Life Satisfaction and Migration Intentions in Urban Romania: The Local-Global Divide
- Open Access
- 01.12.2025
- Research Paper
Erschienen in:
Aktivieren Sie unsere intelligente Suche, um passende Fachinhalte oder Patente zu finden.
Wählen Sie Textabschnitte aus um mit Künstlicher Intelligenz passenden Patente zu finden. powered by
Markieren Sie Textabschnitte, um KI-gestützt weitere passende Inhalte zu finden. powered by (Link öffnet in neuem Fenster)
Abstract
1 Introduction
According to the World Migration Report (2022), Romania ranks as the 17th largest origin country of international migrants globally (in absolute terms) and has the second-largest emigrant population among EU countries living abroad. These statistics reflect a broader regional trend of emigration that has characterized Central and Eastern Europe (CEE) since the fall of the Communist regime. According to the OECD (2019), Romania’s economic prospects and labor market conditions are the main reasons for emigration among Romanians, with young people typically expressing the strongest emigration intentions. Internal migration within Romania remains relatively limited compared to other EU countries. As rural-to-urban migration flows have significantly decreased over the past three decades, migration flows away from urban areas have gained increasing importance. In addition, according to the World Happiness Report (2021), Romania ranks among the bottom 20% of EU countries in terms of life satisfaction. In this context, understanding the factors behind Romanians’ migration intentions, including the relationship between subjective well-being (SWB) and migration intentions, as well as the specific drivers of the local–global divide (i.e., the choice between internal and international migration), is essential for designing more effective policies to mitigate the demographic consequences of migration.
In the literature, migration drivers have been extensively debated, with more recent research focusing on SWB. When analyzing intentions, rather than actual behavior, most studies find a negative relationship between emigration intentions and SWB, using a variety of methodologies and datasets (e.g. Brzozowski & Coniglio, 2021; Chindarkar, 2014; Graham, 2016; Ivlevs, 2015; Mendez, 2020; Otrachshenko & Popova, 2014). However, research comparing internal and international migration to identify similarities and differences in the relationship between SWB and migration intentions remains limited (Otrachshenko & Popova, 2014). As Erlinghagen et al. (2021) recently concluded, it remains unclear “whether empirical findings regarding subjective well-being in the course of internal migration can readily be transferred to international migration and vice versa.”
Anzeige
Although internal and international migration intentions have recently been analyzed, either separately or comparatively, through cross-country (Cai et al., 2014; Otrachshenko & Popova, 2014) or longitudinal studies (Shamsuddin & Katsaiti, 2020), incorporating individual-, regional-, and country-level data (Hendriks & Bartram, 2016; Polgreen & Simpson, 2011; Otrachshenko & Popova, 2014), to the best of our knowledge, the characteristics of lower-level local administrative units have not been included in analyses of international or outward migration from urban areas. Moreover, only a few studies have examined migration intentions among Romanians (Plopeanu et al., 2020; Mitrică et al., 2022), and only Otrachshenko and Popova (2014) analyzed life satisfaction in relation to both emigration and internal migration intentions across several European countries, including Romania. While the latter study analyzed permanent versus temporary migration to identify cross-country patterns, our paper provides an in-depth analysis of the Romanian case, distinguishing between short- and long-distance internal migration alongside emigration.
This paper focuses on urban Romania and examines the relationship between Romanians’ life satisfaction and migration intentions, considering both emigration and internal migration. The analysis combines individual- and county-level characteristics, using multilevel random-intercept multinomial logistic regressions and multilevel regressions estimated via the control function method. This multilevel approach facilitates a comparative exploration of how SWB relates to migration intentions, accounting for the diversity across Romanian counties. The relevance of this study extends beyond its contribution to academic literature. In Romania, where out-migration from urban areas has been increasing, understanding the relationship between SWB and migration intentions in a regional context may yield valuable insights for designing more effective and context-sensitive migration policies.
The paper is structured as follows: Sect. 2 provides a literature review on life satisfaction and migration (2.1), contrasts internal and international migration drivers (2.2), and outlines migration dynamics in post-communist Romania (2.3). Section 3 details data and methodology, Sect. 4 presents empirical results, and Sect. 5 concludes with key findings and policy recommendations.
2 Background and Related Literature
Migration has long been a central theme across the social sciences, studied in relation to both objective factors, such as income, poverty, and economic growth, and more recently, subjective dimensions like life satisfaction and well-being. The increasing availability of data on life satisfaction, along with the development of reliable and harmonized national and international databases, has stimulated a growing body of research analyzing life satisfaction in connection with migration, primarily through cross-sectional studies or short-term panel data. Because longitudinal migration data remain limited in many contexts, researchers frequently rely on migration intentions as a proxy for actual behavior. Although migration intentions may not perfectly predict actual migration behavior, numerous studies have demonstrated a strong and significant relationship between intentions and subsequent migration outcomes (e.g. de Haas & Fokkema, 2011).
Anzeige
2.1 Migration and Happiness
The pursuit of improved well-being is considered a core motivator of migration. Individuals with lower levels of life satisfaction tend to perceive higher expected gains from migration and are therefore more likely to express intentions to move. Empirical findings consistently support this relationship (Cai et al., 2014; Otrachshenko & Popova, 2014; Chindarkar, 2014; Graham, 2016; Brzozowski & Coniglio, 2021).
One theoretical lens for this relationship is the aspiration–capability framework (de Haas, 2021), which posits that migration requires both the desire to migrate and the capability to act. Individuals facing low levels of SWB may express a desire to migrate but lack the necessary resources or agency to realize that goal, as originally noted by Carling (2002).
Another influential explanation is the “frustrated achievers” theory (Graham & Markowitz, 2011), which suggests that individuals with relatively low levels of SWB, despite generally being better off than others, are more likely to intend to emigrate (Graham & Markowitz, 2011). Rooted in the concept of relative deprivation, this theory has become particularly influential in understanding migration intentions. At its core, it suggests that frustration arises not from absolute poverty, but from a widening gap between rising expectations and insufficient outcomes or limited social mobility (Gurr, 1970). “Frustrated achievers” are not necessarily poor, but rather members of the emerging middle class who consider migration as a means of fulfilling unrealized potential elsewhere (de Haas, 2021). The theory also highlights that economic development alone does not guarantee higher life satisfaction; instead, it is the alignment between expectations and reality that matters most. However, broader empirical evidence for the theory has been mixed. For instance, Kratz (2020), analyzing internal migration in Germany, found that economically motivated migrants reported higher life satisfaction before moving than those who remained, implying that dissatisfaction may not be a prerequisite for migration. Similarly, Ivlevs (2015) found that both highly satisfied and highly dissatisfied individuals may express migration intentions, challenging the idea of a simple linear relationship.
The relationship between life satisfaction and migration intentions may also be moderated by cumulative inertia, a concept suggesting that the longer someone lives in a particular location, the less likely they are to relocate. This is due to the increasing depth of social, psychological, and economic ties, which raise the real and perceived costs of moving (Myers et al., 1967). Consequently, residential mobility tends to decline over the life course, with older adults exhibiting a lower propensity to move even when dissatisfied (Coulter et al., 2016; Gillespie et al., 2024). Thus, cumulative inertia emerges as a powerful moderator of the relationship between SWB and migration intentions, especially in later life.
The SWB- migration link has been investigated in a range of methodological settings, including cross-country (Cai et al., 2014; Otrachshenko & Popova, 2014), longitudinal (Shamsuddin & Katsaiti, 2020), and aggregate-level studies (Polgreen & Simpson, 2011). Several researchers have employed multilevel models that account for individual-, regional-, and country-level factors (e.g., Hendriks & Bartram, 2016), yet very few incorporate lower-tier geographic characteristics such as county or city characteristics. This omission is particularly relevant for countries like Romania, where local disparities may significantly influence migration intentions.
2.2 Internal and International Migration
Traditional research on migration generally comprises two distinct strands of literature, addressing internal and international migration separately. While internal migration exceeds international migration in volume globally (Bell et al., 2015), the latter has attracted more scholarly and policy attention in recent decades. Nonetheless, the boundaries between these two forms of migration are often blurred. For instance, internal migration can serve as a first step toward international migration (Skeldon, 2006), while international migration may lead to internal mobility, as seen in the case of returning migrants relocating within their home country (Light & Johnston, 2009).
Over time, increasing attention has been paid to developing a unified theory of migration, based on the idea that national boundaries are becoming less significant, and that distance is no longer a key differentiating factor. However, such an approach must consider geographical, economic, and social dimensions. For instance, Adepoju (1998) argued that in sub-Saharan Africa, international migration is often an extension of internal mobility, as both are driven by similar socioeconomic pressures. Similarly, Molloy et al. (2011) pointed out that country size matters: larger countries typically offer greater internal mobility opportunities, whereas residents of smaller countries may need to cross borders to pursue life improvements. Despite these attempts, internal and international migration remain analytically and institutionally distinct. As King and Skeldon (2010) note, they are often studied using different methodologies and policy frameworks, raising the question: Does it matter conceptually whether migration is internal or international, in terms of explaining the movement or predicting outcomes?
This question becomes even more pressing when considering SWB. In their comparative study of internal and international migrants in Germany, Erlinghagen et al. (2021) found some commonalities in how SWB evolves before and after migration in both cases. Yet, they also highlight unresolved differences in anticipation and adaptation processes, concluding that findings on SWB from one form of migration cannot necessarily be generalized to the other.
2.3 Migration Dynamics in post-communist Romania
To provide a comprehensive understanding of Romania’s migration landscape, this section outlines key statistics and contextual features. Romania currently has the lowest net migration rate in the European Union, and Eurostat projections indicate that this negative trend is likely to persist for at least two more decades.
Romania’s EU accession in 2007, followed shortly by the 2008 global financial crisis, played a pivotal role in shaping outward migration, particularly to OECD countries such as Spain, Italy, and Germany. According to the most comprehensive institutional report on Romanian migration by the OECD (2019), between 2004 and 2016: (1) 62% of Romanian emigrants were under 30 years old; (2) 71% did not hold a job in the year prior to migration; (3) 37% were unmarried at the time of arrival; (4) 70% had a medium or high level of education; and (5) 78% lived in a household without children.
According to the Gallup World Poll, Romanians reported the highest emigration intentions in the EU during the period 2009–2018. While 26% of respondents stated that they would like to move abroad permanently if given the opportunity, only about a third of them (34%) indicated that they had concrete plans to leave the country within a year. The proportion of 8.84% of those intending to emigrate in the following year, based on Gallup data (2009–2018), is therefore quite close to the figure reported by the 2020 Urban Barometer survey used in this paper (5.25%). The OECD (2019), also drawing on Gallup data, found that unemployed individuals are the most likely to express emigration intentions - consistent with the findings presented here.
In addition to external migration, Romania has experienced significant internal mobility following the collapse of the communist regime. The process of market liberalization and uneven economic development, particularly the expansion of urban centers, has contributed to the intensification of internal migration. The internal migration rate reached approximately 10‰ in 2020. While in 1991, 70% of internal migration consisted of rural-to-urban moves, by 2000 this share had dropped to 40% of internal migration. This sustained trend of urban-to-rural migration is primarily driven by the pursuit of a better quality of life, including lower living costs in rural areas, proximity to nature, the expansion of residential zones outside major cities, a slower pace of life, and familial considerations such as caring for aging parents (Ghețău & Sandu, 2018).
Regional economic disparities also play a crucial role in shaping internal mobility. The most developed urban centres have evolved into regional growth poles, attracting intra-county migration from adjacent areas (Mitrică et al., 2022). In such cases, mobility is often motivated by the desire to improve access to economic opportunities while maintaining social and familial ties. By contrast, inter-county migration is more frequently driven by structural inequalities. Individuals residing in economically disadvantaged counties are more likely to relocate to wealthier regions in search of improved living standards, whereas those already living in prosperous areas tend to move within the same regional perimeter.
3 Data and Methodology
3.1 Data
Data are taken from the 2020 Urban Barometer, the first Romanian urban Barometer1, developed using the methodology of Flash Eurobarometer survey2. The Urban Barometer was released as part of a World Bank project aimed at supporting the development of Romania’s first national urban policy. It is a cross-sectional, representative survey of individuals residing in urban areas, covering 13,380 respondents3 across 41 representative Romania cities4. The survey addresses multiple dimensions of urban life, including satisfaction with infrastructure and public services, perceptions of city safety and environmental conditions, levels of social trust, views on local governance, and aspects of personal and financial well-being5.
Most of the data used in this analysis are at the individual level and sourced directly from the Urban Barometer. The only exception is a macroeconomic variable, GDP per capita, aggregated at the county (regional) level, which is provided by the Romanian National Institute of Statistics.
The main variable of interest is the intention to migrate for at least two years (abbr. Migration intentions), which has four categories: (1) I don’t intend to migrate; (2) I intend to relocate in another city within the same county; (3) I intend to relocate in another city within another county; and (4) I intend to relocate in another country. Categories (2) and (3) indicate internal migration, while category (4) refers to emigration. According to our data, 85% of respondents declare that they don’t intend to leave the city, 3.10% intend to relocate in another city within the same county, 6.65% intend to relocate in another city within another county, and 5.25% intend to emigrate.
As the main explanatory variable, we use the question asking respondents to rate their life satisfaction on a four-level scale from “1” (very satisfied) to “4” (very unsatisfied)6. In addition, a set of control variables commonly included in similar studies, along with two other specific variables, are also used. The first specific variable is based on the question: “In general, how much do you trust most people living in the same city as you?” This is rated on a four-level scale from “1” (high trust) to “4” (low trust) and is abbreviated as social trust. The cumulative inertia axiom is tested in the empirical section using the variable How long have you lived in the city?, which is ordinal and has five categories: (“1”- I have lived here since I was born, “2”- I have lived here for more than 10 years; “3”- I have lived here for 5–10 years; “4”- I have lived here for 1–5 years; “5”- I have lived here for less than 1 year).
Figure 1 presents migration intentions in relation to life satisfaction categories to provide preliminary insights. At a glance, life satisfaction appears to be significantly correlated with the intention to emigrate, but not with internal migration intentions. This preliminary finding will be further examined using quantitative methods.
Fig. 1
Life Satisfaction by migration intentions. Note: Data are from the Romanian Urban Barometer 2020
3.2 Methodology
We use two-level random intercept multinomial logistic regression models and standard multinomial logistic regression models to analyze the relationship between life satisfaction and migration intentions. One of the models also includes a correction for endogeneity. In the empirical section, migration intentions are explained using both individual-level and regional-level variables, with the regional level represented by all Romanian counties.
The hierarchical structure of the data, with individuals nested within counties, supports the use of a multilevel model that accounts for the nature of the dependent variable and allows us to distinguish between within-cluster and between-cluster effects. Since our main variable of interest, migration intentions, is nominal, we first apply a multilevel multinomial logistic regression after conducting preliminary tests.
This model also allows the inclusion of regional-level variables at the second level of analysis. However, it does not allow us to address the endogeneity problem, as the complexity of the estimation leads to model convergence issues. Therefore, to address endogeneity, life satisfaction is instrumented using the control function approach, applying multinomial logistic regressions with fixed effects for Romanian counties.
In this section, we present the multilevel multinomial logistic regression model, which has had relatively few empirical applications (Skrondal & Rabe-Hesketh, 2003; Wright & Sparks, 1994). Although the multilevel multinomial logit model can be formulated either using the random utility framework or as a Mixed Generalized Linear Model (GLM), we adopt the GLM formulation here, following Rasbash et al. (2009).
Let’s suppose that Yij is the multinomial response variable for individual i in county j, which takes value 1 for the decision to move in another city within the same county, value 2 for the decision to move in another county within the same country, value 3 for the decision to move in another country (to emigrate), and value 4 for the decision to not leave the residence city. For the four categories of the polytomous response variable, there are 6 pairs of categories, leading to a number of 6 linear predictors. The decision to not leave the residence city is selected as baseline category, so that the multinomial model runs three comparisons to this reference category.
The variable Yij has, conditional on the random effects, a multinomial distribution, taking values in the set of categories {1,…, k}, where k = 4 is the reference category for which all the parameters and the error terms are set to 0, and the conditional probability of Yij=1 is \(\:1/(1+{\sum\:}_{m=2}^{K}exp\left\{{\eta\:}_{ij}^{m}\right\})\).
We define the probability that category k is selected by individual i who belongs to cluster j by \(\:{\pi\:}_{ij}^{\left(k\right)}\), which can be expressed by Eqs. (1) and (2).
$$\:{\eta\:}_{ij}^{\left(k\right)}=log\left({\pi\:}_{ij}^{\left(k\right)}\right)$$
(1)
If we examine the matrix of log probabilities in Eq. (1), ηij can be decomposed into a small number of “contrasts” or effects (McCullagh & Nelder, 1989). The predicted probabilities of different values of x are helpful in explaining the effect of x:
$$\:{\pi\:}_{ij}^{\left(k\right)}=P\left({Y}_{ij}=s|\:{x}_{ij},{\epsilon\:}_{j},{\delta\:}_{ij}\right)=\frac{exp\left\{{\eta\:}_{ij}^{\left(k\right)}\right\}}{1+\sum\:_{m=2}^{K}{\eta\:}_{ij}^{\left(m\right)}}$$
(2)
As Rasbash et al. (2009) notice, a set of t-1 equations are estimated, contrasting each of the remaining response categories with the chosen reference category.
In the framework of GLS models, the general form of the two-level random-intercept multinomial logistic regression model with logit link and a single explanatory variable Xij in each of the t-1 contrasts, can be written:
$$\:log\frac{{\pi\:}_{ij}^{k}}{{\pi\:}_{ij}^{t}}={\alpha\:}^{\left(k\right)}+{\beta\:}^{\left(k\right)}{X}_{ij}+{\xi\:}_{j}^{\left(k\right)}+{\delta\:}_{ij}^{\left(k\right)}$$
(3)
Where, parameter \(\:{\beta\:}^{\left(k\right)}\) indicates the additive effect of a 1-unit increase in x on the log-odds of being in category k rather than category r. \(\:{\xi\:}_{j}^{\left(k\right)}\) and \(\:{\delta\:}_{ij}^{\left(k\right)}\) are vectors of random errors indicating the unobserved heterogeneity at county- and individual level, respectively. They are independent at different levels and fulfil the following assumptions:
$$\:{\xi\:}_{j}^{{\prime\:}}=({\xi\:}_{j}^{\left(2\right)},\:\dots\:,\:{\xi\:}_{j}^{\left(K\right)}){\prime\:}\sim\:N(0,{{\Omega\:}}_{\xi\:})$$
$$\:{\delta\:}_{ij}^{{\prime\:}}=({\delta\:}_{ij}^{\left(2\right)},\:\dots\:,\:{\delta\:}_{ij}^{\left(K\right)}){\prime\:}\sim\:N(0,{{\Omega\:}}_{\delta\:})$$
When considering \(\:{X}_{ij}^{{\prime\:}}\:\)a matrix of individual characteristics (age, income, gender, education, presence of children in household, marital status, unemployment, social trust, number of years lived in the residence city, social trust) and LS as being the level of life satisfaction reported by individuals, Eq. (3) become:
$$\:log\frac{{\pi\:}_{ij}^{k}}{{\pi\:}_{ij}^{t}}={\alpha\:}^{\left(k\right)}+{\beta\:}_{1}^{\left(k\right)}{X}_{ij}^{{\prime\:}}+{\beta\:}_{2}^{\left(k\right)}{LS}_{ij}^{{\prime\:}}+{\xi\:}_{j}^{\left(k\right)}+{\delta\:}_{ij}^{\left(k\right)}$$
(4)
Given that unobserved variables could affect both subjective variables, life satisfaction and the intention to emigrate, estimation of Eq. (4) is likely to be biased. The issue of endogeneity can be addressed by using the average life satisfaction of all individuals in the county where individual i resides, excluding individual i, as a proxy for their own level of life satisfaction, as shown in Eq. (5).
$$\:{LS}_{ij}={\beta\:}_{0}+{\beta\:}_{1\:}{Z}_{ij}+{\beta\:}_{2\:}{W}_{ij}+{\xi\:}_{j}{+\delta\:}_{ij}$$
(5)
In Eq. (5) which models the determinants of life satisfaction, \(\:{W}_{ij}\:is\:\)a matrix of explanatory variables of life satisfaction that could also be included in \(\:{X}_{ij}^{{\prime\:}}\), and \(\:{Z}_{ij}\)is the instrumental variable assumed to be associated with life satisfaction, but not correlated with the intention to leave the city. Equations (4) and (5) are estimated simultaneously.
As in Rasbash et al. (2009), we assume that in our multilevel models random-effects are category-specific in the sense that the intra-county correlation in migration decision may vary by intended destination. However, random effects may be correlated across contrasts, \(\:cov\left({u}_{j}^{\left(k\right)},\:{u}_{j}^{\left(r\right)}\right)=\:{\sigma\:}_{u}^{(s.r)},\:s\ne\:r\). For instance, we expect correlated random effects if there were unobserved county-level factors affecting the choice of more than one category of the migration intentions variable.
Different intercept parameters can be estimated for each category, as indicated by k subscript, or it is possible to constrain some of them to be equal. However, the intercept of the model is assumed to vary across clusters which are represented here by counties.
The estimation method is the maximum likelihood which formally assumes conditional normality, and the normal distribution of latent variables.
The validity of the multinomial logit model should first be examined under the Independence of Irrelevant Alternatives (IIA) assumption, which states that the relative probabilities of choosing between two alternatives are unaffected by the presence or characteristics of other alternatives. The Hausman-McFadden test and the Small-Hsiao test are the most commonly used tests for assessing this assumption. However, a growing body of literature suggests that these tests often perform poorly and may be unreliable in applied research (Cheng & Long, 2006).
Endogeneity is addressed here using the control function approach, applied in two steps (Guevara & Ben-Akiva, 2012). In the first stage, the endogenous variable (life satisfaction) is regressed on exogenous instruments. In the second stage, the residual from this regression is included as an additional explanatory variable in the original model specification.
4 Empirical Analysis
The empirical section aims to examine the impact of life satisfaction of migration intentions, when accounting for both individual- and regional level characteristics. The use of cross-sectional data in this context could raise concerns about endogeneity due to unobserved factors jointly affecting migration intentions and life satisfaction (Brzozowski & Coniglio, 2021), which becomes a challenge per se in selecting the appropriate estimation method. We therefore proceed to develop the empirical analysis in three stages: (1) analyzing the relationship between migration intentions and life satisfaction in a multilevel setting, without addressing the endogeneity issue, (2) estimating the impact of life satisfaction on migration intentions using the control function approach (upon Wooldridge, 2010) and instrumenting life satisfaction to address the endogeneity issue, and (3) confronting the results from the methods above as part of a broader robustness analysis. Step (2) is apparently unnecessary as the endogenous variable “Life satisfaction” could be directly instrumented at step (1) by a Generalized Structural Equation Model, but the model would be too complex involving convergence problems. Therefore, to not rule out the endogeneity, stage (2) is used to provide accurate insights.
In the first stage of our analysis, the hierarchical structure of our data with individuals nested in counties allows examining the impact of both individual and county characteristics on migration intentions. The two-level random intercept multinomial logit model is applied, starting with the null-model which is gradually enriched with new explanatory variables and specifications based on likelihood-ratio tests, with a p-value threshold of 5%. Given the complexity of multilevel multinomial logit model, the slopes are assumed to be constant for all covariates, which basically means that we use a two-level random intercept model. Initially a random effect is included at the county level, and it is constrained to be equal for the first three categories of the variable migration intentions. The fourth category of Migration intentions is chosen to be the baseline one, i.e. the intention to not leave the city. The estimated variance of this random effect is 0.38, which suggests a standard deviation of 0.61. This further means that a 1-standard deviation in the random effects amounts to a exp(0.61) = 1.85 change in the relative change ratio. The effect is therefore practically and statistically significant, so we cannot omit it. In other words, the two-level random intercept logit multinomial model with individuals nested in counties is motivated by our clustered data.
In Table 1, the estimates of the two-level random intercept multinomial logit model are separately reported for the Migration intention categories in columns (1), (2) and (3).
Table 1
Explaining the intention to migrate (marginal effects)
Explanatory Variables | Migration intention: relocation in another city within the same county (1) | Migration intention: relocation in another city within another county (2) | Migration intention: relocation abroad (3) |
|---|---|---|---|
Individual level | |||
Age (centred) | −.0009a*** (0.0001) | − 0.003*** (0.0003) | − 0.002*** (0.0002) |
Absolut income (log) | − 0.003*** (0.001) | − 0.006*** (0.001) | 0.0004 (0.001) |
Children | − 0.005 (0.004) | − 0.02*** (0.005) | − 0.001 (0.004) |
How long have you lived in the city? | −0.002* (0.001) | − 0.005*** (0.002) | 0.002 (0.002) |
Unemployed | − 0.0001 (0.01) | − 0.01 (0.01) | 0.03*** (0.01) |
Education | 0.003*** (0.001) | 0.007*** (0.002) | 0.002* (0.001) |
Married | − 0.01*** (0.003) | − 0.006* (0.003) | − 0.008** (0.004) |
Men | 0.001 (0.003) | 0.002 (0.004) | 0.02*** (0.004) |
Social trust | 0.002* (0.001) | 0.01*** (0.02) | 0.007*** (0.002) |
Life satisfaction | |||
Very satisfied (omitted) | |||
Fairly satisfied | − 0.000 (0.003) | 0.007 (0.04) | 0.02*** (0.04) |
Fairly unsatisfied | 0.005 (0.005) | 0.01** (0.007) | 0.05*** (0.008) |
Very unsatisfied | 0.001 (0.01) | − 0.005 (0.01) | 0.11*** (0.02) |
Regional level | |||
GDP per capita (ln) | − 0.01*** (0.006) | − 0.01 (0.01) | 0.01 (0.008) |
Marginal effects are reported in Tables 1, 2, 3 and 4, while the corresponding multinomial logit coefficients are presented in Appendix (Tables 5, 6, 7 and 8). Marginal effects indicate the change in the predicted probability of an outcome when a predictor increases by one unit. For example, in Table 1, Column (1), the marginal effect of Income on the intention to relocate to another city within the same county suggests that a one-unit increase in Income is associated with a 0.3% lower probability of this outcome. By contrast, the multinomial logit coefficient for a one-unit increase in Income on the same outcome, relative to the intention not to migrate, is − 0.12 (Appendix Table 1), holding all other variables constant. This implies that a one-point increase in Income decreases the log-odds of intending to move to a nearby city (versus not migrating) by 0.12 units. In short, multinomial logit coefficients represent log-relative risk ratios with respect to the base outcome, whereas marginal effects translate these into changes in predicted probabilities, independent of the chosen base outcome.
The most important result derived from the estimates in Table 1 is the strong and significant relationship between the intention to emigrate and life satisfaction (Model 3): the lower a person’s life satisfaction, the higher the likelihood that they intend to emigrate. In contrast, life satisfaction is not significantly associated with internal migration intentions (Models 2 and 3), suggesting that in Romania, people who are dissatisfied with life, regardless of the reason, are more likely to consider emigration rather than moving to another city or county. This finding aligns with a broad body of research on emigration intentions (Brzozowski & Coniglio, 2021; Chindarkar, 2014; Graham, 2016; Ivlevs, 2015). Importantly, this study not only analyzes the relationship between life satisfaction and migration intentions for the first time in Romania, but also highlights the differences in the motivations underlying emigration, short-distance internal migration, and long-distance internal migration.
Some determinants consistently influence migration intentions across models (1)–(3), particularly age, education, marital status, and social trust. In line with Mendez (2020) and Lovo (2014), age is negatively associated with migration intentions, both for internal migration and emigration. This pattern is further supported by OECD (2019), which finds that migration intentions among Romanians are especially high among younger individuals.
Education is positively associated with migration intentions across all types of migration considered (models 1–3). This finding is consistent with OECD (2019), which, based on Gallup World Poll data, reports a similar trend among educated Romanians. The OECD interprets this in the context of the “frustrated achievers” hypothesis: educated individuals are more likely to consider emigration due to the mismatch between their qualifications and the limited opportunities in the domestic labor market. While less-educated individuals also face labor market difficulties, their lower aspirations and greater financial constraints often make emigration an unrealistic option. As a result, their migration intentions tend to focus more on internal rather than international mobility.
Similar to the effects of education and age, social trust exhibits a consistent influence on both internal migration and emigration intentions. Specifically, higher levels of social trust are associated with a stronger inclination to remain in one’s place of residence, an effect that appears to be even more pronounced in the context of long-distance migration (as shown in models 2 and 3). This finding aligns with the existing literature; for instance, Manchin and Orazbayev (2018) argue that individuals with high levels of trust tend to be more socially integrated and embedded in their communities. Similarly, Bjørnskov (2003) highlights the association between trust and pro-social behavior, further supporting the interpretation that social ties can act as a deterrent to migration.
Our results also show that men and unemployed individuals are significantly more likely to express intentions to emigrate rather than to migrate domestically. This finding is consistent with Mendez (2020), who reports that men, traditionally viewed as primary earners, are more likely than women to express a desire to emigrate. The result concerning unemployment mirrors the conclusions of OECD (2019), suggesting that unemployed individuals constitute a key demographic among Romanian emigrants. For this group, emigration may be seen as a more viable or final solution than relocating within the country.
Other control variables, namely income, the presence of children in the household, and the number of years spent in the current place of residence, are significant only for internal migration intentions. These findings suggest that these factors are more relevant for understanding both short- and long-distance internal migration, but not necessarily for international migration. This pattern is consistent with previous research (Lovo, 2014; Otrachshenko & Popova, 2014; Shamsuddin & Katsaiti, 2020; Mendez, 2020), where such variables have been widely employed to capture household constraints and attachment to place in internal mobility decisions.
In models (1) and (2), the cumulative inertia hypothesis (Myers et al., 1967) is confirmed, suggesting that the more years an individual spends in their city of residence, the less likely they are to want to move. This finding aligns with other empirical studies on internal migration (Coulter et al., 2016; Gillespie et al., 2024). However, the hypothesis does not hold in model (3), indicating that the resistance to emigration is outweighed by the costs and risks associated with moving. Moreover, the theory has a weaker empirical basis in the context of international migration, which involves higher institutional constraints.
Income7 is found to significantly correlate only with internal migration intentions: individuals with lower incomes are more likely to consider relocating within the country. However, this relationship does not extend to international migration intentions. This result aligns with findings from Otrachshenko and Popova (2014), but contrasts with other studies such as Mendez (2020), which report a significant role of income in shaping emigration intentions. The explanation could be that poorer people can only afford internal migration, as emigration is more complicated and resource demanding (Stark & Yitzhaki, 1988). Furthermore, migration intentions are shaped not only by financial considerations but also by psychological and social factors, such as hope, trust, aspirations, career opportunities, and dissatisfaction with the broader economic environment. In this regard, the Romanian case reflects a broader regional trend observed across Balkan countries, where non-financial motivations, including education opportunities and family reunification, often outweigh purely economic drivers (Migali et al., 2018).
At the regional level, GDP per capita is significantly associated only with intentions to relocate within the same county, suggesting that short-distance internal migration is influenced by local economic conditions. This finding is consistent with Migali et al. (2018), who also report a weak relationship between regional GDP and emigration intentions in low-income countries. Our results further indicate that short-distance internal migration, specifically within-county moves, is primarily shaped by regional characteristics, whereas individual-level factors play a secondary role. In contrast, long-distance internal migration intentions do not appear to be driven by levels of regional development across any model specification used in this study.
The main conclusion is that internal migration intentions are largely shaped by individual financial and personal circumstances, while emigration intentions reflect broader life dissatisfaction, which is multidimensional and not necessarily rooted in financial constraints. As such, emigration intentions are more complex and less amenable to short-term policy interventions, highlighting the need for long-term strategies.
4.1 Addressing Endogeneity
Although cross-sectional data on migration have inherent limitations, such as potential issues of endogeneity, they often represent the only data available, particularly in contexts like urban migration studies. Despite these methodological challenges, cross-sectional analyses can still offer valuable policy insights. While it is possible to address endogeneity concerns in such studies, doing so remains methodologically challenging and often involves a degree of subjectivity in model specification. In this context, conducting robustness checks becomes crucial for enhancing the credibility of the results and ensuring a stronger empirical foundation for policy recommendations.
To address the potential endogeneity of life satisfaction in relation to migration intentions, this study employs the control function approach, following the specifications outlined by Wooldridge (2015) for nonlinear models. In the first stage, the endogenous variable life satisfaction is regressed on a set of exogenous instrumental variables (Appendix Table 9). In the second stage, the residuals obtained from this regression are included as an additional explanatory variable in the original model specification (Table 2). This two-step procedure allows for a consistent estimation of the impact of life satisfaction on migration intentions within a nonlinear framework.
Following recent studies on migration intentions (e.g., Shamsuddin et al., 2022), the instrument for the endogenous variable life satisfaction is selected using the “leave-one-out” instrumental variable (IV) approach proposed by Angrist et al. (1999). Specifically, we use the average life satisfaction of all individuals within the county of residence, excluding individual i, as a proxy for i’s own life satisfaction. This instrument is expected to be strongly correlated with the individual’s life satisfaction, while remaining exogenous to their migration intentions, thus satisfying the relevance and exclusion restrictions required for valid instrumentation. To further support the validity of our instrumental variable, we control for key county-level economic characteristics, most notably GDP per capita, in all model specifications. This adjustment helps to mitigate concerns that the county-average life satisfaction instrument might be correlated with local economic conditions that independently affect migration intentions. Additionally, robustness checks comparing individuals from poorer versus wealthier counties (in next section) reveal consistent effects of life satisfaction on migration intentions across these heterogeneous regional contexts. These results increase confidence that our instrument captures variation in life satisfaction that is plausibly exogenous to migration decisions.
Compared to the output of the multilevel model reported in Table 1, the model reported below (Table 2) also includes county fixed effects to account for the unobserved fixed county characteristics which represent the second Level of the multilevel setting. To accurately estimate the standard errors for the multinomial logit model, we will use the bootstrapping procedure.
Table 2
Control function approach (marginal effects)
Explanatory Variables | Migration intention: relocation in another city within the same county (1) | Migration intention: relocation in another city within another county (2) | Migration intention: relocation abroad (3) |
|---|---|---|---|
Life satisfaction | |||
Very satisfied (omitted) | |||
Fairly satisfied | 0.0001 (0.005) | 0.006 (0.006) | 0.02*** (0.007) |
Fairly unsatisfied | 0.005 (0.006) | 0.01 (0.01) | 0.05*** (0.01) |
Very unsatisfied | 0.002 (0.01) | − 0.01 (0.01) | 0.11*** (0.01)/ |
Regional level | |||
GDP per capita (ln) | − 0.02** (0.01) | − 0.05* (0.03) | − 0.003 (0.01) |
First stage F = 458 > 10 Stock and Yogo’s test of weak instruments = 373 > 16 | |||
No.obs. 11.979 | |||
When instrumenting Life satisfaction, a set of preliminary tests is used. The relevance of the instruments is tested in the first-stage regression. The F-statistic should be bigger than 10 here as there is a single endogenous regression (Stock et al., 2002). F statistic here is found to be significant, meaning that additional instruments have significant explanatory power for Life satisfaction. In addition, the Stock and Yogo’s test of weak instruments is applied. The null hypothesis is that the set of instruments is weak. Considering that we are willing to tolerate a bias no higher than 5%, we can say that our instrument is not weak because our test value of 373 exceeds the critical value of 16.
This section reports only the estimates for life satisfaction and GDP per capita, as these variables represent the core focus of the analysis. The results indicate that, after accounting for the endogeneity of life satisfaction, this variable remains a statistically significant predictor solely for emigration intentions (Model 3), while it has no significant effect on internal migration intentions (Models 1 and 2). In contrast, regional GDP per capita is found to be significantly associated with both forms of internal migration. Individuals residing in less developed regions are more likely to express intentions to move to another county (long-distance migration) than within the same county (short-distance migration).
Overall, when the endogeneity of life satisfaction is addressed, the results suggest that differences in regional economic development are a key explanatory factor for internal migration intentions in Romania in the year 2020.
4.2 Robustness Analysis
To assess the robustness of our empirical findings, we conduct a robustness analysis by examining whether the model estimates are consistent across different population sub-groups. This analysis focuses exclusively on the models that address the endogeneity of life satisfaction. In order to account for heterogeneity at both regional and individual levels, we perform subgroup comparisons between (i) individuals residing in poorer counties and those in wealthier counties, and (ii) younger and older population segments. While we anticipate variation in the regression coefficients across these subgroups, indicating subgroup-specific effects, our primary interest lies in evaluating whether the estimated impact of life satisfaction remains consistent across these different contexts.
In Table 3, counties with a GDP per capita below the national average are classified as “poorer,” while those above the average are considered “richer.” Overall, most explanatory variables exhibit similar effects on migration intentions across both groups, although minor differences can be observed between poorer and richer counties.
Notably, the association between life satisfaction and migration intentions remains consistent across both subgroups and aligns with the findings reported in Table 2. However, in poorer counties only, county-level GDP per capita shows a significant relationship with long-distance migration intentions. Individuals residing in these counties appear more likely to express intentions to migrate to another county or abroad, potentially reflecting an awareness that long-distance migration may offer better opportunities to improve their socioeconomic conditions.
A notable difference between poorer and richer counties relates to marital status. In poorer counties, married individuals are significantly less likely to express intentions to leave their city compared to those in other types of living arrangements. This may be attributed to the higher financial costs and additional social considerations that families face when planning migration. As expected, individuals with lower income levels are generally more likely to express intentions to migrate internally.
Table 3
Migration intentions characteristics by type of County (marginal effects)
Explanatory Variables- | Migration intention: relocation in another city within the same county | Migration intention: relocation in another city within another county | Migration intention: relocation abroad | |||
|---|---|---|---|---|---|---|
Poorer counties | Richer counties | Poorer counties | Richer counties | Poorer counties | Richer counties | |
Individual level | ||||||
Age (centred) | − 0.001*** (0.0003) | 0.000 (0.000) | − 0.003*** (0.0005) | − 0.002*** (0.0007) | − 0.002*** (0.0004) | − 0.001*** (0.0004) |
Income (log) | − 0.002 (0.002) | − 0.004*** (0.001) | − 0.009*** (0.003) | − 0.005** (0.002) | 0.004 (0.003) | − 0.004 (0.003) |
Children | − 0.003 (0.008) | − 0.005 (0.007) | − 0.04*** (0.009) | − 0.007 (0.008) | − 0.001 (0.008) | − 0.005 (0.01) |
How long have you lived in the city? | − 0.002 (0.004) | − 0.002 (0.003) | − 0.007 (0.006) | 0.0007 (0.006) | 0.003 (0.004) | 0.004 (0.004) |
Education | 0.002 (0.003) | 0.0001 (0.005) | 0.005 (0.005) | 0.004 (0.003) | 0.000 (0.003) | − 0.001 (0.005) |
Unemployed | 0.008 (0.01) | − 0.21*** (0.02) | 0.006 (0.01) | − 0.04 (0.05) | 0.03*** (0.01) | 0.11*** (0.02) |
Married | − 0.01** (0.008) | − 0.003 (0.008) | − 0.01 (0.01) | 0.008 (0.01) | − 0.01** (0.006) | − 0.01 (0.01) |
Man | − 0.0004 (0.004) | − 0.000 (0.003) | 0.007 (0.006) | − 0.003 (0.007) | − 0.02*** (0.004) | 0.03*** (0.01) |
Social trust | 0.003 (0.007) | 0.01*** (0.004) | 0.04*** (0.01) | 0.018* (0.01) | 0.01* (0.008) | 0.03** (0.01) |
Life satisfaction | ||||||
Very satisfied (omitted) | ||||||
Fairly satisfied | 0.004 (0.007) | − 0.007 (0.007) | 0.008 (0.009) | 0.005 (0.007) | 0.03*** (0.01) | 0.02** (0.01) |
Fairly unsatisfied | 0.01 (0.008) | − 0.009 (0.007) | 0.006 (0.01) | 0.01 (0.01) | 0.06*** (0.01) | 0.05*** (0.02) |
Very unsatisfied | 0.003 (0.01) | 0.005 (0.01) | − 0.01 (0.01) | − 0.01 (0.007) | 0.11*** (0.02) | 0.12*** (0.03) |
Regional level | ||||||
County GDP per capita (ln) | 0.05 (0.04) | 0.002 (0.006) | − 0.17*** (0.04) | 0.004 (0.02) | − 0.07* (0.04) | 0.008 (0.02) |
In Table 4, migration intentions are comparatively examined across two age categories: under 35 (abbreviated as “younger”) and over 35 (“older”). The most important findings from Tables 1, 2 and 3 hold here as well; namely, for both younger and older individuals, life dissatisfaction and unemployment remain the most powerful drivers of emigration intentions, without being associated with internal migration intentions. In particular, the more years an older person has spent in the city of residence, the less likely they are to want to migrate within the country. This suggests that the cumulative inertia axiom applies to older individuals, but not to younger ones.
Table 4
Migration intentions characteristics by age categories (marginal effects)
Explanatory Variables | Migration intention: relocation in another city within the same county | Migration intention: relocation in another city within another county | Migration intention: relocation abroad | |||
|---|---|---|---|---|---|---|
Younger | Older | Younger | Older | Younger | Older | |
Individual level | ||||||
Age (centred) | − 0.003*** (0.001) | − 0.0001 (0.0001) | − 0.01*** (0.002) | −0.001*** (0.0003) | − 0.002** (0.001) | − 0.001*** (0.0003) |
Income (log) | − 0.005* (0.003) | − 0.0001 (0.001) | − 0.01*** (0.004) | −0.0002 (0.001) | 0.006 (0.004) | − 0.003* (0.002) |
Children | 0.009 (0.01) | − 0.003 (0.005) | − 0.03** (0.01) | − 0.006* (0.003) | − 0.01 (0.01) | 0.0003 (0.006) |
How long have you lived in the city? | 0.001 (0.004) | − 0.005*** (0.002) | − 0.006 (0.006) | − 0.01*** (0.003) | 0.009 (0.006) | −0.001 (0.002) |
Unemployment | 0.01 (0.01) | − 0.009 (0.02) | − 0.000 (0.02) | − 0.008 (0.02) | 0.06*** (0.01) | 0.03*** (0.01) |
Education | 0.007 (0.005) | 0.002 (0.002) | 0.01* (0.008) | 0.005*** (0.002) | 0.0003 (0.004) | 0.0002 (0.002) |
Married | − 0.017* (0.01) | −0.002 (0.004) | −0.006 (0.02) | −0.003 (0.005) | − 0.01 (0.009) | − 0.01 (0.007) |
Man | 0.003 (0.005) | 0.001 (0.003) | 0.008 (0.008) | 0.002 (0.003) | 0.03*** (0.006) | 0.01*** (0.003) |
Social trust | 0.005 (0.01) | 0.005* (0.003) | 0.06*** (0.01) | 0.008** (0.004) | 0.02* (0.14) | 0.01*** (0.004) |
Life satisfaction | ||||||
Very satisfied (omitted) | ||||||
Fairly satisfied | − 0.0002 (0.009) | 0.0008 (0.005) | 0.01 (0.01) | 0.0002 (0.005) | 0.04*** (0.01) | 0.01*** (0.003) |
Fairly unsatisfied | − 0.001 (0.01) | 0.01 (0.007) | 0.01 (0.02) | 0.009 (0.007) | 0.09*** (0.02) | 0.03*** (0.008) |
Very unsatisfied | − 0.007 (0.02) | 0.01 (0.01) | − 0.01 (0.03) | − 0.006 (0.01) | 0.14*** (0.04) | 0.09*** (0.01) |
County GDP per capita (ln) | − 0.035* (0.02) | −0.006 (0.004) | − 0.12*** (0.05) | 0.0001 (0.01) | − 0.01 (0.02) | 0.002 (0.008) |
Table 5
Explaining the intention to migrate (multinomial logit coefficients)
Explanatory Variables | Migration intention: relocation in another city within the same county (1) | Migration intention: relocation in another city within another county (2) | Migration intention: relocation abroad (3) |
|---|---|---|---|
Individual level | |||
Age (centred) | −0.04a*** (0.004) | −0.06*** (0.003) | −0.05*** (0.003) |
Absolut income (log) | −0.12*** (0.04) | −0.12*** (0.02) | −0.008 (0.03) |
Children | −0.22* (0.13) | −0.40*** (0.10) | −0.09 (0.10) |
How long have you lived in the city? | −0.09* (0.05) | −0.11*** (0.03) | 0.04 (0.04) |
Unemployed | −0.02 (0.37) | −0.24 (0.27) | 0.70*** (0.21) |
Education | 0.14*** (0.05) | 0.14*** (0.03) | 0.07* (0.04) |
Married | −0.35*** (0.11) | −0.15* (0.08) | −0.22** (0.09) |
Men | 0.10 (0.10) | 0.10 (0.07) | 0.52*** (0.08) |
Social trust | 0.11* (0.06) | 0.32*** (0.04) | 0.21*** (0.05) |
Life satisfaction | |||
Very satisfied (omitted) | |||
Fairly satisfied | 0.05 (0.13) | 0.19 (0.11) | 0.78*** (0.12) |
Fairly unsatisfied | 0.28 (0.17) | 0.40*** (0.12) | 1.36*** (0.14) |
Very unsatisfied | 0.24 (0.33) | 0.13 (0.26) | 1.99*** (0.20) |
Regional level | |||
GDP per capita (ln) | −0.54*** (0.21) | −0.23 (0.23) | 0.19 (0.20) |
Table 6
Control function approach (multinomial logit coefficients)
Explanatory Variables | Migration intention: relocation in another city within the same county (1) | Migration intention: relocation in another city within another county (2) | Migration intention: relocation abroad (3) |
|---|---|---|---|
Life satisfaction | |||
Very satisfied (omitted) | |||
Fairly satisfied | 0.06a (0.18) | 0.16 (0.11) | 0.74*** (0.22) |
Fairly unsatisfied | 0.30 (0.19) | 0.33 (0.21)/ | 1.31*** (0.26) |
Very unsatisfied | 0.26 (0.41) | 0.03 (0.18) | 1.93*** (0.27) |
Regional level | |||
GDP per capita (ln) | −0.77*** (0.31) | 0.90* (0.56) | −0.22 (0.35) |
First stage F = 458 > 10 Stock and Yogo’s test of weak instruments = 373 > 16 | |||
No.obs. 11.979 | |||
Table 7
Migration intentions characteristics by type of County (multinomial logit coefficients)
Explanatory Variables- | Migration intention: relocation in another city within the same county | Migration intention: relocation in another city within another county | Migration intention: relocation abroad | |||
|---|---|---|---|---|---|---|
Poorer counties | Richer counties | Poorer counties | Richer counties | Poorer counties | Richer counties | |
Individual level | ||||||
Age (centred) | − 0.04***a (0.007) | − 0.002 (0.01) | − 0.06*** (0.008) | − 0.05*** (0.01) | − 0.05*** (0.006) | − 0.04*** (0.008) |
Income (log) | − 0.09 (0.08) | − 0.24*** (0.08) | − 0.13*** (0.05) | − 0.14** (0.07) | 0.06 (0.06) | − 0.12 (0.09) |
Children | − 0.19 (0.20) | − 0.32 (0.38) | − 0.63*** (0.12) | − 0.19 (0.18) | − 0.13 (0.16) | − 0.14 (0.28) |
How long have you lived in the city? | − 0.06 (0.12) | − 0.10 (0.17) | − 0.11(0.10) | − 0.02(0.13) | 0.04 (0.09) | 0.08 (0.09) |
Education | 0.08 (0.10) | 0.01 (0.19) | − 0.08(0.08) | 0.10 (0.09) | 0.02 (0.08) | − 0.02(0.13) |
Unemployed | 0.32 | −12.07*** | − 0.19 | − 0.97 | 0.66*** | 2.19*** |
Married | − 0.48*** | 0.17 | − 0.31 | 0.14 | − 0.32** | − 0.37 |
Man | 0.06 | 0.03 (0.17) | 0.16* (0.09) | − 0.02 (0.16) | − 0.43*** (0.08) | 0.66*** (0.14) |
Social trust | 0.22 (0.19) | 0.75*** (0.24) | 0.70*** (0.21) | 0.49** (0.25) | 0.39** (0.18) | 0.76** (0.35) |
Life satisfaction | ||||||
Very satisfied (omitted) | ||||||
Fairly satisfied | 0.20 (0.22) | 0.98 (0.52) | − 0.20 (0.14) | 0.16 (0.18) | 0.81*** (0.29) | 0.72*** (0.3) |
Fairly unsatisfied | 0.50 (0.20) | 0.73 (0.45) | − 0.26 (0.22) | 0.46 (0.36) | 1.34*** (0.30) | 1.32*** (0.45) |
Very unsatisfied | 0.28 (0.43) | 0.31 (0.64) | 0.07 (0.23) | − 0.07 (0.18) | 1.88*** (0.34) | 2.02*** (0.42) |
Regional level | ||||||
County GDP per capita (ln) | − 0.85 (1.21) | − 0.16 (0.33) | −2.72*** (0.83) | 0.12 (0.48) | −1.77** (0.96) | 0.20 (0.49) |
Table 8
Migration intentions characteristics by age categories (multinomial logit coefficients)
Explanatory Variables | Migration intention: relocation in another city within the same county | Migration intention: relocation in another city within another county | Migration intention: relocation abroad | |||
|---|---|---|---|---|---|---|
Younger | Older | Younger | Older | Younger | Older | |
Individual level | ||||||
Age (centred) | − 0.09*** (0.02) | − 0.01*** (0.008) | − 0.11*** (0.02) | − 0.04*** (0.01) | − 0.06*** (0.01) | − 0.05*** (0.01) |
Income (log) | 0.13* (0.08) | − 0.01 (0.07) | − 0.13*** (0.05) | − 0.01 (0.06) | 0.04 (0.05) | − 0.12* (0.07) |
Children | 0.11 (0.25) | − 0.18 (0.29) | − 0.37** (0.17) | − 0.23* (0.13) | − 0.26 (0.18) | − 0.003 (0.21) |
How long have you lived in the city? | 0.04 (0.1) | − 0.31*** (0.12) | − 0.04 (0.06) | − 0.37*** (0.1) | 0.11 (0.07) | − 0.07 (0.1) |
Unemployment | 0.40 (0.37) | −0.46 (1.05) | 0.13 (0.25) | − 0.25 (0.80) | 0.86*** (0.21) | 1.28*** (0.37) |
Education | 0.18 (0.12) | 0.13 (0.13) | 0.14* (0.08) | 0.21*** (0.08) | 0.04 (0.06) | 0.02 (0.08) |
Married | − 0.40** (0.21) | − 0.16 (0.23) | − 0.11 (0.18) | − 0.15 (0.2) | − 0.20 (0.13) | − 0.41 (0.27) |
Man | 0.15 (0.12) | − 0.08 (0.17) | 0.15* (0.09) | 0.11 (0.14) | 0.50*** (0.08) | 0.61*** (0.11) |
Social trust | 0.27 (0.23) | − 0.33** (0.17) | 0.70*** (0.18) | 0.33** (0.15) | 0.39** (0.20) | 0.50*** (0.17) |
Life satisfaction | ||||||
Very satisfied (omitted) | ||||||
Fairly satisfied | 0.09 (0.19) | 07 (0.3) | 0.22 (0.18) | 0.03 (0.21) | 0.80*** (0.27) | 0.67*** (0.22) |
Fairly unsatisfied | 0.16 (0.26) | 0.59 (0.3) | 0.35 (0.23) | 0.37 (0.25) | 1.35*** (0.29) | 1.26*** (0.27) |
Very unsatisfied | 0.03 (0.59) | 0.64 (0.53) | 0.08 (0.31) | − 0.09 (0.51) | 1.64*** (0.39) | 2.2*** (0.30) |
Regional level | ||||||
County GDP per capita (ln) | −1.03*** (0.4) | − 0.32 (0.24) | −1.34*** (0.51) | − 0.000 (0.04) | − 0.46 (0.37) | 0.07 (0.3) |
Table 9
Control function approach – first stage Estimation (coefficients)
Explanatory Variables | Coefficients (st.err.) |
|---|---|
Age (centred) | 0.01*** (0.001) |
Absolut income (log) | − 0.03** (0.01) |
Children | − 0.05 (0.04) |
How long have you lived in the city? | 0.04** (0.01) |
Unemployed | 0.47*** (0.12) |
Education | − 0.11*** (0.01) |
Married | − 0.14*** (0.03) |
Men | − 0.003 (0.03) |
Social trust | 0.47*** (0.02) |
Average life satisfaction | 2.30*** (0.1) |
GDP per capita (ln) | 0.05 (0.05) |
Table 10
Migration intentions characteristics by type of county - first stage regression (coefficients)
Explanatory Variables | Poorer (coeff./st.err.) | Richer (coeff./st.err.) |
|---|---|---|
Age (centred) | 0.01*** (0.001) | 0.01*** (0.002) |
Absolut income (log) | − 0.04*** (0.01) | − 0.01 (0.02) |
Children | − 0.12*** (0.05) | 0.10 (0.07) |
How long have you lived in the city? | 0.05** (0.02) | − 0.002 (0.03) |
Unemployed | 0.28** (0.14) | 1.10*** (0.26) |
Education | − 0.10*** (0.02) | − 0.13*** (0.03) |
Married | − 0.09** (0.04) | − 0.27*** (0.07) |
Men | − 0.04 (0.04) | − 0.08 (0.06) |
Social trust | 0.46*** (0.02) | 0.49*** (0.03) |
Average life satisfaction | 2.02*** (0.12) | 3.55*** (0.28) |
GDP per capita (ln) | −25* (0.17) | − 0.07 (0.09) |
Table 11
Migration intentions characteristics by age categories - first stage regression (coefficients)
Explanatory Variables | Younger (coeff./st.err.) | Older (coeff./st.err.) |
|---|---|---|
Age (centred) | 0.01*** (0.006) | 0.008*** (0.002) |
Absolut income (log) | − 0.03(0.02) | − 0.03** (0.01) |
Children | − 0.07 (0.07) | − 0.06 (0.05) |
How long have you lived in the city? | −0.02 (0.02) | 0.12** (0.03) |
Unemployed | 0.16 (0.17) | 0.81*** (0.18) |
Education | − 0.11*** (0.02) | − 0.12*** (0.02) |
Married | − 0.03 (0.05) | − 0.26*** (0.05) |
Men | 0.02 (0.05) | − 0.01 (0.04) |
Social trust | 0.55*** (0.03) | 0.42*** (0.02) |
Average life satisfaction | 2.63*** (0.18) | 2.09*** (0.13) |
GDP per capita (ln) | 0.10 (0.08) | 0.009 (0.06) |
The analysis confirms that the relationship between life satisfaction and migration intentions is robust across all model specifications and population sub-groups. Specifically, the intention to emigrate is consistently associated with lower levels of life satisfaction. In contrast, life dissatisfaction does not appear to significantly influence intentions to migrate internally.
5 Discussion and Conclusions
Although the relationship between life satisfaction and migration intentions has been widely examined in the literature, this topic has not yet been explored using Romanian data. Moreover, the regional dimension has not been incorporated into quantitative studies on Romanian migration. From this perspective, the paper offers new insights to the literature, extending its relevance beyond the Romanian context, as outlined in the Introduction.
A key finding is that, across different specifications and models, life satisfaction is associated only with the intention to emigrate, but not with the intention to migrate domestically. When considering the broader picture, internal migration appears to be primarily driven by individuals who are economically disadvantaged, younger, highly educated, or have personal commitments such as family responsibilities or strong local attachments. In contrast, emigration intentions reflect deeper and more complex circumstances, often involving dissatisfaction or disillusionment, which compel individuals to overcome family-related constraints and leave their local communities despite existing social support, in pursuit of personal or professional goals abroad. Unemployment significantly fuels emigration intentions, but it does not appear to influence internal migration. Social trust acts as a buffer against all types of migration, with this effect being more pronounced in wealthier counties.
Stronger local networks, which tend to develop among long-term residents, reduce the likelihood of domestic migration, particularly among older individuals. In contrast, in the Romanian context, emigration intentions appear less constrained by the psychological and practical costs typically associated with relocation. This suggests that the determinants of emigration are shaped by a distinct set of mechanisms compared to those driving internal migration.
Our findings also show that in Romania, the economic development of one’s county of residence is only weakly correlated with migration intentions. This is not entirely unexpected, as rural–urban migration flows have partially reversed, with many Romanians choosing to leave overcrowded urban areas in favor of rural living, seeking a quieter lifestyle closer to nature. Therefore, it is not the level of economic development in a city that primarily influences migration intentions, but rather other individual-level factors. Nevertheless, residents of poorer counties are more likely to emigrate, while young individuals in those same counties are more inclined toward internal migration.
Despite the Romania’ low unemployment rate compared to other EU countries, unemployment was found to be a major driver of emigration intention and not internal migration intentions as well. This is consistent with the structural weaknesses of Romanian labor market characterized by low wages, poor working conditions, limited upward mobility and the mismatch between workers’ qualifications and the types of jobs available. Addressing the persistent vulnerabilities within Romania’s labor market requires a comprehensive policy approach focused on improving skills alignment and education, fostering regional economic development, enhancing job quality and wage growth, supporting return and circular migration, strengthening labor market institutions, expanding public services and social protection, and restoring institutional trust through governance reforms.
In the context of Romania recording one of the lowest net migration rates and some of the lowest levels of life satisfaction in the European Union, despite sustained economic growth, this study finds that life dissatisfaction significantly predicts emigration intentions but does not appear to influence internal migration intentions. When dissatisfaction with life drives the intention to migrate, individuals are more likely to opt for internal migration if their discontent is tied to specific socioeconomic factors. In contrast, emigration is more likely when dissatisfaction is more complex and extends beyond individual circumstances or the perceived economic progress of their city or county. Emigration offers Romanians improved economic opportunities and a higher quality of life, with barriers to leaving the country reduced by the presence of supportive diaspora communities abroad. By comparison, internal migration tends to be less transformative and is often directed toward already overcrowded and saturated urban centers. This multifaceted dissatisfaction driving emigration should serve as a critical warning to Romanian policymakers. While sustained economic growth typically raises expectations for improved quality of life, the failure to meet these expectations can result in frustration and widespread life dissatisfaction. The persistent gap between economic development and subjective well-being, shaped by chronic institutional distrust, corruption, regional and demographic inequalities, and inefficient public services, reflects a model of growth without gratification, which is neither socially nor politically sustainable in the long term.
Addressing this type of model, where economic growth fails to generate well-being and often leads to emigration, requires a multidimensional policy response focused on rebuilding institutional trust, creating opportunities for all, reducing social inequality, efficiently investing in public services, and targeting regional development. Economic growth should ultimately translate into sustainable, regionally balanced, and inclusive development.
Open Access This article is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License, which permits any non-commercial use, sharing, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if you modified the licensed material. You do not have permission under this licence to share adapted material derived from this article or parts of it. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by-nc-nd/4.0/.
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.