ABSTRACT
Research on happiness finds that rising incomes do not generally lead to increases in happiness. This finding suggests that economic migration – i.e., migration motivated by the prospect of increased income – might not bring greater happiness: when economic migrants believe that migration will improve their lives, that belief might be misguided at least insofar as ‘improvement’ is conceived in terms related to happiness. Perhaps economic migration under certain conditions even results in lower happiness, if it involves sacrifices in other respects that are more consequential for happiness. This paper explores these propositions via comparison of Romanian migrants to non-migrants (using data from the European Social Survey) and finds that returned migrants report lower happiness than non-migrants (controlling for other variables), while migrants who have not returned are not different in happiness from stayers. The cross-sectional analysis cannot directly answer questions about the consequences of migration and return – there are no data on the migrants' happiness prior to migration. But the analysis sharpens the questions that might be asked in future research and considers how various scenarios would be consistent with the findings produced here.
Acknowledgements
I am grateful to Sergiu Bălţătescu for suggestions on an earlier draft.
Notes
1An equivalent mistake among non-migrants would be the relentless pursuit of economic growth and the failure to use increased productivity as a means of reducing working hours (Bartolini Citation2007).
2Life satisfaction is commonly considered equivalent in many ways to happiness, and empirical results usually do not differ for use of happiness vs. life satisfaction questions/variables.
3A significant percentage did not provide valid answers to this question, a point that accounts for the failure of the column totals for the different groups in to sum to the sample size.
4Italy is another common destination but did not participate in Round 4 of the ESS.
5The only potential departure from these conclusions emerges in an ordered probit model that uses robust standard errors: here, the ‘return’ coefficient is significant only at a less demanding threshold, with p=0.054.