Skip to main content

Projections of potential flows to the enlarging EU from Ukraine, Croatia and other Eastern neighbors


This study evaluates potential migration flows to the European Union from its Eastern neighbors and Croatia. We perform out-of-sample forecasts to time series cross-sectional data about post-enlargement migration flows following the EU’s 2004 enlargement. We consider two baseline policy scenarios, with and without accession of sending countries to the EU. Our results show that migration flows are driven by migration costs and economic conditions, but the largest effects accrue to policy variables. In terms of the predicted flows: (i) we can expect modest migration flows in the case of no liberalization of labor markets and only moderately increased migration flows under liberalization; (ii) after an initial increase following liberalization, migration flows will subside to a long run steady state; (iii) Ukraine will send the most migrants; and (iv) the largest inflows in absolute terms are predicted for Germany, Italy and Austria, whereas Ireland, Denmark, Finland and again Austria are the main receiving countries relative to their population.

1. Introduction

The 2013 expansion of the European Union (EU) to include Croatia as its 28th Member State marks the latest move in the process of the EU’s eastern enlargement1. EU expansion is part of a broader process of intensified cooperation with the EU’s Eastern neighbors. This includes countries that have already obtained candidate status, such as Macedonia, Montenegro and Serbia, as well as countries with which the EU initiated a program of intensified cooperation called the Eastern Partnership (EaP) in 20092. The EaP consists of six post-Soviet states, namely Ukraine, Belarus, Moldova, Azerbaijan, Armenia and Georgia, as well as the EU, and is meant to provide an institutionalized forum for the discussion of political and economic topics of joint relevance for all partners. It aims at providing the groundwork for an Association Agreement between the EU and the Eastern partners, which should eventually lead to the establishment of a free-trade zone comprising the 27 EU Member States and the six Eastern partners. In the long run, this might also result in the future membership of these countries.

The expansion of the European Union and the prospect of extending free mobility to workers from the EU’s Eastern neighborhood pose the question of the expected scale of its effect on east–west mobility. The size of the potential sending populations and the economic discrepancies indicate the existence of nontrivial migration potential. According to the IMF’s World Economic Outlook Database (September 2011), Macedonia, Montenegro and Serbia have a total population of slightly more than 10 million, and the six EaP partner countries exhibited a population of almost 76 million people in 2009 (i.e., around 18 per cent of the EU27 total), of which approximately 46 million live in Ukraine. The average GDP per capita (pc) in purchasing-power-parities (PPP) in these countries amounted to slightly more than 9,260 US-$ in the same year. However, there is quite a large heterogeneity within this country group. At the lower end of the distribution, Moldova displays a GDP pc in PPP of around 2,860 US-$, whereas Croatia forms the upper end with around 17,800 US-$, followed by Belarus with slightly more than 12,700 US-$. For comparison, the average GDP pc in PPP of the EU in 2009 amounts to almost 29,700 US-$, and in the Euro-area, it amounts to more than 31,800 US-$. However, also within the EU considerable heterogeneity hides behind these averages.

In the past, EU enlargements entailed controversial discussions with respect to the potential consequences of extending the free movement of labor regulation to the new Member States. This was especially the case in the context of the enlargements towards Central and Eastern Europe in 2004 and 2007, respectively. Consequently, existing Member States implemented different policies towards workers from the accession countries (for a synoptic overview, see the European Commission (2008), p. 11). Whereas the majority of EU-15-countries fully or partially restricted mobility for a transitional period of some years, Ireland, Sweden and the UK allowed free access for workers from countries of the 2004 enlargement round from the outset. By contrast, only Finland and Sweden fully opened their labor markets for Bulgarian and Romanian nationals. In particular, the enlargement towards the eight Central and Eastern European countries (EU-8) in 2004 induced migratory movements from east to west, which from some observers were characterized as being “spectacular” (Kaczmarczyk and Okolski 2008), with the different transitional arrangements displaying substantial effects on observable flows.

Against this background, this paper aims at assessing potential migration flows to the EU from the countries with which the EU is intensifying cooperation – recent entrant Croatia, several candidate countries and members of the EaP program – by utilizing the experiences of the enlargement wave of 2004. In essence, we venture to answer the question of how much (additional) immigration should be expected if EaP countries were allowed to enter the EU and if free movement of workers regulations were extended to them. To this end, we use a well-established model to estimate the determinants of immigration from the accession countries to the EU-15. This model allows for the distinguishing between short- and long-term factors impinging upon observable migration flows. The long-run coefficients are subsequently used to forecast the immigration potential from EaP-countries under different policy scenarios or transitional arrangements. Since this is a double extrapolation exercise – over time and across space – we have to invoke some identification assumptions that must hold to ensure that the forecasts are valid. Whereas the EaP-countries differ from the EU-8 countries in some aspects, the majority of the EU-8 countries share a common history with the EaP-countries in that they are rather young market economies which all underwent a deep economic and political transformation process during the 1990s. Hence, there are reasons to believe that EaP-countries are comparable to EU-8 countries in many social, economic and political aspects. This comparability is of course understood within pre- and post-treatment periods: EU-8 before EU accession compares to EaP in the beginning of 2010s, and EU-8 after EU accession compares to EaP after hypothetical EU accession.

Our empirical results suggest that while economic and demographic variables matter, migration flows are mainly driven by policy variables. From the policy perspective, our key results are that the migration potential from the studied sending countries is modest, the liberalization of migrants’ access to receiving labor markets increases migration flows, albeit only temporarily, and Ukraine will remain the main source country, whereas Germany, Italy and Austria are expected to receive most of these migrants in absolute terms. Relative to the sending countries’ population in 2010, Ireland, Denmark, Finland and again Austria are the main receiving countries,

The rest of the paper is organized as follows. In the next section, we provide some stylized facts and a brief overview of the literature regarding migration within the EU. Section 3 then describes the theoretical model together with its empirical specification and the utilized data before estimation results are presented in section 4. In section 5, we provide different forecasting scenarios, and section 6 concludes.

2. Migration within the EU – stylized facts and review of literature

Despite the free movement of workers regulation, migration streams within the EU were rather low between the 1980s and the beginning of the 2000s. Against the background of rather large and persistent regional differentials in wages and employment prospects, one might even argue that within-EU migration activities were too low during this period (see Fertig, Michael and Christoph M. Schmidt 2002). Even the enlargements of the EU towards Southern Europe (Greece in 1981 as well as Spain and Portugal in 1986) did not induce any remarkable changes in observed migration flows (see, e.g., Bover and Pilar Velilla 2001). To illustrate this for the most recent past, Figure 1 provides intra-EU migration flows for the EU-12 countries3 (without Greece, due to missing data).

Figure 1
figure 1

Net-Migration of EU-12 countries to selected Member States-1998-2010. Note: Nationals from the EU-12 without Greece. Source: Eurostat database, May 2012; own calculations.

From this figure, it becomes transparent that net-migration (i.e., inflows minus outflows) from other core Member States of the EU is negative for the case of Germany throughout all years; that is, emigration of nationals from the EU-11 countries exceeded immigration from these countries in every year. In all Member States, net-migration was positive and varied between a few hundred and more than 40,000 persons per year. However, it was less than 15,000 persons per year in the majority of countries and years. This rather low net-immigration or – in the case of Germany, even substantial net-emigration – stands in stark contrast to the persistent relative income differentials (i.e., ratios of per-capita-incomes in purchasing power parities) between the countries of the EU-12, as illustrated in Figure 2, which indicates that per-capita-incomes in Germany, the Netherlands and UK are substantially and persistently higher than in Portugal and Spain4. Furthermore, we do not even observe any convergence between the considered country pairs. By contrast, the PCI-ratios start to increase after 2004, having been almost constant before. This is especially evident for the case of Portugal. Against this background, the potential average returns of migration within the EU seem to be high and increasing over time, although migration activities remain constantly low. Hence, the (real or perceived) costs of moving to another European Member State seem to be extremely high, despite the free movement of labor regulation.

Figure 2
figure 2

Ratio of per-capita-income in purchasing power parties for selected countries. Source: World Economic Outlook database of the IMF, September 2011; own calculations.

However, the enlargement of the EU-15 towards the eight Central and Eastern European countries plus Cyprus and Malta in 2004, as well as the enlargement of 2007, induced (Bulgaria and Romania) induced considerable migratory movements from east to west within the enlarged EU. In this context, the transitional arrangements with respect to the free movement of labor regulation evidently displayed substantial effects on observable migration flows. According to European Commission (2008, p. 115), the stock of resident foreign nationals from the EU-10 accession countries in the EU-15 countries more than doubled from 924,000 persons in 2003 to 2,016,000 in 2007. Interestingly, the same observation holds for the 2007 accession countries of Bulgaria and Romania. Despite entering the EU as recently as 2007, the number of residents from these two countries in other EU-15 Member States increased from 691,000 in 2003 to 1,331,000 persons in 2006 and further to 1,617,000 people in 2007. Hence, the stock of Bulgarian and Romanian nationals in the EU-15 almost doubled even prior to their EU-accession. Figure 3 illustrates this for the 2004 EU-8 accession countries from Central and Eastern Europe as well as the EU-2 (Bulgaria and Romania) accession countries of 2007.

Figure 3
figure 3

Stock of foreign residents from EU-8 and EU-2 in EU-15. Source: Holland et al. (2011); own calculations.

It becomes apparent from this figure that the stock of EU-8 nationals residing in the EU-15 continuously increased from 1998 to 2008. However, the slope of this increase became considerably steeper after their accession in 2004, with the stock of immigrants from the EU-8 countries increasing by almost 23 percent from 2004 to 2005. The growth rate one year later was slightly less than 32 percent, and between 2006 and 2007 it was around 25 percent. However, growth rates declined considerably after 2007, which might be the result of the financial and economic crisis in Europe and/or an indication for satiation. In contrast to the pattern for the EU-8 countries, the stock of Bulgarian and Romanian nationals residing in one of the EU-15 Member States displays a remarkable increase before 2007, with average annual growth rates of slightly less than 25 percent. The post-accession development was very similar to this pattern, with average annual growth rates of around 24 percent.

As already emphasized by Brücker et al. (2009), post-accession migratory movements not only differ in size but also in their regional distribution with respect to the main destination countries. Table 1 provides the shares of total net-inflows into the EU-15 for each Member State. This table clearly indicates that Germany, the UK and Spain constituted the main destination countries for immigrants from the EU-8 in the pre-accession era. Indeed, after 2004, the UK alone received almost half of all immigrants from the EU-8. Germany was still among the main receiving countries during this period, whereas Spain was replaced by Ireland. However, this picture is considerably different for the EU-2 countries, as the main destinations for nationals from Bulgaria and Romania used to be Spain and Italy prior to accession, and Italy and Spain thereafter. However, a few EU-15 Member States experienced a substantial increase in net-immigration from these accession countries, with Germany, Austria and the UK displaying the highest rates of increase.

Table 1 Regional distribution of net-inflows to EU-15 in percentages

From the sending countries’ perspective, Poland, Bulgaria and Romania experienced the largest net-outflow of inhabitants in the direction of the EU-15. In 2009, almost 1.5 million Polish citizens resided in one of the EU-15 Member States, whereas the corresponding numbers for Bulgaria and Romania are slightly less than 0.5 million and more than two million individuals, respectively. However, Poland and Romania are also the two countries with the largest populations, with Bulgaria ranked fifth, following the Czech Republic and Hungary. Hence, it is unsurprising that a large number of Polish and Romanian nationals live outside their countries. Consequently, it seems to be more sensible to consider the relative outflows – with Figure 4 providing the population share of EU-8- and EU-2-nationals living in one of the EU-15 Member States – thus accounting for the population size in the sending country. The figure indicates that Romanian citizens display the largest group of non-national residents in the EU-15 among the accession countries in relative terms. However, and in contrast to absolute numbers, the Baltic States are now also found in the top group, whereas the share of Polish nationals in the EU-15 is relatively small relative to the population at home. Kahanec, Pytliková and Zimmermann (2015) show that migration flows in an enlarging European Union indeed responded positively to EU enlargement and labor market opening. They also evaluate the effects of economic shocks, arguing that whereas the adverse economic situation in source countries was one of the primary push-factors, short-term fluctuations in the sending countries did not prove to be significant. However, short-term variation in destination countries’ GDP per capita and unemployment rates affected mobility patterns significantly in the expected directions, with weaker economies attracting fewer migrants. This indicates responsiveness of migration flows to the Great Recession.

Figure 4
figure 4

Population share of EU-8 and EU-2 nationals residing in EU-15 countries. Source: Holland et al. (2011); own calculations.

The literature offers some tentative conclusions regarding the economic consequences of these migration flows. Kahanec (2013b) scrutinizes the experiences of the two enlargement waves by assessing actual migration flows and reviewing their effects on the labor markets of receiving and sending countries. The author concludes that the available evidence does not indicate negative effects on the receiving countries’ labor markets or welfare systems. However, sending countries run the risk of skill shortages in certain occupations or sectors, as well as instability of public finances. On the other hand, Elsner (2013a, 2013b) identifies positive effects of out-migration on wages in Lithuania; additionally, outmigration might uncover potential benefits through brain circulation.

Hazans and Kaia Philips (2010) use labor force survey data for the period 2002 to 2007 and several other surveys to compare the profile of Baltic temporary workers abroad before and after EU accession with that of stayers and return migrants. The authors find significant changes in how ethnicity and citizenship affect workers’ mobility. According to the authors’ results, in the first two years after 2004, 11 to 13 percent of migrants from Lithuania and Estonia and 15 percent of their Latvian counterparts were unemployed in the home country in the previous year, while around 7 percent were either students or pupils. These proportions exceed those observed among stayers by a factor of three to four, indicating that work abroad has represented an important coping strategy for the Baltic unemployed or potentially unemployed. Furthermore, Hazans and Kaia Philips (2010) point out that the two-thirds to three-quarters of all migrants had secondary education and that enlargement changed the skill composition of migrants. In the years prior to 2004, Lithuanian migrants had the same skill distribution as stayers, while Latvian and Estonian migrants were on average more educated than stayers. Post-accession migrants from all three countries were significantly less educated than stayers, with this gap tending to increase over time.

Dustmann and Tommaso Frattini (2011) point out that different historical and economic developments resulted in different immigration experiences from different countries of origin. While some European countries have been immigration countries since the 1960s, others emerged as immigration countries around two decades ago. Consequently, European countries presently exhibit very dissimilar immigrant populations with respect to the country of origin, ethnicity and education. Labor market integration of immigrants remains a key policy challenge in Europe, as migrants face significant barriers impeding their access to the labor market or social welfare provisions (Constant et al. 2009; Kahanec et al. 2013a). Human capital gaps are another limiting factor, with some migrant populations exhibiting educational attainment comparable or even exceeding that of the natives, while others are lagging behind (Dustmann and Tommaso Frattini 2011; Kahanec 2013a). Danzer and Dietz (2014) document that immigrant populations from EaP countries have higher educational attainments than natives in the EU and that some seem to attempt to preempt the risk of downskilling by taking training courses before departure. While downskilling into jobs below one’s educational attainment is a widespread problem, temporary migrants tend to compensate for this disadvantage by working longer hours (Kahanec and Shields 2013).

Kahanec and Zimmermann (2010) provide an encompassing account of the early post-accession period, highlighting: (i) the positive role of east–west mobility for allocative efficiency of EU labor markets; (ii) the lack of evidence on negative labor market effects of migration; and (iii) potential benefits, as well as some policy challenges, of brain circulation for the sending countries5.

With respect to the driving forces behind observable international migration flows, the literature suggests a variety of explanations depending on the sample of countries and the time period analyzed. Kim and Cohen (2010) investigate the determinants of international migration flows into 17 Western countries from 230 origin countries during a period ranging from 1950 to 2007. To this end, they regress the logarithm of the number of migrants on a set of demographic, geographic and social explanatory variables. The authors find that demographic and geographic factors are the most important driving forces behind observable inflows. Specifically, the population of origin and destination country, the infant mortality rate of origin and destination, the distance between capitals and the land area of the destination display a significant and substantial impact on immigration activities. Furthermore, a young age structure in the destination was associated with lower inflows, while a young age structure in the origin was associated with higher inflows. By contrast, social and historical determinants proved to be less important. While Borjas (1999), in his seminal article, found some welfare magnet effect on migration flows within the US, the ensuing literature finds rather weak magnet effects in international or European contexts (DeGiorgi and Pellizzari 2009; Pedersen et al. 1990; Giulietti et al. 2013).

A number of studies looking at EaP-migrants in the EU from a policy perspective have appeared recently. Clark and Drinkwater (2014) report that flows from EaP-countries to the UK declined following changes of UK immigration policies. Those remaining in the UK are in many aspects similar to EU-8 migrants, including their relatively high educational attainment, but exhibit lower employment rates and higher occupational status (although not as high as for migrants from the old member states). Biavaschi and Zimmermann (2014), in a study on Germany, document relatively worse labor market outcomes of EaP-migrants vis-à-vis other migrant groups but argue that they possess skills that will be demanded in the German labor market. Marchetti et al. (2014) discuss the Italian experience focusing on migrants from the two EaP-countries that send most migrants to Italy: Moldova and Ukraine. The authors argue that distinct patterns have emerged for both source countries; whereas Ukrainian migrants are mostly mature women with temporary migration plans, Moldovan migrants are younger and with a higher proportion of males, tend to have permanent intentions, and come for family reunification. Farré and Rodríguez-Planas (2014) find that the employment rate of EaP-migrants (mainly Ukrainians) in Spain is similar to that of the natives, mainly because of their higher educational attainment and a quick adjustment that offsets the effects of an employment penalty they face upon arrival. In a similar vein, Duszczyk and Paweł Kaczmarczyk (2013) document the prevalence of women, young, and well educated workers from Ukraine among EaP-migrants in Poland, indicating that their skills tend to complement those of the native labor force. There is little evidence available about migrants from Croatia, Macedonia, Montenegro, or Serbia separately in (the rest of) the EU, but several studies document labor market segmentation between natives and immigrants from the former Yugoslavia in the EU (e.g., Constant and Massey 2005).

Concerning the determinants of migration, Danzer and Dietz (2014) look at the determinants of temporary out-migration from Armenia, Belarus, Georgia, Moldova, and Ukraine in 2006. They find that the incidence of migration to the EU was lower for the less educated and previously unemployed, but higher for those who invested in their human capital prior to departure (including language skills) or were older. Several papers attempt to predict the migration potential of the accession candidates of the previous enlargement rounds (see, e.g., Bauer, Thomas and Klaus F. Zimmermann 1999; Fertig 2001; Fertig and Schmidt 2001; and Orlowski et al. 2000), with the majority of them deriving a rather modest forecast. Zaiceva (2006) provides an encompassing overview of the literature on migration projections with respect to EU-enlargement. To the best of our knowledge, to date there is no study that systematically predicts migration flows from EU’s Eastern neighbors or Croatia to (the rest of) the EU.

3. Theoretical model, empirical specification and data

In order to estimate the structural relationship between migration flows and its determinants, we use an adaption of the model of Hatton (1995) to time series cross-sectional data (for a detailed description, see Fertig (2001). The theoretical model is formulated in terms of individual utility maximization, following the hypothesis that migration is an investment in human capital (Sjaastad 1962). Hence, the individual migration probability depends on the difference in expected utility streams in the country of origin and the destination country minus the costs of migration. Utility streams are assumed to depend on expected income, which is the product of the wage rate and employment probability in each country.

In forming their expectations on utility streams, the model assumes that migrants assign the greatest weight to the most recent past, with this weight declining with time. Thus, the migration decision not only depends on the current difference in utility streams but also on all expected future values. This implies that although the current difference might be negative for some migrants, the net present value of migration might become positive if they were to wait for an additional year.

Furthermore, the model assumes that the employment rates in the destination countries follow a binomial distribution. Hence, the model explicitly accounts for uncertainty in employment prospects, which leads to their greater weight in the destination countries than in the risk-neutral Harris-Todaro model. Finally, in order to estimate the model using aggregate-level migration data, the individual probability concept is approximated by the aggregate migration rate. This implies the assumption that aggregate migration rates reflect the average migration probabilities of all individuals in a specific country of origin.

Putting these pieces together yields the following reduced-form estimation equation of the model (for more details see Hatton 1995; Fertig 2001; and Hatton 2005)

$$ \varDelta {M}_t^{h\to d}={\varepsilon}_h+{\varepsilon}_1\varDelta \ln {\left(\raisebox{1ex}{${w}_d$}\!\left/ \!\raisebox{-1ex}{${w}_h$}\right.\right)}_t+{\varepsilon}_2\varDelta \ln {\left({e}_d\right)}_t+{\varepsilon}_3\varDelta \ln {\left({e}_h\right)}_t $$
$$ +{\varepsilon}_4 \ln {\left(\raisebox{1ex}{${w}_d$}\!\left/ \!\raisebox{-1ex}{${w}_h$}\right.\right)}_{t-1}+{\varepsilon}_5 \ln {\left({e}_d\right)}_{t-1}+{\varepsilon}_6 \ln {\left({e}_h\right)}_{t-1} $$
$$ {\varepsilon}_7{M}_{t-1}^{h\to d}+{\varepsilon}_8MS{T}_{t-1}^{h\to d}+{\varepsilon}_9PO{P}_{t-1}^h+{\varepsilon}_{10}PO{P}_{t-1}^d $$

In this equation, M denotes the migration rate from the country of origin h to the destination country d in year t. The wage rate is denoted by w and employment rates by e. Thus, changes and levels of the economic variables enter the equation separately, providing the possibility to distinguish between short- and long-run determinants of migration flows. Furthermore, the above specification is an extension of the single-destination model in Fertig (2001) to several destination countries. Hence, the model is augmented by the relative wage rate of each single destination country to the rest of the EU-15. This extension should capture the attractiveness of d relative to that of other EU-countries, while the same holds for employment rates.

The stock of migrants (MST) and the population shares of individuals aged 20–40 living in h and d, respectively, enter the equation via migration costs. The model also contains country of origin-specific intercepts ɛ h , which also enter the equation via the modeling of migration costs. Hence, migration costs are, on the one hand, approximated by the stock of migrants from h living in country d, which, from a theoretical perspective, captures network and potential crowding effects. On the other hand, migration costs are meant to be captured by the population shares of individuals aged 20–40 (POP) living in h and d, as well as a sending country-specific term. The latter accounts for country-specific relocation costs (e.g., due to distance) together with differences in the psychological costs of leaving one’s home country (i.e., unobservable sending country-specific migration cost). The population share of individuals aged 20–40 living in h should capture the population group that is typically the least bound to the country due to family ties, and it also similarly exhibits the highest proficiency in foreign languages as well as the highest expected returns to migration due to a relatively long potential working time in the destination country’s labor market. The population share of individuals aged 20–40 living in d approximates the need for immigrants due to demographic factors in the destination country, thus capturing the “openness” of country d towards immigrants or its “willingness to welcome” them simply because they are needed.

Finally, we extent the model by two sets of dummy variables covering policy regimes. The first equals one for all years in which country d allowed workers of country h free access to its labor market, and zero otherwise. The second dummy equals one for all years in which country d partially restricted the access of workers from country h to its labor market, and zero otherwise. Hence, full restriction forms the reference category (see Table A.1 in the appendix for a detailed description of these policy dummies).

The model is estimated by OLS with panel-corrected standard errors using time-series cross-section data for migration from the EU-8 to the EU-15 (for details, see below), following Beck and Katz (1995). The authors demonstrate that this estimation method maintains the advantage of OLS parameter estimates that perform very well in the context of time-series cross-section data and simultaneously avoid the production of inaccurate standard errors, like the OLS- or frequently used FGLS-estimator. Finally, by setting all ∆’s to zero and solving for M, one can derive the long-run steady state relationship for net migration rates. The calculated long-run coefficients are used for the simulations of the expected magnitude of immigration from Croatia and the EaP-countries to the EU-15 under different policy scenarios.

We apply the model to data for migration from the EU-8 to the EU-15 without Luxembourg (EU-14 in what follows) for the period 1999–2009. Within the EU-15, Luxembourg is by far the smallest (in terms of population) and simultaneously also the richest country. Therefore, in order to avoid biased estimation results for the per-capita-income variable, Luxembourg was excluded from the sample of destination countries6. Net-immigration to the EU-14 is calculated as the change in the stock of foreign residents from h in d between t and t-1. The stock of nationals is taken from Holland et al. (2011), while the explanatory variables stem from the IMF’s World Economic Outlook Database (GDP p.c. in PPP) and the statistics database of Eurostat (unemployment and population). The estimation results are summarized in the following section.

4. Estimation results

Table 2 contains the estimation results for three different specifications of our model. In the first specification, we model policy regimes by including the two dummy variables indicating free movement of labor and partial restrictions7. In the second specification, we disentangle the free movement dummy by including a separate dummy variable for each year of free movement. This specification is meant to investigate whether the impact of free movement is higher in the first years of opening up borders, or if it reveals any other time pattern. Finally, the third specification takes into account that sending country-specific migration costs might depend on the policy regime. On the one hand, accession to the EU might open up completely new opportunities with respect to leaving the country, thus reducing migration costs. On the other hand, living in an EU Member State might also advance economic prospects at home due to higher economic integration, hence increasing migration costs. In order to account for this, we add a dummy variable for the post accession period and interact it with the sending country-dummies.

Table 2 Estimation results

In general, our estimation results suggest that economic conditions in the destination countries play an important role in explaining observable migration flows. Both the PCI-ratio between the destination and sending countries as well as employment rates in the EU-8 exhibit a significant impact on net-migration. The higher the PCI-ratio, the higher the observable flows, all other things being equal. The opposite holds for the employment rate in the origin countries. By contrast, the employment rate in the EU-14 seems to be of minor importance. Hence, our results suggest that migrants’ income opportunities in the destination country compared to the home country have a systematic impact on their decision to leave the country.

However, estimation results also clearly indicate that the costs of migration are important. The stock of migrants in the destination country exhibits a statistically significant positive impact on net-migration, which suggests that existing migrant networks in the destination countries help to attract further immigrants. Furthermore, the set of sending country-specific intercepts suggest that migration costs vary by country and react to policy. The results of specification 3 suggest that country-specific migration costs exhibit a largely different pattern before and after the accession of the EU-8 countries. Prior to accession, the Czech Republic, Hungary and Slovenia display significantly and substantially higher migration rates to the EU-14 than Poland, whereas those of all other sending countries do not differ systematically from Poland. However, after accession, we observe significantly higher migration rates (i.e., lower country-specific migration costs) for the Baltic States only, even after controlling for economic conditions and policy regimes. Thus, our results indicate that due to accession, country-specific migration costs increased in all EU-8 countries compared to Poland, except the Baltic States. For the latter, we observe a decrease. This finding suggests that for the accession countries outside the Baltic region, accession to the EU seems to have opened up new opportunities at home (e.g., due to economic integration into a large single market) and hence increased migration costs. Finally, the mean of post-accession country-specific intercepts is around 25 percent higher in specification 3 than in specification 2, in which we do not distinguish between policy regimes.

However, the largest single impact on observable flows is observed for policy indicators. Hence, policy regimes matter more than migration costs and economic conditions. Whereas partial liberalization does not have a significant effect on net-migration, the number of years of free movement creates a significant and quantitatively substantial impact on observable flows. One important result is that this impact follows an inversely u-shaped pattern, i.e., immigration increases in the first years after completely opening up labor markets, reaches its maximum in year four and declines thereafter. Thus, labor market liberalization evidently provides an incentive for nationals of the accession states to leave the country. This incentive is, however, countervailed by increasing migration costs, at least for some countries (see above).

By contrast, short-term variation in economic variables does not seem to greatly matter. The estimated coefficients of the changes in economic indicators are all statistically insignificant, apart from that of the employment rate in the destination countries8. Finally, the significant negative impact of the lagged net-migration rate suggests that immigration to the EU-14 varies around a stable level. Thus, there is no reason to expect immigration to the EU to be ever-increasing in the future.

Naturally, our coefficient estimates are a weighted average of two regimes, namely the migration regime prior to 2004 and thereafter. Table 3 contains separate estimates for both time periods using specification 2 from above (apart from the accession dummy). From this table, it becomes apparent that economic conditions in the “closed border” regime only matter for short-term fluctuations, but not in the long-run. By contrast, the stock of migrants still has a positive impact (probably due to family reunification, facilitation of access to the labor market and institutions, or similar reasons). The observation that there was nontrivial immigration from the EU-8 in the years prior to 2004 when borders were “closed” suggests that net-migration flows can be positive, even in the closed border regime that applies to the countries under scrutiny.

Table 3 Estimation results for pre- and post-enlargement periods

Given that our model allows for distinguishing between short- and long-term factors impinging upon observable migration flows, we have to calculate the long-run coefficients for the final specifications, i.e., specification 3 in Table 2, as well as the pre-accession specification in Table 3. This is achieved in a straightforward manner by setting all ∆s in equation (1) to zero so that the impact of short-run fluctuations is eliminated. These long-run coefficients are then used to forecast the immigration potential from EaP-countries for which no common migration history with the EU is available under different policy scenarios or transitional arrangements. The results of these forecasts are presented in the next section.

5. Forecasting scenarios

Neither Croatia, Macedonia, Montenegro, Serbia or the EaP-countries share a common migration history under different policy scenarios or transitional arrangements with the EU9. Thus, forecasts of expected immigration from these countries are a double extrapolation exercise – over time and across space. This implies that we have to invoke a couple of identification assumptions that must hold a priori to ensure that the forecasts are valid. Clearly the most important assumption is the stability and transferability of the estimated structural relationship between observable flows and its determinants. Hence, we have to assume that the structure that quite accurately describes the relationship between migration from the EU-8 and the EU-14 remains stable over the forecasting horizon and holds for the behavior of future migration from the EaP-countries. This implies that the migration decision of individuals from the EaP-countries must be determined by the same factors as the decision of individuals in our sample countries, at least in the long-run. In addition, we need further assumptions for the development of the exogenous variables in our model (i.e., GDP, employment and population). Finally, we have to invoke an assumption for country-specific migration costs. Clearly, the longer the forecasting period, the more likely that these assumptions are violated.

Specifically, we present the results of two scenarios. The pre-accession scenario describes the status quo situation for all the studied countries (except Croatia, which is already in the EU), and the projections thus produce migration flows that can be expected if there is no policy change10. Next, the accession scenario assumes accession and selective liberalization of EU labor markets (i.e., partial restrictions) until 2014 and full liberalization/free access from 2015 onwards.

  • Pre-accession scenario: Long-run coefficients derived from the estimation results in the left part of Table 3, using medium migration costs (i.e., the mean of the sending country-specific intercepts).

  • Accession scenario: Long-run coefficients derived from specification 3 of Table 2, using low, medium and high sending country-specific migration costs for the time after accession. Low country-specific migration costs are captured by the highest value of the country intercepts, medium migration costs by the mean of all country fixed-effects and high migration costs by the lowest value of the country dummies.

In both scenarios, following Kahanec et al. (2013a), the following assumptions for the exogenous variables in the destination countries (EU-14) are invoked:

  • 2010–2014: European recession, i.e., 0 percent growth of per per-capita incomes together with annual employment growth rates of −0.3 percent for high employment countries, −0.1 percent for the medium group and 0.2 percent for the low employment group11.

  • 2015–2020: EU recovery, i.e., 2 percent annual growth rate of per-capita-incomes together with annual employment growth of 0.15 percent for high employment countries, 0.3 percent for the medium group and 0.6 percent for the low group.

For the sending countries, we took the 2010 population data from the Eurostat population statistics. We assume annual growth rates for per-capita-income presented in Table 4. These growth rate forecasts are based on the IMF World Economic Outlook database and were validated for the EaP countries within the project “Costs and Benefits of Labour Mobility between the EU and the Eastern Partnership Partner Countries,” drawing on the expertise of a network of national experts (Kahanec et al. 2013b). The employment rate forecasts, calculated as 100 – unemployment rate (in %), were also taken from the IMF World Economic Outlook database covering the period up to and including 2017. For the remaining three years (up to and including 2020), employment rates were extrapolated assuming a constant rate of change, calculated as the per annum rate of change from 2012 to 2017. Employment rates were consulted with national experts from the project consortium and, in the case of Belarus, adjusted to better reflect reality (Kahanec et al. 2013a)12. Finally, Libanova (2006) provides population forecasts by age group for Ukraine. Population growth rates from Libanova (2006) were then used for the other EU’s Eastern neighbors (as well as Croatia), as they, in our view and the view of national experts, better reflect the sharply declining fertility rates in the region than the forecasts provided by the United Nations’ World Population Prospects.

Table 4 Assumed annual growth of PCI (in per cent) in EaP-countries, Croatia, Macedonia, Montenegro and Serbia

With respect to the migration forecasts for the EaP-countries, we concentrate on the presentation of results for Ukraine since it is the most important country of this group in quantitative terms. Migration projections for Ukraine are even more interesting given the so-called Euromajdan protests, which started in November 2013, and the political turmoil following them, resulting in displacement of many people. Nevertheless, all the scenarios regarding Ukraine presented in the following rest on the assumption that the political situation is comparable to that in mid-2013. This is because we trust that the current unstable situation is of temporary nature, and we are interested in long-term migration potential. Also, our forecasts should be understood as pertaining to the whole of Ukraine, in its internationally recognized borders. Surely, the violence may result in a changed economic, social and political situation in Ukraine. Given the unpredictable situation as of December 2014, we abstain from speculation on the exact effects of such changes on Ukraine’s migration potential and concentrate on the long-term steady state.

The predictions for Ukraine are confronted with the results for Croatia. Forecasts for the rest of the accession candidates are summarized at the end of this section. Table 5 summarizes the results of the (counterfactual) pre-accession scenario for Croatia, suggesting that there will be net-immigration from Croatia to the different countries of the EU-14, even in the case of closed borders and without accession. However, these inflows will be modest in size and will only total more than 1,000 persons per year for Germany, Italy and Spain.

Table 5 Predicted net-immigration from Croatia to EU-14 – Pre-accession scenario

For the case of Ukraine, Table 6 suggests higher net-immigration in this scenario, given that this sending country is larger in terms of population and less developed with respect to income and employment. However, even for Ukraine, the absolute number of expected migrants only exceeds 100,000 persons for Germany and Italy during the ten years considered.

Table 6 Predicted net-immigration from Ukraine to EU-14 – Pre-accession scenario

Compared to actual net-immigration from Croatia and Ukraine to Germany during 2000–2010, our model produces realistic results. The annual net-inflow from Croatia to Germany in the first decade of the 21st century amounted to around 2,000 persons, while the corresponding number for Ukraine was approximately 6,000 persons. Hence, we slightly under-predict immigration from Croatia, whereas the opposite holds for the case of Ukraine.

Tables 7 and 8 contain the forecasts derived from the accession scenario, with both tables demonstrating that policy and sending country-specific migration costs matter significantly. The importance of policy is demonstrated by the increase in net-migration, especially after allowing free access (2015–2020) for migrant workers; indeed, for some destinations, we even predict positive net-immigration only from Croatia during this period. The importance of migration costs specific to various sending countries is underscored by the variation of projections depending on the assumption about migration costs. For example, forecasts for the case of Ukraine vary between 1.3 million to almost 7 million under various scenarios.

Table 7 Predicted net-immigration from Croatia – Accession scenario
Table 8 Predicted net-immigration from Ukraine – Accession scenario

Table 9 provides a summary of the results for the rest of the countries considered (under the accession scenario medium migration costs post enlargement). From Table 9, it becomes apparent that expected migration flows from these countries are rather modest in size. In absolute numbers, the largest predictions occur for Serbia, Georgia and Armenia, with a little more than 300,000 persons over the entire forecasting period. For Belarus, which is relatively wealthy and exhibits no noteworthy stock of migrants in the EU-14 in 2010, we predict positive net-immigration only to Austria, Denmark, Germany, Ireland, the Netherlands, Sweden and the UK. For the rest of the destination countries, the model predicts no net-immigration13.

Table 9 Summary of forecasts for rest of EaP-countries, Macedonia, Montenegro and Serbia

Figure 5 illustrates total annual inflows to the EU-14 from all countries under study (except Belarus, due to the above-mentioned data problems and resulting negative predictions). Extrapolating the status quo will result in total predicted inflows to the EU-14 countries between 0.1 and 0.2 million migrants per year. Due to the positive impact of the stock of migrants, these annual inflows display a slightly positive trend over time. In the accession scenario, predicted inflows increase to almost 0.3 million migrants per year during the assumed five year period of partial liberalization. After 2015, i.e., if we assume free access to the EU-14’s labor markets, annual inflows amount to around 0.8 million persons per year, although the inflows are expected to go down after the fourth year after liberalization, which is in line with the results from our regression model. In both cases, the majority of them, of course, are predicted to come from Ukraine. Hence, even in the unlikely event of simultaneously liberalizing EU-14’s labor markets to all these countries, there is no reason to expect overwhelming immigration.

Figure 5
figure 5

Total annual inflows to EU-14 for both scenarios.

Table 10 summarizes the predicted inflows by destination for all countries under study, apart from Belarus. Accordingly, the predicted total net-immigration in absolute terms is the highest for Germany, Italy and Austria. Under the accession scenario with medium migration costs (post enlargement), Germany can expect slightly more than 0.6 million immigrants between 2010 and 2020, Italy slightly less than 0.6 million and Austria around 0.4 million. Portugal and Finland are found at the other end of the distribution.

Table 10 Predicted total net-immigration 2010–2020 by destination countries (Croatia, Macedonia, Montenegro, Serbia and all EaP-countries, except Belarus)

However, Finland is among the main receiving countries if predicted inflows are related to the destination countries’ population in 2010, and it is predicted to experience a population growth of around 5.5 percent due to these inflows, which is only exceeded by Ireland and Denmark. Hence, the distribution of net-immigration to the EU-14 in relative terms differs considerably from the distribution in absolute terms. The only destination country that is among the main receiving countries in both dimensions is Austria. By contrast, Germany and Italy can expect relative net-inflows of only around 1 percent.

6. Conclusions

The European Union’s eastern enlargement sparked heated debates in Europe during the 2000s, which continue to date. The discourse revolves around issues such as the effects of migration on wages and employment, the propensity to take up or be attracted by welfare benefits, or social dumping. Given that much of this debate has been politicized, misinformed or based on outright myths, it is important to anchor it in sound analysis based on hard data.

This paper provides an evaluation of the scale of east–west European mobility under two key scenarios – the status quo and the liberalization of access to EU labor markets. Moreover, it also informs the broader debate about the determinants of migration by providing insights into the determinants of east–west migration in Europe following the 2004 enlargement of the European Union.

Using longitudinal data on bilateral flows between the EU-8 and EU-14, we estimate a robust prediction model that exhibits desirable properties. The key result is that while migration costs and economic conditions matter for east–west post-enlargement migration flows, policy variables explain a greater part of the observed variation.

Informed by expert demographic and economic forecasts and assuming two archetypal policy scenarios, we provide out-of-sample projections of migration flows from the Eastern Partnership countries, Croatia, Macedonia, Montenegro and Serbia to the EU15 minus Luxembourg. The predicted migration flows are generally modest, remaining so even under the scenario of liberalized access to receiving labor markets. In fact, the predicted increase due to liberalization appears to be temporary, with the predicted incremental migration flows generally subsiding after several years. Ukraine will remain the country that sends the most migrants, mainly due to its size, while Germany, Italy and Austria will be among the countries receiving the most migrants in absolute terms. Overall, we predict that during 2010–2020, on average, 1.7 percent of populations of the studied countries (except Belarus) will decide to try their fate in the EU-15 minus Luxembourg under the status quo (pre-accession scenario), and 6.7 percent under liberalization (accession scenario). This implies that additional total inflow from all these countries over 2010–2020 corresponds to around only 0.3 percent of the receiving countries’ populations as of 2010 under the pre-accession scenario and 1.5 percent under the accession scenario.

From a policy perspective, a key result is that migration policy frameworks matter, although the effect of liberalization of migrants’ access to receiving labor markets is predicted to be temporary. A further implication is that a non-harmonized timing of liberalization across the receiving countries, as was the case for prior eastern enlargements, may divert migration flows and concentrate them in some receiving countries more than others. In this regard, our projections should be seen as indicators of the migration potential. In any case, based on a sound out-of-sample prediction, we conclude that aggregate migration potential is modest and that fears of mass migration from the EU’s Eastern neighbors and Croatia are unjustified.


1This was preceded by two waves of EU eastern enlargement, the first in 2004 when the Czech Republic, Estonia, Hungary, Latvia, Lithuania, Poland, Slovakia and Slovenia (along with Cyprus and Malta) and the second in 2007 when Bulgaria and Romania joined the EU.

2Iceland and Turkey also obtained candidate status. Albania and Bosnia and Herzegovina are classified as potential candidates.

3EU-12 encompasses the following countries: Belgium, Denmark, France, Germany, Greece, Ireland, Luxembourg, the Netherlands, Portugal, Spain, Italy and the UK.

4 A very similar picture emerges if we compare Germany, the Netherlands and the UK with other Mediterranean countries like Greece or even Italy.

5See also Constant (2012).

6In quantitative terms, this exclusion is negligible since the stock of migrants from the EU-8 living in Luxembourg amounts to merely 0.4 percent in 2009.

7Partial liberalization signifies liberalization of labor market access in specified sectors or occupations, or a combination of them, typically based on job shortages.

8The findings of Elsner (2013a, 2013b) that out-migration increases domestic wages implies that wages, and possibly unemployment rates as well, may be endogenous with respect to their effects of out-migration. If outmigration positively affects income or decreases unemployment rates in sending countries, the true (negative) effects of home income and employment rate variables on outmigration could be larger.

9Of course, the former Yugoslav countries look back on a common migration history with some EU Member States for a rather long time. However, the policy regimes under which this migration history occurred (guest or seasonal worker programs and refugee migration) were completely different from what we are interested in this paper.

10For Croatia, these projections are valid under the (counterfactual) assumption of no EU accession.

11The EU14 countries are divided into three groups according to their labor market situation in 2010: (i) high employment countries: Austria, Denmark and the Netherlands; (ii) medium employment countries: Belgium, Finland, France, Germany, Greece, Ireland, Italy, Portugal, Sweden and the United Kingdom; and (iii) low employment countries: Spain.

12For Belarus, the IMF reports an unemployment rate of 0.7% in 2010. Following the advice of national experts, we assumed the unemployment rate of 10% in 2010 and used the unemployment rate trends from Ukraine to generate forecasts for Belarus until 2020.

13However, we suspect that the underlying GDP and employment figures for Belarus might be biased upward. In such a case, our projections would be underestimated


  • Bauer, Thomas and Klaus F. Zimmermann (1999), Assessment of Possible Migration Pressure and its Labour Market Impact Following EU Enlargement to Central and Eastern Europe. Bonn: IZA and London: CEPR.

  • Beck N, Katz JN (1995) What to do (and not to do) with time-series cross-section data. American Political Science Review 89:634–647

    Article  Google Scholar 

  • Biavaschi C, Zimmermann KF (2014) Eastern partnership migrants in Germany: outcomes, potentials and challenges. IZA Journal of European Labor Studies 3:7

    Article  Google Scholar 

  • Bover, Olympia and Pilar Velilla (2001), Migrations in Spain: Historical Background and Current Trends. Bonn: Institute for the Study of Labor (IZA); IZA Discussion Paper No. 88, IZA-Bonn

  • Brücker, Herbert, Timo Baas, Iskra Beleva, Simone Bertoli, Tito Boeri, Andreas Damelang, Laetitia Duval, Andreas Hauptmann, Agnieszka Fihel, Peter Huber, Anna Iara, Artjoms Ivlevs, Elke J. Jahn, Pawel Kaczmarczyk, Michael E. Landesmann, Joanna Mackiewicz-Lyziak, Mattia Makovec, Paola Monti, Klaus Nowotny, Marek Okólski, Sándor Richter, Richard Upward, Hermine Vidovic, Katja Wolf, Nina Wolfeil, Peter Wright, Krisjane Zaiga and Anna Zylicz (2009), Labour mobility within the EU in the context of enlargement and the functioning of the transitional arrangements. Final Report of the European Integration Consortium on behalf of the Employment, Social Affairs and Equal Opportunities Directorate General of the European Commission, Nürnberg, European Integration Consortium (IAB, CMR, fRDB, GEP, WIFO, wiiw).

  • Clark K, Drinkwater S (2014) Labour migration to the UK from Eastern partnership countries. IZA Journal of European Labor Studies 3:15

    Article  Google Scholar 

  • Constant AF (2012) Sizing It Up: Labor Migration Lessons of the EU Enlargement to 27. In: Gortázar C, Parra C, Segaert B, Timmerman C (eds) European Migration and Asylum Policies: Coherence or Contradiction. Bruylant, Belgium, pp 49–77

    Google Scholar 

  • Constant AF, Massey DS (2005) Labor market segmentation and the earnings of german guestworkers. Popul Res Policy Rev 24(6):5–30

    Google Scholar 

  • Constant A, Kahanec M, Zimmermann KF (2009) Attitudes towards immigrants, other integration barriers, and their veracity. Int J Manpow 30(1/2):5–14

    Article  Google Scholar 

  • Danzer AM, Dietz B (2014) Labour migration from eastern Europe and the EU's quest for talents. J Common Mark Stud 52(2):183–199

    Article  Google Scholar 

  • DeGiorgi G, Pellizzari M (2009) Welfare migration in Europe. Labour Econ 16(4):353–363

    Article  Google Scholar 

  • Dustmann, Christian and Tommaso Frattini (2011), Immigration: The European Experience. Bonn: Institute for the Study of Labor (IZA); IZA Discussion Paper No. 6261, IZA-Bonn.

  • Duszczyk, Maciej, Marek Góra and Paweł Kaczmarczyk (2013), Costs and Benefits of Labour Mobility between the EU and the Eastern Partnership Partner Countries, IZA. Available at

  • Elsner B (2013a) Does emigration benefit the stayers? evidence from EU enlargement. J Popul Econ 26(2):531–553

    Article  Google Scholar 

  • Elsner B (2013b) Emigration and wages: the EU enlargement experiment. J Int Econ 91(1):154–163

    Article  Google Scholar 

  • European Commission (2008), Employment in Europe 2008. Luxembourg; Office for Official Publications of the European Communities 2008.

  • Farré L, Rodríguez-Planas N (2014) Immigrants from eastern partnership (EaP) countries in Spain. IZA Journal of European Labor Studies 3:1

    Article  Google Scholar 

  • Fertig M (2001) The economic impact of EU-enlargement: assessing the migration potential. Empir Econ 26:707–720

    Article  Google Scholar 

  • Fertig M, Schmidt CM (2001) Aggregate-Level Migration Studies as a Tool for Forecasting Future Migration Streams. In: Djajic S (ed) International Migration: Trends. Routledge, Policies and Economic Impact, pp 110–136

    Google Scholar 

  • Fertig, Michael and Christoph M. Schmidt (2002), Mobility within Europe – What do we)still not) know? Bonn: Institute for the Study of Labor (IZA); IZA Discussion paper No. 447, IZA-Bonn

  • Giulietti C, Guzi M, Kahanec M, Zimmermann KF (2013) Unemployment benefits and immigration: evidence from the EU. Int J Manpow 34(1/2):24–38

    Google Scholar 

  • Hatton TJ (2005) Explaining trends in UK immigration. J Popul Econ 18:719–740

    Article  Google Scholar 

  • Hatton TJ (1995) A Model of UK emigration, 1870–1913. Rev Econ Stat 77:407–415

    Article  Google Scholar 

  • Hazans, Mihails and Kaia Philips (2010), The Post-Enlargement Migration Experience in the Baltic Labor Markets. In: Kahanec, Martin and Klaus F. Zimmermann (eds.), Berlin: Heidelberg; EU Labor Markets After Post-Enlargement Migration, Springer, 255–304.

  • Holland, D., T. Fic, P. Paluchowski, A. Rincon-Aznar and L. Stokes (2011), Labour Mobility within the EU: The impact of Enlargement and Transitional Arrangements, NIESR Discussion Paper No. 379, National Institute of Economic and Social Research, London.

  • Kaczmarczyk P, Okolski M (2008) Demographic and labor market impacts of migration on Poland. Oxf Rev Econ Policy 24:600–625

    Article  Google Scholar 

  • Kahanec M (2013a) “Skilled Labor Flows: Lessons from the European Union”, Social Protection and labor discussion paper no. SP 1301. Washington D.C., The World Bank

    Google Scholar 

  • Kahanec M (2013b) Labor Mobility in an Enlarged European Union. In: Constant AF, Zimmermann KF (eds) International Handbook on the Economics of Migration. Edward Elgar, Cheltenham, pp 137–152

    Chapter  Google Scholar 

  • Kahanec M, Shields MP (2013) The working hours of immigrants in Germany: temporary versus permanent. IZA Journal of Migration 2:14

    Article  Google Scholar 

  • Kahanec M, Zimmermann KF (2010) EU Labor Markets after Post-Enlargement Migration. Springer, Berlin

    Book  Google Scholar 

  • Kahanec M, Kim AM-H, Zimmermann KF (2013a) Pitfalls of immigrant inclusion into the european welfare state. Int J Manpow 34(1/2):39–55

    Google Scholar 

  • Kahanec, Martin, Klaus F. Zimmermann, Lucia Kurekova, and Costanza Biavaschi (2013). Labour Migration from EaP Countries to the EU – Assessment of Costs and Benefits and Proposals for Better Labour Market Matching. Bonn: Institute for the Study of Labor (IZA); IZA Research Report No. 56.

  • Kahanec, Martin, Mariola Pytliková and Klaus F. Zimmermann (forthcoming). The Free Movement of Workers in an Enlarged European Union: Institutional Underpinnings of Economic Adjustment. Forthcoming in: M. Kahanec and K.F. Zimmermann (eds.), Labor Migration, EU Enlargement, and the Great Recession, Springer.

  • Kim K, Cohen JE (2010) Determinants of international migration flows to and from industrialized countries: a panel data approach beyond gravity. Int Migr Rev 44:899–932

    Article  Google Scholar 

  • Libanova EM (ed) (2006) Комплексний демографічний прогноз України на період до 2050 р. Український центр соціальних реформ, Kiev

    Google Scholar 

  • Marchetti S, Piazzalunga D, Venturini A (2014) Does Italy represent an opportunity for temporary migrants from the eastern partnership countries? IZA Journal of European Labor Studies 3:8

    Article  Google Scholar 

  • Orlowski, Witold M., Leszek Zienkowski and Leon Podkaminer (2000), Potential Migration Flows from Central and Eastern Europe to Austria After the Enlargement. WIIW Project Nr. 7919, Vienna, Austria.

  • Pedersen PJ, Pytlikova M, Smith N (1990) Selection and network effects: Migration flows into OECD countries. European Economic Review 52(7):1160–1186

    Article  Google Scholar 

  • Sjaastad LA (1962) The cost and returns of human migration. J Polit Econ 70:80–93

    Article  Google Scholar 

  • Zaiceva, Anzelika (2006), Reconciling the Estimates of Potential Migration into the Enlarged European Union. IZA Discussion Paper No. 2519, IZA-Bonn

Download references


This paper expands on a research project entitled “Costs and Benefits of Labour Mobility between the EU and the Eastern Partnership Partner Countries” funded by the European Commission during 2011–2013 under the reference number EuropeAid/130215/C/SER/Multi. The authors thank the project consortium and national experts as well as the European Commission for inspiring this paper and for useful discussions which helped to validate our data and form our ideas about east–west mobility in Europe. Martin Kahanec acknowledges the financial support of the Eduworks Marie Curie Initial Training Network Project (PITN-GA-2013-608311) of the European Commission’s 7th Framework Programme. Furthermore, we are grateful to an anonymous referee for very constructive and helpful comments. Any opinions expressed in this study are of the authors and do not necessarily represent the opinion of the European Commission or any partner in the project consortium. We thank Richard Forsythe for a language check of the manuscript. We remain responsible for any remaining errors.

Responsible editor: Amelie Constant

Author information

Authors and Affiliations


Corresponding author

Correspondence to Martin Kahanec.

Additional information

Competing interests

The IZA Journal of Migration is committed to the IZA Guiding Principles of Research Integrity. All authors declare that they have observed these principles.



Table A.1. Definition of policy regimes for workers from EU-8

Table A.1 presents the definition of policy variables measuring immigration policy regimes. Two policy dummies are distinguished: free movement and partial restrictions. Free movement equals one if the respective country allowed immigrants from the new member states to access the host country's labor market in the respective year, and zero otherwise. The second policy dummy variable equals one if in a given country and year only partial restriction on the inflow of workers to its labor market were applied, and zero otherwise.

Member state

Pre-accession period

Post-accession period


Full restriction

Full restriction


Full restriction

Partial restrictions


Full restriction

Partial restrictions


Full restriction

Free access since 2006


Full restriction

Free access since 2008


Full restriction

Partial restrictions


Full restriction

Free access since 2006


Full restriction

Free access since 2004


Full restriction

Free access since 2006


Full restriction

Free access since 2007


Full restriction

Free access since 2006


Full restriction

Free access since 2006


Full restriction

Free access since 2004

United Kingdom

Full restriction

Free access since 2004

European Commission (2008), Table 1, p. 111.

Rights and permissions

Open Access This article is distributed under the terms of the Creative Commons Attribution 4.0 International License (, which permits use, duplication, adaptation, distribution, and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made.

Reprints and permissions

About this article

Check for updates. Verify currency and authenticity via CrossMark

Cite this article

Fertig, M., Kahanec, M. Projections of potential flows to the enlarging EU from Ukraine, Croatia and other Eastern neighbors. IZA J Migration 4, 6 (2015).

Download citation

  • Received:

  • Accepted:

  • Published:

  • DOI:

JEL codes