REVIEW OF ECONOMIC PERSPECTIVES – NÁRODOHOSPODÁŘSKÝ OBZOR
VOL. 15, ISSUE 1, 2015, pp. 15–34, DOI: 10.1515/revecp-2015-0009
Intra-EU Direct Investment and Enlargement
Andrea Festa1
Abstract: This paper examines the determinants of the intra-EU direct investment (IDI)
into the New Member States (NMS) using a panel dataset of bilateral capital flows for
the period 1993-2013. It is found out by using a simple gravity model that EU membership
is the most important determinant. Unlike previous studies including non-EU countries,
the distance is insignificant, which is caused by proximity of these countries to one
another. A separate analysis focused on subgroups of accession countries gives some
evidence that even when size of their economy, distance, institutional quality and EU
accession are taken into account, Central European countries receive more IDI than the
Baltic and the Balkan states. On the contrary to that, the analysis restricted to the Balkan
countries which have joined the EU shows the inexistence of a negative Balkans effect
in attracting foreign investment. This finding is relevant because previous studies
demonstrate a persistent negative Balkans effect for non-EU Balkan countries and suggests
a crucial impact of the EU accession in determining the intra-EU capital flows.
Key words: Intra-EU direct investment, EU membership, New Member States
JEL Classification: C33, P27, P33
Introduction
The determinants of foreign direct investment (FDI) into the New Member States (NMS)
of the EU have attracted much attention during the past decades, even before the EU
accession. In the recent twenty years, Europe has seen historical developments in the
ex-socialist countries and, in economic terms, a transition from socialism to free-market
economies as well as an increasing FDI into these economies.
The EU enlargement policy helped this transition process, giving the basis of a common
legislation across a wide range of subjects, including common trade and financial rules,
customs union and free movement of workers, capitals and services. This process
moved contemporaneously with the economic transition of these countries which share
similar political and economic features. In fact, they all have faced institutional and
social challenges in the 1990s, followed by economic reforms in order to comply with
the EU legislation.
1
Italian Customs and Monopoly Authority, Regional Directorate of Lombardy, Office of Milan 2,
via Valtellina 1, 20159 Milan, Italy. Email: andrea.festa@agenziadogane.it. I would like to thank
two anonymous reviewers for their valuable comments. The views expressed in this paper are
those of the author and do not necessarily reflect those of the Italian Customs and Monopoly
Authority.
REVIEW OF ECONOMIC PERSPECTIVES
16
The goal of this paper is to analyze the determinants of the intra-EU direct investment
(IDI) into NMS, for which panel data on bilateral capital flows from EU-15 countries to
NMS during the period 1993-2013 will be used. Because the most important part of the
FDI into the transition economies comes from the developed Europe, as showed by
Bevan and Estrin (2004), the analysis is focused on the determinant of IDI to investigate
which are the main factors driving the capital flows in the continental Europe with a
special regard to the EU membership for the transition economies to be considered the
symbol of the re-integration into the developed Europe. In fact, the perspective and,
more importantly, the accession to the EU can be crucial in determining the capital
flows into the NMS.
The paper is organized as follows: Section 1 introduces the issue of IDI into the European
transition economies. Section 2 provides a brief overview of the literature, and
Section three illustrates data and empirical specification. Section 4 highlights the results,
and finally, Section 5 concludes.
Literature
A number of studies investigated the determinants of capital flows into the ex-socialist
European countries during the transition phase (for instance, Baldwin et al., 1997; Bartlett,
2008; Bevan and Estrin, 2004; Bevan, Estrin and Mayer, 2004; Brenton et al, 1997;
Buch et al., 2003; Dikova and van Witteloost, 2007; Di Mauro, 1999; Estrin and Uvalic,
2013; Estrin, Xavier and Brada, 2000; Lankes and Venables, 1996; Meyer and Estrin,
2004; or Richet and Brada, 2000).
Some papers focus on the key determinants of FDI into the transition economies by
analysing forms and origins (Hunya 2011, 2012; Kalotai, 2010). Brada et al. (2006)
show that the Balkan countries which experienced political transition problems receive
less FDI compared to the Central-European and Baltic countries.
Bevan and Estrin (2004) investigate the determinants of FDI into the ex-socialist European
economies and find out that unit labour costs and gravity factors are important in
affecting the capital flows. Furthermore, they came to the conclusion that the announcement
about the EU accession has a positive impact on the capital inflows. In fact,
it is perceived as an important step that is achievable only after substantial economic
and political reforms. In general, the process took time so the announcement itself
should be considered a crucial achievement for a transition country.
Estrin and Uvalic (2013) found out that Western-Balkan countries receive less FDI
compared to countries of Central East Europe (CEEC) even when market size, distance
and EU accession are taken into account. The authors suggest that this difference is due
to the inadequate level of the public institutions, the heritage of a historical fragmentation,
current political risks and the economic environment, too, in which the protection
of property rights has not yet reached the level of more developed economies.
Garibaldi et al. (2001) ascertained that market size, availability of resources, openness
and barriers to investment are significant determinants of capital flows, whereas wages,
initial liberalization and initial condition do not seem to play the key role.
Volume 15, Issue 1, 2015
17
Furthermore, Barrell and Pain (1999) find out that positive initial conditions play a role
in determining capital inflows by attracting significant FDI at the early stage of transition
as well as supporting economic development and growth.
Uvalic (2003) and Brada, Kutan and Yigit (2006) stress the fact that political instability
negatively affects the investment flows, especially for the region of the Balkans, where
there are still important and unresolved political issues. These findings are substantially
confirmed by Demekas et al. (2005).
On the other hand, Kekic (2005) finds out that the determinants of FDI for the Balkans
are equivalent to other ex-socialist economies even if the analysis does not cover the
whole transition phase.
Despite consistent literature on this topic is abundant, there is still a lack in the investigation
of the intra-EU direct investment (IDI) from EU-15 member states into the transition
countries. Because several NMS joined the EU during the last decade, now it is
possible to check the importance of the EU membership in determining the capital inflows
after some time since the accession. IDI, as well as FDI, can be crucial to accelerate
and complete the transition to free market economies (see, for instance, Mayhew
1998). Moreover, because some group of countries share similar social and institutional
features, it is possible to investigate whether some subgroups of NMS have benefited
from the EU membership more than others and, if so, to which extent.
Data and methodology
In this paper, the data have been drawn from the comprehensive and comparable Eurostat
database for 27 EU member states for the period 1993-2013. Observations of IDI
are in thousands of Euros and describe the direct investment flow from an EU-15 country
to a NMS. The dataset includes 2506 observations and covers the full period of transition
so it is possible to have a complete framework of what has been the effect of the
EU accession on intra-EU direct investment. EU-15 countries are the source economies,
ex-socialist countries that were involved in the enlargement process are the host economies.
Current official and potential candidates, as well as the 28th EU member state
(Croatia) have been excluded from the final sample because of insufficient data, especially
for the Western-Balkan countries. In order to study the impact of the EU accession
on IDI, the focus is therefore on transition economies that joined the EU during the
last decade (that is, Bulgaria, Czech Republic, Cyprus, Estonia, Hungary, Latvia, Lithuania,
Malta, Poland, Romania, Slovakia, and Slovenia). In fact, the EU membership can
be viewed as a signal of development and democratization, albeit as regards to market
size, institutional development and the degree of assimilation of the acquis communitaire
(Bevan and Estrin, 2004), there are lingering differences between the old and new
member states. Nevertheless, the EU membership allows to directly investigate on how
accession affected the economies under observation by considering the bilateral investment
flows as an appropriate signal of integration and development.
On the other hand, European association agreements were not taken into account because
of multicollinearity issues. These agreements might be considered a part of the
pre-accession phase. As a consequence, their role is considered to be captured by the
EU membership dummy in the empirical estimates.
REVIEW OF ECONOMIC PERSPECTIVES
18
The idea underlying this paper is that the more the continental Europe will experiment
with large capital flows between old and new member states, the more these economies
will become integrated and developed. One of the main objectives of the enlargement
policy is to establish common grounds of rules allowing the EU enterprises to operate
without barriers in any member state. The goal is to ensure as much as possible same
opportunities to do business all over the European Union, regardless of the enterprises’
country of origin. The EU membership is a symbol of stability of a country, quality of
its institutions as well as implementation of political reforms. Thus, investors consider it
the most important signal that an ex-socialist country can send. In fact, even if the enlargement
process includes several progressive steps, it is only at the moment of the
accession that a candidate country demonstrates that it fulfils all the conditions to be
part of the developed EU. Thus the EU membership reduces the country risk and it is
expected to positively affect the inward capital flows, albeit the accession could be a
better signal for extra-EU rather than intra-EU foreign investors.
However, despite the EU membership the IDI flows can be limited by gravity factors. In
general, the greater is the distance between two countries, the lesser should similarities
be, for instance, at social and cultural level, and the lesser should the bilateral trade and
investment flows be, too. As stated by Bevan and Estrin (2004), the distance is a measure
of transactional cost of doing business abroad. Moreover, the IDI flows are expected
to be influenced by market size of the involved economies as well as by distance. In
order to investigate whether or not the IDI flows are determined by gravity factors, a
simple gravity model was used.
The gravity approach is used widely in literature. It assumes that the size of economies,
the distance between source and host countries and other factors (unit labour cost, institutional
and social stability, or membership in economic associations, for instance) are
the main drivers for trade or investment flows. In line with recent research, this approach
was adopted to investigate the determinants of the bilateral investment flows
between the old and new EU member states. Following Helpman (1984), Brainard
(1997) and Bevan and Estrin (2004), the model controls for different variables typically
used to capture comparative advantages in the NMS, and has the following form [Eq.1]:
݈݊‫ܫܦܫ‬௜௜௧ ൌ ߚ଴݈݊‫ܲܦܩ‬௜௧ ൅ ߚଵ݈݊‫ܲܦܩ‬௝௧ െ ߚଶ݈݊݀݅‫݁ܿ݊ܽݐݏ‬௜௝ ൅ ߚଷ݈݊‫ݏݏ݁݊݊݁݌݋‬௝௧
െ ߚସ݈ܴܷ݊‫ܥܮ‬௝௧ ൅ ߚହ‫4002݌݄݅ݏݎܾ݁݉݁ܯ‬௝ ൅ ߚ଺‫7002݌݄݅ݏݎܾ݁݉݁ܯ‬௝
൅ ߚ଻‫ݏ݊݋݅ݐݑݐ݅ݐݏ݊ܫ‬௝ ൅ ߝ௜௝௧
where the subscript i stands for the source country, j stands for the host country, and
finally t for year. Any variable entering the equation provided above is in logs, except
for the distance variable and the membership dummy. The IDI from a source EU-15
country to a host NMS is the dependent variable. On the right hand side (RHS) of the
equation [1], GDPit is the size of the source country; GDPjt is the size of the host country;
distanceij stands for the distance from the capitals (in kilometers); opennessjt stands
for the sum of total import and export over GDP of the host country; RULCjt stands for
the real unit labor cost of the host country; Membership2004j is the EU accession dummy
variable which assumes the value of 0 before 2004 and then 1 for the Czech Republic,
Cyprus, Estonia, Hungary, Latvia, Lithuania, Malta, Poland, Slovakia and Slovenia;
Membership2007 is the EU accession dummy variable that assumes the value of 0 be-
Volume 15, Issue 1, 2015
19
fore 2007 and then 1 for Bulgaria and Romaniaj; Institutionsj considers a measure drawn
by the Heritage Foundation’s Index of Economic Freedom: the investment freedom
index (Estrin and Uvalic, 2013); and finally εijt is the error term. The model is estimated
with random effects because the Hausman test does not reject the null hypothesis of
hortogonality between regressors and the error term. The model controls for gravity
factors and for factors of cost and market conditions. The variables are in logs to address
non-linearities of the data (Estrin and Uvalic, 2013). Due to the fact that investment
decisions take time and are not influenced by the RHS variables immediately, the
model was estimated with a lag of one period in the independent variables, the only
exception being the constant. This appears to be a reasonable assumption especially as
regards time consuming decisions such as the decision of investing abroad. Moreover, it
helps us address potential endogeneity issues.
With regard to the RHS variables, the coefficients for both GDP and openness variables
are expected to be positive related to IDI because market size and openness of the economies
should be thought of as signals of the capacity to supply and development (Bevan
and Estrin, 2004). On the other hand, the higher the factor costs are in a host country
compared to the factor costs of the source country, the less an enterprise of the source
country is motivated to invest there. Thus the coefficient of the real unit labor cost for
the host country is expected to negatively affect investment flows. The distance variable
is a proxy for transnational costs (in economic and social terms, i.e. different cultures,
languages, etc.). It holds true that the greater the distance between source and host countries,
the higher the transnational costs are expected to be. Hence, the coefficient of this
variable should be negatively related to IDI. Vice versa, the EU membership dummy as
a signal of institutional stability and economic development should positively affect the
dependent variable and the investment freedom index.
In order to detect heteroskedasticity, White’s and Breusch-Pagan/Cook-Weisberg’s tests
for heteroskedasticity were performed. They evidence the presence of heteroskedasticity
in the residuals of the main model. Moreover, as far as autocorrelation is concerned, the
Cumby-Huizinga general test for autocorrelation was performed with the null hypothesis
of no serial correlation or that serial correlation exists, but died out at a known finite
lag (‫ݍ‬ ൐ 0). The test can be used in cases where alternatives such as the BoxPierce/Ljung-Box,
Durbin's h test and the Breusch-Godfrey test are not applicable. For
example, the Box-Pierce/Ljung-Box is not applicable if regressors are not strictly exogenous
and there is heteroskedasticity in the residuals. The Breusch-Godfrey test as well
as the Durbin's h test, relaxes the assumption of strictly exogenous regressors but are
inappropriate in presence of conditional heteroskedasticity. The Cumby-Huizinga test is
advantageous in the fact that it can be applied in the fixed-T large-N panel data context.
The test rejects the null hypothesis of no serial correlation in the residuals. Therefore,
because the residuals may exhibit heteroskedasticity and autocorrelation, the model was
estimated correcting for heteroskedasticity and within-cluster correlation in order to
obtain cluster-robust standard errors and mitigate the problem of autocorrelation. This
method was chosen because the main specification shows some differences between the
cluster-robust and default standard errors. In this case, cluster-robust estimation should
be preferred.
REVIEW OF ECONOMIC PERSPECTIVES
20
On the other hand, the estimator proposed by Driscoll and Kraay (1998) – that is robust
to heteroskedasticity, both within and across cluster correlations and that has the advantage
that correlations across clusters are possible – was not used because the asymptotics
of its standard errors require a larger number of time periods. The Newey–
West/Bartlett kernel (HAC) estimator was not used either because it requires large-T
asymptotics.
The data was clustered at the state level, with residuals assumed to be uncorrelated
across clusters but correlated within a cluster. In other words, the residuals in different
periods for a given state may be correlated, while the residuals for different states are
assumed to be uncorrelated. It should be noted that, in general, if observations are correlated
over time, cluster-robust standard errors account even for such serial correlation.
However, as a robustness check, the data were clustered in the N-dimension, that is, on
panel units in order to obtain standard errors robust to arbitrary serial correlation. The
main findings remain unaltered.
Results
The analysis of determinants of the intra-EU capital flows into the new member states is
based on the assumption that the EU membership positively affects IDI while gravity
factors, such as the distance between the source and host countries negatively affects the
investment flows. Furthermore, factor costs, size and openness of the countries involved
should play a role in this contest (in a negative and positive manner, respectively).
Table 1 shows the annual investment flows into the NMS. At a first glance, it is possible
to see substantial increase in investment flows after 2000 for all the transition economies.
This is a result of macroeconomic and political stability. Despite the fact that the
EU accession is a better signal for extra-EU rather than intra-EU foreign investors, even
the intra-EU investment flows benefited from the EU enlargement. Table 1 shows that
in comparison to 2003, capital flows into the NMS increased twice during the accession
phase (2004 – 2007). Moreover, the year after the second enlargement (Romania and
Bulgaria, 2008), capital flows into the NMS reached their historical peak. These flows
decreased due to global economic crisis after that year. Nevertheless, they were always
larger compared to the last year before the first enlargement (2003). Thus, it is possible
to state that the EU accession reduced the country risk even in case of IDI flows. The
increase in investment flows widened after the first enlargement (2004) for several
countries, albeit with differentiation. It should be underlined that all the NMS attracted
more IDI in comparison to the 1990s. On the other hand, the second enlargement, which
was restricted to Bulgaria and Romania, did not impact investments flows of these
countries in any notable manner, especially for Bulgaria. The reason for that can be seen
in the economic crisis that caused a fall in IDI after 2007-2008 in most ex-socialist
countries.
Table 2 illustrates the top three source countries in each NMS, and provides an analysis
of the latest year available (2012). In the group of countries that represented major investors
in at least three NMS, there is only the Netherlands (Cyprus, Hungary and Slovakia).
Denmark (Czech Republic and Romania), Spain (Malta and Poland) and Sweden
(Estonia and Latvia) are major investors in two NMS, while Luxembourg (Bulgaria),
Volume 15, Issue 1, 2015
21
Belgium (Slovenia) and Germany (Lithuania) are major investors in one country. This
could suggest that distance may play a role in determining the IDI flows. In fact, Spain
and two Nordic countries (Denmark and Sweden) are to be found among the top source
countries. In comparison to mainland Europe, these countries are relatively more distant
to the NMS. On the other hand, it should be noted that the most important EU economy
(Germany) is the top investor in one country only. Here, it should be underlined that an
enterprise can opt to invest as well as to export without investing abroad. This is especially
valid for enterprises of countries that are export-oriented economies, such as
Germany.
Figure 1 illustrates the sectoral distribution of the IDI flows. The data have been aggregated
into the agriculture, mining, manufacturing and services sector of the NMS in
2011. This analysis can be useful in evaluating the long-run impact of the capital flows
on employment and growth (Estrin and Uvalic, 2013). At a first glance, it is possible to
see that the services sector collects substantial capital flows, on average 59 percent of
total. In particular, this sector represents more than 75% in Hungary, while lower shares
were registered in other countries (47% in Romania, 51% in Slovakia, and 52% in the
Czech Republic). As far as manufacturing is concerned, there is an important differentiation
among the NMS. In fact, compared to other ex-socialist economies, the Central
NMS (the Czech Republic, Slovakia and Poland) attracted a substantial amount of IDI
in the manufacturing sector (more than 68% of total).
Table 3 shows the main results. It should be noted that in all specifications, any variable
entering the eq. [1] has the expected sign, with the exception of the investment freedom
index. This index is insignificant, though. Column (1) illustrates the empirical specification
without the EU membership dummies. As expected, the coefficient of the distance
variable (-.018) is negative. However, it is insignificant. This result diverges from the
previous studies – which include extra-EU countries as source economies – but is not
surprising given the proximity of the countries to each other.
The model is estimated including the EU membership 2004 dummy in Column (2). The
coefficients of the other variables do not change significantly. More importantly, it
should be noted that the EU membership dummy is the main determinant of the IDI
flows as shown by its magnitude (.208). The distance, on the other hand, does not play
the key role in determining investment flows into the NMS. In fact, the coefficient
reaches a value close to zero (-.006) and is statistically insignificant. The size of the
source economies is always positive and significant (from .047 to .050; Table 1) while
the size of the host economies and openness of the host countries to world trade seems
not to have any crucial impact on determining the investment flows. The same reasoning
is valid to factor costs measured by real unit labor costs whose coefficient in any
specification is negative but insignificant. It is necessary to say that this was expected.
The model is estimated including both the EU membership 2004 and 2007 dummies in
Column (3). The coefficients of the RHS variables do not change much. Once again, the
EU membership 2004 and 2007 dummies are the main determinants of the intra-EU
capital flows as shown by their magnitude (.253 and .424, respectively) and significance
at 1%.
REVIEW OF ECONOMIC PERSPECTIVES
22
Column (4) presents the results adding the investment freedom index among the independent
variables as a measure of institutional quality. The results remain unchanged
albeit the coefficient of the investment freedom variable is insignificant as in the work
of Estrin and Uvalic (2013). Thus the impact of the institutional qualities is – not surprisingly
– fully absorbed by the EU membership dummies.
Our variable of interest, EU membership dummies, retains its significance and magnitude
(.258 and 424, respectively). If we were to put it another way, even when controlling
for other institution variables, the EU membership is the main determinant for IDI
flows while the distance between the source and the host countries does not play the key
role. Interestingly, the IDI flows to the NMS in any specification do not seem to be
significantly affected by the size market of the host countries while the size market of
the source countries is positive and significant. Real unit labor costs and openness show
insignificant coefficients. These results partly contrast with the previous findings (Bevan
and Estrin, 2004), especially with regard to the factor costs, even if the coefficients
of both GDPs are similar. However, it should be noted that the authors in their paper
investigated the determinant of foreign direct investment into transition economies,
including capital flows from non-EU countries, such as US or Japan.
Furthermore, because table 7 shows that there are some issues of collinearity – for instance,
among the institutional variables and the EU membership as well as among
distance and GDP variables – and in order to investigate the impact of collinearity on
the results, the model was estimated by adding one or several regressors at a time; after
doing this, the main findings did not change.
Table 4 illustrates the results related to the Baltic countries (Estonia, Latvia, and Lithuania).
The recent history of these countries is similar in political, social and economic
terms, which allows us to consider them a sub-group of the NMS, and to proceed with a
separate analysis. By focusing our attention to these young democracies, it is possible to
see that, once again, the coefficients of all the explanatory variables have the expected
sign, with the exception of the investment freedom index because of collinearity. The
EU membership dummies for 2004 and 2007 are significant at 1% and have notable
magnitude. Hence, the EU accession was vital in attracting IDI for this subgroup of
countries, too. Source GDP and openness show significant coefficients at 10% (.046
and .009, respectively). On the contrary, the Baltic dummy is negative (-.028 and -.027,
respectively in column 1 and 2), therefore suggesting that being a Baltic country could
negatively affect the investment flows. Because the distance is statistically insignificant,
this impact should not be attributed to geographic reasons. In fact, it should be noted
that among the NMS, these economies were the only ex-socialist countries which had
once been a part of the USSR. In comparison to other NMS, these disadvantageous
initial conditions can be interpreted as more pronounced political barriers to investment
from the rest of Europe.
Table 5 restricts the analysis to the Balkan countries (Bulgaria, Romania, Slovenia and
Hungary). Once again, the EU membership dummies are the main determinants of the
IDI flows with coefficients significant at 1% while the coefficient of the source GDP is
significant at 10% (.047 and .046, respectively in column 1 and 2) and the distance is
negative but insignificant. More importantly, the coefficient of the Balkans dummy is
close to zero (-.006 and -.007 in column 1 and 2, respectively). Thus, the analysis re-
Volume 15, Issue 1, 2015
23
stricted to the Balkan countries that joined the EU demonstrates the inexistence of a
negative Balkans effect in attracting IDI. This finding is remarkable because previous
research identified a persistent negative Balkans effect for non-EU Balkan countries
(Estrin and Uvalic, 2013).
The study restricted to the countries of Central Europe (the Czech Republic, Slovakia
and Poland) confirmed the results of the main model (Table 6). Even for these economies
the EU membership is the main determinant of the IDI flows with coefficients of
notable magnitude and significant at 1%. Source GDP and openness have coefficients
significant at 10%, whereas the distance is insignificant. In particular, unlike the analysis
focused to the other subgroups, the Central dummy is positive (.016 and .015, respectively
in column 1 and 2), which suggests that even when market size, distance,
factor costs, institutional quality and EU membership are taken into account, countries
of Central Europe receive more IDI in comparison to the Baltic and Balkans ones. As
has been mentioned, the reason could be that the Central NMS attract substantial
amount of IDI in manufacturing sector. This sector usually needs more investment
compared to the services sector. Moreover, it could better support industrial restructuring,
structural changes and employment creation (Estrin and Uvalic, 2013). However,
the Baltic and Balkan countries receive substantial amount of IDI in the services sector.
This could help explain why the Central NMS received more IDI flows than the other
two subgroups of transition countries. However, because all the subgroup dummy variables
show insignificant coefficients, this interpretation should be taken with caution.
Vice versa, the coefficient of zero related to the Balkans dummy demonstrates that being
part of the EU eliminated any negative Balkans effect in attracting intra-EU investment.
This is an important finding because previous research (Estrin and Uvalic, 2013)
suggested that a negative Balkans effect in attracting foreign investment still persisted
for non-EU Balkan countries.
Conclusions
Using a gravity approach covering the period 1993–2013 with data extracted from the
Eurostat database, this paper investigates the determinants of the intra-EU direct investment
flows into the NMS. The analysis is focused on 27 EU countries divided into
source economies (the EU-15 countries) and host economies (the NMS that entered the
EU on 2004 and 2007). The paper extends the period analyzed by previous research and
gives evidence that the main determinants of the investment flows between EU-15
countries and NMS are represented by the EU accession. Thus, the membership in EU is
a crucial signal of institutional and macroeconomic stability in order to plan direct investments
into the NMS by other EU enterprises.
Moreover, in contrast to the existing literature that evidences the importance of openness
as a determinant of foreign direct investment, in this paper this variable is not crucial
in determining the IDI flows because is substantially absorbed by the EU membership
that, among other things, requires free circulation of goods and services, customs
union and the adoption of a common EU legislation with regard to trade rules. This
probably helps explain why this variable is positive, as expected and as it was in previous
studies, but statistically insignificant in the main specification. As far as the market
size is concerned, and in line with previous research, the results suggest that GDP of the
REVIEW OF ECONOMIC PERSPECTIVES
24
source country is important in determining the IDI flows. Vice versa, GDP of the host
countries as well as real unit labor cost do not seem to play a key role.
Differently from previous works, the distance is – not surprisingly – statistically insignificant.
The reason is to be seen in the proximity of these countries to one another. On
the other hand, even this variable is captured by the EU membership for the same reasons
related to openness. This is valid also for measures of institutional quality such as
the investment freedom index that is fully absorbed by the EU membership dummies.
A separate analysis on three subgroups of NMS – sharing similar political, social and
economical features – shows that even when market size, distance, institutional quality
and EU membership are taken into account, the Baltic and Balkan countries continue to
receive less IDI than Central ones. With regard to the Baltic countries, the reason is to
be found in the initial conditions of these economies. In fact, when the transition from
socialism to free market economies started, they are the only states which had been a
part of the USSR. This might suggest more pronounced political barriers that during the
transition process limited the investment flows from Western Europe. With regard to the
Central NMS, they probably receive more IDI compared to the Baltic and Balkan countries
because they attract substantial amount of IDI in the manufacturing sector that
needs more capital compared to the services sector in which the Baltic and Balkan countries
receive the most important part of IDI. However, these results should be taken with
caution because the related coefficients are insignificant.
On the other hand, the analysis restricted to the Balkan countries which have joined the
EU does not evidence any negative Balkans effect in attracting IDI. This finding is
important because previous studies show a persistent negative Balkans effect for nonEU
Balkan countries. Thus, the results suggest that the EU membership is crucial for the
EU Balkan countries in order to eliminate any negative Balkans effect in attracting
intra-EU capital flows.
References
BALDWIN R., FRANCOIS J. and PORTES R. (1997). The costs and benefits of eastern
enlargement: The impact on the EU and Central Europe. Economic Policy 24, 125–
170.
BARRELL, R. and PAIN N. (1999). Domestic institutions, agglomerations and foreign
direct investment in Europe. European Economic Review 43, 925–934.
BEVAN A. and ESTRIN, S. (2004). The determinants of foreign direct investment into
European transition economies. Journal of comparative economics 32 775–787.
BEVAN A., ESTRIN, S. and MEYER K. (2004). Foreign investment location and institutional
development in transition economies. International Business Review 1, 43–64.
BOX G.E.P. and PIERCE D.A. (1970), Distribution of residual autocorrelation in autoregressive-integrated
moving average time series models, Journal of the American Statistical
Association, 65, 1509-1526.
BLONINGEN B. A. (2005). ‘A review of the empirical literature on FDI determinants’,
Atlantic Economic Journal, 33, pp. 383-403.
Volume 15, Issue 1, 2015
25
BRADA J.C., ALI M. KUTAN and TANER M. YIGIT (2006). ‘The effects of transition
and political instability on foreign direct investment inflows: Central Europe and
the Balkans’, Economics of Transition, 14 (4), pp. 649-680.
BRAINARD, S. L. (1997). An empirical assessment of the proximity-concentration
trade-off between multinational sales and trade. American Economic Review 87, 520–
544.
BRENTON P., DI MAURO F. and LÜCKE M. (1999). Economic integration and FDI:
An empirical analysis of foreign investment in the EU and in Central and Eastern Europe.
Empirica 26, 95–121.
BREUSCH T.S. and GOFFREY L. (1981), A review of recent work on testing for autocorrelation
in dynamic simultaneous models, Macroeconomic Analysis, ed. by Currie D.,
Nobay R. and Peel London D., Cron-Helm.
BREUSCH T.S. and PAGAN A.R. (1979). A Simple Test for Heteroskedasticity and
Random Coefficient Variation, Econometrica 47 (5), 1287–1294.
BUCH C., KOKTA R. and PIAZOLO D. (2003). Does the East get what would otherwise
flow to the South? FDI diversion in Europe. Journal of Comparative Economics 31,
94–109.
COOK R. D. and WEISBERG S. (1982). Residuals and influence in regression. Chapman
and Hall, New York-London.
CUMBY R. E. and HUIZINGA J. (1990), Testing the autocorrelation structure of disturbances
in ordinary least squares and instrumental variable regressions NBER Technical
Working Paper 90.
DEMEKAS, D.G., BALAZS H., RIBAKOVA E. and YI WU (2005). ‘Foreign Direct
Investment in Southeastern Europe: How (and how much) can policies help?’, IMF
Working Paper 05/110
DIKOVA D. and VAN WITTELOOSTUIJN A. (2007). ‘Foreign direct investment
mode choice: entry and establishment modes in transition economies’, Journal of International
Business Studies, 38, pp. 1013–1033.
DI MAURO F. (1999). The effects of economic integration on FDI flows: an empirical
analysis and comparison with trade. Document paper No. 134. Center for European
Policy Studies, Brussels.
DRISCOLL J.C. and KRAAY, A.C. (1998). Consistent Covariance Matrix Estimation
with Spatially Dependent Panel Data. The Review of Economics and Statistics, 80(4),
549–560.
DURBIN J. (1970). Testing for serial correlation in least squares regression models
when some of the regressors are lagged dependent variables, Econometrics, 21, 333-55.
ESTRIN S. and UVALIC M. (2013). Foreign direct investment into transition economies:
Are the Balkans different? LSE ‘Europe in Question’ Discussion Paper Series.
ESTRIN, S., XAVIER R. and BRADA J. (2000). Foreign Direct Investment in Central
Eastern Europe: Case Studies of Firms in Transition, Armonk and London: M. E. Sharpe.
REVIEW OF ECONOMIC PERSPECTIVES
26
GARIBALDI, P., MORA, N., SAHAY, R., AND ZETTELMEYER, J. (2001), What
Moves Capital to Transition Economies?, IMF Staff Papers, Vol. 48, Special Issue.
HUNYA G. (2011). ‘Diverging patterns of FDI recovery’, in WIIW Database on 2011
Foreign Direct Investment in Central, East and Southeast Europe, Vienna: The Vienna
Institute for International Economic Studies.
HUNYA, G. (2012). ‘Short-lived recovery’, in WIIW Database on 2012 Foreign Direct
Investment in Central, East and Southeast Europe, Vienna: The Vienna Institute for
International Economic Studies.
JANICKI, P.H. and WUNNAVA V.P. (2004). ‘Determinants of foreign direct investment:
empirical evidence from EU accession candidates’, Applied Economics, 36, pp.
505-509.
KALOTAY K. (2010). ‘Patterns of inward FDI in economies in transition’, Eastern
Journal of European Studies, 1 (2), pp. 55-76.
KEKIC L. (2005). ‘Foreign direct investment in the Balkans: recent trends and prospects,
Journal of Southeast European and Black Sea Studies, 5.
LANKES H.K. and VENABLES A.J. (1996). ‘Foreign direct investment in economic
transition: the changing pattern of investments’, Economics of Transition, 4 (2), pp.
331–347.
MAYHEW A. (1998). Recreating Europe. Cambridge Univ. Press, Cambridge.
RESMINI L. (2001). The determinants of foreign direct investment into the CEECs:
new evidence from sectoral patterns. Economics of Transition 8, 665–689.
UNCTAD. World Investment Report 2011, available on-line.
UVALIC M. (2003). ‘Economic Transition in Southeast Europe", Southeast European
and Black Sea Studies, 3 (1), pp. 63-80.
WHITE H. (1980), A Heteroskedasticity-Consistent Covariance Matrix Estimator and
Direct Test for Heteroskedasticity, Econometrica, 48(4 – May), 817-838.
Volume 15, Issue 1, 2015
27
Appendix 1: Tables and Figures
Table 1 Annual IDI Flows to the New Member States
1993 1994 1995 1996 1997 1998 1999 2000 2001 2002
Bulgaria 10 112 10 70 91 193 127 509 187 1014
Cyprus 5 28 1 42 -90 126 -206 73 -224 211
Czech Republic 854 1210 2548 1582 2010 2298 1620 2876 2709 2024
Estonia 5 -3 11 20 47 77 313 213 246 188
Hungary 1335 928 3148 1871 1859 3332 -110 2892 1870 149
Latvia 3 12 26 19 48 71 94 55 85 -22
Lithuania 5 7 27 16 31 99 235 125 175 198
Malta 18 12 5 300 36 -45 103 8 63 23
Poland 749 906 1472 3455 3325 5945 6913 8230 5985 1865
Romania 26 26 99 137 445 530 523 638 770 537
Slovakia 90 182 123 399 331 437 456 1460 1033 3899
Slovenia 65 38 121 98 208 203 258 234 429 1063
Source: Author’s elaborations based on Eurostat data. Values in millions of euros.
Table 1 Contiuned
2003 2004 2005 2006 2007 2008 2009 2010 2011 2012
Bulgaria 113 162 1301 846 2173 2827 1671 633 466 170
Cyprus 25 504 2112 2214 6088 5955 43411 1605 1480 30751
Czech Republic -691 1195 3889 4480 5289 4983 2248 4121 1760 3949
Estonia 233 520 2510 605 448 1346 1177 891 -2287 232
Hungary 2953 4893 16187 4107 13600 6720 -13557 7710 173 -3403
Latvia -174 235 21 105 319 759 -24 314 906 175
Lithuania 233 228 138 258 132 871 101 466 532 216
Malta 522 725 4213 12174 1866 3469 4570 492 -4543 -3629
Poland 1663 6827 3485 8451 9529 7667 5760 6818 8781 5278
Romania 558 2744 2002 8669 4827 10756 1758 1913 1113 2012
Slovakia -1052 1017 1422 1763 1544 2117 1682 1240 948 -2914
Slovenia 643 269 575 507 760 2859 -35 169 532 -678
Source: Author’s elaborations based on Eurostat data. Values stated in millions of €.
REVIEW OF ECONOMIC PERSPECTIVES
28
Table 2 IDI Flows by Source Country (2012)
Country of destination Top 3 source countries
Bulgaria Luxembourg – Netherlands – Denmark
Cyprus Netherlands – Belgium – France
Czech Republic Denmark – Ireland – Netherlands
Estonia Sweden – Ireland – Netherlands
Hungary Netherlands – Spain – Finland
Latvia Sweden – UK – Ireland
Lithuania Germany – Sweden – Spain
Malta Spain – UK – Luxembourg
Poland Spain – Netherlands – Denmark
Romania Denmark – Spain – Netherlands
Slovakia Netherlands – Spain – Luxembourg
Slovenia Belgium – Denmark - Luxembourg
Source: Eurostat.
Volume 15, Issue 1, 2015
29
Figure 1 Foreign Direct Investment by Economic Activity, 2011
Clockwise from the top: Agriculture (blue), Mining (red), Manufacturing (green) and Services (purple)
Bulgaria Czech Republic
Estonia Latvia
Lithuania Hungary
REVIEW OF ECONOMIC PERSPECTIVES
30
Poland Romania
Slovenia Slovakia
Source: Author’s elaborations based on data provided by the Eurostat database. NACE R2. Sectors:
Agriculture, forestry and fishing; Mining and quarrying; Manufacturing; Services. Partner:
All countries of the world.
Volume 15, Issue 1, 2015
31
Appendix 2: Empirical estimates
Table 3 Determinants of Intra-EU Capital Flows Into the New Member States
The estimated equation is:
݈݊‫ܫܦܫ‬௜௜௧ ൌ ߚ଴݈݊‫ܲܦܩ‬௜௧ ൅ ߚଵ݈݊‫ܲܦܩ‬௝௧ െ ߚଶ݈݊݀݅‫݁ܿ݊ܽݐݏ‬௜௝ ൅ ߚଷ݈݊‫ݏݏ݁݊݊݁݌݋‬௝௧ െ ߚସ݈ܴܷ݊‫ܥܮ‬௝௧
൅ ߚହ‫4002݌݄݅ݏݎܾ݁݉݁ܯ‬௝ ൅ ߚ଺‫7002݌݄݅ݏݎܾ݁݉݁ܯ‬௝ ൅ ߚ଻‫ݏ݊݋݅ݐݑݐ݅ݐݏ݊ܫ‬௝ ൅ ߝ௜௝௧
Dependent variable: lnIDI (1) (2) (3) (4)
lnGDPi .014 (.044) .019 (.030) .003 (.026) .005 (.025)
lnGDPj .048* (.023) .050** (.022) .047* (.023) .047* (.023)
lndistance -.018 (.025) -.006 (.027) -.021 (.024) -.022 (.024)
lnopenness .007 (.007) .007 (.007) .009 (.005) .009 (.005)
lnRULC -.000 (.006) -.000 (.005) -.002 (.005) -.002 (.005)
Membership 2004 .208**(.070) .253***(.060) . 258***(.066)
Membership 2007 .424*** (.052) . 424***(.049)
Investment freedom -.044 (137)
Constant 3.67*** (.178) 3.46*** (.197) 3.56*** (.192) 3.76***(.569)
Obs. 2506 2506 2506 2506
R2 .002 .014 .024 .024
(i) In the estimated empirical model, lnIDIiit denotes intra-EU direct investment from an EU-15
country j into a NMS-12 i at time t; (ii) lnGDPit denotes GDP of the source country i at time t; (iii)
lnGDPjt denotes GDP of a host country j at time t; (iv) lndistanceij denotes the distance between
the source and host capitals; (v)lnopennessijt denotes the openness of the host country j at time t;
(vi) lnRULCjt denotes the real unit labour cost of host country j at time t; (vii) Enlargementj denotes
the EU membership dummy; (vii) Institutions. The estimation sample contains 27 European
countries over the period 1993-2013. The model was estimated with a lag of one period in the
independent variables, with the exception of the constant. Cluster-robust standard errors in parentheses.
*denotes significant at 10%; ** at 5%; *** at 1%.
REVIEW OF ECONOMIC PERSPECTIVES
32
Table 4 Determinants of Intra-EU Capital Flows Into the Baltic Countries
The estimated equation is:
݈݊‫ܫܦܫ‬௜௜௧ ൌ ߚ଴݈݊‫ܲܦܩ‬௜௧ ൅ ߚଵ݈݊‫ܲܦܩ‬௝௧ െ ߚଶ݈݊݀݅‫݁ܿ݊ܽݐݏ‬௜௝ ൅ ߚଷ݈݊‫ݏݏ݁݊݊݁݌݋‬௝௧ െ ߚସ݈ܴܷ݊‫ܥܮ‬௝௧
൅ ߚହ‫4002݌݄݅ݏݎܾ݁݉݁ܯ‬௝ ൅ ߚ଺‫7002݌݄݅ݏݎܾ݁݉݁ܯ‬௝ ൅ ߚ଻‫ݏ݊݋݅ݐݑݐ݅ݐݏ݊ܫ‬௝ ൅ ߝ௜௝௧
Dependent variable: lnIDI (1) (2)
lnGDPi .003 (.026) .003 (.025)
lnGDPj .046* (.023) .046* (.023)
Lndistance -.027 (.023) -.027 (.023)
Lnopenness .009* (.004) .009* (.004)
lnRULC -.002 (.005) -.002 (.005)
Membership 2004 .257***(.059) .258*** (.065)
Membership 2007 .412*** (.055) .412*** (.056)
Baltics dummy -.028 (.021) -.027 (.025)
Investment freedom -.152 (.169)
Constant 3.62***(.187) 2.90***(.624)
Obs. 2506 2506
R2 .025 .025
(i) In the estimated empirical model, lnIDIiit denotes intra-EU direct investment from an EU-15
country j into a NMS-12 i at time t; (ii) lnGDPit denotes GDP of the source country i at time t; (iii)
lnGDPjt denotes GDP of a host country j at time t; (iv) lndistanceij denotes the distance between
the source and host capitals; (v)lnopennessijt denotes the openness of the host country j at time t;
(vi) lnRULCjt denotes the real unit labour cost of host country j at time t; (vii) Enlargementj denotes
the EU membership dummy; (vii) Institutions. The estimation sample contains 27 European
countries over the period 1993-2013. The model was estimated with a lag of one period in the
independent variables, with the exception of the constant. Cluster-robust standard errors in parentheses.
*denotes significant at 10%; ** at 5%; *** at 1%.
Volume 15, Issue 1, 2015
33
Table 5 Determinants of Intra-EU Capital Flows Into the Balkan Countries
The estimated equation is:
݈݊‫ܫܦܫ‬௜௜௧ ൌ ߚ଴݈݊‫ܲܦܩ‬௜௧ ൅ ߚଵ݈݊‫ܲܦܩ‬௝௧ െ ߚଶ݈݊݀݅‫݁ܿ݊ܽݐݏ‬௜௝ ൅ ߚଷ݈݊‫ݏݏ݁݊݊݁݌݋‬௝௧ െ ߚସ݈ܴܷ݊‫ܥܮ‬௝௧
൅ ߚହ‫4002݌݄݅ݏݎܾ݁݉݁ܯ‬௝ ൅ ߚ଺‫7002݌݄݅ݏݎܾ݁݉݁ܯ‬௝ ൅ ߚ଻‫ݏ݊݋݅ݐݑݐ݅ݐݏ݊ܫ‬௝ ൅ ߝ௜௝௧
Dependent variable: lnIDI (1) (2)
lnGDPi .003 (.026) . 005 (.025)
lnGDPj .047* (.024) .046* (.023)
lndistance -.023 (.026) -.025 (.026)
lnopenness . 009 (.005) . 009 (.005)
lnRULC -.002 (.004) -.002 (.005)
Membership 2004 .252*** (.060) .256*** (.066)
Membership 2007 .428*** (.051) .429*** (.049)
Balkans dummy -.006 (.014) -.007 (.013)
Investment freedom -.048 (.139)
Constant 3.58***(.213) 3.79***(.578)
Obs. 2506 2506
R2 .024 .024
(i) In the estimated empirical model, lnIDIiit denotes intra-EU direct investment from an EU-15
country j into a NMS-12 i at time t; (ii) lnGDPit denotes GDP of the source country i at time t; (iii)
lnGDPjt denotes GDP of a host country j at time t; (iv) lndistanceij denotes the distance between
the source and host capitals; (v)lnopennessijt denotes the openness of the host country j at time t;
(vi) lnRULCjt denotes the real unit labour cost of host country j at time t; (vii) Enlargementj denotes
the EU membership dummy; (vii) Institutions. The estimation sample contains 27 European
countries over the period 1993-2013. The model was estimated with a lag of one period in the
independent variables, with the exception of the constant. Cluster-robust standard errors in parentheses.
*denotes significant at 10%; ** at 5%; *** at 1%.
REVIEW OF ECONOMIC PERSPECTIVES
34
Table 6 Determinants of Intra-EU Capital Flows Into the Countries of Central Europe
The estimated equation is:
݈݊‫ܫܦܫ‬௜௜௧ ൌ ߚ଴݈݊‫ܲܦܩ‬௜௧ ൅ ߚଵ݈݊‫ܲܦܩ‬௝௧ െ ߚଶ݈݊݀݅‫݁ܿ݊ܽݐݏ‬௜௝ ൅ ߚଷ݈݊‫ݏݏ݁݊݊݁݌݋‬௝௧ െ ߚସ݈ܴܷ݊‫ܥܮ‬௝௧
൅ ߚହ‫4002݌݄݅ݏݎܾ݁݉݁ܯ‬௝ ൅ ߚ଺‫7002݌݄݅ݏݎܾ݁݉݁ܯ‬௝ ൅ ߚ଻‫ݏ݊݋݅ݐݑݐ݅ݐݏ݊ܫ‬௝ ൅ ߝ௜௝௧
Dependent variable: lnIDI (1) (2)
lnGDPi .007 (.027) .007 (.026)
lnGDPj .047* (.023) .047* (.023)
lndistance -.019 (.026) -.020 (.026)
lnopenness .009* (.005) .009* (.005)
lnRULC -.002 (.005) -.002 (.005)
Membership 2004 .251*** (.060) .254***(.065)
Membership 2007 .429*** (.054) .428*** (.051)
Central dummy .016 (.026) .015 (.025)
Investment freedom -.024 (.133)
Constant 3.53*** (.218) 2.82***(.644)
Obs. 2506 2506
R2 .024 .024
(i) In the estimated empirical model, lnIDIiit denotes intra-EU direct investment from an EU-15
country j into a NMS-12 i at time t; (ii) lnGDPit denotes GDP of the source country i at time t; (iii)
lnGDPjt denotes GDP of a host country j at time t; (iv) lndistanceij denotes the distance between
the source and host capitals; (v)lnopennessijt denotes the openness of the host country j at time t;
(vi) lnRULCjt denotes the real unit labour cost of host country j at time t; (vii) Enlargementj denotes
the EU membership dummy; (vii) Institutions. The estimation sample contains 27 European
countries over the period 1993-2013. The model was estimated with a lag of one period in the
independent variables, with the exception of the constant. Cluster-robust standard errors in parentheses.
*denotes significant at 10%; ** at 5%; *** at 1%.
Table 7 Correlation Matrix
lnGDPi lnGDPj lndistance lnRULC lnopen lnvestfree Enlarg. Enlarg2.
lnGDPi 1
lnGDPj 0.0000 1
lndistance -0.1201 -0.2524 1
lnRULC 0.0707 -0.0000 0.0548 1
lnopen 0.0856 0.0000 0.0718 0.0425 1
Investfree 0.0056 0.0000 -0.0832 -0.0752 -0.0773 1
Enlargement -0.0074 0.0000 -0.0637 -0.1009 -0.0485 0.2694 1
Enlargement2 -0.0015 -0.0000 0.0955 0.1115 0.0088 -0.0592 -0.2719 1