‘Bloom Where You’re Planted:’ Explaining Public Opposition to (E)migration

Authors: Alexander Kustov

Journal: Journal of Ethnic and Migration Studies (2022)

DOI: 10.1080/1369183X.2020.1754770

Resources

Figures & Tables

Abstract

Why is migration unpopular? A vast literature argues that voters oppose immigration because of threatened interests and prejudice. This paper is among the first studies of opposition to emigration—the other side of the issue salient in many countries. Departing from existing public opinion research, I develop a number of tests comparing emigration and immigration attitudes and then exploit relevant survey data from 30 countries, as well as original experimental and qualitative evidence. Overall, I document high opposition to both emigration and immigration in many countries and show that respondents are unlikely to confuse these issues. I then show that individual emigration and immigration attitudes are significantly correlated and have similar predictors, which is reflected in respondents’ own open-ended explanations. While consistent with sociotropic accounts, this new evidence suggests that many natives may exhibit an aversion to human mobility between countries in general, not immigration or emigration in particular.

Keywords: Emigration; Immigration; Public Opinion; Political Psychology \

Word count: 8600 words

Introduction

When surveyed about the right of their own citizens to freedom of movement in the EU, the majority of UK voters were positive. However, only a fraction agreed with a similar question about the right of other EU citizens to live and work in the UK, according to the YouGov survey from November 19-24, 2015. While such self-serving inconsistencies are rather common in public opinion, it is instructive that far from everyone is even willing to let their fellow citizens go and live abroad. Why do many people shun human mobility between countries? A large literature across social sciences has argued that voters strongly oppose migration because of threatened personal and group interests or deep predispositions related to dislike of ethnic outgroups [@Hainmueller2014].

Nonetheless, due to its current politicization in Europe and the United States, most research is focused only on attitudes toward immigration—not international mobility in principle. This paper argues that a full account of migration politics and public opinion should also consider emigration—the other side of the issue salient in many middle- and even high-income countries. As I demonstrate below, concerns about issues such as “brain drain” are not limited to pundits and policy-makers. In fact, in some countries the overwhelming majority thinks of emigration as a serious problem and wants their government to reduce it, which may include limiting people’s own economic opportunities and political rights. Furthermore, these attitudes may have real-world consequences. In Lithuania—where more than 10% of population left for other EU countries since 2004—the surprising victory of the Peasants and Green Union (LPGU) party in 2016, for instance, was attributed to their anti-emigration stance, according to Telegraph from October 24, 2016. Most important, however, exploring the roots of these sentiments may in turn also shed a light on some of controversies in the immigration politics literature.

Departing from existing public opinion theories, I develop a number of empirical tests juxtaposing predictors of emigration and immigration attitudes and then exploit the relevant Gallup World Poll and Transatlantic Trends data from 30 middleand high-income countries, as well as the original survey experiment and qualitative evidence from the UK. In Study 1, I document high, salient public opposition to both emigration and immigration by majorities or pluralities in most middle-income countries and demonstrate that respondents are unlikely to confuse these issues. I further establish that emigration and immigration attitudes have systematically distinct contextual predictors—respondents are expectedly more likely to dislike emigration (immigration) in countries with higher emigrant (immigrant) shares.

I then show that emigration and immigration attitudes are nonetheless strongly correlated and have similar individual predictors in nearly all countries, indicating a substantial common component. Given that the relative extent of opposition to inor out-migration is context-dependent, however, it is unlikely that one attitude is just an artifact of another. As further indicated by multivariate regression analysis, the strong relationship between migration attitudes does not change even after accounting for major demographic covariates.

In Study 2, I conduct a follow-up population-based survey experiment in the UK—one of the few high-income countries with considerable emigration flows and the related public concerns about it. In line with sociotropic accounts, I show that voters are much more opposed to high-skilled than low-skilled emigration regardless of their own skills or racial prejudice. A further analysis of open-ended responses corroborates the idea that most voters prioritize national interest over personal concerns or ethnic animus in their stated emigration policy preferences. The qualitative evidence, however, also reveals that many simply think people have a moral obligation to stay where they are born regardless of any consequences.

Overall, these analyses of public opinion reveal that many voters are averse to any human mobility between countries in general, not immigration or emigration in particular. While in line with existing group interest accounts and some notion of general anti-migration bias, this evidence challenges the widespread idea that people oppose migration primarily due to self-interest or anti-foreigner prejudice. Given that most governments restrict immigration but almost never emigration—despite the strong will of their citizens—the study results also have implications for democratic theory and migration politics in general.

The other side of migration politics

Historically, most states were more interested in controlling exit rather than entrance. They thus had significant and harsh out-migration controls—prosecuting potential emigrants, forcing them to pay high exit fees, refusing to issue identification documents, preventing departure with personal property, and even renouncing their citizenship [@Fitzgerald2006]. After the fall of the “Iron Curtain” and the Berlin Wall, however, emigration restrictions for either political or economic reasons have become rather rare [@DeHaas2011]. After all, prompted by the ideas embodied in the American and French revolutions [@Green2007a], the right to exit and emigration—as opposed to entrance and immigration—was internationally recognized in the UDHR in 1948. (Note: “Everyone has the right to leave any country, including his own, and to return to his country” (Universal Declaration of Human Rights, Article 13).)

Nonetheless, while the outright use of coercion to restrict emigration have almost disappeared, at least one out of four governments worldwide today have policies to discourage emigration [@UN2013]. States—especially those in the developing world—use a variety of instruments to reduce the number of their (disproportionately young and educated) emigrants, encourage return migration, and work with their diasporas abroad [@DeHaas2011, Weinar2017]. Since the famous proposal of @Bhagwati1973, calls to combat so-called “brain drain” by taxing high-skilled emigrants or the receiving governments, as well as imposing professional quotas and punishing recruiters, have become common among pundits and policymakers. At the same time, some political theorists point out that most accepted normative reasons to restrict immigration (such as related to its distributive consequences) should imply similar reasons to restrict emigration [@Ypi2008a]. (Note: Many scholars are, however, increasingly skeptical that “brain drain” is a problem to be solved pointing out to a number of practical and ethical issues, as well as to the countervailing effects of incentivizing education [@Clemens2014, Sager2014a]. For a general review of economic effects of emigration, see @Leeson2015. For normative considerations, see @Stilz2016.) But despite the everyday reality of emigration and the continuing policy attempts to regulate it in many countries, the literature has largely ignored public opinion on the issue—perhaps due to its low political salience in the United States and other advanced democracies.

Comparing emigration and immigration attitudes

How many people are actually concerned about emigration across different, more or less developed countries? To the extent voters are concerned about emigration, do they think it should be increased or decreased? While these questions may be of interest in themselves, I argue it would be even more informative to juxtapose them with well-established explanations of immigration attitudes. Although a comprehensive theory of emigration attitudes is beyond this paper’s scope, examining the ways people think about regulating out-migration can potentially help better understand widespread resistance to in-migration and international human mobility in general. The study is thus aimed at comparing immigration and emigration attitudes, as well as their individual and contextual predictors.

In doing so, I first briefly discuss the similarities and differences of the two demographic processes which are potentially relevant to people’s opinion about the respective policy issues. As for similarities, both processes relate to human mobility between countries and thus imply some compositional population change at the societal level. Accordingly, migration in either direction can substantially impact the economy and other aggregate outcomes in the long run through changing agglomeration effects, as well as the demand for and supply of labor and housing. Unless there is a significant shock in migration patterns, however, the average economic effects of most population movements are marginal [@Leeson2015].

As for differences, while immigration naturally leads to the increase of population size, emigration leads to its decrease. Accordingly, the public often perceives immigration to increase the competition for economic resources [@Muste2013], which is arguably less likely in the case of emigration. At the same time, as explained above, individual rights to immigration are much less widely recognized than emigration rights around the world.

Despite these differences, however, both emigration and immigration can in principle generate either positive or negative social welfare effects depending on the context and the policy choices involved [@Clemens2018a]. Consequently, both processes can be viewed by voters as social problems to be addressed by the government.

Building on the recent public opinion literature, I further distinguish between two major components of migration attitudes?preferences and their issue salience. While migration preferences refer to comparative evaluations of various positions regarding human mobility, migration salience captures the importance that individual voters attach to these issues [@Hatton2017].

When it comes to immigration, it has been increasingly acknowledged that people’s preferences on the issue have been much more stable and robust to contextual shocks compared to its political salience [@Dennison2018, Kustov2019b]. In this sense, the literature suggests that people are not necessarily more negative—but are more concerned—about immigration in contexts with increasing foreign-born population [@Pottie-Sherman2017, Dennison2019b]. Similarly, one can expect emigration to be a larger political issue in contexts where it occurs more frequently.

At the extreme, a non-negligible level of immigration or emigration is a necessary condition for people to have an opinion about these issues. As mentioned earlier, for instance, this may explain why people rarely voice their concerns about emigration in the developed (largely immigrant-receiving) countries. At the same time, it is also true that people can have meaningful policy preferences on issues that are not politically salient. Accordingly, there is a large variation of more or less restrictive immigration preferences even among lower-income countries where it is not an issue significant to most voters [@IOM2015].

As a result, while migration levels and other contextual variables such as policy environment [@Citrin2012, Heizmann2015] may be important for explaining the prevalence of public concerns about emigration across countries, they are likely less useful for understanding the underlying individual drivers of migration preferences. To that end, the study further discusses the general expectations about the within-individual relationship between immigration and emigration preferences based on the widely established theories of political behavior.

Expectations of uncorrelated migration preferences

The first most widely explored predictor of individual immigration preferences—or any political attitude for that matter—is (perceived) self-interest [@Weeden2017]. According to the logic of labor market competition, for instance, voters are expected to oppose immigration of similarly-skilled individuals. But who would be more or less interested in keeping their own citizens from leaving their countries? On the surface, it should be the reverse of this basic political economy explanation. That is, voters can be expected to support emigration of similarly-skilled individuals. Given that immigrants tend to be lower-skilled—and emigrants to be higher-skilled—than the average population, however, the predictions are less clear. This is especially true considering migrants’ fiscal and other indirect impacts, as well as the fact that natives’ skills are often correlated with non-economic reasons to support or oppose migration [@Hainmueller2014]. Nonetheless, it is unlikely that particular individuals can be made better off by decreasing (or increasing) both immigration and emigration simultaneously. Consequently, emigration and immigration preferences are not expected to be correlated if self-interest is a major driver of migration attitudes.

Another prominent explanation of opposition to immigration is related to ethnic and racial prejudice or dislike of foreigners in particular. While this may encompass a variety of cultural and demographic factors, the basic premise is that some people hold antipathy against certain (immigrant, largely non-white) ethnic outgroups and thus want government to restrict immigration to their country [@Ceobanu2010]. It is not clear, however, why people who are biased against foreigners would oppose emigration and limit mobility of their (ingroup) fellow citizens. Of course, it is possible that those who want to preserve the ethnic demographic composition of their communities would want to limit (native) emigration. But while sometimes equated with prejudice, such diversity concerns can also be viewed as a separate factor which may or may not itself be caused by prejudice [@Kaufmann2019]. Emigration and immigration preferences are thus not expected to be correlated if anti-foreign prejudice is a major driver of attitudes.

Expectations of correlated migration preferences

Outgroup prejudice, however, is sometimes hard to disentangle from various group interest motivations to support restrictive policies [@Newman2019]. Most prominently, the ideas of “sociotropic politics” and “group threat” suggest that—unrelated to dislike of foreigners—people support or oppose immigration because they think it is good or bad for others in their national ingroup [@Kustov2019a]. Perhaps even more so than in the case of other policy issues, it has been increasingly acknowledged that public attitudes toward immigration are much more driven by its perceived social rather than personal impacts [@Hainmueller2014]. While the widespread perceptions of immigration’s negative impacts are well-documented [@Muste2013], it is likely that emigration is also generally perceived in a negative way by the public. After all, in part due to the pervasive `brain drain’ narrative, government policies to discourage emigration are much more common than to encourage it [@UN2013]. Consequently, one would expect people—who care more about the well-being of their countries and compatriots—to be especially opposed to both immigration and emigration. (Note: Although self- and group-interest explanations are empirically distinct, in theory voters may advance their collective interests as a heuristic for their self-interest without being altruistic [@Weeden2017].) In turn, this implies a positive correlation between migration preferences.

Finally, immigration and emigration attitudes could also be in principle positively correlated if people exhibited moralistic anti-migration bias with no regard to migration’s economic and social consequences. Indeed, the significance of moralistic or categorical moral judgments (i.e., rules of right and wrong applied to oneself and others) to political behavior has been widely documented [@Baron2003, Haidt2007, Ryan2017]. While this factor has been somewhat overlooked in the case of migration preferences, @Wright2016a, for instance, find that much of the opposition to (undocumented) immigration is “rooted in rigid moralistic convictions about the importance of strict adherence to rules and laws.” Similarly, but even more generally, people may be categorically averse to international migration of any form above and beyond the economic and cultural motivations described above.

Are people more supportive of emigration or immigration? Do immigration and emigration attitudes correlate at the individual level? As can be seen from the discussion above, depending on the particular theory, the general expectations regarding the two sides of the migration debate may either converge or diverge. While it is likely that individual attitudes are driven by all of the factors described above, disentangling between their relative strength has important implications for our understanding of migration politics and public opinion. Overall, conditional on some non-negligible level of both emigration and immigration, we would expect the following relationship between emigration and immigration attitudes (in the case of self-interest or prejudice and group interest or anti-migration bias respectively):

[1] Opposition to emigration has a negligible correlation with opposition to immigration [self-interest or prejudice]

[2] Opposition to emigration has a positive correlation with opposition to immigration [group interest or anti-migration bias]

Study 1

Data

To provide a general test of these theoretical expectations, this paper exploits the unique and previously unexplored Gallup World Poll data from 19 European and CIS countries (2013 and 2015) on attitudes toward regulation of both immigration and emigration. (Note: The dataset consists of representative samples including Armenia (2013), Azerbaijan (2013, 2015), Belarus (2013, 2015), Bulgaria (2015), Croatia (2015), Czech Republic (2015), Georgia (2013, 2015), Greece (2015), Hungary (2015), Kazakhstan (2013, 2015), Kyrgyzstan (2013, 2015), Moldova (2013), Poland (2015), Romania (2015), Russia (2013, 2015), Slovakia (2015), Tajikistan (2013, 2015), Ukraine (2013, 2015), Uzbekistan (2013, 2015). Given the absence of a proper survey infrastructure, the results from Turkmenistan (2013, 2015)—though seemingly not significantly different from neighboring countries—are excluded from the analysis.) The two main dependent variables are defined as follows (Note: While these items do not specify particular government interventions, I follow the immigration literature and assume that they are indicative of related policy attitudes.):

Emigration preferences: “In your view, should emigration from this country be kept at its present level, increased, or decreased?”
Immigration preferences: “In your view, should immigration in this country be kept at its present level, increased, or decreased?”

To test the extent to which migration attitudes are context-dependent, I complement the survey data with World Bank data (2015) on immigrant and emigrant population shares. While I do not aim to directly test the underlying individual-level theories of migration preferences, I also include the standard demographic indicators used in the literature and available in the Gallup data as control variables in some regression specifications. (Note: This includes gender, age, marriage status, residency, nativity, education, income, and unemployment.) All variables are normalized on a 0-1 scale. For the detailed variable descriptions, see Appendix.

To make a stronger empirical case and also consider generalizability across more or less developing contexts, I also complement the analysis with the Transatlantic Trends Survey (TTS) of 15 countries from 2013 and 2014. (Note: The full list of countries is as follows: France (2013, 2014), Germany (2013, 2014), Greece (2014), Italy (2013, 2014), the Netherlands (2013, 2014), Poland (2013, 2014), Portugal (2013, 2014), Romania (2013), Russia (2014), Slovakia (2013), Spain (2013, 2014), Sweden (2013, 2014), Turkey (2013, 2014), the United Kingdom (2013, 2014), and the United States (2013, 2014).) Importantly, it provides an alternative item—modified for clarification in 2014—capturing individual emigration salience (or “concerns”):

Emigration concerns (2013): “Do you think that emigration in [COUNTRY] is a very serious problem, a somewhat serious problem, not a serious problem, or not a problem at all for [COUNTRY]?”
Emigration concerns (2014): “Do you think that emigration in [COUNTRY], that is the number of [NATIONALITY] who are leaving to live in other countries, is a very serious problem, not a serious problem, or not a problem at all for [COUNTRY]?”

While the TTS has no otherwise equivalent immigration questions, I use the following set of relevant items to capture immigration concerns (for details, see Appendix):

Immigration concerns: “Can you tell me if you are worried or not worried about $?$ immigration?”

Accordingly, the main advantage of the TTS dataset is that it offers a much more explicit item on emigration which can help rule out potential measurement error. At the same time, in addition to some of the Gallup countries, it also covers more developed and immigrant-receiving contexts. Given that the TTS items indicate migration salience' rather than preferences,’ however, the results cannot be directly compared to the ones from Gallup.

Analysis and results

First, I estimate the overall level of opposition to human migration across the available representative samples in the Gallup data. Given the remarkable aggregate stability of both emigration and immigration attitudes in all ten countries surveyed by Gallup twice (no change exceeds the margin of error), the descriptive analysis below combines their 2013 and 2015 samples. As can be seen from Figure fig:antimigration, across most countries the majority or an overwhelming plurality wants to reduce both emigration and immigration. (Note: For public opposition to migration in all samples separately, see Figure fig:time in Appendix.)

To rule out potential language confusion, I also calculate the level of emigration concerns and its change depending on the explicit wording across countries in the TTS data. As Figure fig:TT indicates, all of the Gallup countries (Greece, Poland, Romania, Slovakia, Russia) exhibit a similarly high level of concerns about emigration. Furthermore, with the sole exception of Turkey, these attitudes are very stable from one wording to another (and from year to year). In sum, explicitly defining emigration to a respondent and thus minimizing the potential for measurement error does not seem to impact the results.

Figure. Public opposition to emigration and immigration (Gallup)

Figure. Emigration attitudes and their sensitivity to wording (Transatlantic Trends)

Descriptively, we can see that the opposition to emigration almost always exceeds the one to immigration (Figure fig:antimigration). While somewhat surprising, as explained earlier, this may be a function of the particular high-emigration context of the middle-income countries sampled by Gallup. According to the simple correlation of migration stocks and attitudes, this indeed appears to be the case: respondents are more likely to oppose emigration (immigration) in countries with higher emigrant (immigrant) shares, but not vice versa (see Figure fig:scatter in Appendix). Although a fuller examination of contextual predictors is beyond the scope of this paper, this systematic relationship is important because it further suggests that emigration attitudes is a meaningful construct distinct from immigration attitudes. Overall, however, these correlations are rather small and not necessarily indicative of a causal relationship. As emphasized by some scholars [@Hopkins2010], the elites first have to successfully connect a more or less observable demographic change with politics for it to have any influence.

Examining individual correlations

After establishing the meaningfulness and distinctiveness of attitudes toward (e)migration at the societal level, I move to examining their relationship at the individual level. First, I correlate emigration and immigration preferences both across and within Gallup countries. Overall, the relationship between the two variables as measured in a bivariate correlation in the whole sample of countries is substantial ($r = 0.37$). (Note: Azerbaijan and Armenia are henceforth excluded from the individual-level analysis since both of these emigrant-sending countries have virtually no immigration and the related opposition to it (see Figure fig:antimigration).) While there is some between-country variation in the strength of the coefficient, the relationship is positive and strong in nearly all countries (see Figure fig:corr). Furthermore, this holds true even after accounting for major demographic covariates (see Table tab:individual). Finally, migration preferences are equally correlated among collegeand non-college educated respondents ($r = 0.34$ and $r = 0.38$), further indicating that the relationship is unlikely to be a result of linguistic confusion. (Note: Another potential concern about this analysis is that the respondents in all samples are by definition the ones who did not emigrate. It is unclear, however, whether this might bias the observed correlation between emigration and immigration attitudes upward or downward. In the end, while it is true that migrants appear to be slightly more supportive of both emigration and immigration (see Table tab:individual), migration preferences are equally correlated among the native-born and foreign-born populations ($r = 0.37$). Consequently, the absence of current emigrants among the respondents is unlikely to significantly bias the correlation estimates.)

Figure. The relationship of anti-migration preferences by country (Gallup)

Second, I conduct a similar correlational analysis based on the alternative attitudinal measures in the TTS data. As in the case of preferences, the relationship between emigration and immigration concerns is substantial ($r = 0.2-0.3$, depending on the sampling year and operationalization). In other words, those who view emigration as a problem are more likely to worry about immigration as well (Figure fig:corr3). Importantly, this holds across different contexts, including both poorer and richer countries within the European Union.

Although the results regarding migration preferences and salience (based on the Gallup and TTS data respectively) are not directly comparable, they are very much in line with each other. All in all, these analyses suggest that those who dislike immigration and those who dislike emigration are largely the same people (and, in some countries, they are in overwhelming plurality). In particular, across the Gallup countries, about 45% on average would like to see both emigration and immigration to be decreased, while only less than 5% would want the opposite. Similarly, across the EU countries in the TTS data, about 42% are at least somewhat concerned about both emigration and immigration, while less than 12% are not at all concerned about either of these demographic processes.

Figure. The relationship of anti-migration concerns by country (TTS)

Exploring individual predictors

According to the analysis above, emigration and immigration attitudes are significantly correlated. However, there can still be important differences in the individual predictors of these attitudes. Although the direct test of psychological mechanisms is beyond the scope of Study 1, this section further compares the explanatory power of major demographic factors previously highlighted in the immigration literature. To that end, I regress both emigration and immigration preferences on a number of prominent covariates available in the Gallup data using a standard OLS specification with fixed country and year effects.

As can be seen from Table tab:individual (1-2), most standard demographic covariates are more or less similarly related to opinion on either issue. Most prominently, younger respondents and those born abroad are much less likely to oppose human mobility between countries of any kind. Somewhat surprising, but consistent with previous evidence from Eastern Europe [@Bessudnov2016], urban residents are not more, but less, likely to support migration. Albeit of seemingly various magnitude, however, the coefficients for education, income, and unemployment are not statistically different between emigration and immigration attitudes.

Overall, while immigration attitudes are slightly better predicted than emigration attitudes by the same variables, this difference can in part be explained by the greater between-country variation of the former in the Gallup data (country fixed effects alone explain 11% and 5% of variation respectively). Importantly, however, the relationship between these attitudes does not change after accounting for these demographic covariates (see Table tab:individual).

Table. Correlation of anti-emigration and anti-immigration preferences (Gallup)

Discussion

The fact that emigration and immigration attitudes are highly correlated and have similar individual predictors indicates a substantial common component underlying both attitudes. As mentioned earlier, this result is more in line with some theories of public opinion than others. Most important, it seems hard to explain why most people feel similarly about immigration and emigration by just appealing to self-interest or outright prejudice (at least in terms of anti-foreigner sentiments). At the same time, the results are more in line with the increasingly accepted group interest accounts, according to which people may oppose both emigration and immigration because they view them as social problems. Considering that the correlation between emigration and immigration attitudes does not diminish much even after including all explanatory covariates, however, it is clear that group interest and other prominent accounts cannot be exhaustive explanations. While the data do not allow for a direct test of other mechanisms, consistent with these results is also the idea that many people may simply have an anti-migration—as opposed to mere anti-foreign—bias.

One of the most important concerns about the study design is perhaps a potential measurement error related to the confusion of emigration with immigration by the respondents. Nonetheless, there are several reasons why it should not significantly affect the results. First, as emphasized earlier, opposition to emigration (or immigration) is predictably higher in countries with greater emigrant (or immigrant) shares. Second, it is hard to argue that self-reported opposition to emigration attitudes is just a function of opposition to immigration since the former is significantly higher in many of the sampled countries. Third, it is instructive that the percentage of “don’t know” answers is actually lower for emigration than immigration in most of the Gallup countries (13% vs 15% on average), which is in line with the overall emigrant-sending demographic context there. Fourth, the correlation between immigration and emigration attitudes is similar among low-educated and high-educated respondents. Fifth, there is little evidence that the observed correlation can be explained as a language artifact (Note: While in most languages the terms “immigration” and “emigration” are pronounced and spelled similarly to English (with a difference in a prefix or a suffix), some languages use either the same or a completely different word for these processes. Depending on one’s language, people may thus be more or less likely to confuse emigration and immigration. Nonetheless, language differences can only explain a part of between-country—not within-country—variation.), especially given the evidence on differential wording from Transatlantic Trends. Finally, the analysis of qualitative responses (see below) revealed only a few cases ($< 1%$) where people confused the issues.

Study 2

Study 1 was one of the first to document high public opposition to emigration salient in many contexts and to examine its strong relationship with public opposition to immigration. Nonetheless, given the limitations of large cross-national surveys and secondary data analyses, the drivers of these attitudes may still be subject to various interpretations. Furthermore, quantitative data alone arguably cannot tell us with certainty whether the stated emigration attitudes in large-scale surveys are indeed politically meaningful to people. Relatedly, given the novelty of the topic to public opinion research, we still do not know whether these attitudes can translate into actual, albeit seldom realized, policy preferences for the government regulation of migration. To address these concerns, Study 2 is designed to provide a more direct test of the group interest mechanism (vs self-interest and prejudice) behind the public opposition to (e)migration along with an exploratory qualitative inquiry into the issue.

According to the established logic of group interest outlined in the theoretical section, people tend to support policies that they think are good for their country and compatriots. Consequently, public attitudes toward immigration and other issues are often driven by their perceived national rather than personal impacts. Consistent with this idea, and contrary to some of the explanations based on economic self-interest, most natives prefer high-skilled to low-skilled immigration regardless of their own skills, because the former is perceived to be more socially beneficial [@Hainmueller2010]. Applied to emigration attitudes, this idea implies that people should be more opposed to high-skilled than low-skilled emigration since the former is likely to be perceived as more socially detrimental. Importantly, if group interest rather than self-interest is at work, the relative preference for low-skilled emigration should be independent of people’s own skills (Hypothesis 3). Since it is also possible that these skilled-based preferences are attributable to prejudice [@Newman2019], I also aim to test whether these preferences interact with racial attitudes. If the preference for low-skilled emigration is also indicative of prejudice rather than just group interest, we would expect the skill difference to be larger among those of high prejudice (Hypothesis 4).

Data

One important finding that emerges from Study 1 is that many people are concerned about emigration even in the predominantly immigrant-receiving, high-income countries such as the United Kingdom. The amount of negativity in these views is instructive, since the emigration issue—unlike immigration—is arguably not politicized in rich countries with positive net-migration. Though it is mostly rightly viewed through the lens of immigration politics (Note: The widespread popular opposition to migration in the UK has been often linked to the relative success of the euroskeptic UK Independence Party (UKIP) and the “Leave” vote in the 2016 EU referendum [@Goodwin2017].), however, the United Kingdom also has one of the largest emigrant populations among high-income countries. Accordingly, as revealed in Study 1, the UK also exhibits the highest correlation between emigration and immigration attitudes, at least among developed countries. To provide a more direct test for the mechanisms underlying my argument, I thus complement the cross-national analysis in Study 1 with an original survey in the UK.

In particular, I administered a quasi-representative survey of 2008 British citizens conducted online by Qualtrics in May 2018. (Note: As a part of larger research project, this diverse national sample targeted to match census demographics was obtained from the initial pool of 2050 respondents after accounting for response quality (attention check and survey completion) and excluding non-citizens. The inclusion of all respondents in the analysis, however, does not affect the results (not shown). The employed Qualtrics panel was representative of the population across most important demographic and political characteristics (see Table tab:Descr for summary statistics).) In doing so, I was able to include explicit measures of emigration preferences and racial prejudice (measured as negative group affect via feeling thermometer), as well as an optional open-ended emigration item so that people can elaborate on their stated preferences (for details, see Appendix). Most important, I also made an explicit distinction between high-skilled and low-skilled emigration in a (between-subjects) survey experiment framework to test Hypotheses 3 and 4:

Low-skilled emigration condition: “Now, please consider British citizens without university degree who are leaving to live in other countries. Do you think low-skilled emigration from Britain should be increased a lot, increased a little, left the same as it is now, decreased a little, or decreased a lot?”
High-skilled emigration condition: “Now, please consider British citizens with university degree who are leaving to live in other countries. Do you think high-skilled emigration from Britain should be increased a lot, increased a little, left the same as it is now, decreased a little, or decreased a lot?’

Analysis and results

In line with the results from the Transatlantic Trends data, the majority of British respondents do not seem to have a particular preference about emigration levels. After all, while emigration and immigration levels have recently been quite comparable in the UK [@Vargas-Silva2017], the former is much less discussed and debated than the latter.

However, it is important to differentiate between the existence of meaningful attitudes themselves—whether positive or negative—and their salience in the public discourse [@Dennison2018]. Indeed, it appears that a significant fraction of the UK population does oppose emigration and, similar to other countries, this opposition correlates with anti-immigration attitudes at 0.24. (Note: Relatedly, those who oppose emigration are also more likely to report voting Leave in the 2016 EU referendum (r = 0.18), even after accounting for major demographic covariates (not shown).) Moreover, this opposition is also expectedly related to emigrants’ skill level in question (see Figure fig:skill). In fact, twice as many people are willing to decrease high-skilled compared to low-skilled emigration (34% vs. 17%).

Figure. Emigration attitudes by skill-level (UK)

How can we explain these significant differences? Building on the research design of @Hainmueller2010, I regress anti-emigration attitudes on education (university degree), treatment condition (low-or high-skilled), and their interaction. As can be seen from Table tab:skill, while respondents’ education and emigrants’ skill have significant effects, these factors do not interact with each other. Put differently, all respondents are less supportive of high-skilled emigration regardless of their own skills and educated respondents are more supportive of any emigration regardless of emigrants’ skills. In turn, these results are more consistent with group interest rather than self-interest accounts of public attitudes.

Table. The effect of emigrants’ skills on anti-emigration attitudes by respondents’ education

Can the relative preference for low-skilled emigration be also indicative of racial prejudice at work? To test this idea, I also look at the heterogeneous treatment effects by respondents’ prejudice levels. As can be seen from Table tab:prej, the relative popular preference for low-skilled emigration does not depend on one’s attitudes toward racial and ethnic minorities. Furthermore, while prejudice is positively correlated with general anti-emigration attitudes (similar to Study 1), this relationship disappears after controlling for anti-immigration attitudes. Importantly, the same results hold even after accounting for various demographic factors, religiosity, and political ideology. In the end, similar to the results from Gallup, British emigration and immigration attitudes are among the strongest predictors of each other.

Table. The effect of emigrants’ skills on anti-emigration attitudes by respondents’ prejudice

Exploring the content and meaning of emigration attitudes

While we have a good understanding of immigration attitudes, it is much less clear what people think about emigration. Do voters have meaningful concerns about others leaving their country and how different are those from common immigration concerns? To address these questions, I also asked my UK respondents to elaborate on their emigration attitudes stated earlier in an optional open-ended form. Although some of these opinions may be viewed as just `socially desirable’ rationalizations of underlying predispositions [@Lodge2013], open-ended responses are informative inasmuch as they represent the way individuals justify their political preferences to others. As may be expected from the comparatively low salience of the emigration issue in the UK (and the fact that the form was optional), about half of the respondents chose not to comment on their issue stance. Nonetheless, this does not at all imply that the stated preferences observed in the survey experiment are just artifacts with no substantive political meaning. Indeed, the analysis of the given majority of responses indicates a predominance of meaningful political attitudes on emigration. (Note: While more qualitative and quantitative text-based research on (e)migration attitudes is warranted, a more systematic text analysis of open-ended responses used for theory testing is beyond the scope of this article.)

First, similar to what one may expect in the case of any other political issue, many respondents were alluding to a variety of sociotropic and group interest concerns. (Note: Although few respondents also referenced self-interest and personal choice arguments, they were more likely to prefer status quo levels to increased or decreased emigration.) In the high-skilled condition, the most common explanation for anti-emigration attitudes was related to the idea that “we [the British people] need them [the workers] here” to help the economy or avoid “brain drain.” Interestingly, it was also common to state group interest reasons in favor of greater low-skilled emigration (e.g., “[Low-skilled workers are] not contributing to society” or “[w]e have too many people”). Consequently, one should be careful with interpreting pro-emigration preference as a principled embrace of “freedom of movement.” Furthermore, a few people also mentioned that they want to decrease emigration due to their concerns about increased immigration or, specifically, “population replacement.” But while this might explain why negative attitudes toward racial minorities correlate with opposition to either type of migration, it is unclear whether such preferences for ethnic homogeneity can themselves be considered a part of prejudice or group interest concerns.

Finally, group interest and prejudice aside, many of the expressed reasons to restrict emigration were suggestive of the categorical anti-migration bias described in the previous study. In particular, many respondents alluded to the idea that people ought to live and work in their own countries regardless of consequences. Interestingly, this universal moral responsibility to stay put as perceived by the respondents, which is in principle violated by both emigrants and immigrants, has been rarely qualified. Some respondents, however, further alluded to special national obligations (e.g., “[p]eople should remain loyal to their country”) or fairness (e.g., “we shouldn’t emigrate to another country and be a burden on that country”). Overall, it appears immigration and emigration views may go together due to voters’ deep moral intuitions about the wrongness of international migration to both receiving and sending societies (in addition to their sociotropic concerns).

Discussion

Study 2 provided an original UK survey with an embedded experiment and an open-ended policy item about emigration, which can be considered a harder case to test my argument. Overall, the study confirmed that voters have systematically different views toward emigration based on its skill composition. In turn, the revealed strong preference for low-skilled emigration across all voters regardless of their own skills or outgroup attitudes is more in line with group interest rather self-interest or prejudice-based explanations. Nonetheless, both quantitative and qualitative evidence in Study 2 also makes it clear that group interest alone is hardly sufficient to explain popular opposition to migration. Specifically, a substantial relationship between immigration or emigration attitudes independent of other factors implies that voters tend to resist human mobility in general, not immigrants or emigrants in particular. The analysis of open-ended responses, many of which simply stated that people should remain where they are born, further supports this notion.

Of course, it is possible that the results are partly a function of the UK case or the choice of skill manipulation. After all, other emigration distinctions can be as or even more consequential depending on a particular context. Most notably, apart from conducting original surveys in countries where emigration is a highly politicized issue, future research can look at emigration attitudes based on emigrants’ ethnic background and other demographic characteristics, as well as emigration reasons and duration. For instance, it is possible that voters would be more supportive of ethnic minority emigration, politically-motivated emigration, or temporary emigration (especially in countries that are highly reliant on remittances).

Conclusion

As of now, many scholars view immigration and emigration as separate demographic processes that are relevant in vastly different national contexts. However, this overlooks the fact that pluralities of voters in many countries dislike both types of international migration despite the important contextual differences. By systematically comparing opposition to emigration and immigration, the paper further contributes to the literatures on migration politics, nationalism, and public opinion more generally.

Overall, I provide new empirical evidence that (1) many voters around the world express salient and meaningful political opposition to emigration; and that (2) emigration and immigration attitudes, albeit being distinct, are both significantly driven by common factors. Taken together, these novel tests from Study 1 and Study 2 corroborate that sociotropic concerns and perceived national interests are key, though not exclusive, predictors of migration attitudes [@Hainmueller2014]. In line with the recent literature on the role of morality in politics [@Ryan2017], however, people’s own explanations of their attitudes further suggest that some may simply view international human mobility of any form as morally wrong.

Compared to the more consequentialist concerns about national interest that can lead to either supporting or opposing migration depending on the context [@Kustov2019], categorical anti-migration bias reflected in open-ended responses is more about an unconditional preference against any type of human mobility between countries. Consequently, while being largely overlooked in the literature, such moralistic aversion to international migration in general can potentially explain a large share of anti-immigration sentiments beyond group interest concerns, threatened personal interests, and prejudice against foreigners. Nonetheless, since it is still unclear how widespread these moral concerns are and to what extent they can predict migration attitudes in the general population, scholars should examine these questions in more detail. Future research could also explore alternative, less proximate causes for general anti-migration bias revealed in this study.

Furthermore, it is important to provide a more comprehensive theory and collect other data on emigration attitudes in the future. Scholars could also elaborate on what “self-interest” and “prejudice” imply in the case of emigration, as well as measure how people actually perceive the effects of emigration in different contexts. Finally, future studies can benefit from examining the interrelationship of group interest concerns, anti-foreign bias, and moralistic aversion to human mobility described here.

All in all, the evidence of widespread anti-emigration attitudes has important implications for democratic theory and politics. Most obviously, the fact that most governments (including the countries studied here) rarely restrict or even merely discourage emigration despite the strong will of their citizens presents both an empirical and a normative challenge. Normatively, theoretical accounts that defend immigration restrictions by appealing to sovereignty and citing public opinion, should also be able to explain why it would not apply to emigration [@Ypi2008a]. Empirically, scholars may want to consider why emigration has become much freer than immigration despite the similarities in public attitudes [@Peters2015].

References

See source bibliography.

Online Appendix

Gallup World Poll

The Gallup World Poll dataset consists of representative samples including Armenia (2013), Azerbaijan (2013, 2015), Belarus (2013, 2015), Bulgaria (2015), Croatia (2015), Czech Republic (2015), Georgia (2013, 2015), Greece (2015), Hungary (2015), Kazakhstan (2013, 2015), Kyrgyzstan (2013, 2015), Moldova (2013), Poland (2015), Romania (2015), Russia (2013, 2015), Slovakia (2015), Tajikistan (2013, 2015), Ukraine (2013, 2015), Uzbekistan (2013, 2015). Given the absence of a proper survey infrastructure, the results from Turkmenistan (2013, 2015)—though seemingly not significantly different from neighboring countries—are excluded from the analysis.

Emigration attitudes (WP14599): “In your view, should emigration from this country be kept at its present level, increased, or decreased?”
Immigration attitudes (WP1328): “In your view, should immigration from this country be kept at its present level, increased, or decreased?”
Demographic covariates
Education (WP3117), Gender (WP1219), Age (WP1220), Marriage Status (WP1223), Urban residence (WP14), Foreign-born (WP4657), Income quintile (INCOME 5), Employment Status (EMP 2010)
“Plans to emigrate” (WP1325): “Ideally, if you had the opportunity, would you like to move PERMANENTLY to another country, or would you prefer to continue living in this country?”
“Receives remittances (from abroad)” (WP9086): “In the past 12 months, did this household RECEIVE help in the form of money or goods from another individual living inside this country, living in another country, both, or neither?”
“Dislikes foreign governments” (8-item index): “Do you approve or disapprove of the job performance of the leadership of the United States (United Kingdom, Germany, France, Russia, China, Japan, India?”
“Worries about national issues” (6-item index):
(WP4941) “Do you think the government is doing enough to fight terrorism, or not?”
(WP6763) “Do you think the government of your country is doing enough to fight corruption, or not?”
(WP7503) “In your opinion, is the government doing enough to fight domestic violence?”
(WP131) “In this country, are you satisfied or dissatisfied with efforts to deal with the poor?”
(WP132) “In this country, are you satisfied or dissatisfied with efforts to preserve the environment?”
(M30) “How would you rate economic conditions in this country today? as excellent, good, only fair, or poor?”

Transatlantic Trends Survey

The Transatlantic Trends Survey dataset consists of representative samples including France (2013, 2014), Germany (2013, 2014), Greece (2014), Italy (2013, 2014), the Netherlands (2013, 2014), Poland (2013, 2014), Portugal (2013, 2014), Romania (2013), Russia (2014), Slovakia (2013), Spain (2013, 2014), Sweden (2013, 2014), Turkey (2013, 2014), the United Kingdom (2013, 2014), and the United States (2013, 2014).

Emigration attitudes (2013): “Do you think that emigration in [COUNTRY] is a very serious problem, a somewhat serious problem, not a serious problem, or not a problem at all for [COUNTRY]?” [A very serious problem (3), a somewhat serous problem (2), not a serous problem (1), not a problem at all (0)]
Emigration attitudes (2014): “Do you think that emigration in [COUNTRY], that is the number of [NATIONALITY] who are leaving to live in other countries, is a very serious problem, not a serious problem, or not a problem at all for [COUNTRY]?” [A very serious problem (3), a somewhat serous problem (2), not a serous problem (1), not a problem at all (0)]
Immigration concerns (2013) [an average of the two items]:
“Can you tell me if you are worried or not worried about Legal immigration?” [Worried (1), not worried (0)]
“Can you tell me if you are worried or not worried about Illegal immigration?” [Worried (1), not worried (0)]
Immigration concerns (2013) [an average of the two items]:
“Can you tell me if you are worried or not worried about immigration from within the EU?” (only asked in the European Union) [Worried (1), not worried (0)]
“Can you tell me if you are worried or not worried about immigration from outside the EU?” (only asked in the European Union) [Worried (1), not worried (0)]

Qualtrics Survey (UK)

Low-skilled emigration condition (0-1): “Now, please consider British citizens without university degree who are leaving to live in other countries. Do you think low-skilled emigration from Britain should be increased a lot, increased a little, left the same as it is now, decreased a little, or decreased a lot?”
High-skilled emigration condition (0-1): “Now, please consider British citizens with university degree who are leaving to live in other countries. Do you think high-skilled emigration from Britain should be increased a lot, increased a little, left the same as it is now, decreased a little, or decreased a lot?”
University degree (0 or 1): “What is the highest degree or level of school you have completed? (If currently enrolled, highest degree received)”
Racial prejudice (0-1): “We’d like to get your feelings toward a number of groups on a feeling thermometer. A rating of 0 means you feel as cold and negatively as possible toward the group. A rating of 100 means you feel as warm and positively as possible toward the group. You would rate the group at 50 if you feel neither positively nor negatively toward the group. How do you feel toward? Whites, Blacks, Asians, Arabs” (constructed as the average feeling toward non-Whites)
Other covariates
Ideology: “In politics people sometimes talk of left' and right.’ Using this scale, where would you place yourself on this scale, where 0 means the left and 10 means the right?” [0-10 scale]
Religiosity: “Regardless of whether you belong to a particular religion, how religious would you say you are?” [0-10 scale]
Female, Age, Married, Foreign-born, Income, Urban residence, Unemployed

Figures and Tables

Afigure

Figure. Public opposition to emigration and immigration across time

Afigure

Figure. The relationship of migrant shares and migration attitudes

Atable

Table. Descriptive statistics—Gallup sample (n = 27,067)

Atable

Table. Descriptive statistics—UK Qualtrics sample (n = 2008)