDP per capita and by 12% for HE per capita. In these
countries the government expenditures on health reached a maximal level of 78.9% while the household
out-of-pocked was minimal (14.6%). The high living standard has been manifested in a very high LE
for women (84.0) and men (79.7). Very positively should be rated the lowest SDR for diabetes mellitus,

Sustainability 2018,10, 4554 17 of 22
and diseases of the CS. The high LE can be the reason for higher SDR for diseases of the nervous
system and SDR for MBD.
Cluster 2 (two countries—Switzerland, United States).
Cluster 2 contains only two countries. For both countries are typically extremely high levels of
GDP per capita ($60,112) and expenditure on health per capita ($8906). The GDP per capita increased
in the time span from 2009 till 2016 in real terms only moderately by 6.9%, but the HE per capita
increased by 20% which was the second highest growth among the clusters. The difference between
both real changes ended with the highest value of 13.1 p.p. Health expenditure as % of GDP stood
at 14.8%, this was the maximal ratio of all clusters. The LE at birth for females and males was in the
“middle” of the clusters, not the best, but not the worst. According to the health expenditures per
capita it was expected that these countries will achieve better results in the case of LE at birth for both
sexes and also in the case of the standardized death rates. Between all clusters, Cluster 2 had the
highest levels of SDR for MBD and SDR for the NS, third worst SDR for malignant neoplasms. On the
other side positively can be rated the low death rates for diseases for RS, CS and diabetes mellitus.
Cluster 3 (six countries—Greece, Israel, Italy, Spain, Portugal, Slovenia).
In this cluster interestingly were joined together the most problematic countries of the EU,
the so-called PIGS countries: Portugal, Italy, Greece, Spain with Slovenia and Israel. The GDP per
capita and HE per capita of this cluster are in the “middle” of the clusters, but both characteristics
between 2008 and 2016 did not change very positively. The real GDP per capita on average for this
cluster stayed unchanged and the HE per capita in real terms declined by about 3%. The average
decline for this cluster was very negatively influenced by strong declines especially in Greece. The LE
at birth for females was the highest one (84.5) while for men the second highest (79.2). The high LE
was reached also due to relatively lower SDR for diseases for the RS, NS, or CS. A bit worse were the
SDR for malignant neoplasm.
Cluster 4 (three countries—Chile, South Korea, Turkey).
The emerging countries of the OECD merged together into Cluster 4. Only Mexico as another
emerging country created a separate, solo cluster. These countries had a lower GDP per capita and,
HE per capita than the former communist countries joined together in Cluster 6. Positively should be
rated the highest average real increase of GDP per capita by almost 30% and real growth of HE per
capita by 37.3%. The LE at birth was higher in this cluster compared with Clusters 5 and 6. Low were
the SDR for malignant neoplasms and SDR for MBD. On the other hand, higher were the death rates
for DM, in the case of other selected SDR the result of Cluster 4 can be rated as “middle”.
Custer 5 (one country—Mexico).
In both periods of cluster analysis only one single—solo cluster was created. It means that
according to the analyzed indicators Mexico is not similar to any other OECD country at all. Mexico
can be considered as an “extreme country”. The main reason for the separation are some extremely
low or extremely high levels of selected indicators. For example, the GDP per capita ($18,583) and
the HE per capita ($1080) were at extremely low levels. The real changes of both characteristics are
moderate, so the assumption is not very positive for a greater growth in the nearest future. The health
expenditures as a percent of GDP were minimal and as low as 5.8%. The lowest was the government
expenditure on health (53%) but the household out-of-pocket payments on health were extremely
high (41.9%). So, high household out-of-pocket payments are dangerous in an emerging country like
Mexico. The health can be a luxury product-service for a lot of ordinary people. In Mexico, the SDR
for DM, and diseases of the RS and CS, were the highest among the clusters. The lowest SDR were
achieved in the case of SDR for MN, MBD, or diseases of the NS.
Cluster 6 (six countries—Czech Republic, Slovakia, Estonia, Poland, Hungary, Latvia).
The former communist countries joined again together into one separate cluster. These six countries
created together a cluster also in the first period of the analysis, fortunately with much better results
in the second period. For example, their average GDP per capita in the first period ($11,227) was the
lowest one, it jumped to $29,202 in the second period, with a much better position between the clusters.

Sustainability 2018,10, 4554 18 of 22
The HE per capita reached $2003. In these group of countries, the increase of the GDP per capita was
higher than the increase of HE per capita. The LE at birth was the second lowest for females and also for
males. The lower LE at birth is associated with the extremely high SDR for diseases of the circulatory
system and malignant neoplasms. Positively should be rated the lowest SDR for diseases of the RS.
Table 8. Cluster centroids of OECD countries—second period.
Sustainability 2018, 10, x FOR PEER REVIEW 18 of 23
for DM, and diseases of the RS and CS, were the highest among the clusters. The lowest SDR were
achieved in the case of SDR for MN, MBD, or diseases of the NS.
Table 8. Cluster centroids of OECD countries—second period.
Source: OECD.Stat database [23]—own calculations based on OECD database.
Cluster 6 (six countries—Czech Republic, Slovakia, Estonia, Poland, Hungary, Latvia).
The former communist countries joined again together into one separate cluster. These six
countries created together a cluster also in the first period of the analysis, fortunately with much
better results in the second period. For example, their average GDP per capita in the first period
($11,227) was the lowest one, it jumped to $29,202 in the second period, with a much better position
between the clusters. The HE per capita reached $2003. In these group of countries, the increase of
the GDP per capita was higher than the increase of HE per capita. The LE at birth was the second
lowest for females and also for males. The lower LE at birth is associated with the extremely high
SDR for diseases of the circulatory system and malignant neoplasms. Positively should be rated the
lowest SDR for diseases of the RS.
4. Discussion
The GDP per capita in PPP current prices jumped from an average level of USD 23,616.5 to USD
42,429.2. This variable belongs to the indicators of SD Goal 10 and the increase of the GDP per capita
in OECD can be rated positively. In 2016 Mexico had the lowest while Luxembourg the highest living
standard between the OECD Member States. The convergence of OECD countries from the GDP per
capita point of view developed positively, the coefficient of variation decreased from a level of 43%
in 2000 to 38.8% in 2016. It is a good signal of a convergence process of the GDP per capita of the
OECD population. The average amount spent on health per capita increased from $1788 in 2000 to
$3997 in 2016. The relative variability of HE per capita is higher compared to the CV of GDP/capita.
In 2000 the CV of current expenditures on health per capita was as high as 55% and declined to 50.7%
in 2016. High expenditure on health per capita was typical for countries which faced also a very high
level of GDP per capita. The highest HE per capita in 2016 was reached in the United States ($9892).
The assumption that between GDP and HE per capita exists a positive association was fulfilled. In
2016 the Pearson’s correlation coefficient between both variables was very high and positive (rxy =
Clusters of OCED countries cluster 1 clust er 2 clus ter 3 clus ter 4 clu ster 5 clu ster 6
Number of countries 17 26316
GDP pe r capita
(PPP, curr. p.) in 2016
52 025 60 112 33 757 28 011 18 583 29 202
Real growth rate, GDP per capita,
(2009 - 2016)
1.099 1.069 0.998 1. 294 1.143 1.210
HExpend. per capita
(PPP, curr . P.) in 2016
5 060 8 906 2 862 1 931 1 080 2 003
Real growth rate, Hexpend. per
capita, (2009 - 2016)
1.120 1.200 0.969 1. 373 1.080 1.171
HExpend. (% of GDP) in 2016 9.8 14.8 8.5 6.8 5.8 6. 8
HExpend. (government
compulsory of total, %) in 2016
78.9 56.4 67.4 65. 5 51.7 72.1
HExpend., Out of pocket
pam ynets (% of total ) in 2015
14.6 19.7 24.5 28. 7 41.4 24.8
LE_females (2015) 84.0 83.2 84.5 82.5 77.7 80.7
LE_males (2015) 79.7 78. 6 79.2 76.9 72.3 72.9
SDR _MN (2013) 202.9 182.7 203.2 180.4 113.1 247.3
SDR_DM ( 2013) 15.3 18. 9 22.5 34.1 143.0 20.1
SDR_MBD ( 2013) 35.1 49.1 15. 3 12.3 6.6 14. 0
SDR_NS (2013) 38.8 41. 1 23.7 34.0 14.3 17.8
SDR_CS (2013) 230.1 234. 2 244.4 264.7 269.7 545.0
SDR_RS (2013) 66.1 63.0 64.8 85. 5 93.0 53.0
Source: OECD.Stat database [23]—own calculations based on OECD database.
4. Discussion
The GDP per capita in PPP current prices jumped from an average level of USD 23,616.5 to USD
42,429.2. This variable belongs to the indicators of SD Goal 10 and the increase of the GDP per capita
in OECD can be rated positively. In 2016 Mexico had the lowest while Luxembourg the highest living
standard between the OECD Member States. The convergence of OECD countries from the GDP per
capita point of view developed positively, the coefficient of variation decreased from a level of 43% in
2000 to 38.8% in 2016. It is a good signal of a convergence process of the GDP per capita of the OECD
population. The average amount spent on health per capita increased from $1788 in 2000 to $3997
in 2016
. The relative variability of HE per capita is higher compared to the CV of GDP/capita. In 2000
the CV of current expenditures on health per capita was as high as 55% and declined to 50.7% in 2016.
High expenditure
on health per capita was typical for countries which faced also a very high level of GDP
per capita. The highest HE per capita in 2016 was reached in the United States ($9892).
The assumption
that between GDP and HE per capita exists a positive association was fulfilled. In 2016 the Pearson’s
correlation coefficient between both variables was very high and positive (
rxy = 0.83
) and it means that in
countries with high GDP per capita we can expect a high level of HE per capita and vice versa.
One of the assumptions was not fulfilled. The assumption was that the GDP per capita, as a
very rough estimate of the of overall country’s productivity, will grow more strongly than the HE
per capita.
The expectation
that the productivity generates the growth of HE has not been confirmed.
The real
change GDP per capita belongs to the indicators of SD Goal 8. Almost in all OECD countries
the increase of real GDP per capita was lower than the real change in HE per capita between 2000
and 2016
. Most notable was the difference between both characteristics in Korea where the real GDP
per capita increased in the selected time span by 69.7%, but the real HE per capita grow by 222.9%.
In only two countries the real GDP per capita growth trespassed the real health expenses per capita

Sustainability 2018,10, 4554 19 of 22
changes (Iceland, Turkey). Even in countries most affected by the economic crisis the HE per capita
grows moderately. For example, in Greece where the real GDP per capita between 2000 and 2016
declined by 2.7% the HE increased by 9.7% and in Italy the GDP per capita dropped by 5.3%, but the
real HE per capita changed oppositely and grow by 11.6%.
The strong growth of HE should positively affect the health outcome of the population in OECD
countries. That is why selected indicators of public health were chosen for analysis, these indicators
belong to the SD Goal 3. According to the expectation the LE at birth or at age 65 should increase and
on the other hand the standardized death rates of some diseases should decline. LE at birth increased
progressively in the OECD countries. The average LE for females was 73 years in 1970 and jumped
to 83.2 in 2015 while for males the average value increased from 66.7 years to 77.9 years respectively.
The LE for males was lower by 6.3 years compared to women’s life expectancies in 1970 but the gap
is closing. A very positive development was discovered for the LE at age 65. While the average LE
at birth for women increased by about 14% and for man by 16.8%, the LE at age 65 increased more
rapidly. For women the growth reached 35% and for men the increase achieved 40%.
The increasing life expectancy, especially the LE at higher ages belongs to the main reasons for
increases of the HE per capita. The aging of the population is nowadays an issue that is often discussed
from the perspective of the economic aspects of aging
For the multivariate analysis the PCA and cluster analysis were used. In the case of a strong linear
association between the indicators the most important principal components were used instead of the
original strongly correlated dataset. The OECD countries were grouped into clusters based on their
similarity. For classification 15 indicators were used: Some indicators of the economic situation in the
countries, indicators of health expenditures, life expectancy at birth and standardized death rates of
noncommunicable diseases. The selected variables are part of the SD goals 3, 8 and 10.
In the first period of CA the 35 OECD countries were split into five clusters. The created clusters
separated the most developed countries into two clusters, one cluster with 15 objects and one with
eight objects. On the other hand, the less developed OECD countries were classified into a cluster with
four objects, the former communist countries were separated into a cluster with six objects. One cluster
content only two countries (Ireland and the United Kingdom). The 15 very developed OECD countries
in Cluster 1 can be rated as “middle” regarding some characteristics, like GDP or expenditures on
health per capita. The LE for women and men was as high as in the countries in Cluster 3, but one
very significant difference can be visible between Clusters 1 and 3. The very high life expectancies at
birth in Cluster 1 were achieved with much lower current expenditures on health per capita. In the
second cluster, only two countries were placed, both are geographically very closely located, and both
are island lying (Ireland, United Kingdom). Their governments’ expenditure on health as % of total
health expenditures was the highest while household out-of-pocket payments on health (11.8%) was
the lowest one among all clusters. Cluster 2 had the worst position for SDR related to the diseases of
the respiratory system. In the third cluster, the LE for females and males was the highest and this can
be the reason for the highest SDR for mental and behavioral disorders and highest levels of SDR for
diseases of the NS. In Cluster 4, the emerging and the least developed countries of the OECD were
joined together. Typical for these countries are very low levels of GDP per capita combined with the
lowest HE per capita. These countries had the lowest government expenditure on health as % of health
expenses (53%) and surprisingly the highest household out-of-pocket payments on health (41.9%).
The high
household out-of-pocket payments in less developed countries is a risky condition for the
overall health of the population. The health expenses can be financed only by richer persons while the
poorer population can be at risk of health and health care. Very negatively should be rated the highest
SDR for diabetes mellitus (especially in Mexico). In the last cluster the former communist countries
joined together. These countries are similar not only due to their former communist regimes, due to
their territory closeness, but also due to the chosen indicators. These countries had in the first period
the lowest GDP per capita level and second lower HE per capita. The low levels were combined with
the highest real GDP growth that averaged at 54.6% and the highest real change in the HE per capita

Sustainability 2018,10, 4554 20 of 22
(75.8%). Former communist countries had the lowest levels of LE for women (77.1) and men (68.0).
The low LE can be a result of an extremely high SDR for diseases of the CS (more than 770 deaths
per 100,000) and the highest SDR for malignant neoplasms. These countries should focus on better
national screening programs for early detection of initial cancer stages that could be treated with a
more successful result.
In the second period of CA the OECD countries were split into six relatively isolated clusters.
Seventeen most developed countries created one large cluster—Cluster 1. These countries with a
very high living standard had the highest government expenditures on health (78.9. The high living
standard has been manifested in a very high LE for women (84.0) and men (79.7) and the lowest
SDR for diabetes mellitus, and diseases of the CS. The second cluster consists of Switzerland and the
United Stated. For both countries are typically extremely high GDP per capita ($60,112) and HE per
capita ($8906). The GDP per capita increased in the time span from 2009 till 2016 in real terms only
moderately by 6.9%, but the HE per capita increased by 20%. HE as % of GDP stood at 14.8%, this was
the maximal ratio of all clusters. According to the extremely high HE per capita it was expected that
these countries will achieve better results in the case of LE and SDR for noncommunicable diseases.
The PIGS countries (Portugal, Italy, Greece, Spain) together with Slovenia and Israel joined together
into Cluster 3. The GDP per capita and HE per capita of this cluster are in the “middle” of the clusters,
but the growth rates of both characteristics between 2008 and 2016 were not very positive. The real
GDP per capita on average for this cluster stayed unchanged and the HE per capita in real terms
declined by about 3% mainly as a result of a strong decline in Greece. The LE at birth for females
was the highest one (84.5) while for men the second highest (79.2). Clusters 4 and 6 consists of the
less developed countries of the OECD. The emerging countries of the OECD merged together into
Cluster 4 (except Mexico that created solo cluster). These countries had a lower GDP per capita and,
HE per capita than the former communist countries joined together in Cluster 6. Positively should be
rated the highest average real increase of GDP per capita by almost 30% and real growth of HE per
capita by 37.3%. Cluster 5 is a solo cluster that includes only one object—Mexico. Mexico is, according
to the used data set, not similar to any other OECD country, therefore Mexico can be considered as an
“extreme country”. The main reason for the separation are extremely low or extremely high levels of
selected variables. The GDP per capita ($18,583) and the HE per capita ($1080) were
extremely low
.
The HE as a percent of GDP were minimal and as low as 5.8%, lowest were also the government
expenditures on health (53%), but the household out-of-pocket payments on health were extremely
high (41.9%). This very high household out-of-pocket payments are dangerous in an emerging country,
like Mexico. Health care can be a luxury product-service for a lot of ordinary people. This may be this
is the reason for the highest SDR for DM, and diseases of the RS. The former communist countries
joined again together into one cluster, Cluster 6. The same six countries created a cluster that was also
in the first period of the CA. In the second period they reached much better results compared with
the first period. Their average GDP per capita in the first period ($11,227) was the lowest one and it
jumped to $29,202 in the second period. In these countries the increase of the GDP per capita was
higher than the increase of HE per capita. The LE at birth was the second lowest for females and males.
The lower LE at birth is associated with the extremely high SDR for diseases of the CS and MN.
The analyses of status and development of indicators that are part of the SDGs are very important
for achieving better results for sustainable growth. These analyses could help governments find
weaknesses, react, and find the best solutions to reach the goals of Agenda 2030.
Author Contributions:
S.M. and V.L. contributed equally to the development of the present paper, all the phases
have been discussed and worked by S.M. and V.L.
Funding: The research was financed by the VEGA 1/0376/17 project.
Acknowledgments:
This paper was supported by the Slovak Scientific Grant Agency as part of the research
project VEGA 1/0376/17.
Conflicts of Interest: The authors declare no conflict of interest.

Sustainability 2018,10, 4554 21 of 22
References
1.
United Nations. General Assembly. Resolution Adopted by the General Assembly on 25 September 2015: 70/1.
Transforming Our World: The 2030 Agenda for Sustainable Development; United Nations: New York, NY,
USA, 2015.
2.
The Sustainable Development Agenda—United Nations Sustainable Development. Available online: https:
//www.un.org/sustainabledevelopment/development-agenda/ (accessed on 7 May 2018).
3.
Earth Summit: Press Summary of Agenda 21; Department of Public Information, United Nations: New York,
NY, USA, 1992.
4.
United Nations. General Assembly. Resolution Adopted by the General Assembly: 55/2. United Nations Millennium
Declaration; United Nations: New York, NY, USA, 2000.
5.
Report of the World Summit on Sustainable Development: Johannesburg, South Africa, 26 August–4 September 2002;
United Nations: New York, NY, USA, 2002.
6.
United Nations. General Assembly. Resolution Adopted by the General Assembly on 27 July 2012: 66/288.
The Future We Want; United Nations: New York, NY, USA, 2012.
7.
Bebbington, J.; Unerman, J. Achieving the United Nations Sustainable Development Goals. Account. Audit.
Account. J. 2018,31, 2–24. [CrossRef]
8.
About the Sustainable Development Goals—United Nations Sustainable Development. Available online:
https://www.un.org/sustainabledevelopment/sustainable-development-goals/ (accessed on 2 May 2018).
9.
Sustainable Development in the European Union: A statistical Glance from the Viewpoint of the UN Sustainable
Development Goals; Eurostat: Luxembourg, 2016.
10.
OECD. Better Policies for Better Lives. Measuring Distance to the SDG Targets. An Assessment of Where OECD
Countries Stand; OECD: Paris, France, 2017.
11.
Better Policies for Sustainable Development 2016: A New Framework for Policy Coherence; OECD Publishing: Paris,
France, 2016.
12.
European Commission. Eurostat. EU SDG Indicator Set. Indicator for Monitoring the Sustainable Development
Goals (SDGs) in an EU Context; European Commission: Brussels, Belgium, 2017.
13.
European Commission. Eurostat. EU SDG Indicator Set 2018. Result of the Review in Preparation of the 2018
Edition of the EU SDG Monitoring Report; European Commission: Brussels, Belgium, 2018.
14.
Sustainable Development in the European Union: Monitoring Report on Progress towards the SDGs in an EU Context;
Publications Office of the European Union: Luxembourg, 2017.
15.
OECD. Better Policies for 2030. An OECD Action Plan on Sustainable Development Goals; OECD Publishing:
Paris, France, 2016.
16.
Development, OECD and the Sustainable Development Goals: Delivering on Universal Goals and Targets.
Available online: http://www.oecd.org/development/sustainable-development-goals.htm (accessed on
16 November 2018).
17.
European Commission. Available online: https://ec.europa.eu/commission/index_en (accessed on
17 November 2018).
18.
EU Approach to Sustainable Development. Available online: https://ec.europa.eu/info/strategy/
international-strategies/global-topics/sustainable-development-goals/eu-approach-sustainable-
development_en#documents (accessed on 17 November 2018).
19.
European Commission. Communication from the Commission to The European Parliament, The Council,
The European Economic and Social Committee and The Committee of The Regions. Next Steps for a Sustainable
European Future—European Action for Sustainability; COM/2016/0739; European Commission: Brussels,
Belgium, 2016.
20.
Unerman, J.; Bebbington, J.; O’Dwyer, B. Corporate reporting and accounting for externalities.
Account. Bus. Res. 2018,48, 497–522. [CrossRef]
21.
Palmi, P.; Morrone, D.; Miglietta, P.P.; Fusco, G. How Did Organizational Resilience Work Before and after
the Financial Crisis? An Empirical Study. Int. J. Bus. Manag. 2018,13, 54–62. [CrossRef]
22.
Venturelli, A.; Caputo, F.; Leopizzi, R.; Pizzi, S. The EU Directive on non-financial and diversity disclosure:
Constraint or opportunity. In Proceedings of the EURAM Conference, Glasgow, Scotland, 21–24 June 2017.
23. OECD. OECD Statistics. Available online: https://stats.oecd.org/ (accessed on 1 March 2018).
24. Loveday, R. Statistics; Cambridge University Press: Cambridge, UK, 2016.

Sustainability 2018,10, 4554 22 of 22
25. Witte, R.S.; Witte, J.S. Statistics; Wiley: Hoboken, NJ, USA, 2017.
26. Vu, K. A note on interpreting the beta-convergence effect. Econ. Lett. 2013,118, 46–49. [CrossRef]
27.
Das, R.C. Handbook of Research on Global Indicators of Economic and Political Convergence; Business Science
Reference: Hershey, PA, USA, 2016.
28.
Janssen, F.; Hende, A.V.D.; Beer, J.A.A.D.; Wissen, L.J.V. Sigma and beta convergence in regional mortality:
A case study of the Netherlands. Demogr. Res. 2016,35, 81–116. [CrossRef]
29.
Hofer, H.; Wörgötter, A. Regional Per Capita Income Convergence in Austria. Reg. Stud.
1997
,31, 1–12.
[CrossRef]
30.
Pauhofova, I.; Zelinsky, T. On the Regional Convergence of Income at District Level in Slovakia. Èkon. Cas.
2017,65, 918–934.
31.
Cluster Analysis. Available online: https://www.sciencedirect.com/topics/medicine-and-dentistry/cluster-
analysis (accessed on 7 June 2018).
32.
Conduct and Interpret a Cluster Analysis. Available online: http://www.statisticssolutions.com/cluster-
analysis-2/ (accessed on 7 June 2018).
33.
Statsoft. How to Group Objects into Similar Categories, Cluster Analysis. Available online: http://www.
statsoft.com/Textbook/Cluster-Analysis (accessed on 8 June 2018).
34. Milligan, G.W. Cluster Analysis; College of Business, Ohio State University: Columbus, OH, USA, 1995.
35. Uprichard, E.; Byrne, D.S. Cluster Analysis. Logic and Classics; SAGE: London, UK, 2012.
36.
Statistics Solutions. Conduct and Interpret a Factor Analysis. Available online: http://www.
statisticssolutions.com/factor-analysis-2/ (accessed on 8 June 2018).
37.
Dillon, W.R.; Goldstein, M. Multivariate Analysis: Methods and Applications; John Wiley & Sons: New York,
NY, USA, 1984.
38. Dunteman, G.H. Principal Component Analysis; Sage: Newbury Park, CA, USA, 1989.
39. Jollife, I.T. Principal Component Analysis; Springer: New York, NY, USA, 2010.
40.
Statistics Explained. Available online: http://ec.europa.eu/eurostat/statistics-explained/index.php/Main_
Page (accessed on 9 June 2018).
41.
Your Key to European Statistics. Available online: https://ec.europa.eu/eurostat/web/gdp-and-beyond
(accessed on 9 July 2018).
42.
Carnicky, S.; Megyesiova, S.; Conkova, M.; Zavadsky, C. Productivity development and convergence across
the EU Member States. Econ. Ann. XXI 2017,162, 13–17. [CrossRef]
43.
Monfort, P. Convergence of EU Regions Measures and Evolution; Regional Policy—European Commission:
Brussels, Belgium, 2008.
44.
Grau Grau, A.J.; Ramírez López, F. Determinants of Immigration in Europe. The Relevance of Life Expectancy
and Environmental Sustainability. Sustainability 2017,9, 1093. [CrossRef]
45.
Săndică, A.-M.; Dudian, M.; ¸Stefănescu, A. Air Pollution and Human Development in Europe: A New Index
Using Principal Component Analysis. Sustainability 2018,10, 312. [CrossRef]
46.
Death Rate Due to Chronic Diseases by Sex. Available online: http://ec.europa.eu/eurostat/cache/
metadata/en/sdg_03_40_esmsip2.htm (accessed on 29 June 2018).
47.
Megyesiova, S. The Relationship between the Health Status and Economic Situation in the EU Member
States. In SGEM 2014 Conference on Psychology and Psychiatry, Sociology and Healthcare, Education; STEF92
Technology Ltd.: Albena, Bulgaria, 2014.
48.
Loster, T. Determining the optimal number of clusters in cluster analysis. In Proceedings of the 10th
International Days of Statistics and Economics, Prague, Czech Republic, 8–10 September 2016; pp. 1078–1090.
49.
Loster, T. Evaluation of coefficients for determining the optimal number of clusters in cluster analysis on real
data sets. In Proceedings of the 9th International Days of Statistics and Economics, Prague, Czech Republic,
10–12 September 2015; pp. 1014–1023.
50.
Janotka, M.; Gazda, V.; Horvath, D. Migration Trends Among Regional clusters in Slovakia. Eng. Econ.
2014
,
24, 437–446. [CrossRef]
©
2018 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access
article distributed under the terms and conditions of the Creative Commons Attribution
(CC BY) license (http://creativecommons.org/licenses/by/4.0/).