MIGRATION, RETURN, AND HAPPINESS IN ROMANIA Open Access

Migration from a poor country to a wealthy one can improve migrants' economic circumstances – but does it raise their happiness? In general, increased income does not result in greater happiness, and this core finding of ‘happiness studies’ can be read to imply that economic migration would be no more effective in raising happiness than other means of increasing one's income. Many migrants believe that gaining entry to a wealthier country will improve their lives, but insofar as ‘improvement’ would include greater happiness this belief might simply be misguided on average – especially when migration involves sacrifices in relation to other factors that are more important for happiness than income.¹ More generally, while people in wealthy countries are on average happier than people in poorer countries, this cross-sectional fact does not by itself mean that migration to the former leads reliably to increased happiness for migrants; while prima facie plausible, the assumption that it does amounts to an ecological fallacy.

Empirical research that does justice to the ‘longitudinal’ nature of this question carries stringent data demands, including measurement of happiness (and other variables) prior to migration, not just afterwards. Most existing large-scale data collection efforts do not satisfy this condition, primarily because most panel data-sets are constructed as ‘national’ entities, e.g., the British Household Panel Survey. If a panel member emigrates then he or she is a casualty of ‘attrition’. In destination countries, immigrants – to the extent they are represented in panels at all – do not contribute data collected prior to their immigration. There are, then, significant data limitations inhibiting the study of international migration, particularly with respect to consequences for the migrants.

In lieu of a direct test of the question regarding migrants generally, this paper therefore considers the happiness of two Romanian migrant groups: those living in another European country and those who have returned home after working abroad, in comparison to Romanians who have not migrated. Romanians gained access to the labor markets of Britain, Ireland and Sweden with EU accession in 2007, but there is a longer history of undocumented migration and large numbers of migrant workers employed in Italy, Spain and Germany (Culic 2008). The data available allow only a cross-sectional analysis and cannot provide definitive conclusions about the happiness consequences of this migration. Still, the results of the comparison help sharpen the questions that might be addressed in future research. The main empirical finding of the paper is that Romanians who emigrated and then returned home are less happy, controlling for other variables, than those who did not leave; migrants abroad, however, are not different in happiness from stayers.

As in much research on happiness, the question posed here is rooted in the ‘Easterlin paradox’ and the finding that increases in income generally do not result in increased happiness (Easterlin 1974). In a cross-sectional analysis, those who earn more are generally happier than those who earn less, and people in wealthier countries are happier than people in poorer ones (though the association is not particularly strong). But economic growth – at least in relatively wealthy countries – appears to contribute little or nothing to happiness. This conclusion is particularly evident in Japan, where several decades of rapid growth saw no change in average levels of happiness (Easterlin 2005). This argument informs the dominant perspective regarding income in happiness studies, and although it has not gone unchallenged (e.g., Stevenson and Wolfers 2008), there is no indication that happiness researchers generally have abandoned the view that economic growth contributes little to happiness (e.g., Easterlin et al.2010), though again some dispute it.

Economic growth might not even be merely benign in its consequences for happiness. Growth is not only a matter of ‘more’ in a quantitative sense – growth leads to qualitative changes in individuals’ lives, changes that might be experienced as a ‘dislocation’ in perhaps uncomfortable ways (Graham 2009). Deaton (2008) and Lora and Chaparro (2009) find evidence of an ‘unhappy growth paradox’: controlling for GDP levels, a higher growth rate predicts lower satisfaction with life² (as well as lower satisfaction in more specific domains e.g., jobs and housing). A key mechanism emerges in the way economic growth results in increased aspirations. That mechanism resonates also with Graham and Pettinato's (2002) notion of ‘happy peasants and frustrated achievers’: poorer people might moderate their material aspirations while relatively wealthy people experience rising expectations.

Rising income in general, then, appears not to contribute to happiness, and life choices emphasizing pursuit of income might lead to some disappointment (Kahneman et al. 2006). Several interrelated psychological processes are important here. One is adaptation: we might initially gain greater satisfaction from a higher income and the purchases it facilitates, but this gain quickly erodes (Frederick and Loewenstein 1999; cf. Scitovsky 1992). A second process is aspiration: what matters for happiness in this regard is the degree to which income matches income aspirations, and income increases are typically followed by an increase in aspirations, i.e., for even greater income (Easterlin 2001; Stutzer 2003). The third process, rooted in the notion that money/income is in part a positional good, is social comparison – and many people are inclined to give more weight to upward comparisons, to their detriment (Boyce et al.2010). Following an increase in income, people sometimes alter their reference groups for comparison: instead of gaining satisfaction from giving salience to one's improved position relative to a stable reference group, increased income leads some to compare themselves to an even wealthier reference group (Clark et al.2008). These processes help account for the finding that those inclined towards materialist pursuits generally report lower levels of psychological well-being (including happiness) than those who prioritize other types of goals (Richins and Dawson 1992; Frank 1999).

These findings, derived from the study of happiness and income generally, suggest that a higher income gained via migration might be no more effective in raising happiness than a higher income gained by other means. A fundamental problem for migrants is the likelihood of discrepancy between prior expectations and realities after migration (Vohra and Adair 2000; Knight and Gunatilaka 2010). The evolution of aspirations might pose particular difficulties for migrants insofar as they experience direct exposure to the consumption standards of wealthier societies (as against indirect exposure, e.g., via television): any increase in income might be outweighed by an increase in income aspirations following migration. A similar prediction emerges for comparisons: migration might be disadvantageous insofar as one's relative position in the destination country is likely to be lower than in the origin country, particularly since difficulties with language and unrecognized qualifications often mean that those who were middle-class in the origin encounter obstacles to gaining middle-class jobs in the destination (Portes and Bach 1985). At an early stage economic migrants are often willing to endure lower status (Piore 1979), but over time one's reference group will likely come to include (wealthier) people in the destination, with unfavorable consequences for happiness. Happiness is determined in part by personality/temperament, and if the migration decision was rooted in a sense of frustrated ambitions then one might find that that feeling persists after migration even if one achieves economic gains (Graham 2009).

Given the data limitations noted earlier, research on migration and happiness typically compares immigrants to natives in the destination country. Cross-sectional data show that immigrants in wealthy countries typically report lower happiness than natives, controlling for other determinants (e.g., Bălţătescu 2007a; Bartram 2011). In research on immigrants in Europe the happiness disadvantage of immigrants does not diminish over time, contrary to what one would expect from an assimilation paradigm (Safi 2010). Knight and Gunatilaka (2010) analyze rural-to-urban migrants in China and find that happiness among migrants was lower than among stayers in rural villages and lower also in comparison to urban non-migrants (despite migrants having earnings comparable to those of urban non-migrants). Their analysis suggests that migrants have false expectations about life in urban settings, and that migrants fail to anticipate that their own aspirations (e.g., about income) will rise after migration.

It is perhaps unsurprising that research in this mode finds that immigrants are less happy than natives, given difficulties of economic integration and the likelihood of a downward trajectory of relative position. But a different approach to comparison considers the situation of migrants relative to those who remain in the country of origin. In particular, temporary migration to a wealthier country followed by return might be a more promising path to greater happiness: if migrants can gain skills and accumulate savings while abroad, the result upon returning home might be a higher relative position (in comparison to their position prior to migration). This possibility is all the more relevant for migration originating in poorer countries, given that the cross-sectional association between income/status and happiness is stronger there than in wealthier countries, as evident in Delhey's (2004) research on Central/Eastern Europe (cf. Bălţătescu 2007b). That point also suggests (in contrast to the implications derived from the Easterlin paradox) that migrants from eastern European countries might gain happiness via migration to wealthier countries in the west. The remainder of this paper then considers Romanian-origin migrants of both types in comparison to Romanians without migration experience.

To evaluate the association between migration and happiness among Romanians, I analyze data on Romanians from the European Social Survey (Jowell 2007), drawing on the fourth round (2009); the overall sample size was 2,167.

The common practice for measurement in happiness studies is self-report on surveys (Diener et al.1999); confidence regarding validity of such measures is high (Frey and Stutzer 2002; Pavot 2008), and the concerns that do exist are likely not consequential for the analysis presented here. Data on happiness here come from answers to the question, ‘Taking all things together, how happy would you say you are?’. Options for answers were given on an 11-point scale, with 0 indicating ‘extremely unhappy’, 10 indicating ‘extremely happy’, and the intervening nine options unlabeled.

The main independent variable of interest is migration experience, particularly migration motivated by the prospect of economic gain. In regard to returned migrants, the analysis here turns on answers to a question asking whether during the last 10 years respondents have done any paid work in another country for at least 6 months. The yes/no options enable a direct comparison between those with migration experience involving paid employment and those lacking such experience (variable labeled ‘returnee’). Roughly 5% of those providing valid answers in this sample answered yes to this question.³

The other group of interest for comparison is emigrants from Romania currently living abroad. The sample analyzed here thus includes residents of other ESS countries who indicate they were born in Romania, labeled ‘migrant’; this group is similar in size to the group of returnees. Most Romanian emigrants in this sample (82%) are located in western European countries, particularly Spain and Germany⁴ – not a surprising fact if one expects migration to be motivated mainly by desire for the higher incomes available in wealthier countries. Most of those migrating to another eastern European country are in Hungary, reflecting long-standing trans-border ties in that region.

A number of variables are included as controls. Prior research (summarized e.g., in Diener and Seligman 2004 and Dolan et al.2008) establishes that happiness is associated with a variety of factors, among which the most important are health, unemployment, religiosity, marriage (or other forms of partnering), and participation in social activities. Respondents reported the state of their health on a 5-point scale (1 is excellent, and so the regression coefficient is expected to be negative). For unemployment, a variable was created from answers to a question regarding ‘main activity in the last 7 days’, where unemployment was coded 1 if main activity involved unemployment whether actively seeking a job or not; all other responses were coded 0 (i.e., not unemployed).

Religiosity is measured on an 11-point scale in response to the question, ‘How religious are you?’; 0 is ‘not religious at all’ and 10 is ‘very religious’. For marriage/intimate relationships, what matters is not being married per se but living with an intimate partner. ESS data offer a derived variable that indicates directly whether the respondent is living with a spouse or partner. Frequency of participation in social activities is gauged with the question, ‘how often do you meet socially with friends, relatives or work colleagues?’.

Happiness in cross-sectional comparisons is also associated with one's economic situation, and the analysis here includes household income as a control. The ESS income question asks respondents to select from a set of 10 ranges, deriving from country-specific deciles. This approach facilitates consideration of income in a relative sense (e.g., Ball and Chernova 2008), particularly when considering people living in different countries: Romanians living in other countries have to meet expenses in their local context, and they might also compare their economic situation to that of the people around them. The analysis also includes variables on gender and age; usually an additional age-squared term is added to reflect non-linear association, i.e., lower happiness during middle age with rising happiness subsequently – but this term was not significant in any model here and was therefore dropped.

Characteristics of the sample are given in Table 1 (showing means/standard deviations or percentages as appropriate), with separate values provided also for non-migrants (‘stayers’), returnees, and migrants. Table 2 shows the distribution of responses in the happiness variable for each group.

The dependent variable is happiness, originating in a question with 11 options for response. This variable is ordinal, and in principle the appropriate form of analysis would be an ordered probit model. However, given the relatively large number of response options ordinary least squares (OLS) regression can be expected to produce equivalent results (Ferrer-i-Carbonell and Frijters 2004 make this point with particular reference to research on happiness), and indeed for results below the issue of OLS versus ordered probit makes no difference, with one partial exception as noted. OLS regression is therefore adopted for ease of interpretation and the availability of standardized coefficients.

In a bivariate comparison, Romanians who live abroad and those who returned did not report significantly different levels of happiness from those without migration experience (6.4 and 5.9 vs. 6.2). But this comparison is misleading in failing to control for age in particular: migrants are generally younger, younger people are generally happier, and any association of migration with happiness might be masked by a countervailing influence of age when age is not controlled. If we control for age alone (Model 2 in Table 3) we see that returnees are more than half a point less happy than stayers on average (p=0.034), while there is no difference in the happiness of migrants and stayers.

That pattern persists in Model 3, with a full set of control variables. Other variables behave as expected. Happiness is higher among those who have better health and higher incomes; those who report being more religious are also happier, as are those who live with a spouse or partner, those who participate more frequently in social activities, and those who are not unemployed. There is evidence of heteroskedasticity for this model, but in a ‘robust regression’ analysis (not shown) employing a more realistic approach to calculating standard errors given heteroskedasticity one arrives at identical conclusions for hypothesis tests at conventional significance levels.⁵

A key issue arises in a comparison between Models 3 and 4, where the latter excludes income as a control variable. We can learn more from the difference between these models than from an attempt to determine which model is ‘right’. If our goal were to explain as much variation in happiness as possible, then we would want to emphasize a model that includes income. But the goal here is to gauge the association between migration and happiness. In that context – a context in which many individuals are attempting to gain more income via migration – it is not clear that income should be controlled. Many people become migrants in hopes of raising their incomes. The Easterlin paradox suggests that gaining more income does not in general bring greater happiness, but for the analysis here we should allow for the possibility that migrants might be different in this respect (cf. Bartram 2011). Particularly in a cross-sectional analysis in which income is demonstrably associated with happiness, it is not obvious that income is something one should control for when trying to understand the happiness consequences of economic migration, given the possibility that higher incomes might have been achieved via migration. In particular, returned migrants might have achieved a higher relative position in the Romanian income ranking – perhaps via the accumulation of skills and savings (e.g., capital for small business) while abroad – with favorable consequences for happiness.

The cross-sectional data available here cannot provide a definitive answer to these questions, but it is illuminating all the same in the way it narrows the range of possible answers. In Model 4 where income is not controlled, the coefficient for ‘returnee’ is negative but not statistically significant: returnees in the population might be less happy than stayers in this model, but we cannot be confident that that difference is not merely a consequence of sampling error.

Still, in a model that does not control for income a coefficient for ‘returnee’ that is not appreciably different from zero is notable for what it is not. Given the generally positive and significant cross-sectional association between income and happiness, and given that returnees have higher incomes than stayers (as in Table 1), in a model that omits income as a control one would expect returnees to have greater happiness than stayers, ceteris paribus. In other words, the returnee variable ought to proxy for the income variable when the income variable is omitted. But again the ‘returnee’ coefficient is not significantly different from zero. Even though we do not have data about happiness or income prior to migration, by comparing the two models we see a noteworthy difference between hypothesized and actual findings: returnees have higher incomes, but they are nonetheless not happier than stayers.

Romanians currently living in other countries, however, earn incomes that place them in a lower position in ‘local’ income distributions: at a mean decile position of 4.8, they are a full ‘point’ lower than Romanian non-migrants (5.9). That finding is consistent with more general scholarship on migration, as noted earlier: migrants often earned well in local terms prior to migration and then find it difficult to achieve a comparable position in the destination country. Here, whether one controls for income or not, there is no significant difference in the happiness of migrants vs. stayers. Migrants might have been expecting to gain from earning higher (‘absolute’) incomes in a wealthier country, but their position in relative terms is relatively low (at least in comparison to other Romanians). Another key point is their high unemployment rate, at 14% (perhaps partly attributable to very high unemployment rates in Spain, a significant destination for Romanian migrants).

The results described earlier emerge from a cross-sectional analysis and do not allow us to offer a direct answer to the question motivating this paper. That question asks about the consequences of migration: does migration/return lead to an increase (or decrease) in happiness? Lacking data on happiness prior to migration, we cannot know whether the lower happiness of returnees (relative to stayers) represents a decrease over time.

In research on a different context (Latin America), Graham and Markowitz (2011) show that people who express an intention to migrate are generally less happy than others. (They also find reasons to be confident that these intentions are frequently realized in migration itself.) If that pattern were replicated among migrants from (and returning to) Romania, the implication for findings here would be that returned migrants have not improved their happiness (as against a conclusion that migration followed by return results in lower happiness). Returnees might also have been earning more prior to migration than stayers: Graham and Markowitz's (2011) ‘frustrated achiever’ concept indicates that people who become migrants tend to earn more than others even before migration (but are less happy than others, i.e., ‘frustrated’). If true for Romanians – if migration did not result in increased post-return earnings – then the common-sense basis for expecting an increase in happiness is absent. But even a more optimistic scenario in which migration has facilitated improvement in migrants’ post-return economic situation apparently does not extend to optimism about the happiness consequences of this improvement. If improved earning ability does have favorable consequences for returned migrants, the fact that they are less happy than stayers suggests that there might also be negative consequences, perhaps even to the point of outweighing the positive ones.

Panel data representing all three groups (non-migrants, emigrants, and returned migrants) would greatly enhance prospects for research determining the prevalence of the different choices and their association with changes (if any) in happiness. While producing data of this sort poses logistical challenges, it ought to be possible on a reasonably general way at least for migration within Europe, via research mechanisms of the European Union.

Motivations for migration vary; this article has considered migration rooted in economic motivations, but some migration choices are informed by different goals, including relationships and lifestyle (e.g., Benson and O'Reilly 2009). The difficult economic situation faced by many Romanians in recent years supports the assumption that economic issues would have featured prominently in motivations for emigration from Romania. But the points about income and happiness considered here might be tangential to migration undertaken for family reunification or other relationships, for example. A broader research goal would be to determine the different circumstances and contexts that account for different happiness consequences of migration.

I am grateful to Sergiu Bălţătescu for suggestions on an earlier draft.

David Bartram is Senior Lecturer in sociology at the University of Leicester. His current research investigates the relationship between immigration and happiness, in particular the assumption that migration to a wealthy country is advantageous to the migrants themselves. His earlier work focused on non-Jewish migrant workers in Israel and migrant worker policies in Japan, with articles appearing in International Migration Review, Politics & Society and Ethnopolitics as well as a book published by Palgrave. He taught previously at the University of Reading and at Haverford College (USA); he holds a PhD from the University of Wisconsin–Madison and a BA from Kenyon College.