Causes of the “brain drain” Problem in Selected
Western Balkan Countries
ANA TOPALOVIĆ, DAVID HAMPEL
Faculty of Business and Economics,
Department of Statistics and Operation Analysis,
Mendel University in Brno,
Zemědělská 1, 613 00 Brno,
CZECH REPUBLIC
Abstract: - This paper deals with the identification of the factors that influence the emigration of young and
highly educated people from Western Balkan countries. Indicators of the quality of economic, political, and
educational systems in Western Balkan countries and target countries were used for this purpose. A comparison
of Western Balkan countries with EU countries was provided via a cluster analysis. Cross-sectional and panel
data regression point to important indicators affecting emigration. An important finding was that for highly
educated people not only economic indicators but also political environment and educational system quality are
significant factors, which influence emigration.
Key-Words: - Brain drain, Cluster Analysis, Emigration, Regression Analysis, Western Balkans.
Received: April 27, 2023. Revised: September 28, 2023. Accepted: October 3, 2023. Published: October 13, 2023.
1 Introduction
The Western Balkans (WB) are characterised by a
long history of emigration. Political instability
followed by considerable poverty and
unemployment rate are one of the reasons for the
high interest in moving abroad. The transition from
a centralised to an open economy as well as wartime
events caused a wave of emigration from this area in
the 1990s, [1]. According to, [2], although the
countries of south-eastern Europe are characterised
by abundant emigration dating back several
decades, the change that took place at the end of the
last century is that the emigration of unskilled
workers has mostly turned into emigration of highly
skilled workers.
As a consequence of political instability and an
economically underdeveloped society, education in
WB countries was also unable to progress to the
level of the education systems of European
countries. The Bologna reforms and other education
reforms, which began in WB countries in the 2000s,
depending on the state, mostly only affected
legislative changes, while the implementation of
these reforms in practice was very inefficient (it
remained almost unchanged), [3]. Consequently,
ambitious students, who aim to raise their
knowledge and skills to a higher level, seek
“refuge” in more developed countries to have
opportunities for better education systems, training,
research, etc. These differences in education
systems called the schooling gap, depend on many
factors. The main ones are summarised in, [4], as
the level of development, religious fractionalization,
geographical distance, and the number of people
that have already emigrated and may be considered
as a strong pull factor.
In order for a country to economically develop,
it requires an educated and highly qualified
population. However, due to an inefficient labour
market, poor political system, expensive education,
etc., fewer young and educated people are interested
in participating in the economic development of
their country of origin and an increasing number of
young people meet their needs for a functional
system in other, more developed countries. The
paper, [2], states that the emigration of a highly
educated population is a serious threat to democracy
and elections due to the lack of educated residents in
the country. He also describes the brain drain as a
consequence of the absence of basic human rights,
such as the right to work or the right to an
education.
Moreover, there is a two-way causal link
between the lag in economic growth and emigration.
Due to poor living conditions (poverty, political
instability, low wages, etc.), people tend to leave a
country, which has a negative effect on
development, [5]. Ignoring this problem would only
lead to a vicious circle of these two matters. The
WSEAS TRANSACTIONS on BUSINESS and ECONOMICS
DOI: 10.37394/23207.2023.20.198
Ana Topalović, David Hampel
E-ISSN: 2224-2899
2300
Volume 20, 2023
work, [6], shows that for WB countries in particular,
the main source of GDP growth is external debt
increase.
According to, [1], modern theories of
endogenous growth, besides human capital, also
consider the effect of migration on development.
Therefore, emigration of the highly educated
population may be detrimental to development by
slowing it down. Overproduction of professional
manpower is therefore used for the development of
already developed countries and thereby slows
down the development of underdeveloped countries,
[7]. The less developed a country is, the more it is
affected by the brain drain.
When asked whether brain drain increases, in,
[8], we can read that although skilled migration in
absolute terms increases, it grows relatively at the
same rate in regard to overall education levels. In,
[5], emigration growth to population growth are
compared and the conclusion is given that the
emigration rate has not increased drastically in the
past few decades when population growth is taken
into account. However, if only high-skilled
migration is considered, it has grown at a much
faster pace and may be considered one of the major
aspects of globalisation.
In, [1], the authors found high unemployment as
a strong push factor and a high number of migrants
abroad as a strong pull factor. Both indicators
positively affect migration. According to, [8], less
populated, religiously fractalised, and politically
unstable countries with a low level of human capital
have a higher proportion of brain drain (the positive
relationship between emigration and political
instability and religious fractionalisation was also
presented in, [4], while government effectiveness
has not proven to be a significant variable). At a
micro level, factors that affect emigration are career
concerns and lifestyle and family reasons in addition
to higher income, [8]. Authors of, [9], came to the
conclusion that high wages in developed countries
attract migrants from less developed areas. The
smaller the wage gap, the lower the motivation to
emigrate. Another interesting fact determined in,
[4], is that the brain drain is higher in those
countries with a lower proportion of natives in the
educated population, which is why certain poor
regions of Africa and Asia are characterised by high
levels of brain drain.
Geographical distance has a negative impact on
emigration (although it should be noted that skilled
migrants are less sensitive to distance), while former
colonies, as well as bilingual states, are more open
to migration, [4]. Moreover, [10], came to the
conclusion that “Cultural similarities, colonial
legacies, and physical distance are often more
important determinants of educational selectivity
than wage incentives or selective immigration
policy”. Yet, the countries most affected by brain
drain are small island states, e.g., [4], [8].
Contrary to the brain drain stands a
phenomenon we call brain gain. One idea of how
the brain gain effect may be achieved is that not all
people who increase their human capital in order to
migrate will actually leave their country, [8]. The
final effect of this is an increase in the human
capital in the home country. This effect would be
even stronger if political barriers restricted
migration, which is not a common practice
nowadays. In, [11], states that remittances, the
creation of diasporas, and return migration are
appropriate compensation for skilled emigration.
Authors of, [12], determine remittances as a
significant factor positively affecting the GDP
growth rate in WB countries. In, [8], concluded that
remittances are large enough to cover the fiscal
costs of skilled emigrants. High-skilled migrants
that return to their home country are bringing with
them newly learned skills and experience (human
capital) in addition to the money acquired abroad,
[1]. In, [8], takes India as an example of a country
that in the past decade experienced the benefits of
brain gain through investments and expertise from
the Indian diaspora. Brain gain as opposed to brain
drain is reported by, [13], in the case of Nepalese
students. On the other hand, [7], states that
emigrants returning home have already completed
their most productive years and are therefore not
eligible to participate in development.
The main reason for emigration in almost all
WB countries is the high level of unemployment,
which is the highest among young residents. It
usually takes them several years to find a job, even
though the job here is not considered to be the job
they were educated for but any job. When it comes
to the inability of young people to find employment
in their profession, another problem, called brain
waste, appears given that this educated population
mostly ends up doing jobs that do not require their
level of knowledge and qualifications.
Unemployment among people aged 15 to 24 reaches
30% in most WB countries and in some countries, it
even exceeds this percentage (North Macedonia,
Bosnia and Herzegovina (BiH)). In Montenegro,
three-quarters of the young population said they
feared unemployment and a majority stated that
employment is generally not found thanks to
qualifications and knowledge but mostly thanks to
links to people in power, political party
membership, or personal relations, [14].
WSEAS TRANSACTIONS on BUSINESS and ECONOMICS
DOI: 10.37394/23207.2023.20.198
Ana Topalović, David Hampel
E-ISSN: 2224-2899
2301
Volume 20, 2023
According to, [15], in Serbia, there is a
widespread belief that human rights are not
respected. Furthermore, there is party employment,
an autocratic political system, and discrimination in
the labour market, especially in the case of women,
people over the age of 55, persons with disabilities,
and other vulnerable groups. Therefore, most people
emigrating abroad do not plan their return, at least
not before they retire.
Neither minimum nor average wages cover the
costs of a respectable lifestyle. Average earnings as
well as GDP per capita in the most popular
destination countries are several times higher than in
WB countries from which people decide to
emigrate. These are all push factors that cause
emigration.
According to, [16], the high rate of emigration
from Vojvodina (the northern province of Serbia) to
neighbouring Hungary is due to economic reasons,
and especially high unemployment. Although this
reduces the unemployment rate, it also has a major
negative impact on human capital in the home
country, as explained above. In general, destination
countries for migration are not only Western
European countries but also countries from central
and parts of Eastern Europe, where various specifics
emerge, see the case of Poland discussed in, [17].
Due to the emigration of young people who
have just finished schooling, the state loses most of
the money spent on their education. Depending on
the level of education achieved, the state invests in
the education of a young person for eight to over 20
years and when persons leave the state, this
investment, which becomes an expense, turns into
an investment in the state that receives these
educated people without taking any costs for their
education. Furthermore, the reduction in GDP may
be seen as another undesirable effect because people
who leave no longer contribute to the state budget,
both by consuming and paying income taxes, [14].
The long-term effect of education level on GDP is
empirically presented in, [18]. However, all WB
countries benefit the most from remittances relative
to their GDP (from around 6% in North Macedonia
to around 14% in BiH), including foreign pensions
and other personal and social transfers.
Unfortunately, this benefit cannot be used for
further development of the economy, given that in
all the countries a negligible part of these funds goes
to investments. Authors of, [19], studied the
effectiveness of migration policy in the context of
sustainable development in EU and non-EU
countries.
BiH experienced a large wave of emigration
during the war period from 1992 to 1995, although
the number of emigrants is still growing. In, [20],
made several recommendations for exploiting the
potential of the Bosnian diaspora to support brain
gain. Firstly, it would be desirable to alleviate
legislative barriers for those considering returning to
the country so that they are able to enter the labour
market unhindered. The diaspora should further be
included in all state decision-making bodies
concerning them, whose initiatives should be
supported by the government. Finally, the
authorities should take the necessary steps to attract
the diaspora.
As, [1], states, and as may be found in most
papers on a similar topic, the lack of data is a
serious problem in researching these phenomena in
WB countries. This fact may be considered
surprising given that these countries have a long
history of emigration and do not have enough data
on this topic as well as enough research, [21].
On the basis of the conducted literature review, it
can be concluded that the brain drain problem is
currently being researched and is considered a
serious problem in countries where there is an
outflow of young and educated people. However,
the research is conducted separately for each
country and economic sector, focusing only on
specific factors. In particular, there is a noticeable
lack of comprehensive results for the region of WB
countries, which are needed, among other reasons,
for the correct formulation of policies for the WB
countries as candidates for EU membership.
The aim of this paper is to identify factors that affect
the emigration of young people from certain
countries of the WB. We will consider persons of
the age from 15 to 24, for which the indicator “Not
in Education, Employment, or Training” is
provided. We will pay most attention to indicators
of the political and economic position of these
countries. Furthermore, using the appropriate
analyses, we will examine the relationships between
the proposed indicators, try to determine their
influence on migration rates in these countries, and
based on these indicators compare them to countries
to which people mostly emigrate, and finally
compare the WB countries with each other.
The paper is organised as follows: Section 2
introduces the indicators used as potential regressors
of emigration and the statistical methods used in the
paper. Section 3 provides a comparison of the WB
countries with the EU via selected indicators and the
results of the selected methods. Finally, Section 4
contains a discussion related to the results achieved
by other methods for different countries, and
Section 5 provides a conclusion.
WSEAS TRANSACTIONS on BUSINESS and ECONOMICS
DOI: 10.37394/23207.2023.20.198
Ana Topalović, David Hampel
E-ISSN: 2224-2899
2302
Volume 20, 2023
2 Material and Methods
The term “Western Balkans” generally refers to
Albania and the former Yugoslavia countries
excluding Slovenia. These countries are mostly
surrounded by EU Member States and are in the
initiated process of joining the EU, i.e., Albania,
BiH, Croatia, Kosovo, North Macedonia,
Montenegro, and Serbia. Since 2013, when Croatia
became a Member State, we could exclude it from
the WB, but considering the historical background
of this country, it is often still included in this group.
Although many studies research Kosovo as a
separate country, there is a lack of the data we need,
so Kosovo will not be included in the further
analysis. Given that the most recent reliable
migration data are from 2015, we will take other
indicators for this year as well.
To gain an insight into the economic position of
the WB countries in relation to the EU, several
economic indicators will be used. Gross Domestic
Product (GDP) per capita, measured in Purchasing
Power Standards (PPS), was taken from the
European Commission’s database Eurostat. GDP
is one of the primary indicators of a country’s
economic activity and is therefore an important
representative of the economic development gaps
between the examined countries. Expressing GDP in
PPS eliminates differences in price levels between
countries. The wage level is considered to be one of
the major motivations for migration. Two indicators
of wage level, statutory nominal gross monthly
minimum wage and mean nominal monthly earnings
of employees (both converted to U.S. dollars as the
common currency in order to make it easier to
compare countries), were taken from the
International Labour Organization’s database
(ILOSTAT). Another group of economic indicators,
measuring the underutilisation of the labour supply,
which is considered to be a strong push factor when
reaching high values, includes total unemployment,
the unemployment rate of a highly educated
population, and the share of youth not in employed,
education, or training (NEET). The unemployment
rate of the highly educated population and the share
of youth NEET is of particular importance when
taking into account the emigration of the young and
highly educated population. The values of the
unemployment rate and share of youth NEET were
also taken from the ILOSTAT database, while the
values of the unemployment rate of the highly
educated population were taken from the Vienna
Institute for International Economic Studies (wiiw)
databases concentrating data on central, eastern, and
south-eastern Europe.
To include the level of political stability in the
possible causes of high emigration rates, two
dimensions of the Worldwide Governance
Indicators (WGI) were taken: political stability and
absence of violence/terrorism, and government
effectiveness. These indicators, created as a
combination of the views of a large number of
surveyed companies, citizens, and experts in over
200 countries and territories, were taken from the
World Bank’s database. The measured value of
these indicators is expressed on a scale from −2.5 to
2.5 for both.
As explained above, a poor education system
may be a strong motive for highly ambitious
students to emigrate, so it is necessary to consider
some of the indicators of the quality of the
education system. For this purpose, the results of
PISA tests in reading, science, and mathematics
were taken from the Organization for Economic Co-
operation and Development (OECD).
In order to measure emigration, several different
indicators will be presented. The emigration rate,
taken from a database of the Institute for
Employment Research (IAB), measures the number
of emigrants per 1000 inhabitants of the pre-
migration population (age 25+). Emigration rates by
education level were taken from the same dataset,
where the most attention should be paid to the
emigration rate of the highly educated population.
However, in order to prohibit emigration rates by
the education level from being distorted, we will
also take into account the structure of the population
according to the level of education, taken from the
Eurostat database. The net migration rate is another
important migration indicator, which represents the
net number of migrants (number of immigrants
minus the number of emigrants) per 1000
inhabitants. This indicator, along with the estimates
of migrant stock, which stands for the number of
people born in a country other than that in which
they live (including refugees), was taken from the
United Nations database. Another source of data
was the database of the Migration Data Portal. One
indicator taken from this database is the share of
international migrants between 15 and 24 years
residing in the country/region at mid-year. This age
group is common for the share of youth NEET and
hence it is useful to include it in the analysis. The
last indicator found, also taken from the same
database, is so-called public opinion (data were
taken for the year 2016), which stands for the
percentage of adult respondents who reported plans
to move permanently to another country in the next
12 months, will help us to gain insight on the
attitude of citizens towards leaving the country.
WSEAS TRANSACTIONS on BUSINESS and ECONOMICS
DOI: 10.37394/23207.2023.20.198
Ana Topalović, David Hampel
E-ISSN: 2224-2899
2303
Volume 20, 2023
Cluster analysis based on the selected indicators
will be used with the main aim to find groups of
countries that are characterised by similarity within
themselves and diversity among countries from
other groups (for details [22]). Dividing countries
into different clusters will help us to evaluate the
current position of the examined countries in
relation to EU countries in all the selected areas. For
the purposes of this analysis, the computational
system MATLAB R2021a will be used. Another
statistical method examined will be regression
analysis, discussed in detail by, [23]; regression of
the cross-sectional data coming from the year 2015
and regression of panel data from 1995 to 2020 with
a five-year periodicity. Generally, the maximal
model can be described as follows:
𝑌
𝑖𝑡 = 𝛽0+ 𝛽1𝐺𝐷𝑃𝑝𝑐𝑖𝑡 + 𝛽2𝑁𝐸𝑖𝑡 + 𝛽3𝐺𝐸𝑖𝑡 +
𝛽4𝑁𝐸𝐸𝑇𝑖𝑡 + 𝛽5𝑃𝐼𝑆𝐴𝑖𝑡 + 𝛽6𝑈𝑁𝑖𝑡 + 𝛽7𝑈𝑁𝐻𝐸𝑖𝑡 +
+𝛽8𝑃𝑆𝑖𝑡 + 𝑎𝑖+ 𝑢𝑖𝑡,
where Y is gradually substituted by different
migration characteristics, GDPpc means GDP per
capita, NE average nominal monthly earnings, GE
government effectiveness, NEET Share of youth
NEET, PISA means average PISA results, UN
unemployment rate, UNHE unemployment rate of
highly educated people, PS political stability, ai
mean the individual effect of countries, and u
remains for random error. In order to eliminate
spurious regression, the stationarity of residuals is
tested by the Im-Pesaran-Shin unit root test. The
model was estimated in fixed effects (FE) and
random effects (RE) forms and the Hausman test
was employed to decide between them. Using this
analysis, we will be able to examine the dependence
of migration indicators on selected economic,
political, and educational indicators. For the
regression analysis, the Gretl 2021b program will be
used.
3 Results
In order to gain a closer insight into the position of
the WB countries in relation to the EU, we will first
make a comparison of these countries in several
selected indicators that will be further used in the
analysis. Croatia is also included in the analysis as a
peer country, as this country is an EU Member State
but was also considered to be part of the WB until
recently, and sometimes is classified as a WB
country even today. Figure 1 shows the
development of GDP per capita (in PPS) across WB
countries in the period between 2010 and 2019.
Even though GDP per capita is growing in the
long run in these countries, this growth is very slow,
except in Montenegro, which experienced the
highest growth in a given period, and values in most
WB countries are more than two times lower than
the EU average. For the entire period, the highest
values were reached by Montenegro (50 in 2019)
and the lowest by Albania (around 30 for the entire
period). On the other hand, Croatia achieves
significantly higher values than all WB countries,
even in the period before joining the EU. However,
Croatia’s GDP per capita is still below the EU
average with the highest value of 65 in 2019. Two
indicators of wage level: statutory nominal gross
monthly minimum wage and mean nominal monthly
earnings of employees for 2015 are shown in Figure
2. As it is possible to see from the Figure 2, the
wage gap between all WB countries and the EU
average is significant. Albania reaches the lowest
values, while Croatia is again the closest to the
average of the EU.
Fig. 1: Development of GDP per capita in WB
countries
Source: Data processed from the Eurostat database
Comparisons of the total unemployment rate,
the unemployment rate of the tertiary educated
population, and the share of youth NEET in WB
countries for 2015 are shown in Figure 3. Again, all
the surveyed countries achieve far worse results
than the EU in all the selected indicators. In 2015,
the highest unemployment rate was in BiH (almost
30%), followed by North Macedonia. Croatia,
Albania, Montenegro, and Serbia achieve very
similar results in terms of the total unemployment
rate (about 17%), so, in this case, Croatia is far
closer to its neighbouring WB countries than to the
EU average (10%).
Two components of the WGI (political stability
and absence of violence/terrorism, and government
effectiveness) for 2015 are presented in Figure 4. In
WSEAS TRANSACTIONS on BUSINESS and ECONOMICS
DOI: 10.37394/23207.2023.20.198
Ana Topalović, David Hampel
E-ISSN: 2224-2899
2304
Volume 20, 2023
terms of these indicators, Croatia again achieves the
best results of all the WB countries. The value of
political stability in Croatia is very close to the EU
average, while the value of government
effectiveness is two times lower than the EU
average. BiH achieves the worst results, with
extremely negative values of both indicators.
Fig. 2: Wage gap between WB countries and the EU
average
Note: For Cyprus, data were taken for 2016; for
Croatia, Denmark, Latvia, Malta, Netherlands,
North Macedonia, Romania, and Serbia, data were
taken for 2014.)
Source: Data processed from the ILOSTAT database
Selected indicators of the quality of the
education system (the results of PISA tests in
reading, mathematics, and science) in Table 1 show
that all the WB countries (excluding Croatia)
achieved results far below the EU average in all
areas, while Croatia has results closest to the EU
average, and even achieved a result of 1 point above
the EU average in reading. The correlation
coefficients of the PISA results in all EU countries
are above 95%, so there is no need to include each
of them in the analyses. Instead, we will use the
average results of all three subjects.
Fig. 3: Indicators of underutilisation of the labour
supply
Note: Indicator 1 Total unemployment rate;
Indicator 2 An unemployment rate of the tertiary
educated population; Indicator 3 Share of youth
NEET Source: Data processed from wiiw DATABASES
and ILOSTAT
Fig. 4: Worldwide Governance Indicators
Source: Data processed from the World Bank database
Table 1. Results of PISA tests held in 2015
PISA
Results
Reading
PISA
Results
Math
Albania
405
413
BiH
403
406
Croatia
487
464
Montenegro
427
418
North Macedonia
352
371
Serbia
439
448
EU average
486
488
Note: For BiH and Serbia, data were taken for 2018.
Source: Data processed from the OECD database
WSEAS TRANSACTIONS on BUSINESS and ECONOMICS
DOI: 10.37394/23207.2023.20.198
Ana Topalović, David Hampel
E-ISSN: 2224-2899
2305
Volume 20, 2023
We will further graphically present and describe
some of the selected migration indicators. Figure 5
shows the population structure and emigration rates
by education level in some WB countries for 2015.
It should be first noted, that in all the WB countries
the highest emigration rates are among the highly
educated population. In BiH in 2015, more than 4%
of the highly educated population were emigrants.
Fig. 5: Emigration rates and population structure
by education level
Note: Albania and BiH were excluded from the right plot
due to a lack of data. In the illustrated database, the
values for Serbia and Montenegro are listed as one
country in the left plot.
Source: Data processed from the IAB brain-drain
database and Eurostat
These rates are slightly lower in Albania,
Croatia, and North Macedonia, while Serbia and
Montenegro reach more than two times lower values
of this indicator (about 1.5%). The right plot shows
the population structure by education level. If the
division of the population by levels of education is
taken into account, it is possible to state that the
emigration rates of each of these groups are
somewhat equal. However, this does not diminish
the fact that the group of highly educated population
is most vulnerable to emigration.
Figure 6 shows the development of the net
migration rate and migrant stock in WB countries in
the period between 1990 and 2020. Net migration
has constantly changed over the given period,
varying from country to country. The largest change
in this indicator occurred in BiH and Albania in the
1990s. BiH experienced excessive emigration as a
result of the largest military conflict of all in this
area that lasted four years, while in Albania the
biggest reason was the fall of communism.
Fig. 6: Development of migration indicators in WB
countries
Source: Data processed from the UN database
This growing emigration trend is also
pronounced in the right plot, which shows a
significant increase in migrant stock in these two
countries. In Bosnia, the trend subsided in 1995,
when the war ended, while in Albania the migrant
stock continued to grow at a slower pace until 2010.
In all countries, the migrant stock grew between
1990 and 2000. The reasons for this phenomenon
are numerous, but the main one is considered to be
the breakup of Yugoslavia and its consequences
WSEAS TRANSACTIONS on BUSINESS and ECONOMICS
DOI: 10.37394/23207.2023.20.198
Ana Topalović, David Hampel
E-ISSN: 2224-2899
2306
Volume 20, 2023
numerous wars fought in this region in the 1990s,
hyperinflation, conflicts in Kosovo followed by the
bombing of the Federal Republic of Yugoslavia
(FRY) in 1999. Since 2000, there has been a
significant decline in the migrant stock in Serbia,
due to both the end of the wars and the significant
changes following the overthrow of Slobodan
Milosevic and the socialist regime and the
beginning of the rule of democracy. It is also worth
noting that despite a comparable population with
other countries (except Serbia, which is the most
populous of all the countries surveyed), BiH reached
the highest migrant stock of all.
The first statistical analysis conducted is
hierarchical cluster analysis. EU Member States
together with WB countries were included. A
standardised Euclidean distance was used due to the
difference in the units of measurement, together
with the Ward criterion. The clusters were formed
as shown in Figure 7.
According to the results of the non-hierarchical
cluster analysis using a k-means algorithm, the
optimal number of clusters seems to be four or six.
The setting of six clusters would lead to the
formation of two extra clusters than we obtained, of
which one cluster would include only one country;
therefore, we will maintain the optimal number of
four clusters. Based on the selected indicators, all
the WB countries (except Croatia) were classified in
the same cluster, together with their neighbouring
countries (Romania, Bulgaria, and Greece). All the
other clusters were formed by EU countries
only.The results obtained through cluster analysis
allow us to describe the position of the WB
countries in the EU context, both by directly
characterizing the individual clusters and by
calculating selected migration indicators for the
obtained clusters. Table 2 shows the average values
of the indicators of the economic, political, and
educational systems. As expected, the cluster
formed by the WB countries achieved the worst
results in all the selected indicators. In contrast,
cluster 2, which consists of the most developed
European countries (Scandinavian countries,
Luxembourg, Netherlands, etc.) achieved the best
results in all the indicators. The other two clusters
achieved very similar results; however, differences
may be seen in some indicators (cluster 3 reached
slightly better results in terms of the unemployment
rate, the share of youth NEET, and average PISA,
while Cluster 4 reached better results in terms of
GDP per capita, mean nominal monthly earnings
and political stability).
Fig. 7: Dendrogram presenting clustering of the discussed countries
Source: Own calculation
WSEAS TRANSACTIONS on BUSINESS and ECONOMICS
DOI: 10.37394/23207.2023.20.198
Ana Topalović, David Hampel
E-ISSN: 2224-2899
2307
Volume 20, 2023
Table 2. Average values of selected indicators
GDP per
capita in
PPS
Mean
nominal
monthly
earnings
Unemployment
rate
Political stability
and absence of
violence/
terrorism
Government
effectiveness
Share of
youth
NEET
Average
PISA results
Cluster 1
44
645
0.18
0.0
0.0
0.22
423
Cluster 2
147
4174
0.07
1.0
1.8
0.07
502
Cluster 3
86
1733
0.10
0.6
1.0
0.12
489
Cluster 4
96
1882
0.11
0.8
1.0
0.13
485
Source: Own calculation
Table 3. Average values of selected migration indicators
Emigration rate
Emigration rate of a highly
educated population
Net migration rate
Cluster 1
11.1
23.0
−2.3
Cluster 2
5.0
10.0
5.8
Cluster 3
4.3
10.2
−0.7
Cluster 4
18.7
30.6
1.9
Source: Own calculation
Table 3 shows the average results in terms of
migration indicators. From this table, we see that the
cluster most affected by the emigration of the
population is cluster 4. This is even more
pronounced when it comes to the emigration of a
highly educated population. The average net
migration rate in this cluster, however, is not nearly
as low, so it is possible to assume that although
these countries have a large outflow of population,
they also have a high inflow of immigrants. The
lowest value of this indicator is reached by the
cluster consisting of the WB countries. Cluster
immigration rates calculated for immigrants from
WB countries only (Albania, BiH, Montenegro,
North Macedonia, and Serbia) are shown in Figure
8. These rates are calculated using a standard
formula for calculating emigration/immigration
rates that divides the number of
emigrants/immigrants by the total population of the
country (in this case, the cluster of countries). The
Figure 8 shows that people from WB countries
migrate in large numbers within the Balkans as well,
especially in the period between 1990 and 2000 for
cluster 1 and in the period between 1990 and 1995
for cluster 4, when major political and military
conflicts were taking place in the region. The sharp
increase in the number of immigrants in cluster 4 in
the early 1990s was mainly caused by the huge
number of BiH citizens who emigrated en masse to
neighboring Croatia. The high immigration rate in
cluster 1 is mainly caused by the large-scale
emigration of the Bosnian population to Serbia, but
also by the huge number of Albanians who
emigrated to Greece in the same period. This
finding is in line with, [24].
Based on the immigration rates in Cluster 2 and
Cluster 3, it is possible to conclude that emigrants
going to other parts of Europe tend to choose more
developed countries, as Cluster 2 is made up of the
most developed EU countries. The slight decrease in
migrants from WB countries after 2000 in cluster 2
is probably caused by the return of migrants from
Serbia (but also from North Macedonia and
Montenegro) back home due to the end of the war
and regime changes as described above. The
significant decline in the number of migrants from
Serbia in Germany is also possibly caused by
changes in the conditions and methodology for
including migrants in official statistics.
The dependence of migration indicators on
selected indicators of the economic, political, and
educational system was examined using regression
analysis on both cross-sectional and panel data.
Cross-sectional data models were applied to all EU
and WB countries for 2015 data. Due to data
limitations and the small number of states, it was not
possible to model even one multivariate model.
Table 4 contains univariate models with the net
migration rate as the dependent variable. This
migration indicator proved to be affected by most of
the indicators examined: GDP per capita, average
earnings, government effectiveness, and The PISA
results had a positive effect on net migration rates,
while the share of young people without
employment, education, or training, and the overall
unemployment rate had a negative effect. The
WSEAS TRANSACTIONS on BUSINESS and ECONOMICS
DOI: 10.37394/23207.2023.20.198
Ana Topalović, David Hampel
E-ISSN: 2224-2899
2308
Volume 20, 2023
highest coefficients of determination are achieved
for GDP per capita (60% of the explained
variability), and average earnings (around 50% of
the explained variability).
As another dependent variable, the total
emigration rate was used. For this dependent
variable, only one univariate model, presented in
Table 5, was found for the regressor of the share of
youth NEET, which positively affected the
emigration rate, with a relatively low coefficient of
determination of only 15%. On the emigration rate
of the highly educated population, the share of
youth NEET proved to have a positive impact, while
government effectiveness, PISA results, and average
earnings had a negative impact. The results of the
models are presented in Table 6.
Fig. 8: Immigration rates of clusters immigrants
from WB countries only
Source: Own calculation
Table 4. Regression models of cross-sectional data with the dependent variable: net migration rate
Regressor
Constant
Coefficient
p-value
R2
Model 1
GDP per capita
−6.38
0.079
<0.001
0.611
Model 2
Mean nominal monthly earnings
−3.881
0.002
<0.001
0.494
Model 3
Government effectiveness
−2.824
3.921
0.002
0.281
Model 4
Share of youth NEET
5.382
−33.281
0.012
0.190
Model 5
Average PISA results
−22.019
0.048
0.039
0.110
Model 6
Unemployment rate
4.070
−28.127
0.037
0.137
Source: Own calculation
Table 5. Regression model of cross-sectional data with the dependent variable: emigration rate
Regressor
Constant
Coefficient
p-value
R2
Model 1
Share of youth NEET
3.447
39.500
0.026
0.154
Source: Own calculation
Table 6. Regression model of cross-sectional data with the dependent variable:
emigration rate of the highly educated population
Regressor
Constant
Coefficient
p-value
R2
Model 1
Share of youth NEET
4.782
90.200
0.005
0.236
Model 2
Government effectiveness
25.150
−8.579
0.006
0.228
Model 3
Average PISA results
89.170
−0.152
0.006
0.200
Model 4
Mean nominal monthly earnings
23.100
−0.003
0.041
0.132
Source: Own calculation
Table 7. Regression models of cross-sectional data with the dependent variable: public opinion
Regressor
Constant
Coefficient
p-value
R2
Model 1
The unemployment rate of the highly
educated population
0.172
28.300
<0.001
0.339
Model 2
Average PISA results
20.597
−0.038
0.002
0.253
Model 3
Government effectiveness
4.241
−2.056
0.004
0.248
Model 4
Unemployment rate
0.306
17.500
0.019
0.171
Model 5
Political stability
3.708
−2.349
0.029
0.149
Source: Own calculation
WSEAS TRANSACTIONS on BUSINESS and ECONOMICS
DOI: 10.37394/23207.2023.20.198
Ana Topalović, David Hampel
E-ISSN: 2224-2899
2309
Volume 20, 2023
Table 8. Comparison of the coefficients of regression models of cross-sectional data
Dependent variable
Regressor
Net
migration
rate
Emigration rate
Emigration rate of
highly educated
Public
opinion
GDP per capita
0.079
Mean nominal monthly earnings
0.002
−0.003
Government effectiveness
3.921
−8.579
−2.056
Share of youth NEET
−33.281
39.500
90.200
Average PISA results
0.048
−0.152
−0.038
Unemployment rate
−28.127
17.500
The unemployment rate of highly educated
28.300
Political stability
−2.349
Source: Own calculation
Table 9. Regression models of panel data with the dependent variable: net migration rate
Regressor
Constant
coefficient
p-value
Hausman test: p-
value
Model 1
Statutory nominal gross monthly
minimum wage
−12.634
0.029
<0.001
0.607
Model 2
Government effectiveness
−3.970
6.911
0.007
0.928
Source: Own calculation
The last indicator of migration for which an
influence of the examined variables was found in
public opinion. This indicator was shown to be
positively affected by the unemployment rate of the
highly educated population and the total
unemployment rate, and negatively affected by
PISA results, government effectiveness, and
political stability. Again, the results of the models
are presented in Table 7. Table 8 provides a
comparison of all the regression models found for
the cross-sectional data. The table shows that the
indicators that positively affect the net migration
rate, which at high values indicates a high
attractiveness for migrants, also negatively affect
the other three migration indicators, which at high
values indicates a high outflow of migrants.
Therefore, the results for GDP per capita, average
earnings, government effectiveness, and average
PISA, which positively affect net migration rates
and some of which negatively affect the other three
indicators of emigration, confirm that better
economics, political conditions, and quality
education determine high attractiveness for
migrants. On the other hand, the high share of youth
NEET and the high unemployment rate point to a
high outflow of migrants. Furthermore, public
opinion is positively influenced by the
unemployment rate of the highly educated
population and negatively influenced by political
stability, which means that people are more likely to
emigrate when the unemployment rate of the highly
educated population is high and less likely to
emigrate when the country has better political
stability.
For the panel data analysis, only a limited set of
variables was available: net migration rate and
emigration rate as dependent variables and share of
young NEET, political stability, government
effectiveness, unemployment rate, minimum wage,
and GDP per capita in PPS as regressors. Using
these data, two significant regression models were
estimated for the WB countries only with the
dependent variable of net migration rate. Two
regressors were found to have a positive effect on
the net migration rate, i.e., minimum wage, and
government effectiveness, confirming that higher
wages and better government effectiveness make a
country more attractive to migrants. In both cases,
models with random effects were employed
according to the Hausman test, which is shown in
Table 9. When using the other migration indicators
described above, no regression model was found
with either the cross-sectional data or the panel data.
4 Discussion
Similar work on this topic is lacking in the studied
region, which is surprising given the extent of the
problem of emigration of young people from the
WB. The poor availability of data needed for its
research certainly contributes to this. In, [25],
WSEAS TRANSACTIONS on BUSINESS and ECONOMICS
DOI: 10.37394/23207.2023.20.198
Ana Topalović, David Hampel
E-ISSN: 2224-2899
2310
Volume 20, 2023
provides an overview of the factors, motivations,
and trends concerning youth emigration from the
WB in relation to other Southeast European
countries. The papers, [14], [26], based on sample
surveys focused on an examination of the causes of
emigration from WB countries lead to a similar
conclusion that the main cause of emigration is poor
economic and political systems, as well as better
education systems in the destination countries.
Paper, [27], also used a questionnaire-based survey
to examine physician migration between Romania
and France. They concluded that mobility is a
response to professional goals, but also to the
instability of the work situation, as well as to
personal and family goals. After snowball sampling,
[28], used a factor analysis to identify four main
determinants of Lithuanian high-skilled migration:
occupational attractiveness abroad, socioeconomic
conditions, state academic system and cooperation,
and macroeconomic conditions and government
policy.
Authors of, [29], used a pooled panel regression
model to analyse the factors influencing migration
flows from central and eastern European countries
to Germany in the period between 1998 and 2016.
The author indicated that the GDP gap has a
significant impact on the decision of the population
to migrate to Germany. The unemployment rate has
been shown to be a reason for the decline in
migration flows, but this has several reasons (the
main one being the economic crisis, during which
unemployment rates rose in all EU countries). Other
positive influences on emigration were relatively
low expenditure on education in the countries of
origin, the increase in the diaspora population, and
the enlargement of the EU in 2004. Similar results
were reached in, [30], where is used regression
analysis to arrive at the key determinants of
emigration of university-educated people from
eastern European countries in the period between
1980 and 2010, which were wages and education
expenditures in the sending countries. Results of the
paper, [31], point to the fact, that the East-West
European migration rate in the period from 2000 to
2017 responds quickly to the changes of GDP per
capita and unemployment rate of the young
population. The paper, [32], concluded that well-
educated people from poorer countries are the most
likely to emigrate. Using a regression model, he
found that the level of development, captured by
GDP per capita, is a negative determinant of
emigration, unemployment is negatively correlated
with the number of migrants, and average wage is
positively correlated with the number of migrants.
Authors of, [33], deal with the possibilities of how
to reduce youth emigration and assess the
relationship between youth participation in
entrepreneurship promotion initiatives and
emigration attitudes. A study conducted in the Utena
region of Lithuania highly affected by youth
unemployment did not show any correlation
between the analysed elements.
Another commonly used method for
investigating emigration is the gravity model. For
example, [34], estimated a gravity model to explore
the channels through which OECD countries attract
foreign physicians from abroad. The results showed
that the main drivers of physician outflows are
lower unemployment rates, good salaries, an aging
population, and a high level of medical technology
in the destination country. Distance has also been
shown to have a negative effect on emigration,
while colonial relations, language, and EU and
Schengen membership have a positive effect. Paper,
[24], applied the gravity model for Eastern
European countries (including WB countries) and
concludes that emigration increases with relatively
low income in the country of origin, which is in line
with our findings; moreover, results show that
emigration from autocracies is significantly higher
than from democracies. However, for employing a
gravity model, it is necessary to have an appropriate
range of countries and data, which precludes the
application of this method limited to the countries
studied in this paper.
Overall, our results provide new insights based
on national-level data for the region of WB
countries. The comparisons of the key determinants
of brain drain with those for EU countries and the
causal relationships suggested by the models add to
the knowledge in this area. Comparison of the
results with similar work in other countries allows
us to conclude that our results are in general
agreement with existing research, but also point to
facts not reported elsewhere, in particular the
considerable impact of differences in education
systems and political culture on brain drain.
5 Conclusions
The results of our work show that in most of the
selected indicators, there are wide gaps in the WB
countries compared to more developed countries
(specifically the EU, which is the most popular
destination for emigrants from the WB region).
GDP per capita in most of these countries is below
50% of the EU average, and the wage gap and
unemployment rate are several times higher than the
EU average. Political system indicators also lag well
behind the EU average and average PISA results are
WSEAS TRANSACTIONS on BUSINESS and ECONOMICS
DOI: 10.37394/23207.2023.20.198
Ana Topalović, David Hampel
E-ISSN: 2224-2899
2311
Volume 20, 2023
below the EU average in all the WB countries.
Cluster analysis showed that based on selected
indicators of economic, political, and educational
systems, as well as emigration indicators, the WB
countries, together with their neighbouring countries
Romania, Bulgaria, and Greece, perform worst in all
the selected indicators and have the lowest net
migration rates. By using regression analysis, it was
shown that the selected emigration indicators are
affected by most of the economic, political, and
educational system indicators. The results of GDP
per capita, mean nominal monthly earnings,
government effectiveness, share of youth NEET,
unemployment rate, and average PISA has the most
significant influences. It was also shown that people
from WB countries are more likely to move to more
developed countries, but also to countries within the
Balkan Peninsula region itself.
An important warning for the WB countries is
the finding that the inhabitants of these countries
emigrate not only for economic reasons but also
because of dissatisfaction with the functioning of
the state and the lower level of the education
system, which is especially true for emigrants with
higher education.
In terms of data collection and processing, the
lack of data was generally visible in the WB region,
which may be considered a serious problem to
overcome. This not only hinders research on this
topic but may also cause a distorted picture of the
emigration situation in these countries. Therefore,
extra attention should be paid to the interpretation of
the data and the results themselves. The most
important task must be directed to the responsible
institutions: collect and process the data properly in
the first place and agree on the methodology used
for this purpose.
This paper focuses on the causes of the brain
drain problem but does not consider the other side
of this phenomenon its consequences. The
discussion of the consequences of brain drain goes
both ways (positive and negative), showing that
emigration may even contribute to a developing
country in terms of economic gains (mainly through
remittances, which make up a significant percentage
of a country’s GDP), but also in terms of improved
human capital. Further research should focus on
assessing brain gain in the WB region and the
overall outcome of skilled emigration.
References:
[1] Landesmann, M. & Mara, I., Massive
Migration and its Effect on Human Capital
and Growth: The Case of Western Balkan and
Central and Eastern European Countries, wiiw
Balkan Observatory Working Paper, Vol.124,
2016.
[2] Horvat, V., Brain Drain. Threat to Successful
Transition in South-East Europe?, Alumni
Association of the Center for
Interdisciplinary, University of Sarajevo,
Sarajevo, 2004.
[3] Dolenec, D., Baketa, N. & Maasen, P.,
Europeanizing Higher Education and
Research Systems of the Western Balkans,
University of Zagreb, 2014.
[4] Docquier, F., Lohest, O. & Marfouk, A.,
Brain Drain in Developing Countries, The
World Bank Economic Review, Vol.21, No.2,
2007, pp. 193-218.
[5] Docquier, F. & Rapoport, H., Globalization,
Brain Drain, and Development, Journal of
Economic Literature, Vol.50, No.3, 2012, pp.
681-730.
[6] Dauti, B. & Voka, I., External Debt and
Economic Growth in the Western Balkan
Countries, with special focus to Albania,
Kosovo, and North Macedonia in the course
of the pandemic COVID-19, WSEAS
Transactions on Business and Economics,
Vol.19, 2022, pp. 1303-1317.
[7] Iravani, M.R., Brain Drain Problem: A
Review, International Journal of Business
and Social Science, Vol.2, No.15, 2011, pp.
284-289.
[8] Gibson, J. & McKenzie, D., Eight Questions
about Brain Drain, Journal of Economic
Perspectives, Vol.25, No.3, 2011, pp. 107-
128.
[9] Lawan Ngoma, A. & Wana Ismail, N., The
determinants of brain drain in developing
countries, International Journal of Social
Economics, Vol.40, No.8, 2013, pp. 744-754.
[10] Belot, M.V.K. & Hatton, T. J. Immigrant
Selection in the OECD. The Scandinavian
Journal of Economics, Vol.114, No.4, 2012,
pp. 1105-1128.
[11] Commander, S., Kangasniemi, M. & Winters,
L.A., The Brain Drain: A Review of Theory
and Facts, Brussels Economic Review, Vol.47,
No.1, 2004, pp. 29-44.
[12] Ziberi, B. & Alili, M. Z., Economic Growth in
the Western Balkans: A Panel Analysis, South
East European Journal of Economics and
Business, Vol.16, No.2, 2021, pp. 68-81.
[13] Bhattarai, K., Brain Gain (Drain),
Immigration and Global Network: Nepalese
Students in the UK, International Journal of
WSEAS TRANSACTIONS on BUSINESS and ECONOMICS
DOI: 10.37394/23207.2023.20.198
Ana Topalović, David Hampel
E-ISSN: 2224-2899
2312
Volume 20, 2023
Economic Policy in Emerging Economies,
Vol.4, No.4, 2011, pp. 345-365.
[14] WFD, Cost of Youth Emigration: Serbia,
Westminster Foundation for Democracy,
London [online], 2019, https://www.wfd.org/
what-we-do/resources/alienation-politics-
fuels-emigration-serbia-wfd-survey (Accessed
Date: 10 May 2022).
[15] Radonjić, O. & Bobić, M., Brain Drain
Losses A Case Study of Serbia. University
of Belgrade, Beograd, 2020.
[16] Gabrić-Molar, I. & Slavić, A., The Impact of
Emigration from Serbia to Hungary on the
Human Resources of Vojvodina, Zbornik
Matice srpske za drustvene nauke, Vol.148,
2014, pp. 571-581.
[17] Zwiech, P. International migration in Poland.
International Journal of Economic Policy in
Emerging Economies, Vol.1, No.2-3, 2008,
pp. 299-313.
[18] Bouzahzah, M., Quality of the education
system and economic growth. Projections in
the case of Morocco, WSEAS Transactions on
Business and Economics, Vol.18, 2021, pp.
949-961.
[19] Tkachova, N., Krushelnytska, T., Marchenko,
O. & Kuznetsova, N., Migration Policy in the
Context of Sustainable Development, WSEAS
Transactions on Business and Economics,
Vol.18, 2021, pp. 619-627.
[20] Zeneli, B, Cosic, E., Dzebo, A. & Zmejkoski,
R., Brain Gain Policies and Practices in the
Western Balkans, Group 484 Centre for
Migration, 2013.
[21] Predojević-Despić, J.R. & Penev, G.D.,
Emigration Zones in Serbia: 2011 Census
Results, Institute of Social Sciences,
Demographic Research Centre, Belgrade,
2014.
[22] Martinez, W.L. & Martinez, A.R.
Computational Statistics Handbook with
MATLAB, Boca Raton, Fl: CRC Press, 2016.
[23] Greene, W.H., Econometric analysis, Boston:
Prentice Hall, 2012.
[24] Parker, E., Migration Patterns in Eastern
Europe and the World: A Gravity Approach,
South East European Journal of Economics
and Business, Vol.15, No.1, 2020, pp. 66-79.
[25] Lavrič, M., Youth Emigration from the
Western Balkans: Factors, Motivations, and
Trends, in Valeska Esch et al. (eds.),
Emigration from the Western Balkans, Aspen
Institute Germany, 2021.
[26] WFD, Cost of Youth Emigration: Bosnia and
Herzegovina, Westminster Foundation for
Democracy, Sarajevo [online], 2020,
https://www.wfd.org/what-we-do/resources/
cost-youth-emigration-bosnia-and-
herzegovina-wfd-research (Accessed Date: 10
May 2022).
[27] Séchet, R. & Vasilcu, D., Physicians’
Migration from Romania to France: A Brain
Drain into Europe?, Cybergeo: European
Journal of Geography, Space, Society,
Territory, document 743, 2015.
[28] Kazlauskienė, A. & Rinkevičius, L.,
Lithuanian “Brain Drain” Causes: Push and
Pull Factors, Engineering Economics, Vol.46,
No.1, 2006, pp. 27-37.
[29] Přívara, A., Explaining emigration patterns in
Estonia, Lithuania, Slovenia and Slovakia,
Geographical Journal, Vol.71, No.2, 2019,
pp. 161-180.
[30] Ionescu, L., Emigration from Eastern Europe
with a Focus on Brain Drain, Journal of
Social and Economic Statistics, Vol.4, No.2,
2015, pp. 54-74.
[31] Franc, S., Časni, A.Č. & Barišić, A.,
Determinants of Migration Following the EU
Enlargement: A Panel Data Analysis, South
East European Journal of Economics and
Business, Vol.14, No.2, 2019, pp. 13-22.
[32] Zizi, G., Main Determinants of Romanian
Emigration. A Regional Perspective, Journal
of Social and Economic Statistics, No. Special
number 2016, 2016, pp. 1-13.
[33] Petronyte, A. & Ulbinaite, A.,
Entrepreneurship promotion as a tool to
counter youth emigration, International
Journal of Economic Policy in Emerging
Economies, Vol.14, No.4, 2021, pp. 415-442.
[34] Botezat, A. & Ramos, R., Physicians’ Brain
Drain a Gravity Model of Migration Flows,
Globalization and Health, Vol.16, No.7,
2020, pp. 1-13.
WSEAS TRANSACTIONS on BUSINESS and ECONOMICS
DOI: 10.37394/23207.2023.20.198
Ana Topalović, David Hampel
E-ISSN: 2224-2899
2313
Volume 20, 2023
Contribution of Individual Authors to the
Creation of a Scientific Article (Ghostwriting
Policy)
The authors equally contributed to the present
research, at all stages from the formulation of the
problem to the final findings and solution.
Sources of Funding for Research Presented in a
Scientific Article or Scientific Article Itself
This work was supported by the Internal Grant
Agency PEF MENDELU, No. PEF_TP_2021007.
Conflict of Interest
The authors have no conflicts of interest to declare.
Creative Commons Attribution License 4.0
(Attribution 4.0 International, CC BY 4.0)
This article is published under the terms of the
Creative Commons Attribution License 4.0
https://creativecommons.org/licenses/by/4.0/deed.en
_US
WSEAS TRANSACTIONS on BUSINESS and ECONOMICS
DOI: 10.37394/23207.2023.20.198
Ana Topalović, David Hampel
E-ISSN: 2224-2899
2314
Volume 20, 2023