Aktuelles

Focus on European Economic Integration Q3/22

Call for applications: Klaus Liebscher ­Economic Research Scholarship

Please e-mail applications to scholarship@oenb.at by the end of October 2022. Applicants will be notified of the jury’s decision by end-November 2022.

The Oesterreichische Nationalbank (OeNB) invites applications for the “Klaus ­Liebscher Economic Research Scholarship.” This scholarship program gives out­standing researchers the opportunity to contribute their expertise to the research activities of the OeNB’s Economic Analysis and Research Department. This ­contribution will take the form of remunerated consultancy services.

The scholarship program targets Austrian and international experts with a proven research record in economics and finance, and postdoctoral research ­experience. Applicants need to be in active employment and should be interested in broadening their research experience and expanding their personal research ­networks. Given the OeNB’s strategic research focus on Central, Eastern and Southeastern Europe, the analysis of economic developments in this region will be a key field of research in this context.

The OeNB offers a stimulating and professional research environment in close proximity to the policymaking process. The selected scholarship recipients will be expected to collaborate with the OeNB’s research staff on a prespecified topic and are invited to participate actively in the department’s internal seminars and other research activities. Their research output may be published in one of the ­department’s publication outlets or as an OeNB Working Paper. As a rule, the ­consultancy services under the scholarship will be provided over a period of two to three months. As far as possible, an adequate accommodation for the stay in ­Vienna will be provided. 1

Applicants must provide the following documents and information:

  • a letter of motivation, including an indication of the time period envisaged for the consultancy
  • a detailed consultancy proposal
  • a description of current research topics and activities
  • an academic curriculum vitae
  • an up-to-date list of publications (or an extract therefrom)
  • the names of two references that the OeNB may contact to obtain further information about the applicant
  • evidence of basic income during the term of the scholarship (employment ­contract with the applicant’s home institution)
  • written confirmation by the home institution that the provision of consultancy services by the applicant is not in violation of the applicant’s employment ­contract with the home institution

1 We are also exploring alternative formats to continue research cooperation under the scholarship program for as long as we cannot resume visits due to the ­pandemic situation.

Studies

The role of public services quality in shaping migration intentions in Central, Eastern and Southeastern Europe

Anna Katharina Raggl 2

What role does the quality of public services play in shaping migration intentions? Using OeNB Euro Survey data collected in 2018, we study the impact of individual perceptions of public services quality on individuals’ migration intentions in ten Central, Eastern and Southeastern European (CESEE) countries. We apply ordinary least squares (OLS) as well as instrumental variable (IV) estimations, using externally merged infrastructure-related variables and individuals’ opinions on the adequacy of public spending on services as instrumental variables. Our findings suggest that dissatisfaction with the quality of public services in the home countries increases the likelihood of individuals having migration intentions. Broken down by the type of public service, we find that dissatisfaction with social security, health, public infrastructure and with services that target businesses and regional development is associated with higher migration aspirations. Furthermore, for people with young children, we see a higher effect on migration intentions resulting from dissatisfaction with education, health and public safety. For self-employed individuals, the effect of dissatisfaction with public services that address companies and regional development is particularly important. The results further confirm that sociodemographic characteristics, economic factors and network effects are closely associated with the aspiration to move abroad.

JEL classification: J61, F22, O52

Keywords: migration intentions, quality of public services, Central, Eastern and Southeastern Europe

In Europe, long-term demographic trends continue to follow different patterns reflecting past geopolitical divisions: we see population growth in Western, Southern and Northern Europe and population decline in Central, Eastern and Southeastern Europe (CESEE). The decrease in population in CESEE since 1990 results from a combined effect of natural population decrease (i.e. deaths exceeding births) and emigration. While some countries recently experienced moderate population growth (Hungary, Czechia), projections suggest that the observed declining population trends will continue over the next decades due to both natural change and net migration (VID and IIASA, 2020).

In a recent IMF report, Batog et al. (2019) comprehensively assess the implications of demographic developments in CESEE on labor supply, age-related fiscal spending and on productivity and the prospects for economic growth in the region. They conclude that without mitigating policies, growth and convergence toward Western European living standards would slow down considerably. Implementing policies that increase labor supply is one way to counteract demographic developments. But as it is unlikely that this is sufficient, a more comprehensive approach that also addresses emigration, immigration and return migration is needed.

This study attempts to contribute to a better understanding of why individuals intend to emigrate from their home country and to provide insights for designing policies that increase people’s willingness to stay in their home countries (and/or to return or move to CESEE from abroad). Such policies can be important tools to counteract labor force declines, human capital deterioration and challenged public finances, and, as a consequence, to remove obstacles to continued and sustainable convergence to Western Europe. Following up on previous research dedicated to understanding why people in CESEE want to emigrate, we address the link ­between the perceived quality of public services – social security, public infrastructure, education, health, defense and public safety, regional development ­measures – and individuals’ migration intentions.

We use individual-level data from the 2018 wave of the OeNB Euro Survey, which covers ten CESEE countries. Making use of a series of questions on public debt, we investigate the relationship between the perceived quality of public services and individual migration intentions. Our data allow us to control for a rich set of covariates that have previously been identified as relevant for the emergence of ­migration intentions and that range from sociodemographic and economic factors to networks and the trust in local and EU institutions. In addition, we take advantage of the fact that the data are geo-referenced and use matched data on nighttime light combined with regional averages of unemployment and income to control for the level of development in the surroundings of respondents’ places of residence. We use instrumental variable estimations in order to address identification issues that can stem from the simultaneity of migration intentions and contentment with public services and from the omission of factors that influence both these variables. As instrumental variables we use exogeneous data on road density as well as land coverage in individuals’ neighborhoods as measures of infrastructure, and we further employ individuals’ assessment of the adequacy of public spending on public services, a variable that is covered in the survey (an instrumental variable).

The remainder of this paper is structured as follows. Section 1 provides an overview of related literature. The empirical setting is explained in section 2. ­Section 3 summarizes the data and provides descriptive statistics of the key variables. In section 4, the estimation results are presented, and section 5 concludes. Additional material is included in the annexes.

1 Literature

Traditionally, differences in incomes and labor market opportunities across countries are seen as key determinants of migration, and a large body of literature focuses on this link. Also, the relationship between education and migration has received a lot of attention (Borjas, 1987; Chiswick, 1999; Chiquiar and Hanson, 2005), and the importance of networks abroad has repeatedly been established (Docquier et al., 2014; Manchin and Orazbayev, 2018).

More recently, nonpecuniary factors feature more prominently in studies on the determinants of migration intentions. Otrachshenko and Popova (2014), for example, relate life satisfaction measures to individual migration intentions. Using data from Western and Central Europe, they find that individuals that are dissatisfied with life are more likely to have the intention to emigrate. Similarly, van Dahlen and Henkens (2013) show for the Netherlands that discontent with the quality of the public domain (with regard to mentality, space and overcrowdedness, nature, pollution, crime, etc.) is an important set of factors for explaining migration intentions. In a recent study, Williams et al. (2018) use data from nine European countries – among them Romania as the only country also covered in our analysis – and find that although socioeconomic factors have strong explanatory power, nonpecuniary factors also play a certain role. Several related studies work with different waves and country sets of Gallup World Poll (GWP) data to understand what drives individuals’ migration intentions. Dustmann and Okatenko (2014) use GWP data for sub-Saharan Africa, Asia and Latin America (2005, 2006) to study migration intentions. They find that contentment with local amenities, such as public services and public security, are key determinants of migration intentions and explain a considerable share of variation in migration intentions. Manchin and Orazbayev (2018) focus predominantly on the impact of networks but confirm that satisfaction with local amenities and local security decreases the probability of people moving away from the current region of residence (150 countries, GWP data, 2010-2013). In its flagship Transition Report (2018), the EBRD highlights the role of satisfaction with local amenities, using GWPs data for the EBRD region (2010–15). Further related studies addressing this link often use only one country and study internal migration (see for example Chen and Rosenthal, 2008). Studies by Tran et al. (2019, 2021) also address factors that are related to political outcomes and institutions, focusing on the impact of institutional quality and institutional quality differentials between host and origin countries on return migration to Vietnam. Similarly, Etling et al. (2018) look into the relationship between political discontent and migration intentions in the Arab Mediterranean region. 3

This paper contributes to the literature in several ways. It concentrates on a region that has been faced with particular challenges related to demographic change and emigration. Furthermore, the data we use allow us to put a clear focus on public services, that is, factors that can be changed by policymakers, as opposed to studying the impact of fairly vague concepts, such as amenities. In addition, we take into account the possible endogeneity of the variable measuring dissatisfaction with public services in an attempt to limit biased coefficient estimates.

2 Empirical setting

2.1 OLS estimations

We use ordinary least square (OLS) estimations to address the impact of individuals’ assessment of public services quality on their migration intentions, controlling for a range of different factors that have been shown to be relevant in this setting (see Raggl, 2019, for example).

In particular, we estimate the following basic relationship:

m_i=α_r+β_d d_i+∑_(j=1)^J▒〖x_ij^Socio β_j^Socio 〗+∑_(k=1)^K▒〖x_ik^Econ β_k^Econ 〗+∑_(l=1)^L▒〖x_l^Region β_l^Region 〗+∑_(m=1)^M▒〖x_im^Network β_m^Network 〗+〖∑_(p=1)^P▒〖x_ip^Trust β_p^Trust 〗+ϵ〗_i

mi is a binary variable that takes a value of 1 if an individual has the intention to move abroad within the next 12 months, di is a measure of dissatisfaction with public services and the key variable of interest. xij Socio are J variables that belong to the group of sociodemographic characteristics, xik Econ represents the group of K economic factors, xl Region the L regional characteristics, xim Network the M variables capturing network effects and xip Trust are P variables capturing trust in institutions. In addition, a constant and a full set of country dummies, denoted in the equation by a country-­specific constant αr, is included in all specifications. The country dummies control for all factors that are the same for all individuals in a country, such as institutional characteristics, the political environment, historical ties to other countries, geographic location and the like. ϵi is the remaining error term. Standard errors are clustered at the regional level in countries where the regions are defined according to NUTS 2 or a finer classification (HR, BG, MK).

We would expect individuals who are dissatisfied with the provision of public services to have stronger migration intentions ceteris paribus. These findings would be in line with other empirical studies that highlight the relevance of nonpecuniary factors for the emergence of migration intentions.

The data allow us to study coefficient heterogeneities along different dimensions. First, the magnitude of the effects might differ by country. The ten countries in the sample differ in size, average income levels, EU membership, etc., and the importance of public services quality for migration intentions could vary across countries. We assess these possible heterogeneities by interacting the dissatisfaction variable with country dummies. Second, distinguishing by the type of service – social security, public infrastructure, education, health, defense and public safety as well as economic development – allows us to study whether the perceived quality of certain services has a higher impact on migration intentions than that of other services. We would expect ex ante that the magnitude of the effects could be related to the exposure of an individual to a specific service, however, and therefore ­heterogeneous effects with respect to the individual and household situation can be assumed. This exposure can be empirically approximated with interaction terms that allow, for example, the effect of dissatisfaction with education, health or social security services to differ between individuals with (school-aged) children and those without. Similar interactions are to be tested between being unemployed or retired and ­reacting to the quality of social security services or between being self-employed and public economic development (and business support) services.

Regarding the sign of the effects of the control variables, no large deviations from Raggl (2019) are expected: Migration intentions are expected to be stronger among the young and among men, and weaker for respondents who are married and have children (represented by a principal component based on marital status and the number of children, see below). Furthermore, being unemployed is expected to raise migration intentions, and, ex ante, a similar effect would be expected for ­income (the latter has not been found in Raggl, 2019). In addition, we anticipate that having networks abroad increases migration intentions, and trust in domestic (foreign) institutions is associated with weaker (stronger) migration intentions. Apart from being interesting in its own right, the trust variables can also help ­capture people’s general satisfaction with the institutional situation in the home country and can act as an important proxy for this otherwise unobservable factor.

As there is a large number of covariates and some of them are highly correlated with each other, we use (Polychoric) Principal Component Analyses, (P)PCA, to reduce the dimensionality of the data (see box below). For a full list and description of the variables included in the estimations, please refer to table B1 in annex B.

(Polychoric) principal component analysis, (P)PCA

As in an earlier study on migration intentions (Raggl, 2019), we use (polychoric) principal component analysis to reduce the dimensionality of certain groups of covariates. Some groups of covariates contain variables that are highly correlated with each other; including all of them in a regression could cause multicollinearity issues. If we only use a selection of variables that are believed to be relevant, potentially important information might be omitted from the analysis. (P)PCA is a method that reduces the dimensionality of the data while keeping a large part of the information they contain. The method dates back to works by Pearson (1901) and Hotelling (1933). It identifies the linear combination of the variables that accounts for the greatest variance in the data. The first principal component is the linear combination of the original variables that accounts for the largest share of the variance. The second principal component is orthogonal to the first and contains the largest part of the remaining variance, etc. The analysis identifies as many components as variables are used, and if all components are used in a subsequent regression, nothing would be gained vis-à-vis including all the variables. There is no binding rule on how to decide how many components should be used in a subsequent analysis, but a general rule of thumb is that components with an eigenvalue (EV) greater than 1 should be included (Kaiser rule, scree test). As many of our variables are discrete, either binary or based on a Likert-type scale, we use polychoric PCAs (PPCAs) for these cases. Kolenikov and Angeles (2004) developed this method for discrete variables.

The table below lists the groups of variables for which we perform (P)PCAs and the ­components we use in the regressions.

Table : Groups of variables and components used in the regressions  
Group of variables Variables included Component(s) used Eigenvalue Share of variation
explained
Contentment with
public services
Dissatisfaction with social
­security, public infrastructure,
education, health, defense and
public safety, and economic
­development
Component that
­represents dissatisfaction
with public services
3.6 60%
Household demographics Household size, marital status
of respondent, number of
­children aged 6 or younger,
number of children aged 6 to 15
Component that
­represents large families
2.4 61%
Proxy for household wealth Ownership of the main residence,
a secondary residence, other
real estate, other land and
car ownership
Component that
­represents wealth
­holdings
2.3 47%
Direct networks
abroad
Having friends and family
abroad, receiving money from
abroad
Component that
­represents direct
­networks
1.8 88%
Indirect networks
abroad
Share of respondents in the
primary sampling unit (PSU)
and in the region that have
friends and family abroad, share
of respondents in the PSU and
the region that receive remittances
Component that
­represents indirect
­networks
2.5 62%
Trust in institutions Demeaned trust in the government,
the police, domestic banks,
foreign banks, the ECB and
the EU
Component 1 that
­represents trust in local
institutions
1.9 32%
Component 2 that
­represents trust in EU
­institutions
1.4 23%
Source: Author’s compilation.

Further (P)PCAs are performed to reduce the dimensionality of the instrumental variables (see table B1 in the annex).

2.2 Endogeneity issues and instrumentation

Establishing a causal relationship between personal perceptions of the quality of public services and migration intentions is not a trivial task. Individuals might be more dissatisfied with public services in their home countries if they have migration intentions and/or third – unobservable – factors might drive both the ­perceived quality of public services and the intention to emigrate. The result can be biased OLS estimates. 4

We use instrumental variables (IV) as sources of exogeneous variation and two-stage least squares (TSLS) estimations in addition to the basic OLS estimations to take into account potential endogeneity. In particular, we pursue two avenues for instrumentation. First, we use external measures of infrastructure in the close proximity of respondents’ residences: road density (Meijer et al., 2018) and land use (artificial continuous and discontinuous urban fabric surfaces, based on CORINE land cover nomenclature). Broadly speaking, these variables can serve as proxies for general infrastructure in close proximity of individuals’ residences and contribute to explaining contentment with the quality of public services based on the actual infrastructure in the neighborhood. These measures of infrastructure might also represent the employment opportunities in the region where the respondents live and which could compromise the exogeneity (and therefore validity) of the instruments if these employment opportunities are not controlled for in the main equation. But the main equation does control for regional economic development. It does so by including (PCA) measures that are based on nighttime light intensity in a respondent’s neighborhood (measured for different radii around the respondents’ residences) as well as on the average unemployment and income of respondents living nearby. Second, we make use of a question in the survey that asks about ­respondents’ opinion on the adequacy of state spending on specific public services 5 . The idea is that the view of whether state spending on a particular service should be increased, maintained at the same level or decreased should not influence ­migration intentions per se, except through its influence on the perceived quality of public services. We use Kleibergen-Paap statistics and Hansen-J statistics to ­assess the relevance and exogeneity of the instruments and present first-stage ­results of the estimations in table D1 in the annex.

3 Data and descriptive evidence

We use the 2018 fall wave of the OeNB Euro Survey to address the link between the perceived quality of public services and migration intentions. The OeNB Euro Survey is an individual-level dataset created from a survey that has been conducted on behalf of the Oesterreichische Nationalbank (OeNB) in ten CESEE countries 6 since 2007. It covers approximately 1,000 randomly selected individuals per country and wave and focuses on topics such as (euro) cash holdings, saving behavior and debt, economic opinions and expectations, and experiences. In addition, sociodemographic and economic characteristics are collected, as well as information on individuals’ migration intentions 7 . The 2018 survey wave contains a special module on public debt. Individuals were asked about their satisfaction with public services, in particular with social security, public infrastructure, education, health services, defense and public safety, and economic development (i.e. support for small and medium enterprises, etc.). In this study, we also employ data from external sources, which are merged with the Euro Survey: nighttime light, urban fabric measuring the biophysical coverage of the Earth (CORINE land cover data) and road density (Global Road Inventory Project, GRIP, dataset; Meijer et al., 2018) at individuals’ places of residence. See table B1 in annex B for a complete list and ­description of the variables used for this study.

3.1 Migration intentions

The data show that in the ten CESEE countries covered, the share of individuals aged 25 to 64 who intend to move abroad within the next 12 months varies between 3% in ­Czechia and 20% in North Macedonia (see table A1 in annex A). When weighted by population size, the share of individuals with migration intentions in CESEE is approximately 6%, indicating that 6% of the population in CESEE intend to move abroad within the next 12 months. This population-weighted average takes into account that the ten countries differ greatly in size, but this also implies that the average is strongly driven by Poland and, to a smaller degree, Romania.

Pooling the data for all ten CESEE countries, the data indicate that approximately 9.1% of the respondents aged 25 to 64 indicate that they intend to move abroad within the next 12 months. This average refers to an “average CESEE ­country” and does not correct for the differences in the size of countries. We can further see in the data that the share of individuals with migration intentions varies considerably with age: among those aged 29 and younger, over 20% have migration intentions, among the 30- to 39-year-olds, 12% to 13% intend to emigrate, and in the working-age population above 40 years of age, just above 5% intend to leave their home country. When we look at gender differences, the data indicate that migration intentions are more common among men than among women. This holds for both the pooled and the population-weighted CESEE averages as well as for most individual countries. With respect to educational attainment, the differences are less well defined. While, driven by Poland, the population-weighted average indicates higher emigration intentions among the highly skilled, this does not hold for individual countries (except for Serbia and Poland) or for the pooled average of respondents (see table A1 in annex A). 8

Chart 1displays a population pyramid for an average CESEE country. The vertical dimension of the chart represents the age groups (starting from group aged 20 to 24 years up to the group aged 90 to 95 years). The horizontal dimension indicates the share of the respective subgroup in the total population in %. This horizontal dimension differs between women (right part) and men (left part). The vertical bars in the chart indicate through different colors the educational attainment and the migration intentions of each age group. The pyramid shows that if individuals who have the intention to emigrate left the country, the remaining population structure would be more constrictive, i.e. narrower at the bottom.

Source: Author's calculations based on OeNB Euro Survey data (2018) using pooled data for the ten CESEE countries.

Chart 1 shows the gender-age-education-migration nexus using a population pyramid for an average CESEE country (not population-weighted). The colors represent the level of education of the gender-age group and the hatched areas the part of the subgroup that has the intention to move abroad. The chart further indicates the shape of the pyramid that we would expect if the individuals with migration intentions actually left the country. The stylized pyramid drawn by the black line is more constrictive, and the tapering at the bottom indicates the shrinking and aging population structure prevalent in CESEE.

3.2 Dissatisfaction with the quality of public services

The special module on public debt in the 2018 wave of the OeNB Euro Survey contains a question on respondents’ satisfaction with public services in the areas of social security, public infrastructure, education, health, defense and public safety, and economic development. 9

Chart 2 shows the share of respondents who say that they are dissatisfied with a specific public service. It shows that 54% are dissatisfied with social security, 57% with public infrastructure, 52% with education, 66% with health, 51% with public safety, 54% with economic development services.

Source: Author's calculations based on OeNB Euro Survey data (2018).
Chart 3 shows respondents’ average dissatisfaction with public services by country. The shares of respondents dissatisfied with public services by country are: Bosnia and Herzegovina: 67%; Romania: 66%; North Macedonia: 62%; Bulgaria: 61%; Croatia: 60%; Hungary: 58%; Serbia: 49%; Albania: 45%; Czechia: 45%; Poland: 43%; total CESEE sample: 56%.

Source: Author's calculations based on OeNB Euro Survey data (2018).

The level of dissatisfaction with public services is high in CESEE. If we look at the average across all public service categories, more than 55% of respondents say that they are either strongly dissatisfied or dissatisfied with the public services provided in their country. Dissatisfaction is high for all service types, but particularly so for health services (see chart 2). As regards the differences across countries (chart 3), the data reveal that in Poland and Czechia, but also in Albania and Serbia, dissatisfaction is below the sample average. In Bosnia and Herzegovina, dissatisfaction is highest; here, an average of close to 70% of respondents are not satisfied with the provision of public services.

Chart 4 provides first descriptive evidence on the relationship between people’s contentment with public services and individuals’ migration intentions after pooling these data for all ten countries. It shows that migration intentions are stronger among individuals that are dissatisfied with public services. This finding holds for all service categories and is particularly striking for education and economic development services. Among those not dissatisfied with education services, for example, 7.7% intend to emigrate, while among those dissatisfied with the public provision of education, over 10% intend to leave their home countries.

As such a correlation can be driven by other (observable or unobservable) ­factors, we estimate OLS regressions that control for sociodemographic, economic and regional characteristics, network effects, etc. as well as TSLS regressions to mitigate a possible bias in the estimates due to reverse causality and/or omitted variables.

Chart 4, Migration intentions and dissatisfaction with public services, shows a pair of bars for each of the six service categories – social security, public infrastructure, education, health, public safety, and economic development. One bar indicates average migration intentions among those who are dissatisfied with public services, and the other bar indicates average migration intentions among those who are not dissatisfied with public services. For all the service categories, the following holds: individuals who are dissatisfied with the quality of a service have stronger migration intentions than those who are not dissatisfied.

4 Estimation results

4.1 OLS estimation

Table 1 shows the standardized coefficients of OLS estimations when the data for all countries are pooled. Starting out with a parsimonious specification in (1) that contains an index of the perceived quality of public services, sociodemographic characteristics and a full set of country dummies, we successively add sets of ­further covariates. The (standardized) coefficients of the PPCA index that ­measures the degree of dissatisfaction with public services are positive and significant in all specifications. They range between 0.046 and 0.064, i.e., a 1 standard deviation increase in the dissatisfaction index increases the likelihood of having migration intentions by 0.05 standard deviations. The standard deviation of the migration intentions variable in the sample is approximately 0.28 (and the mean approximately 0.08), i.e., a 0.05 standard deviation increase is equivalent to an approximately 1.4 percentage point increase in migration intentions. With an average of 8% of respondents in the sample having migration intentions, this is a nonnegligible effect.

With regard to other covariates, the results closely mirror Raggl (2019), a study that relies on OeNB Euro Survey data collected a year earlier, i.e. in 2017. Women are less likely to have migration intentions, just like respondents who have small children or large families and who are married. Also, the likelihood of migration intentions declines with respondents’ age 10 . Unemployment is a strong predictor of migration intentions, while (equivalent) income is not statistically significant. Respondents’ wealth, approximated by a PPCA of the ownership of the main residence, a second or other residence, land and/or a car, is negatively associated with migration intentions. By contrast, migration intentions increase significantly if people have networks abroad, either direct and indirect ones, with the latter being measured by the networks of respondents living in close proximity (primary sampling unit and region).

4.2 Heterogeneous effects

The estimates discussed so far present results that are based on the pooled sample of all ten countries in the survey, with country fixed effects having been controlled for. As the ten countries are very different, we interact the variable indicating ­dissatisfaction with public services with country dummies. The resulting country-­specific coefficients are plotted in chart 5. 11 While the estimated coefficients in many countries are close to the pooled estimate (indicated by the horizontal line), for some countries they are insignificant. With the exception of Bosnia and Herzegovina, the effects in non-EU CESEE countries are above the pooled average. By far the highest impact is estimated for Albania, where the coefficient estimate is twice as high as for Romania and Serbia, the countries with the second- and third-highest coefficients.

This high effect for Albania might give rise to the presumption that the overall effect in the pooled sample might be driven by a single country (e.g. Albania). For robustness, we run ten pooled regressions, omitting one country in each regression. The resulting regression coefficients of the dissatisfaction with public services variable are plotted in chart C1 in annex C. It shows that while the overall coefficient is smaller when we omit Albania, the estimates remain statistically significant and positive in all specifications.

Table 1: OLS estimation (standardized coefficients)  
(1) (2) (3) (4) (5) (6) (7)
Socio-dem Economic Wealth Regional Networks Trust PSU-FE
PPCA: dissatisfaction 0.049** 0.062*** 0.064*** 0.061*** 0.047*** 0.047*** 0.048***
with public services (2.18) (2.82) (2.83) (2.92) (3.14) (3.03) (3.32)
Female –0.053*** –0.056*** –0.056*** –0.054*** –0.055*** –0.049*** –0.046***
(–4.67) (–4.74) (–4.65) (–4.53) (–4.48) (–4.17) (–3.80)
Age –0.498*** –0.457*** –0.438*** –0.436*** –0.439*** –0.424*** –0.378***
(–5.69) (–5.04) (–4.88) (–4.85) (–4.99) (–4.86) (–4.75)
Age sq. 0.251*** 0.219*** 0.203** 0.202** 0.212*** 0.200** 0.181**
(3.25) (2.78) (2.60) (2.58) (2.71) (2.53) (2.44)
Medium education 0.004 0.011 0.011 0.017 0.041** 0.045** 0.058***
(0.22) (0.73) (0.72) (1.08) (2.43) (2.60) (2.70)
High education –0.019 –0.011 –0.010 –0.005 0.029* 0.031* 0.035*
(–0.83) (–0.50) (–0.49) (–0.27) (1.78) (1.88) (1.74)
PPCA: Large family –0.066*** –0.085*** –0.084*** –0.084*** –0.079*** –0.080*** –0.055***
(–6.02) (–6.44) (–5.91) (–5.91) (–5.80) (–5.43) (–3.22)
Log(size of town) 0.007 0.031 0.029 0.033 0.010 0.011
(0.41) (1.62) (1.55) (1.37) (0.44) (0.47)
Log(equiv. income) –0.164 –0.163 –0.110 –0.189 –0.231 –0.326
(–0.59) (–0.58) (–0.39) (–0.73) (–0.91) (–1.17)
Log(equiv. income) sq. 0.092 0.090 0.093 0.129 0.158 0.287
(0.35) (0.34) (0.35) (0.52) (0.65) (0.99)
Unemployed 0.101*** 0.099*** 0.090*** 0.093*** 0.089*** 0.081***
(5.77) (5.72) (5.05) (5.56) (5.23) (4.78)
PPCA: wealth –0.002 –0.005 –0.039** –0.029* –0.029*
(–0.08) (–0.25) (–2.14) (–1.67) (–1.73)
PPCA: direct networks 0.151*** 0.148*** 0.141***
(6.97) (6.51) (5.71)
PCA: indirect networks 0.137*** 0.137***
(4.09) (4.00)
PCA: trust in local 0.002
inst. (0.13)
PCA: trust in EU 0.037*** 0.017
(2.95) (1.05)
R^2 0.092 0.108 0.106 0.108 0.151 0.153 0.080
N 9,407.000 7,123.000 7,055.000 7,029.000 6,969.000 6,568.000 6,593.000
Source: Author’s calculations.
Note: The table contains standardized beta coefficients. t statistics in parentheses.
The dependent variable is binary and takes a value of 1 if the respondent intends to move abroad within the next 12 months. The addition “PCA”/”PPCA” in a variable name ­indicates
that the variable is taken from a (polychoric) principal component analysis.
All specifications contain a constant. Specifications (1) to (6) include country fixed effects, specification (7) includes PSU fixed effects. Specifications (4) to (6) include three principal
components representing regional economic development.
* p < 0.1, ** p < 0.05, *** p < 0.01

The country-specific estimates point toward differences in the average effects between EU and non-EU CESEE countries, with Romania and Bulgaria being ­notable exceptions. When we interact the dissatisfaction variable with an EU dummy (in the pooled sample), we see this confirmed: The relationship between dissatisfaction with public services and migration intentions is statistically significantly lower in EU CESEE countries.

Chart 5 shows bars representing O L S coefficient estimates for each country, which are: Albania: 0.029 (statistically significant); Romania: 0.014 (statistically significant); Serbia: 0.013 (statistically significant); Bulgaria: 0.009; North Macedonia: 0.008; Croatia: 0.006 (statistically significant); Czechia: 0.006 (statistically significant); Bosnia and Herzegovina: 0.005; Hungary: -0.006; Poland: -0.006.

Source: Author's calculations based on OeNB Euro Survey data (2018).
Chart 6 shows bars representing O L S coefficient estimates for specific types of public service, which are: social security: 0.0157 (statistically significant); public infrastructure: 0.0118 (statistically significant); education: 0.0074; health: 0.0143 (statistically significant); defense and public safety: 0.0076; economic development: 0.0138 (statistically significant). 

Source: Author's calculations based on OeNB Euro Survey data (2018).

Distinguishing by the six different types of public services, the estimates plotted in chart 6 show that dissatisfaction with the public social security system, health services, services related to economic development and the public infrastructure are associated with statistically significant increases in migration intentions. Here, the variables capturing dissatisfaction with each service are dummy variables, hence we refrain from portraying standardized coefficients and show the unstandardized ­coefficient estimates instead. Dissatisfaction with a public service type is associated with a 1.2 to 1.6 percentage point increase in migration intentions. Dissatisfaction with education and ­defense and safety does not show up as significant.

We further investigate heterogeneous effects with respect to individuals’ exposure to public services. 12 More specifically, we interact respondent-­specific characteristics that indicate that a respondent may be particularly exposed to an individual service with a dummy that indicates dissatisfaction with this service. We find that having children under the age of 6 leads to a significant effect of dissatisfaction with education services and increases the ­effect of dissatisfaction with health and defense and public safety. Furthermore, among the self-employed, the impact of dissatisfaction with public services related to economic development is higher than among non-self-employed. We also interact dissatisfaction with social security with having small children, being unemployed or retired, but do not find statistically significant interactions.

4.3 IV estimation

In order to consider possible endogeneity issues that can cause biases in OLS estimates, we perform instrumental variable (IV) estimations using different IVs. Table 2 shows the results and indicates the IVs used in the bottom panel (the corresponding first-stage results are displayed in table D1 in annex D).

Table 2: IV estimations  
(1) (2) (3) (4) (5) (6)
OLS IV IV IV IV IV
PPCA: dissatisfaction 0.00755*** 0.124** 0.0925** 0.0145*** 0.0161*** 0.0160***
with public services (3.80) (2.48) (2.11) (2.71) (3.00) (2.98)
Observations 6,615 6,604 6,604 6,615 6,604 6,604
Kleibergen-Paap F 4.400 3.173 27.33 24.11 22.50
Hanson-J 1.633 7.073 24.25 31.58 36.34
Hanson-J p 0.652 0.215 0.390 0.248 0.164
Instrumental variables
PCA road density (4 components) Yes Yes No Yes Yes
PCA urban fabric (2 components) No Yes No No Yes
State spending inadequate No No Yes Yes Yes
Source: Author’s calculation.
Note: All specifications include a list of covariates and a full set of country dummies.
t statistics in parentheses. * p < 0.1, ** p < 0.05, *** p < 0.01.

The results remain positive and statistically significant after instrumentation, and the coefficients increase in magnitude. When only externally merged geo-referenced data are used as IVs (road density and urban fabric), the IVs are weak, as indicated by low Kleibergen-Paap statistics. The high coefficient estimates in columns (2) and (3) should thus be disregarded due to weak instruments. The first-stage results, displayed in table D1 in the annex, confirm that urban fabric has no significant impact on the perceived quality of public services, and also the PCAs based on different road types are only weakly associated with the rating of public services quality. The first-stage F statistic improves considerably, however, when the variable measuring the respondents’ view of the adequacy of public spending on public services is added as instrument. The first-stage results show that respondents who think that state spending on public services should be maintained at the current level (as opposed to increased) exhibit a lower degree of dissatisfaction with public services. This confirms that the opinion on the adequacy of state ­spending can be related to the perceived quality of public services. Columns (4) to (5) display the results for different combinations of IVs, all including the adequacy of state spending as instrumental variables, and indicate that coefficients are considerably larger than when estimated with OLS. When we standardize the coefficients in (4) to (6), they average at approximately 0.1 and are thus twice as high as the OLS standardized coefficients. A 1 standard deviation increase in dissatisfaction with public services thus increases migration intentions by approximately 3 percentage points.

While contentment with public services is not the most important factor for explaining migration intentions, with sociodemographic factors, unemployment and networks explaining a large share of the variation in the data, it is certainly an important nonpecuniary factor that robustly relates to individuals’ aspirations to emigrate.

5 Conclusions

Data from a recent wave of the OeNB Euro Survey suggest that across CESEE, between 3% and 20% of individuals aged between 25 and 64 intend to leave their home countries within the next 12 months. The population-weighted CESEE ­average, which takes into account the considerable differences in country size, indicates that around 6% of individuals in the region intend to emigrate. Average migration intentions in the pooled sample, which represents an “average CESEE country,” amount to 9%. Across the ten countries in the sample, average migration intentions vary considerably and are highest in North Macedonia (20%) and lowest in Czechia (less than 3%). Descriptively, the data confirm that young people and men have stronger migration intentions.

Ongoing emigration from CESEE, especially of young people, adds to ­population aging, declining labor forces and increasing dependency ratios, and poses further demographic challenges to the countries in the region. Many factors that have been identified as relevant drivers of migration intentions are difficult to directly ­address by policymakers, such as sociodemographic factors or networks abroad. For this reason, we think it is particularly important to study contributing factors that can be tackled by policymakers. Therefore, we focus on individuals’ contentment with public services in their home countries, studying its relationship to migration ­intentions. Descriptively, we see that migration intentions are stronger among ­respondents who are dissatisfied and weaker among those who are not ­dissatisfied with the quality of public services. This finding is confirmed in an OLS framework, where we control for sociodemographic and economic factors, proxies for wealth holdings, economic development in the home region, network effects and trust in local and EU institutions. Respondents who are more dissatisfied with public services are more likely to have the intention to move abroad. This effect holds for our index of dissatisfaction that combines the responses of the six public services categories distinguished in the survey: social security, public ­infrastructure, education, health, defense and public safety, and economic development. Apart from education and public safety, the relationship is also significant when we ­distinguish between public services categories. We allow the effects of ­dissatisfaction with public services to differ across countries and find that in many countries, the effect is similar to the average effect in the sample. A notable exception is Albania, where the coefficient estimate is more than twice as high as in Romania and Serbia, the two countries that rank second and third in terms of effect size. Robustness checks confirm, however, that the overall effect is not driven by Albania, or another single country in the sample. Instrumental variable estimations, which we carry out to address possible endogeneity issues, reassure the OLS findings: The relationship between dissatisfaction with public services and migration intentions is positive and statistically significant. The IV estimates indicate a higher coefficient estimate than the OLS framework.

The link between people’s contentment with public services quality and migration bears the risk of a vicious cycle: With more individuals emigrating, public finances can get under increasing pressure, which leaves less room for improving the quality of public services or may even lead to a further deterioration. This, in turn, may further add to emigration pressures. At the same time, there is also the chance of creating a virtuous cycle: Appropriate policies would lead to a strengthening of public services in CESEE countries. Sound social security systems, high quality education and health care, good infrastructure, safe living conditions and publicly supported regional economic development initiatives would provide for an environment that may reduce individuals’ aspirations to move abroad, thereby strengthening public finances and creating space for public service improvements that further incentivizes people to stay. At the same time, these developments can have a positive impact on return migration and/or immigration to CESEE. Public services quality is certainly not the only factor that influences migration intentions, but it is a relevant one and, importantly, one that policymakers have the power to change.

References

Basu, D. 2015. Asymptotic bias of OLS in the presence of reverse causality. Working Paper
No. 2015-18. University of Massachusetts, Department of Economics.

Batog, C., E. Crivelli, A. Ilyina, Z. Jakab, J. Lee, A. Musayev, I. Petrova, A. Scott, A. Shabunina, A. Tudyka, X. C. Xu and R. Zhang. 2019. Demographic Headwinds in ­Central and Eastern Europe. European Departmental Paper Series No. 19/12. International Monetary Fund.

Borjas, G. 1987. Self-Selection and the Earnings of Immigrants. In: America Economic Review 77(4). 531–553.

Chen, Y. and S. S. Rosenthal. 2008. Local amenities and life-cycle migration: Do people move for jobs or fun? In: Journal of Urban Economics 64. 519–537.

Chiquiar, D. and G. H. Hanson. 2005. International Migration, Self-Selection, and the Distribution of Wages: Evidence from Mexico and the United States. In: Journal of Political Economy 113(2). 239–281.

Chiswick, B. 1999. Are Immigrants Favorably Self-Selected? In: American Economic Review 89(2). 181–185.

Docquier, F., G. Peri and I. Ruyssen. 2014. The Cross-country Determinants of Potential and Actual Migration. In: International Migration Review 48(s1). 37−99.

Dustmann, C. and A. Okatenko. 2014. Out-migration, wealth constraints, and the quality of local amenities. In: Journal of Developing Economics 110(C). 52–63.

EBRD. 2018. Work in Transition. Transition Report 2018-19. European Bank for Reconstruction and Development.

Etling, A., L. Backeberg and J. Tholen. 2018. The political dimension of young people’s ­migration intentions: evidence from the Arab Mediterranean region. In: Journal of Ethnic and Migration Studies. 1–17.

Giulietti, C. and J. Wahba. 2013. Welfare migration. In: International Handbook on the ­Economics of Migration. Edited by A. F. Constant and K. F. Zimmermann. Edward Elgar Publishing. 489–504.

Hotelling, H. 1933. Analysis of a complex of statistical variables into principal components.
In: Journal of Educational Psychology 24. 417–441 and 498–520.

Kolenikov, S. and G. Angeles. 2004. The Use of Discrete Data in PCA: Theory, Simulations, and Applications to Socioeconomic Indices. CPC/MEASURE Working Paper No. WP-04-85.

Liebig, T. and A. Sousa-Poza. 2004. Migration, Self-Selection and Income Inequality: An International Analysis. In: Kyklos 57(1). 125–146.

Manchin, M. and S. Orazbayev. 2018. Social networks and the intention to migrate. In: World Development 119(C). 360–374.

Meijer, J. R., M. A. Huijbregts, K. C. G. J. Schotten and A. M. Schipper. 2018. Global patterns of current and future road infrastructure. In: Environmental Research Letters 13(6). 1–10.

Otrachshenko, V. and O. Popova. 2014. Life (dis)satisfaction and the intention to migrate: ­Evidence from Central and Eastern Europe. In: The Journal of Socio-Economics 48. 40–49.

Pearson, K. 1901. On lines and planes of closest fit to systems of points in space. In: Philosophical Magazine 2. 559–572.

Raggl, A. K. 2019. Migration intentions in CESEE: sociodemographic profiles of prospective emigrants and their motives for moving. In: Focus on European Economic Integration Q1/19. Vienna: OeNB. 49–67.

Tran, N. T. M., M. P. Cameron and J. Poot. 2019. Local Institutional Quality and Return ­Migration: Evidence from Viet Nam. In: International Migration 57(4). 75–90.

Tran, N. T. M., M. P. Cameron and J. Poot. 2021. Perception of Institutional Quality Difference and Return Migration Intention: The Case of the Vietnamese Diaspora. Discussion Paper Series CDP 14/21. Centre for Research and Analysis of Migration (CReAM).

Vienna Institute of Demography (VID) and International Institute for Applied Systems Analysis (IIASA). 2020. European Demographic Datasheet 2020. Key Findings. Wittgenstein Centre for Demography and Global Human Capital. Vienna.

van Dalen, H. P. and K. Henkens. 2013. Explaining Emigration Intentions and Behaviour in the Netherlands, 2005-10. In: Population Studies 67(2). 225–241.

Williams, A. M., C. Jephcote, H. Janta and G. Li. 2018. The migration intentions of young adults in Europe: A comparative, multilevel analysis. In: Population, Space and Place 24(1). 1–16.

Annex A: Migration intentions by gender, education and by country

Table A1: Share of individuals aged 25 to 64 with migration intentions  
Gender Difference Education Difference
All Men Women Men vs. women Low Medium High High vs. low + medium
%
Czechia 2.7 3.3 2.0 10.9 2.0 4.8
Poland 3.7 5.8 1.9 ** 1.9 2.9 7.6 *
Hungary 4.3 3.6 5.0 3.3 3.9 6.0
CESEE average
(population-weighted)1
6.0 7.5 4.7 *** 4.8 5.8 7.8 **
Croatia 6.8 8.3 5.5 2.1 6.3 9.5
Romania 6.9 7.6 6.3 0.0 7.7 4.4
Bulgaria 7.5 8.1 7.0 14.9 6.7 7.7
Average over 10 countries2 9.1 11.0 7.3 *** 9.4 8.8 9.8
Serbia 11.1 13.9 8.4 ** 7.8 9.4 18.7 ***
Bosnia and Herzegovina 12.9 16.7 8.8 *** 5.2 16.4 10.8
Albania 14.2 17.1 11.7 ** 17.7 15.1 12.4
North Macedonia 20.0 24.2 16.0 ** 20.6 22.0 13.3 **
Source: Author’s calculations based on OeNB Euro Survey data (2018).
Notes: Survey weights applied. The columns labeled “Difference” indicate whether the mean is statistically different between two subgroups. Statistical significance is based on t-tests from
robust OLS estimations of migration intentions on gender and education dummies, respectively. *, ** and *** indicate a 10%, 5% and 1% level of significance, respectively.
1 Survey-weighted country averages are weighted by the population aged 25 to 64 of each country.
2 Simple average over survey-weighted averages of the ten countries.

Annex B: Variables used in the OLS and IV estimations

Table B1: List of variables used in OLS and IV estimations  
Variable Description
Dependent variable
Migration intentions Dummy variable that takes a value of 1 if respondent intends to move abroad within the next 12 months;
respondents stating “don’t know” or “no answer” are excluded from the analysis.
Quality of public services
PPCA: dissatisfaction with public services Principal component that represents low satisfaction with the six categories of public services; the PCA is
based on the six original variables asking about satisfaction with public services (Likert-type scale of answers).
Dissatisfaction with… Dummy variable that takes a value of 1 if respondent is (very) dissatisfied with…
… social security … social security (unemployment compensation, public pension, benefits for families and children)
… public infrastructure … public infrastructure (e.g. public road and town construction, railway network, public transport)
… education services … education (e.g. public kindergartens, schools, universities)
… health services … health (e.g. public hospitals)
… defense and public safety … defense and public safety (e.g. police, justice system)
… economic development … economic development (e.g. support for small and medium-sized companies, investment allowances,
­financial support for disadvantaged regions)
Sociodemographic factors
Age, age squared Age of respondent and its square
Medium education Dummy variable that takes a value of 1 if respondent has medium education (i.e. lower and upper
­secondary, post-secondary but non-tertiary)
High education Dummy variable that takes a value of 1 if respondent has high education (i.e. first and second stage of tertiary)
Female Dummy variable that takes a value of 1 if respondent is female
PPCA: large family Principal component that represents members of large families, i.e. married individuals from large
households, with small children
Log(size of town) Logarithm of the size of the residence town
Individual economic factors
Log(equiv. household income) [sq.] Logarithm of the equivalized household income [and its square]; equivalized household income is
­computed using a weight of 1 for the first adult in the household, 0.5 for each additional person aged 13
and over and 0.3 for each child under the age of 13. (It is more common to use the age of 14 as a cutoff
­between a weight of 0.5 and 0.3, but this is not possible in our data, and we use 13 instead.)
­Non-­response in the income variable enters as missing values.
Unemployment Dummy variable that takes a value of 1 if respondent is not working but seeking a job
PPCA: wealth Principal component that represents real estate ownership (ownership of residence, secondary residence,
other real estate and other land, including car ownership)
Regional development
PCA: prosperous region Principal component that represents individuals living in regions with low unemployment, moderate
­income, high activity and low growth in activity (prosperous, stable region)
Regional income/unemployment is calculated as the (survey-weighted) average of equivalized household
income/individual unemployment using survey weights. Economic activity is measured as the logarithm of
night light intensity in 2015, 2016, 2018 (data source: Earth Observation Group, Visible and Infrared Imaging
Suite, VIIRS). Growth in activity is measured as the log-difference in night light intensity between 2005
and 2013 (data source: Earth Observation Group, Defense Meteorological Satellite Program –
Operational Linescan System, DMPS-OLS). All variables are calculated at different levels of regional
aggregation: For night light data, we use the 10km and 20km radius around respondents’ residences and
the NUTS 2 level; average income and unemployment are aggregated to the PSU and the regional level.
PSU is the ­primary sampling unit and represents households in close proximity of the respondent’s
residence, the ­regions are defined based on NUTS 2 classifications – or finer in some countries
(HR, BG, MK).
PCA: depressed region Principal component that represents individuals living in regions with high unemployment, low income,
moderate activity and moderate growth in activity
PCA: developing region Principal component that represents individuals living in regions with moderate unemployment, high
­income, low activity but high growth in activity
Source: Author’s compilation.
Note: Unless otherwise noted, the source of all variables is the OeNB Euro Survey carried out in fall 2018.
Table B1 continued: List of variables used in OLS and IV estimations  
Variable Description
Network effects
PPCA: direct networks Principal component that represents individuals with direct networks abroad. PPCA contains a dummy
variable that takes a value of 1 if respondent and/or their partner receives remittances from abroad and a
dummy variable that indicates close family living abroad.
PCA: indirect networks Principal component that represents individuals with indirect networks abroad. PCA contains the share of
remittance receivers in the PSU and in the region and the share of respondents in the PSU and in the
region that has family living abroad.
Trust in institutions
PCA: trust in local institutions Principal component that represents trust in national institutions (trust is measured on a Likert-type scale;
trust variables are demeaned before they enter the PCA)
PCA: trust in the EU Principal component that represents trust in the EU
(The PCA is performed based on trust in the government/cabinet of ministers, the police, domestically
owned banks, the national central bank, foreign owned banks, and the EU.)
INSTRUMENTAL VARIABLES
PCA road density (4 comp.) The first four components of a PCA on the following indicators of road density: The density (m/km2) of
five types of roads (highways, primary, secondary, tertiary and local roads) based on 8km/8km raster data
as well as based on vector data calculated for 5km/10km/20km radii around the place of residence
(data source: Global Road Inventory Project (GRIP) dataset, Meijer et al., 2018).
PCA urban fabric (2 comp.) The first two components of a PCA on the following indicators of urban fabric: CORINE continuous and
discontinuous urban fabric in 500m/1km/2km/5km/10km/20km radii around the place of residence
(data source: CORINE landcover data).
PPCA: inadequate state spending on
public services
Principal component that represents individuals that consider state spending on the six types of public
services as inadequate. The six original variables enter the PPCA.
Source: Author’s compilation.
Note: Unless otherwise noted, the source of all variables is the OeNB Euro Survey carried out in fall 2018.

Annex C: Robustness

Chart C1, Robustness of coefficient estimate to omission of countries, shows eleven bars representing O L S coefficient estimates of the dissatisfaction variable. The first bar is estimated using the full sample and the value is 0.007. The other ten bars are estimated excluding one of the countries each to test the robustness of the results to the omission of a country. When omitting Albania, the coefficient estimate is 0.005; when omitting Bosnia and Herzegovina, it is 0.007; when omitting Bulgaria, it is 0.007; when omitting Czechia, it is 0.007; when omitting Croatia, it is 0.007; when omitting Hungary, it is 0.009; when omitting North Macedonia, it is 0.007; when omitting Poland, it is 0.009; when omitting Romania, it is 0.007; when omitting Serbia, it is 0.006. All coefficient estimates are statistically significant. 

Source: Author's calculations based on OeNB Euro Survey data (2018).

Annex D: First-stage results of IV estimations

Table D1: IV estimations – first-stage results corresponding to columns (2) to (6) in table 2  
(2) (3) (4) (5) (6)
1st stage IV 1st stage IV 1st stage IV 1st stage IV 1st stage IV
PCA: high road density –0.0164 –0.0200 0.00433 –0.00350
(–0.87) (–0.89) (0.24) (–0.16)
PCA: highways –0.0344** –0.0357** –0.0192 –0.0201
(–2.41) (–2.49) (–1.45) (–1.51)
PCA: primary roads 0.0419** 0.0425** 0.0393** 0.0399**
(2.46) (2.49) (2.43) (2.46)
PCA: few medium roads –0.0171 –0.0206 –0.0312 –0.0314
(–0.85) (–1.00) (–1.64) (–1.60)
PCA: high urban fabric –0.0167 0.00225
(–0.57) (0.08)
PCA: high urban fabric
close to home
–0.0209 –0.0237
(–0.69) (–0.83)
State spending on social security
Maintained –0.242*** –0.243*** –0.243***
(–4.83) (–4.82) (–4.83)
Lowered 0.218*** 0.216*** 0.214**
(2.62) (2.58) (2.55)
Do not know 0.234* 0.226* 0.224*
(1.87) (1.81) (1.79)
No answer 0.0810 0.0856 0.0796
(0.12) (0.13) (0.12)
State spending on public
infrastructure
Maintained –0.309*** –0.310*** –0.309***
(–6.41) (–6.41) (–6.39)
Lowered 0.0559 0.0601 0.0616
(0.78) (0.84) (0.86)
Do not know 0.236 0.242 0.244
(1.56) (1.60) (1.61)
No answer 1.766** 1.770** 1.762**
(2.46) (2.45) (2.42)
State spending on education
Maintained –0.306*** –0.300*** –0.299***
(–5.99) (–5.82) (–5.80)
Lowered 0.0280 0.0270 0.0266
(0.27) (0.26) (0.26)
Do not know 0.268* 0.277* 0.277*
(1.85) (1.90) (1.90)
No answer 1.110*** 1.121*** 1.125***
(2.85) (2.85) (2.88)
State spending on health
Source: Author’s calculations.
Note: t statistics in parentheses .
* p < 0.1, ** p < 0.05, *** p < 0.01
Adequacy of the state spending on the different public services: All answer categories are included in the estimations, with the response
“state spending should be increased” being the reference category.
First-stage estimations also include all control variables of the second stage, but they are omitted from this table.
Table D1 continued: IV estimations – first-stage results corresponding to columns (2) to (6) in table 2  
(2) (3) (4) (5) (6)
1st stage IV 1st stage IV 1st stage IV 1st stage IV 1st stage IV
Maintained –0.303*** –0.304*** –0.304***
(–5.17) (–5.16) (–5.15)
Lowered 0.130 0.122 0.124
(1.32) (1.24) (1.25)
Do not know 0.416** 0.414** 0.415**
(2.26) (2.25) (2.25)
No answer 0.404 0.389 0.399
(0.56) (0.53) (0.55)
State spending on defense
Maintained –0.0630 –0.0645 –0.0645
(–1.30) (–1.33) (–1.33)
Lowered 0.135** 0.134** 0.135**
(2.10) (2.07) (2.08)
Do not know 0.555*** 0.543*** 0.543***
(4.83) (4.74) (4.73)
No answer –0.248 –0.278 –0.272
(–0.75) (–0.86) (–0.85)
State spending on
economic devel.
Maintained –0.0269 –0.0293 –0.0285
(–0.54) (–0.59) (–0.57)
Lowered 0.156* 0.150* 0.152*
(1.80) (1.72) (1.74)
Do not know 0.591*** 0.590*** 0.590***
(6.07) (6.08) (6.08)
No answer 0.403 0.435* 0.432
(1.53) (1.65) (1.64)
Observations 6,604 6,604 6,615 6,604 6,604
Source: Author’s calculations.
Note: t statistics in parentheses .
* p < 0.1, ** p < 0.05, *** p < 0.01
Adequacy of the state spending on the different public services: All answer categories are included in the estimations, with the response
“state spending should be increased” being the reference category.
First-stage estimations also include all control variables of the second stage, but they are omitted from this table.

2 Oesterreichische Nationalbank, International Economics Section, anna.raggl@oenb.at. Opinions expressed by the authors of studies do not necessarily reflect the official viewpoint of the OeNB or the Eurosystem. The author would like to thank Doris Ritzberger-Grünwald (OeNB) for initiating this research and providing useful inputs, two anonymous referees, the participants of the Women in Economics (WinE) Networking & Mentoring Retreat 2019 in Manchester, the participants of the Wittgenstein Centre (WIC) Colloquium in February 2020 as well as colleagues in the OeNB’s Economic Analysis and Research Department for helpful comments and valuable ­suggestions.

3 A related strand of the literature deals with so-called “welfare migration,” where the location choice of migrants is related to the generosity of the welfare system in the country of destination (see Giulietti and Wahba, 2013, for an overview). The empirical evidence for the existence of this so-called welfare magnet hypothesis is mixed, and when such magnet effects are found, they are often very small. In contrast to this literature strand, which deals with the welfare state’s role as a possible pull factor for migration, this study focuses on public services quality acting as a possible push factor.

4 In case of a simultaneity bias, the sign of the bias is not straightforward to assess. Even under simplifying assumptions, for this particular application it cannot be determined, as for the assessment not only the expected signs of the ­impact of the quality assessment on migration intentions and the impact of migration intentions on the quality ­assessments need to be known, but also whether the product of the two exceeds unity (see for example Basu, 2015).

5 The survey question refers to the following services: social security, public infrastructure, education, health, defense and public safety, and economic development; it reads as follows: “And in which of these areas, from your point of view, should the level of state spending be increased, maintained or lowered over the next 10 years?”

6 The survey covers six EU countries (Bulgaria, Croatia, Czechia, Hungary, Poland, and Romania) and four non-EU countries (Albania, Bosnia and Herzegovina, North Macedonia and Serbia). For further information, please refer to https://www.oenb.at/en/Monetary-Policy/Surveys/OeNB-Euro-Survey.html.

7 The exact wording of the question is: “Do you intend to move abroad within the next 12 months?”

8 We would caution against interpreting this finding as evidence that there is no (continued) brain drain. Migration intentions among the highly skilled are sizeable, albeit not statistically significantly higher than among those with lower levels of education. Also, the data describe migration intentions and not actual emigration. To the ­extent that the highly skilled are more likely to act on their intentions, actual migration can be relatively more frequent among those.

9 The exact wording of the question is the following: “Consider the following areas where the state spends money. How satisfied are you with the delivery of public services in these areas in [YOUR COUNTRY]? Social security, public infrastructure, education, health, defense and public safety and economic development.” The possible ­answers are the following: “Very satisfied, satisfied, dissatisfied, very dissatisfied, don’t know, no answer.”

10 The relationship between migration intentions and age is nonlinear, and the decline slows down with increasing age.

11 Please note that these are coefficient estimates and not standardized coefficients as they are based on interactions between variables.

12 The regression outputs are not displayed in this paper due to space limitations, but they are available upon ­request.

The e-motion of car manufacturing in CESEE: the road ahead

Tomáš Slačík 13

Central, Eastern and South-Eastern European (CESEE) countries have benefited considerably from the grand-scale relocation of car production sites to emerging markets over the last two decades. On the back of strong foreign direct investment, the automotive industry has thus become a major economic pillar in several countries and firmly integrated into global, predominantly European, value chains. More recently, the automotive industry has seen some challenging times, though. Global and particularly European car production has been losing steam, and the industry has been hit by major black swan events, most painfully the coronavirus pandemic and Russia’s invasion of Ukraine. What is more, the global automotive industry has been undergoing unprecedented structural shifts on the demand and supply side, such as autonomous driving, shared mobility, connectivity and, most notably, the transition to electric vehicles.

Against the background of these great changes and challenges, the present paper explores the emergence of the largely foreign-owned automotive industry in CESEE and its level of preparedness for managing the risks and uncertainties and seizing the opportunities implied by the ongoing development of the automotive industry. After collecting and analyzing relevant qualitative information we find that the CESEE car industry will be walking a thin line between adopting new technologies and sticking to the internal combustion engine for longer than Western countries. For CESEE countries to maximize the benefits and minimize the risks of the technological transformation in the car industry the key priority is to preserve close links with Germany, stay tuned for battery production and focus on activities and promising industries with higher value added.

JEL classification: F15, F60, L62

Keywords: automotive, electric, battery, transformation, CESEE

Over the last two decades, the frontiers of automotive manufacturing have been shifting toward emerging markets worldwide. The relocation of production from advanced countries has been a big asset for Central, Eastern and South-Eastern European (CESEE) countries in particular. The automotive industry has thus come to play an important role in several countries in the area, particularly Czechia, ­Slovakia, Slovenia (now among the top countries in terms of car production per capita worldwide), Hungary, Poland and Romania. Backed by strong inflows of foreign direct investment (FDI) in recent decades, the industry has been integrated into European and global value chains, and the sector has become a key growth driver for these economies.

However, the automotive industry has been hit hard by the COVID pandemic globally and in Europe in particular. In Europe, motor vehicle production plummeted by nearly a quarter year on year in 2020 (–13% worldwide) and dropped by another 4% from a record low base in 2021 (+3% worldwide). 14 Aside from subdued demand and the impact of numerous lockdowns and (cross-border) mobility constraints, this collapse was the result of major supply chain disruptions. In particular, semiconductor shortages have been slowing down car production significantly since late 2020. Meanwhile, Russia’s invasion of Ukraine and the ensuing economic sanctions have added to the strain on already battered automotive supply chains. Disruptions stem from the suspension of car production by several manufacturers, including Czech Skoda, and from manufacturers being cut off from key supplies of automotive inputs by Russian and Ukrainian firms such as wire harnesses and raw materials. In addition to the real ramifications, the war in Ukraine has put further upward pressure on already elevated prices of crucial raw materials and energy.

The pandemic- and war-triggered shocks to the automotive industry arrived amid a cyclical slowdown as car production had stagnated or even contracted for several years even before the pandemic across the globe, including Europe (especially Germany). Nonetheless, CESEE countries with a strong automotive focus largely defied these developments and navigated the challenges comparatively well. On top of these cyclical trends, the global automotive industry has been undergoing a fundamental transformation driven by unprecedented structural shifts on the ­demand and supply side. These include, in particular, autonomous driving, shared mobility, connectivity, new players entering the automotive arena and, last but not least, electrification as the most widespread means to address the ever-stricter CO2 emission targets. The COVID-19 crisis as well as the war in Ukraine are likely to accelerate many of these trends reflected in, inter alia, changes in supply chains, more rapid digitalization as well as acceleration of the electrification process to fast-track independence from fossil fuels.

Indeed, the automotive sector’s future is very much oriented toward electric mobility, at least in Europe. Apart from mounting peer and market pressures, this trend is largely driven by strengthening global efforts to address climate change and regulate carbon emission. With respect to the latter, Europe has been the world’s trend setter and frontrunner. Since the ever-stricter CO2 emission regulation standards refer to tailpipe emissions only, electric vehicles are carbon-neutral by definition. As a result, all major traditional automakers keep announcing ambitious electrification targets and time schedules. With car producers expanding and accelerating the provision of new electric vehicle models, their adoption has gained pace and the European market has moved into the driving seat regarding electrification. While pure electrics, plug-in hybrids and hybrids 15 accounted for some 4.5% of all new passenger car registrations across the EU in 2017, this share climbed to about 40% in 2021. 16 However, despite the recent boom of electric cars sales, internal combustion engine vehicles will not vanish overnight and will continue to play a role. Especially in CESEE and other emerging countries, the transition to electric vehicles is expected to be much slower than in advanced markets.

Given the advancing powertrain transformation, the development of battery technology and battery production will play a crucial role in the future. Battery packs and their main features such as size, weight and driving range are not only set to become the most important performance component of electric vehicles, creating differentiation among competitors, but also the key cost determinant. With China accounting for about three quarters of the Li-ion battery 17 production capacity, the European automotive industry needs to develop a competitive and innovative ­battery industry with all up- and downstream stages. To cover the rising battery demand, it will likely take some 20 battery pack production sites (so-called gigafactories) in Europe in 2030 and about 35 in 2040 (Deloitte, 2021).

Against this background, the present paper seeks to sketch out where the largely foreign-owned automotive industry in CESEE is coming from and where it is heading in light of the current dynamics and high uncertainties in the sector. By collecting and analyzing relevant qualitative economic arguments we aim to shed some light on what these historical structural and cyclical developments in the ­automotive industry imply for the CESEE economies and how they will walk the thin line between seizing the opportunities and managing the risks associated with electrification on the one hand and continuing to meet demand for (not) outgoing internal combustion engines on the other. 18 The paper is structured as follows. While the next section recapitulates the development of the automotive industry in CESEE over the last 30 years, section 2 collects evidence to assess how the car industry in the region is braced for the big trends of the near future. Section 3 complements the macroeconomic view with a firm-level perspective based on rather unique firm survey data collected by the European Investment Bank. Section 4 discusses the potential future impact of the big trends in the automotive sector, particularly of car electrification, on major macroeconomic variables in CESEE before last section concludes.

1 Dawn and heyday of the automotive industry in CESEE during transition

Building on the long tradition in mechanical engineering and a well-educated workforce, Western automotive companies grasped the historic opportunity brought about by the collapse of communism in 1989. They thus not only revitalized local brands such as Skoda (Czechia) or the Dacia (Romania) but also shifted their own production eastward. The largest share of CESEE automotive production is held by Volkswagen, which allocates almost a third of its European manufacturing to the region, closely followed by Stellantis 19 and Renault. In addition, some Korean and Japanese automakers have a substantial presence in the region. Altogether, more than every fifth motor vehicle factory in the EU is located in CESEE. Along with the automakers, a whole myriad of their suppliers shifted their production eastward. Furthermore, while the absolute number of start-ups in the automotive sector in Europe and in the CESEE region in particular is well below the number of US start-ups (chart 1), CESEE countries are home to several young and innovative automotive companies. Indeed, in relative terms the share of automotive start-ups is not only higher in CESEE than in other European regions but also higher than in the USA (chart 2). One of the most prominent examples of a successful automotive newcomer is Rimac, the Croatian producer of electric hyper-cars, which recently took over the iconic brand Bugatti.

1.1 The presence of CESEE start-ups in the automotive sector

Chart 1, Number of start-ups in the automotive sector, is a column chart which shows the absolute number of start-ups in the automotive sector (vertical axis) for five different regions displayed as columns on the horizontal axis: CESEE (176 start-ups), Southern Europe (247 start-ups), Western and Northern Europe (1021 start-ups), UK (418 start-ups) and USA (2232 start-ups), respectively. 

The aggregates are comprised as follows: CESEE: Bulgaria, Croatia, Czechia, Estonia, Hungary, Latvia, Lithuania, Poland, Romania, Slovakia and Slovenia. Southern Europe: Cyprus, Greece, Italy, Malta, Portugal and Spain. Western and Northern Europe: Austria, Belgium, Denmark, Finland, France, Germany, Ireland, Luxembourg, the Netherlands and Sweden.

The figures relate to firms that were founded between January 2008 and June 2021 and are still active in the automotive sector.

Source: Crunchbase, Eurostat, US Census Bureau, authors’ calculation.
Chart 2, Share of start-ups in the automotive sector, is a column chart which shows the share of start-ups in the automotive sector in % (vertical axis) among all start-ups for five different regions displayed as columns on the horizontal axis: CESEE (1.6%), Southern Europe (1.3%), Western and Northern Europe (1.5%), UK (1.1%) and USA (1.2%), respectively. 

The aggregates are comprised as follows: CESEE: Bulgaria, Croatia, Czechia, Estonia, Hungary, Latvia, Lithuania, Poland, Romania, Slovakia and Slovenia. Southern Europe: Cyprus, Greece, Italy, Malta, Portugal and Spain. Western and Northern Europe: Austria, Belgium, Denmark, Finland, France, Germany, Ireland, Luxembourg, the Netherlands and Sweden.

The figures relate to firms that were founded between January 2008 and June 2021 and are still active in the automotive sector.

Source: Crunchbase, Eurostat, US Census Bureau, authors’ calculation.

The automotive industry thus gained an important role in several CESEE countries over the last three decades. The combined value added by the sector in the ten CESEE countries represents more than 12% of the total automotive value added in the EU. Some CESEE countries have come to be highly specialized in automotive manufacturing. Indeed, in Czechia, Hungary and Slovakia, motor vehicle manufacturing is the lead manufacturing industry, like in Germany, generating about one-fifth of the manufacturing gross value added. In terms of aggregate output and employment, the automotive sector in CESEE is also comparable to Germany with which the industry in CESEE is closely intertwined. The sector thus employs about 1 million people in the CESEE region, some 37% of total EU automotive employment, despite the dramatic increase in robotization in recent years.

The automotive sector in particular and manufacturing in general has been the main target of the FDI flows that started pouring to CESEE countries in the 1990s, with Hungary having attracted the highest manufacturing share (43% of FDI stocks). Over the last two decades, the investment rate in the automotive sector measured as the share of gross fixed capital formation relative to gross value added averaged more than 36% in CESEE, nearly twice as much as in the rest of the EU.

However, as chart 3 shows, FDI stocks have remained rather flat in most CESEE countries since the global financial crisis in 2009. This echoes the fact that the strong relocation of automotive production to the CESEE region slowed down substantially after the global financial crisis. As a result, in contrast to the EU as a whole, the automotive industry in most CESEE countries experienced a markedly higher growth in value added between 2000 and 2008 than in the period after the financial crisis (chart 4). In the wake of strong FDI inflows, the automotive sector in CESEE countries has become one of the region’s key export drivers. Automotive exports thus accounted for a record high of 34% of total exports in Slovakia in 2019 and for more than 20% in Hungary, Czechia and Romania.

1.2 Growth of FDI stocks and value added by the automotive sector in selected CESEE countries

Chart 3, Growth of inward FDI stocks, is a line chart which shows the development of inward FDI stocks as a share of GDP (vertical axis) over the time span 1990 to 2020 (horizontal axis) for the following countries:  Bulgaria, Czechia, Croatia, Hungary, Poland, Romania, Slovenia, Slovakia and Serbia. All country lines start at nearly zero in the early 1990s (in 2000 in case of Serbia) and rise throughout the observed period, some rather gradually such as the line for Slovenia, others rather steeply such as the line for Bulgaria. However, since 2009 the lines for most countries have either leveled off or even declined (e.g. in Bulgaria and Romania stocks have been declining from a peak of 91% and about 80%, respectively. Hence, in 2020 overall inward FDI stocks ranged between about 36% of GDP in Slovenia and 91% of GDP in Serbia.

Note: Special-purpose entities excluded.

Source: wiiw FDI database.
Chart 4, Annual average growth rates of value added by the automotive sector, is a bar chart displaying two sets of horizontal bars for the following countries or regions (vertical axis): Poland, Czechia, Slovakia, Croatia, Hungary, Bulgaria, Serbia, Slovenia, EU, Romania. The first set of bars shows the average annual growth rates of value added by the automotive sector (on the horizontal axis) for the period 2000-2008. The second set of bars shows the average annual growth rates for the period 2011-2018. The countries are ordered according to the results for 2000-2008, with the bar values ranging between 21% in Poland and 2.2% for Romania. In the  2011-2018 period, the bar values range between 2.4% in Slovenia and 15% in Romania. For all countries but Serbia and Romania and for the EU as a whole the first bar value is (significantly) larger than the second bar value. 

Note: Automotive sector defined as NACE rev. 2, C29 (motor vehicles, trailers and semi-trailers). The data for the EU include the EU-27 and the UK.

Source: wiiw FDI database.

Moreover, FDI inflows were crucial in integrating the CESEE automotive sector into global and regional value chains, as a result of which the share of foreign value added in exports of transport equipment has risen to levels as high as 70% in ­Hungary, Slovenia and Slovakia, compared to less than 30% in Germany (Reiter and Stehrer, 2021). While the automotive sector has some of the longest value chains of any industry, it is, at the same time, organized into three main blocks that generally source within their respective regions: the EU, North America and Asia (OECD, 2012). In line with this stylized fact, the lion’s share of the foreign value added in CESEE countries’ exports of transport equipment comes from the EU, particularly Germany (chart 5). The German value added content constitutes a major share of exports in Hungary (19%) and Czechia (13%). The EU value added share in automotive exports is highest in Slovenia (24%) and Hungary (23%).

The deep integration of the CESEE automotive industry in global and particularly regional value chains and the high interlinkages with Germany notwithstanding, it has to be borne in mind that the automotive sector has a very different functional profile in Germany than in CESEE countries. The latter tend to specialize in value chain functions with lower value added and serve as factory economies in the European production networks (Stöllinger, 2021). Hence, the sector in CESEE typically shows a revealed comparative advantage in production while other more skill- and knowledge-intensive pre- and post-production functions – including headquarter services and R&D, but also sales and business services – are underrepresented. Conversely, Germany’s automotive industry has a comparative disadvantage in production, but substantial comparative advantages in headquarter services, R&D and business services, i.e. functions which tend to generate higher value added and pay higher wages.

Chart 5, Foreign value-added content of exports of transport equipment, is a stacked collumn chart showing on the vertical axis the foreign value-added content of exports of transport equipment in % of total exports for the following countries on the horizontal axis: Bulgaria, Czechia, Croatia, Hungary, Poland, Romania, Slovenia, Slovakia, North Macedonia, Serbia. For each country there are two stacked columns, one for 2010 and one for 2018. Each column shows the foreign value-added content of exports of transport equipment from the following regions: Germany, EU-27, Asia and other. 

Note: EU-27* denotes the EU-27 without domestic manufacturing and Germany. Asia includes China, South Korea and Taiwan. Transport equipment comprises the scope defined by NACE rev. 2, CL, C29 (motor vehicles, trailers and semi-trailers) and C30 (other transport equipment).

Source: wiiw multi-country input-output database (wiiw MC-IOD).

2 Automotive industry in CESEE in e-motion?

Turning to electrification, the picture across the region is somewhat varied, and automotive players in CESEE are largely dependent on decisions taken at their mostly foreign headquarters. Nonetheless, in general, CESEE countries are to a large extent involved in the electrification process. Moreover, many appear to have the potential to reap significant benefits from the electrification trend or even to become its key focal points, at least when it comes to the projected production of electric vehicles. For instance, Slovenia, Slovakia and Czechia are expected to have the highest level of battery electric vehicle production per capita in Europe in 2030 (Transport & Environment, 2021). Moreover, vehicle production in some countries in the region is projected to become exclusively (Slovenia) or predominantly (Poland for instance) focused on battery electric vehicles by 2030. In contrast, battery electrics will account only for about 50% of car production in Germany and for about 30% in Romania where the shift to electric vehicle production will be less significant and/or happen at a slower pace. In most of the CESEE countries, we see an emerging shift toward electric vehicles. In Czechia, all automotive players producing in the country (Skoda, Hyundai and Toyota) have launched at least some full electric vehicle or hybrid production. Particularly Skoda (Volkswagen Group) – which is investing EUR 2 billion in e-mobility – wants to turn Czechia into an e-mobility hub. In Slovakia, Stellantis has EUR 180 million in investment plans with the aim of gradually launching mostly hybrid and electric vehicles. Volkswagen has been producing its electric e-Up in Slovakia since 2013. Jaguar Land Rover is adding plug-in hybrid electric cars to its production portfolio in Slovakia. Stellantis’ FCA is investing EUR 165 million in its Polish plant to produce new hybrid and electric vehicles. Moreover, Volkswagen and MAN produce electric vans in the country and Poland is already the leader in manufacturing of electric buses. Furthermore, in 2020 the Polish government announced plans to create a state-owned electric vehicle company, ElectroMobility Poland, which is expected to launch its e-car production in fall 2024. In Hungary, the BMW plant to be opened in 2025 will exclusively produce electric cars, making it the key focal point of BMW’s strategy for e-mobility. Daimler announced plans to invest EUR 141 million in Hungary to add fully electric vehicles to its Hungarian production. The electric drives for the Audi e-tron have been manufactured in Hungary since 2018. The electric models for the Premium Platform Electric (PPE), developed jointly with Porsche, are also to be assembled at the Hungarian plant. In Slovenia, the Renault Twingo EV accounted for one-third of the Revoz plant’s output in 2021. In addition, the Slovenian plant also manufactures the Smart Forfour EV under a partnership with Daimler. While Romanian Dacia (Renault Group) has unveiled its first 100% electric model (Dacia Spring), the market’s most affordable electric vehicle will not be made in Romania but rather at the parent company Renault’s facility in China. Ford has announced plans to invest USD 300 million in Romania, aiming to electrify the entire commercial vehicle range from 2024 on. The Bulgarian manufacturing facility (EUR 143 million investment) of German electric car manufacturing startup Next.e.Go Mobile is scheduled to begin operations in 2024. Rimac, the Croatian e-hyper-car producer, takes on a special role in the CESEE region, contributing to innovation in the electric vehicle branch and producing supplies for already established producers.

Going hand in hand with the strengthening production of electric vehicles in CESEE has been the rising significance of electric powertrain technologies in the region’s exports in recent years. The year 2020 saw a particularly strong jump, with the share of electric and hybrid vehicle exports in total car exports climbing as high as 30% in Slovakia and Romania (see chart 6).

Chart 6, Exports of electric and hybrid cars, is a column chart showing on the vertical axis the export share of electric and hybrid cars in % of total car exports for the following countries listed on the horizontal axis: Germany, France, Spain, Italy, UK, Czechia, Slovakia, Poland, Romania, Hungary and Slovenia. For each country there are four bars for the years 2017, 2018, 2019 and 2020, respectively. Nearly all countries have seen a rise in the share of electric and hybrid car exports in total car exports over the last years, with a particularly large jump in 2020. The increase was especially pronounced in Slovaka, Romania, Poland and Czechia.

Note: Total cars denotes motor cars and other motor vehicles principally designed for the transport of less than ten persons (CN8703). Alternative cars include hybrid cars, plug-in models and pure electric cars. 2020 data for the United Kingdom not available.

Source: Eurostat Comext.

In addition to the production of electric vehicles, CESEE countries are making significant efforts to secure a role in battery production. On the back of the strongly rising demand for batteries, Deloitte (2021) estimates that 16 to 22 gigafactories will have to be built in total in Germany, Czechia, Hungary, Poland and Slovakia to keep electric vehicle assembly in this region. This is because the heavy weight of batteries renders long-distance logistics and shipping costly so that it is reasonable to place the production of batteries close to the assembly of electric vehicles. Against this background, Poland, Hungary, Slovakia and Czechia have already taken action to attract investment in gigafactories. While there has been a gigafactory in operation in Poland since 2017 and there are two in Hungary, one more is currently planned in each of these two countries. In addition, one gigafactory is planned in Slovakia and (up to) two in Czechia. Poland, whose exports of Li-ion batteries already amount to 2% of total exports, has attracted investments in gigafactories by means of direct financial support, free land transfer and tax incentives. Similarly, the Hungarian government has incentivized investments in its two existing gigafactories and planned future ones via direct financial support as well as free provision of the enabling infrastructure or some utilities. In addition to some financial support in favor of a planned gigafactory, Slovakia has set up the Slovak battery alliance, an independent advocacy group, to kick-start its battery industry. In ­Czechia, the government has so far only signed a memorandum of understanding with the electricity company CEZ concerning support for a gigafactory project.

Yet despite the strong involvement of the CESEE region in electrification, the implications are far more complex. The shift in production toward electric vehicles entails risks and opportunities for CESEE and the respective firms. Car manufacturers will need to find a right balance between electric and internal combustion engine vehicles, depending on the market they serve. This entails not only large investments in new and possibly also still in outgoing technologies but in light of the high uncertainty also optimization challenges when it comes to organizing R&D and production. Hence, some car producers in the CESEE region such as Skoda and Dacia are planning a much slower transition to electric vehicles, as they have been and will be serving markets where the onset of e-mobility will be slower. Skoda for instance has been mandated to manage activities and to strengthen the position of the entire Volkswagen Group in India and South-East Asia. This involves the launch of a slew of competitively priced mass-market cars with an internal combustion engine. Moreover, Skoda has also been charged with developing cars on a common platform (MQB-A0) for the entire Volkswagen Group. Based on this platform, individual car brands in the Volkswagen Group will produce vehicles destined in particular for India, Russia, South America and Africa. Overall, the reasons why internal combustion engine production will stay longer in CESEE are mainly the following: (i) the tendency of older technologies to persist longer in peripheral locations; (ii) the CESEE automotive industry’s relatively weak innovation capacities and (iii) its continued (labor and energy) cost advantage in the more labor-intensive internal combustion engine production (see Pavlínek, 2021; and CLEPA/PwC, 2021).

CESEE’s relatively limited innovative power strongly focused on the automotive sector, especially electric vehicles

Europe is a global leader in R&D investment in the automotive industry. One euro out of three spent in R&D in the EU goes to the automotive sector. However, R&D expenditures by car manufacturers are heavily concentrated in some EU countries, notably Germany and France. As a result, the EU is in pole position for innovative development in the automotive sector, as the steady upward trend for automotive patents – well ahead of the USA and China – suggests (chart B.1).

Chart B.1, Evolution of automotive patents in the EU, the USA and China, is a line chart that shows two sets of lines for the three mentioned regions for the period 2009 to 2020. The right vertical axis relates to the count of automotive patents, the left vertical axis to the share of automotive patents in total domastic patent portfolio in %. The EU outperforms the USA and China with respect to both the absolute number of automotive patents and the share in total domestic patents.

Source: Authors’ calculations based upon PATSTAT (PCT) data in collaboration with ECOOM (Centre for R&D Monitoring, Leuven, Belgium).

Yet in line with the above-mentioned functional specialization, the innovative power of the CESEE region’s automotive sector is limited. The absolute number of patents related to automotive innovation in the region is rather low compared to the other European regions. However, in relative terms, automotive patents account for close to 8% of all patents in CESEE, which is in line with the rest of Europe and points to an important relative specialization in this area (chart B.2). In particular, the CESEE region has seen a major uptake of climate-friendly patents in the automotive sector over the last 15 years. These patents are mainly related to the development of electric vehicles and electric vehicle charging (see charts B.3 and B.4). This relative specialization in electric vehicles is stronger than in other European regions. However, while the CESEE region has evolved into a significant knowledge center for vehicle electrification, it noticeably lags behind with respect to innovation in renewable energy and the related supportive technologies which are indispensable to making the electrification of vehicles truly climate-friendly. To sum up, CESEE countries do have a strong relative presence in innovation in the automotive sector as the region engages in cutting-edge research, development and innovation in addition to basic car assembly and part manufacturing. However, the large players in Western and Northern Europe clearly remain the technological leaders.

Chart B.2, Number and share of automotive patents in Europe, is a column and dot chart consisting of three bars, for CESEE, Southern Europe, and Western and Northern Europe, which are plotted on the horizontal axis. The right-hand vertical axis relates to the automotive patent count between 2017 and 2019. The left-hand vertical axis displays the share of automotive patents in the domestic patent portfolio. Whereas the automotive patent count in Northwestern Europe outnumbers the other two regions (about 11.000 vs. 820 in Southern Europe and 360 in CESEE), Northwestern Europe and CESEE are comparable with regard to relative shares (just below 8%), with Southern Europe lagging behind (4.3%).

Note: CESEE: Bulgaria, Croatia, Czechia, Estonia, Hungary, Latvia, Lithuania, Poland, Romania, Slovakia and Slovenia. Southern Europe: Cyprus, Greece, Italy, Malta, Portugal and Spain. Western and Northern Europe: Austria, Belgium, Denmark, Finland, France, Germany, Ireland, Luxembourg, the Netherlands and Sweden.

Source: Authors’ calculations based upon PATSTAT (PCT) data in collaboration with ECOOM (Centre for R&D Monitoring, Leuven, Belgium).

Green patents in the transport sector in 2018

Chart B.3, Number of green transport patents, is a column chart which shows on the vertical axis the absolute number of green patents in 2018 by subcategory in three different regions: Western and Northern Europe, Southern Europe and CESEE. The subcategories - depicted on the horizontal axis - for which the bars are shown are the following: electric vehicles, electric vehicle charging, conventional vehicles, hybrid vehicles, fuel efficiency-improving technologies, hydrogen technologies. Western and Northern Europe thus had more than 500 patents each in the categories electric vehicles and conventional vehicles, about 300 patents in the subcategory electric vehicle charging and 240 in the category hybrid vehicles. In the remaining two categories, i.e. fuel efficiency-improving technologies and hydrogen technologies, the number of patents was 75 and 60, respectively. In the two other regions the number of patents was in the range of 20-30 for the categories electric vehicles, electric vehicle charging and conventional vehicles. For the other categories the number of patents was negligible. 

Note: The classification of “green” domains is broadly based upon the methodology of Haščič and Migotto (2015).

Note: CESEE: Bulgaria, Croatia, Czechia, Estonia, Hungary, Latvia, Lithuania, Poland, Romania, Slovakia and Slovenia. Southern Europe: Cyprus, Greece, Italy, Malta, Portugal and Spain. Western and Northern Europe: Austria, Belgium, Denmark, Finland, France, Germany, Ireland, Luxembourg, the Netherlands and Sweden.

Source: Authors’ calculations based upon PATSTAT (PCT) data in collaboration with ECOOM (Centre for R&D Monitoring, Leuven, Belgium).