Abstract
What factors shape attitudes toward immigration? Previous studies have typically debated whether citizens oppose immigration more for economic or cultural reasons. We broaden this debate by exploring how different segments of the citizenry feel about immigration. Our original surveys conducted in Japan reveal two separate axes along which many citizens view immigration: (1) its cultural and economic effects, and (2) its positive and negative effects. Even in Japan, whose relatively closed policy toward immigration is conventionally believed to reflect widespread public intolerance of outsiders, over 60 percent of our respondents favor widening the doors to immigrants for economic or cultural reasons or both.
Introduction
Attitudes toward immigration have played a potent role in the electoral politics of rich democracies in recent decades, creating an urgent need to understand these views. According to some studies, citizens fear the dilution of prized cultural norms and values (Card, Dustmann, and Preston 2012; Hainmueller and Hopkins 2015; Newman et al. 2013; Sides and Citrin 2007a). Other studies point to economic considerations, such as the concern that immigrants will drain the country’s welfare resources (Burgoon and Rooduijn 2021; Dustmann and Preston 2006; Gerber et al. 2017; Hanson, Scheve, and Slaughter 2006; Mewes and Mau 2013) or that they will compete for jobs (Goldstein and Peters 2014; Mayda 2006; O’Rourke and Sinnott 2006; Scheve and Slaughter 2001a). But beyond the question of why citizens oppose immigration, recent research also reports a considerable degree of pro-immigrant sentiment across developed democracies (Card, Dustmann, and Preston 2012; Hainmueller and Hopkins 2015). We broaden the debate to consider whether different segments of the citizenry may approach the issue from different angles.
Drawing on original surveys conducted in Japan, we demonstrate the importance of understanding both (1) who cares about which kinds of immigrants and (2) why they exhibit these preferences. Japan’s restrictive immigration policy is generally taken to reflect public distaste for outsiders, but we find that over 60 percent of our respondents favor widening the doors to immigrants for economic or cultural reasons or both. Our findings are consistent with recent research on other developed democracies that show sizable pro-immigrant sentiment (Card, Dustmann, and Preston 2012; Hainmueller and Hopkins 2015).
We excavate which factors shape whether individuals value or oppose immigration on cultural and/or economic grounds. As we would expect, education is associated with more open attitudes toward immigration, but there are differences among educated people depending on material or cultural circumstances. All else equal, corporate managers are likely to favor immigration, but only for economic reasons. People with foreign friends (currently or in the past) are more likely to appreciate immigrants for cultural reasons, compared with people who have never had foreign friends. These are examples of the more general phenomenon that attitudes often rest on conditions that could potentially change, suggesting that sentiment toward immigration can shift in the medium to long term.
We also find that the kinds of immigrants people favor depend on whether they oppose or favor immigration for economic or cultural reasons. People who appreciate immigrants for cultural reasons but not for economic ones are the least concerned about immigrants’ countries of origin or sector of employment. People who favor immigration only for economic reasons express similar preferences as those who welcome newcomers for both economic and cultural reasons, except that they want to admit a smaller number. These findings suggest that people with cross-cutting motivations cannot simply be understood as a jumbled midway point between different clear positions, but instead exhibit distinct preferences of their own.
The fact that some countries are more immigration-friendly than others may reflect business strategies to limit labor costs (Peters 2017). Even so, public attitudes toward immigration have become an important determinant of voting decisions in recent decades (Burscher, van Spanje, and de Vreese 2015; Kaufmann 2017; Rydgren 2008), and our disaggregation of types among the general public helps to illuminate additional sources of pressure on government policy from the public at large. This is an important issue for developed democracies in general but especially for Japan, which has gradually been retreating from its highly restrictive immigration policies in recent years (Roberts 2017; Strausz 2019) and will need more immigrants as the graying of its population accelerates. Indeed, Oishi (2020) argues that the recent relaxing of Japan’s immigration restrictions in late 2018 was driven at least in part by favorable public opinion.
By uncovering a multidimensional configuration of attitudes toward immigrants, we raise the possibility of larger coalitions for immigration reform than had previously been theorized. Citizens who favor immigration for only economic reasons and those who favor immigration for both economic and cultural reasons may belong to different political and social circles. Nevertheless, both groups are likely to favor immigrants in labor-scarce industries. If this is the case, people who value immigration only for cultural reasons and those who favor immigration for both cultural and economic reasons would support a more liberal immigration policy, regardless of immigrants’ geographic origin or potential labor market contributions. Pro-immigration policymakers may have more room for maneuver than has been supposed.
We offer preliminary evidence that Japanese attitudes toward immigration are not unique; they appear to be in the range of views in other rich democracies. The importance of these findings for comparative politics is considerable: although the distribution of the four types of individuals we find in Japan may vary by country, the general framework we develop is broadly applicable. A sharper delineation of these types in each country could reveal latent coalitions favoring immigration elsewhere, as in Japan.
Four Types of Attitudes toward Immigrants
A large and growing literature explores the varied bases of anti-immigration feeling. One group of scholars finds evidence of popular anxiety about immigration’s negative effects on a country’s ethnic, religious, and linguistic unity and on shared cultural traditions and norms (Card, Dustmann, and Preston 2012; Hainmueller and Hopkins 2014; Newman, Hartman, and Taber 2012; Sides and Citrin 2007a; Sniderman, Hagendoorn, and Prior 2004).
Another body of scholarship focuses on economic concerns, whether for themselves (self-interest) or for the economy as a whole (sociotropic preferences). Citizens may fear for their jobs (Mayda 2006; Scheve and Slaughter 2001a) or for their welfare checks (Dancygier and Donnelly 2013; Goldstein and Peters 2014; Hanson, Scheve, and Slaughter 2006; Malhotra, Margalit, and Mo 2013), or they may consider the effects of immigration, positive or negative, on the national economy as a whole (Bansak, Hainmueller, and Hangartner 2016; Citrin et al. 1997; Kehrberg 2007; Wilkes, Guppy, and Farris 2008). 1
Self-interested and sociotropic motivations are different in principle, but they are hard to disentangle in practice. For instance, workers in labor-scarce industries such as the construction industry in Japan may favor opening the door to immigrants because consistent with self-interest, new workers would not pose a job threat. If they value an influx of immigrants for the sake of sustaining the country’s construction boom, their motivations would be considered sociotropic. To the extent that the construction boom benefits workers directly and indirectly, favoring immigration for selfish and pro-social reasons push in the same direction. This paper therefore steps away from efforts to differentiate between these kinds of motivations.
We consider perceptions of threat associated with immigrants, but we also build on recent studies showing that many citizens across the developed world are willing to accept a larger number of immigrants, especially of certain types (Card, Dustmann, and Preston 2012; Hainmueller and Hopkins 2015). Even if citizens oppose immigration for cultural reasons, they may appreciate them for economic ones, or vice versa.
Rather than asking whether economic or cultural concerns more powerfully shape attitudes toward immigration, we recognize that citizens’ attitudes may be shaped by both or neither. Theoretically, there is no reason why people who worry about immigration for economic reasons may not also worry for cultural reasons. Conversely, it is also conceivable that those who appreciate immigration for economic reasons may also value immigration for cultural ones.
If citizens may be positively or negatively disposed toward immigrants for economic and/or cultural reasons, we can conceptualize four types of attitudes toward immigrants (Table 1): Integrators who are positively disposed toward immigrants on both economic and cultural dimensions, Exploiters who appreciate immigrants for their economic contributions but not for their cultural impact, Diversifiers who admire immigrants culturally but worry about economic effects, and Resisters who dislike immigration on both cultural and economic grounds.
Typology of Interacted Economic and Cultural Impacts.
Where Do the Different Types Come from?
Previous research has shown that different factors shape the likelihood of an individual supporting or opposing immigration for economic and cultural reasons. According to these studies, the sole factor affecting public support for immigration on both economic and cultural grounds is education. Better educated people should be more open to immigration for economic reasons because their own human capital insulates them from downward mobility. Education also cultivates a cosmopolitan vantage point that contextualizes the pluses and minuses of national culture (Hainmueller and Hiscox 2007). It is not surprising, then, that education is typically found to be the most consistent predictor of public support for immigration (Dancygier and Donnelly 2013; Hainmueller and Hiscox 2007; Hainmueller and Hopkins 2015; Mayda 2006; Scheve and Slaughter 2001b). In expectation, Integrators who are supportive of immigration both economically and culturally should be more educated, while Resisters will cluster at the opposite pole on the education continuum, and Exploiters and Diversifiers should fall somewhere in between.
Scholars have pointed to a number of factors that affect the likelihood of opposing immigration on economic grounds. Those with higher incomes have typically been found to be less averse to immigration (Malhotra, Margalit, and Mo 2013; Sides and Citrin 2007a). This is because, as with education, higher incomes shield individuals from economic competition with immigrants. 2 We therefore expect that people with higher incomes are more likely to be Exploiters or Integrators, and those with lower incomes are more likely to be Resisters or Diversifiers.
Occupational status and skill levels are also likely to affect citizens’ expectations of immigrants’ economic costs and benefits. We build on the insights of Dancygier (2010) and Dancygier and Donnelly (2013) that public attitudes toward immigration are driven at least in part by sector-level concerns. Those who are employed in labor-scarce industries should be less concerned about an influx of immigrants, whether because they need not compete with immigrants for jobs or because the economy will benefit from additional labor in those sectors. To the extent this is true, citizens employed in labor-scarce industries are, all else equal, less likely to be Resisters or Diversifiers and more likely to be Exploiters or Integrators. Attitudes toward immigration should also depend on individuals’ positions within the economy. Firm managers by dint of their insulation from low-skilled immigrants are more likely to be Exploiters or Integrators. Unemployed people, at the opposite end of the vulnerability ladder, are more likely to be Resisters or Diversifiers.
Other factors shape the likelihood that citizens appreciate or oppose immigrants for cultural reasons. The contact hypothesis suggests that personal experiences with outgroups diminish prejudice by increasing cultural understanding and by reducing fear of strangers (Allport 1954; Pettigrew and Tropp 2006). This leads us to expect that people who have (or have had) foreign friends will be more likely to be Diversifiers or Integrators. Conversely, people who have never had foreign friends should be more likely to be Resisters or Exploiters.
Existing studies typically find that young people are more accepting of immigrants than older people (Hainmueller and Hiscox 2006; Sides and Citrin 2007b). Social psychologists often attribute this difference to people’s diminishing capability, as they age, to inhibit their prejudices (Gonsalkorale, Sherman, and Klauer 2009; Von Hippel, Silver, and Lynch 2000). We therefore expect older respondents to be more likely to be Resisters or Exploiters, and less likely to be Integrators or Diversifiers.
Culturally conservative citizens, including known conservatives among supporters of Japan’s ruling Liberal Democratic Party (hereafter LDP), are more likely to be Exploiters or Resisters (Takenaka 2014). Those who support other parties or independents may be more likely to cluster among Diversifiers or Integrators.
Finally, scholars have found right-wing authoritarianism (RWA) to be associated with greater levels of prejudice (Hetherington 2009; Hetherington and Weiler 2009; Newman 2013). RWA has also been linked to ethnocentrism, which scholars have also found to shape hostility toward members of outgroups (Kinder and Kam 2009). We therefore expect people with authoritarian tendencies less likely to be Diversifiers or Integrators, and more likely to be Exploiters or Resisters. Of course, psychological attributes themselves may be partly shaped by macro-level circumstances; alternatively, people of certain personality types could disproportionately select the macro-circumstances in which they find themselves.
Different Preferences for Immigrants by Four Types
Different types of citizens should exhibit different preferences as to what kinds of immigrants to accept. First, different types of citizens may prefer immigrants from certain countries over others. Immigrants’ countries of origin may proxy for levels of human capital (Boyd and Thomas 2002). Exploiters and Integrators, who value immigrants for their potential contributions to the overall economy, should favor immigrants from developing countries, because the Japanese economy faces greater scarcity in unskilled compared with skilled labor. In recent years, labor shortages in the care industry has led Japan to accept care workers from the Philippines and Indonesia, which might cause Japan’s Exploiters and Integrators to associate workers in labor-scarce industries with immigrants from Southeast Asia. Because Integrators also value the cultural diversity of immigrants, they should be more indifferent than Exploiters as to immigrants’ countries of origin. By contrast, Resisters, who wish to preserve Japanese culture, should oppose immigrants irrespective of their countries of origin. Diversifiers, who value cultural diversity, should not markedly prefer immigrants from one country or another.
Second, different types of citizens should prefer immigration in particular sectors over others. Following Bansak, Hainmueller, and Hangartner (2016), we expect that individuals who value immigrants’ contributions to the economy, that is, Exploiters and Integrators, may welcome more immigrants in labor-scarce sectors and be less supportive of immigration in labor-abundant sectors. Diversifiers and Resisters, who think less about the economic contributions of immigrants, should be indifferent as to sectors.
Third, different types of natives should prefer to admit different numbers of immigrants and for different durations. Following Card, Dustmann, and Preston (2012) and Sides and Citrin (2007a), we expect that citizens worried about negative cultural or economic impacts of immigrants should want fewer immigrants and for shorter durations. Citizens who value cultural diversity should also favor more immigrants and to stay longer. Those who value immigrants for their economic contributions should prefer more immigrants and to stay longer, although their preferences should depend on the extent of labor shortage in the economy. In our formulation, this means that Resisters should want to admit fewer immigrants than Integrators. Exploiters or Diversifiers, who have one or another reason to be leery of “too many immigrants,” may be located in the middle of the continuum as to numbers of immigrants. Integrators and Diversifiers might want immigrants to stay for a longer time than would Resisters. Exploiters are likely to be somewhere in the middle because, for example, the economic impacts of immigrants should vary over time depending on macroeconomic conditions. Table A in the online appendix summarizes our main theoretical expectations.
The Survey
The Case of Japan
To test the utility of our proposed typology, we conducted surveys in Japan. Immigration has not been a politically salient issue in Japan if only because of its relative absence. According to the 2015 Japanese census, foreigners comprise only 1.4 percent of the overall population (Somusho Tokeikyoku [Statistics Japan], n.d.), compared with 7.0 percent in the United States, 7.7 percent in the United Kingdom, and 9.3 percent in Germany (OECD Data, n.d.). However, as in other developed democracies, Japan has recently seen a rise in its immigrant population: between 2000 and 2015, the number of foreign-born residents in Japan increased by 36.3 percent, a rate that outpaced the United States, which saw its foreign-born population increase by 34.3 percent over the same period (calculated from OECD Data, n.d.). Despite this growth in immigration, the scholarly literature on Japanese public opinion toward immigration is fairly thin (Kobayashi et al. 2015; Kojima 1996; Mazumi 2015). Some scholars such as Kobayashi et al. (2015) address public attitudes toward naturalization. Richey (2010) and Green (2017) are valuable studies that carefully explore Japanese public opinion toward immigration, but both rely on observational data. Our study is among the first to employ experimental methods to assess Japanese attitudes toward immigration more broadly. 3
Survey Design
We undertook an original three-wave survey in Japan in 2016 upon approval by the Institutional Review Board of the Graduate School of Arts and Sciences, University of Tokyo. The first wave was conducted in April to gauge respondents’ baseline attitudes toward immigrants, followed by a second wave in July in which we undertook conjoint experiments. We conducted a third wave in September 2016 that replicated the first wave and examined the stability of the typology. In total, 2,111 respondents completed all three waves of the survey. The sample was randomly drawn by a Japanese survey firm, Nikkei Research, from their opt-in online panel, stratified on key demographic variables, including age, gender, and residential locations. The sample is not a probability sample, but it is a reasonably good approximation of the general population. The English-language translation of the question wording for all of the items is provided in the online appendix (pp. A25–A26), and the original Japanese-language version of the survey is available upon request. Table B in the online appendix provides summary statistics of the respondents and also discusses our sample in more detail.
The first wave of the survey asked respondents whether they generally favored or opposed immigration into Japan. 4 After asking a series of demographic questions, to distinguish between cultural and economic reasons for respondents’ attitudes, we asked two separate multiple-choice questions about why they favored or opposed an increase in foreign workers. The available choices for economic reasons were the positive/negative impacts of immigrants on Japanese jobs and positive/negative impacts on fiscal conditions, and cultural reasons included the positive or negative impacts on Japanese social norms as well as on crime rates. We included the possible impact on crime because surveys show that Japanese view good manners and respect for others—the opposite of crime—as core attributes of “Japanese-ness” (e.g., Hirai 1999). 5
It is possible that crime would be better included in the economic dimension than on the cultural one, and it is also possible that crime may lie on an entirely different dimension than either. We therefore conducted two robustness checks for our two-dimensional typology. First, we excluded the crime items from the cultural measure. The results will be shown in the discussion of the conjoint analyses below. Second, we conducted a factor analysis to explore how many factors can be identified in the cultural dimension items. Figure F in the online appendix shows that the slope of the graph drops most sharply at the second component, suggesting the plausibility of our positive–negative dichotomy that includes the crime item in the cultural dimension.
After asking respondents to choose as many of the expected impacts of immigration with which they agreed, we then asked them to select which one they agreed with the most. We use the dichotomous measures to construct our four types of variables; for example, we classify those who indicate positive impacts on both economic and cultural measures as Integrators.
To examine the determinants of each type of respondent, we collected covariates that we highlighted in the previous theoretical section, such as age, income, education, unemployment status, and partisanship. To test the labor market competition hypothesis, we measure job security in three ways. First, we follow existing studies and use education as a measure of skill level and, by extension, of job security. Second, as noted earlier, we build on the insights of Dancygier (2010) and Dancygier and Donnelly (2013), and use labor scarcity at the sectoral level as a measure of job security. But because there was limited variation in labor scarcity across different Japanese industries at the time of the survey (Ministry of Health, Labour and Welfare 2016b), we could not exploit variation in labor availability. 6 We single out restaurant/hotel workers because on one hand, they face direct competition with foreign manual workers, but on the other, acute labor shortages may attenuate their fear. At the time of the survey, the restaurant/hotel industry, which employed just over 130,000 foreign workers, was the third largest employer of foreign workers in Japan (Ministry of Health, Labour and Welfare 2016a). 7 This industry also experienced the highest level of labor shortage among all industries in Japan at the time of the survey (Ministry of Health, Labor and Welfare 2016b). The restaurant/hotel industry thus offers an intriguing lens through which to examine whether labor market concerns prevail over the need for more help. In addition, we also measure job security using a binary variable between those who are corporate managers and those who are not. As discussed earlier, those in managerial positions may be more likely to favor immigration because their immigrant workers are less likely to pose an economic threat to them (although skilled workers could, in fact, challenge their authority).
We included questions to measure RWA. Both RWA and ethnocentrism have been linked to outgroup hostility, but the high correlation between the two variables made it difficult to measure them separately in the analyses below. We therefore present the results using RWA, but we also show the results of analyses using ethnocentrism in the online appendix. In addition, we included questions that would allow us to test the contact hypothesis. Specifically, we asked whether respondents had foreign friends or had had foreign friends in the past. Friendship with foreigners may be endogenous, given that people who are generally more open-minded are more likely to have foreign friends. We therefore asked respondents whether they had contact with foreigners at the workplace. We use this measure because Mutz and Mondak (2006) find that the workplace, more than any other venue in contemporary life, offers opportunities for citizens to interact with those with dissimilar perspectives and social backgrounds. Contact with foreigners at the workplace is thus less likely to be driven by self-selection than friendship. Of course, this measure does not entirely eliminate endogeneity issues, but it provides a robustness check of the contact hypothesis.
Larger percentages of immigrants into Japan live in urban areas (foreigners comprise 1.34% of Japan’s urban population, as opposed to 0.78% of remaining areas [calculated from Somusho Tokeikyoku (Statistics Bureau, Ministry of Internal Affairs and Communications 2010)]), prompting us to conduct an identical survey that oversampled rural areas that have percentages of immigrants at rates comparable with Japan’s urban areas to assess whether rural-ness has an effect independent of social contact with foreigners. Table C in the online appendix shows the summary statistics for this sample. Whereas in the main sample, the average population size for respondents’ municipalities of residence was, on average, just under 400,000 people, in the rural sample it was just over 75,000. 8 In the main sample, 0.49 percent of the population in respondents’ municipalities of residence were foreign residents, on average, compared with 0.56 percent of the rural sample. The results from this second sample will also be discussed in greater detail below.
The second wave of surveys included a conjoint experiment to examine which attributes of immigrants Japanese people prefer (Hainmueller, Hopkins, and Yamamoto 2014). Increasingly popular in the field of political science, the conjoint method is useful both for understanding multidimensional preferences and for reducing social desirability bias (Ballard-Rosa, Martin, and Scheve 2017; Hainmueller and Hopkins 2015; Horiuchi, Smith, and Yamamoto 2018; Kage, Rosenbluth, and Tanaka 2019). Conjoint experiments allow us to vary different attributes randomly, making it possible to assess the impact of each attribute on respondent choice. Because the conjoint design neither asks respondents for their views on sensitive issues nor to provide reasons for their choice, it has been used to study potentially sensitive issues such as immigration and gender (Hainmueller and Hopkins 2015; Horiuchi, Smith, and Yamamoto 2018; Kage, Rosenbluth, and Tanaka 2019; Teele, Kalla, and Rosenbluth 2018).
Because our variables of interest included the number of immigrants as well as the kinds of immigrants respondents might favor, we asked respondents to compare hypothetical immigration policies rather than hypothetical individual immigrants. 9 We showed respondents five pairs of hypothetical immigration policies with a randomized combination of six attributes: country of origin, industry to absorb immigrant workers, immigrants’ skill level, immigrants’ gender, length of stay, and number of immigrants (see Table D in the online appendix for a complete list of attributes and Table E for an example). Each respondent was asked to choose a preferred policy. To examine what kinds of immigrants the four types of respondents favor, we follow existing studies (Bansak, Hainmueller, and Hangartner 2016; Horiuchi, Smith, and Yamamoto 2018) in interacting policy attributes with each of the four respondent types. Because interaction analyses by subgroups may yield different results depending on the baseline level of each attribute, we report conditional marginal means (Leeper, Hobolt, and Tilley 2019).
The six attributes for the conjoint experiment were chosen not only because they have often been used in previous studies of immigration (e.g., Clayton, Ferwerda, and Horiuchi 2019; Hainmueller and Hopkins 2015) but also because they are the main policy dimensions that policymakers consider in formulating immigration policy. For instance, under the new rules introduced in 2014, nurses from Indonesia have been permitted to work in Japan for three years. If, during this time, they pass Japan’s national certification exam, they may work in Japan indefinitely, whereas if they fail three times, they are required to return to Indonesia (Ministry of Health, Labour and Welfare, n.d.). There is a cap on the number of nurses who can be admitted into Japan every year, and in 2017, for instance, the cap was three hundred (Japan International Corporation of Welfare Services, n.d.). During the debate over the revision of the Immigration Control and Refugee Recognition Act in 2018, which partially liberalized restrictions against foreign workers, the Japanese government stated that it expected to accept roughly 340,000 workers in the first five years across fourteen designated industries for up to five years; if the workers successfully passed the national certification exam in their respective fields, they could be permitted to stay longer (Jiji Press 2018). In reality, the overwhelming majority of citizens will never be in a situation to select one or another type of policy, but this forced-choice design sheds light on which type of immigration policy is preferred by the citizenry, while also reducing the potential concern of social desirability bias.
Findings
Descriptive Statistics of Types
As Table 2 shows, the largest percentage of respondents, 39 percent, fell into the category of Resisters: the dislike of immigrants on any grounds. These were followed by Exploiters and Integrators, with 32 and 24 percent of respondents, respectively. 10 Diversifiers, who like immigration on cultural grounds, comprised only 6 percent of all respondents. Still, it is striking that a sizable majority of Japanese favor immigration: almost two-thirds of respondents appreciate immigrants, whether for their economic contributions, cultural contributions, or both.
Distribution of Four Types in Wave 1 (N = 2,046).
Table 2 underscores the importance of cross-cutting preferences. Exploiters and Diversifiers together comprised almost 40 percent of respondents. These groups of citizens appreciate immigrants for either economic or cultural reasons. Earlier studies have overlooked these immigration-positive citizens, perhaps because their motivations differ. But they represent a sizable proportion of our sample. Figure A in the online appendix confirms that the distribution of the four types is relatively stable after six months, in the third wave of the survey.
As noted earlier, we conducted an identical survey on a separate sample that focuses on those rural municipalities with percentages of foreign residents commensurate with more urban areas. As shown in Figure B in the online appendix, the distribution of the four types is remarkably similar, suggesting that contact matters more than the urban–rural context.
Determinants of Types
What factors shape the likelihood that one becomes one or another “type” of citizen? Table 3 shows results of multinomial logit regressions with Resisters as the reference category, based on the first wave of the survey. Coefficients are marginal effects. 11 We show the results of the same analyses with odds ratios in Table H in the online appendix. We find that education, job positions, age, contact with foreign friends, and right-wing authoritarian predispositions affect the likelihood of becoming one type over another. Consistent with previous studies (Dancygier and Donnelly 2013; Hainmueller and Hiscox 2007; Hainmueller and Hopkins 2015; Mayda 2006; Scheve and Slaughter 2001b), we find that people with at least a college degree are almost 50 percent more likely to be Integrators who favor immigration on both economic and cultural grounds than Resisters, all else equal. 12
Determinants of Types: Multinomial Logit Estimates.
The sample includes respondents who completed all three waves of the survey. Coefficients are marginal effects. LDP = Liberal Democratic Party; AIC = Akaike information criterion.
p < .1. **p < .05. ***p < .01.
Economic incentives may be responsible for the greater likelihood that corporate managers, who are responsible for labor costs, to be Exploiters or, to a lesser extent, Integrators, than Resisters. Indeed, managers especially were more than twice as likely to be Exploiters than Resisters. Workers in the hotel/restaurant businesses might have been less enthusiastic than their bosses about immigrants for reasons of job and wage competition, but possibly because of the severe labor shortage in Japan in these industries that may cancel out the labor market competition concern, we instead find null findings. 13
Noneconomic factors also affect the likelihood of becoming one or another type in our typology. Consistent with contact theory (Allport 1954; Pettigrew and Tropp 2006), we find that respondents with foreign friends, now or in the past, were unlikely to be Resisters. 14 Furthermore, as the literature on RWA suggests, respondents with (self-identified) authoritarian values were less likely to favor immigration for any reason, economic or cultural. 15 Our results show that age affects prejudice in Japan. But in contrast to many existing studies that find younger individuals to be more accepting of immigrants than older individuals (Hainmueller and Hiscox 2006; Sides and Citrin 2007b), we find that older respondents were more likely to be Integrators or Exploiters than their younger counterparts. While further research is necessary to untangle this finding, the discrepancy may be explained by the greater job security that older individuals typically enjoy in Japan. In addition, older respondents may be more concerned about the acute labor shortage in elder care. LDP supporters are more culturally conservative than supporters of other parties or independents (Takenaka 2014), but we did not find LDP support to significantly increase or decrease the likelihood of becoming one type over another. 16
A number of robustness checks are in order. First, because friendship with foreigners may reflect the very preferences that instigate contact with foreigners, we re-tested the contact hypothesis by asking whether or not respondents engaged with foreigners in their workplace. The results are shown in Table O in the online appendix. The coefficient for workplace contact with foreigners remains statistically significant for Integrators and Exploiters, but the magnitude of the effect is smaller than in Table 3 and no longer significant for Diversifiers, suggesting either some endogeneity in the contact measure or that types of contact have different impacts on types of people. The coefficients on the other variables remain largely unchanged.
Tables I and L in the online appendix show results of analyses that include the percentage of foreign population in the respondent’s municipality of residence. The coefficients on the foreign population variable fail to reach statistical significance for any of the types. Although our contact measures at the municipality level are admittedly rather crude, this result is consistent with findings by Hjerm (2009) and Pichler (2010) that higher or lower percentages of immigrants in the respondent’s immediate vicinity do not systematically change attitudes toward immigrants. In Table I, the coefficient on the education variable became significant at the 10 percent level, but otherwise the rest of the coefficients remained largely similar.
Changes in the foreign population, rather than absolute levels per se, may be more important for public attitudes toward immigration (Hopkins 2010; Kaufmann 2014; Newman 2013). We therefore tested the impact of changes in the percentage of foreign population at the municipality level between 2005–2015 and 2010–2015. The results are shown in Tables J and K in the online appendix. As shown in the tables, neither the change in the percentage of foreign population between 2005–2015 and 2010–2015 makes it more or less likely that an individual becomes one more type relative to a Resister. In both tables, the coefficient on the college variable became statistically significant for Exploiters, but otherwise the results remained largely unchanged. 17
Finally, because the Japanese government’s Annual Survey of Employment Trends [Koyo Doko Chosa] (Ministry of Health, Labor and Welfare 2016b) shows that white-collar office workers (administrators) are labor-abundant, we tested whether these workers exhibit hostility to immigration for economic reasons. The results are shown in Table P in the online appendix. The table shows that clerical workers are not especially more or less likely to favor immigration for either economic or cultural reasons. Although this result contradicts what we would expect on the basis of the labor market competition hypothesis, because clerical workers in our sample work in different industries, future studies should test whether clerical workers in labor-abundant industries are more likely to oppose immigration for economic reasons than those in labor-scarce industries.
In summary, these findings reveal that different factors shape why individuals may become one or another “type.” Compared with Resisters, Exploiters are likely to be older, to have had foreign friends, and to serve in managerial positions. Relative to Resisters, Diversifiers are more likely to have had foreign friends in the past. Integrators are likely to be better educated, older, to have had foreign friends in the past, and less likely to be right-wing authoritarians than Resisters.
These findings also illuminate why attitudes toward immigrants may shift over time and for whom. While the four types may be relatively stable, they are driven by personal experiences and responsibilities, and they are therefore systematically subject to change. For example, ceteris paribus aging (if a life-cycle effect dominates a generational effect) could make individuals more likely to embrace immigrants on the economic dimension under certain conditions. The experience of interacting with foreign friends is likely to move a Resister toward becoming an Integrator, Exploiter, or Diversifier. Promotion to a managerial position may turn a Resister into an Exploiter. The extent to which individuals may change types over the course of their lives bears further empirical exploration.
Preferences over Different Immigrants
If our types reflect relatively stable values and calculations of Japanese citizens, it stands to reason that different types of respondents would favor immigrants with different characteristics. We find this is indeed the case. Figure 1 shows the marginal means in preferences for the four types. 18 Differences in conditional marginal means are visualized in Figure D in the online appendix, using Integrators as the baseline subgroup.
Conjoint analysis on respondents’ preferred policies, by respondents’ types.
First, we do not claim that the two axes of cultural and economic appreciation or opposition are all-encompassing. As Hainmueller and Hopkins (2015) found in the American context, citizens tend to shun immigrants from countries with which there are diplomatic tensions (see also Kalicki 2019). In line with that expectation, all four types in Japan frown on immigrants from East Asian countries, perhaps reflecting Japan’s diplomatic tensions with its neighbors over the wartime legacy. All four types are indifferent toward those of Eastern European origin, possibly because of unfamiliarity.
Beyond a general anti-Chinese and anti-Korean bias, the four types differ in important respects (see Table F in the online appendix for a summary of statistically significant variables). Resisters are particularly negative toward immigrants from East Asia, and, perhaps to compensate, show a strong preference for Western European immigrants. Diversifiers are less reluctant to see East Asian immigrants and are indifferent toward those of other geographic origin. Integrators and Exploiters favor immigrants from Southeast Asia. The preference for people from Southeast Asia and those who will work in the care industry are likely both a cause and a consequence of Japan’s 2014 agreement to admit care workers and nurses from Indonesia, the Philippines, and Vietnam. This last point also indicates that Exploiters and Integrators may not particularly value the economic human capital of immigrants (i.e., immigrants from developed countries; Boyd and Thomas 2002), instead favoring immigrants from developing countries, especially if they can move into labor-scarce industries such as care work. Exploiters are more leery of (other) East Asians than Integrators, although this difference just misses statistical significance at the 10 percent level.
As for the industries in which immigrants would be employed, we find that as expected, Diversifiers are indifferent. By contrast, as expected, Integrators and Exploiters, who both value immigrants’ contributions to the economy, welcome immigrants who would work in industries that are known to be labor-scarce, such as care work and agriculture. Even Resisters show some measure of sociotropic preference in favoring immigration in labor-scarce industries, but this inclination is more pronounced among Integrators (who welcome all but also see the point of economic need) and Exploiters (who are interested principally in addressing labor shortages). The difference among Integrators, Diversifiers, and Resisters for immigration into the care industry is statistically significant at the 10 percent level.
Resisters show a preference for skilled over manual labor, while the other three types are indifferent as to immigrants’ skill level (although the difference between Resisters and Integrators misses statistical significance). In contrast to the United States, where Hainmueller and Hopkins (2015) found high levels of public support for high-skilled immigrants, Japanese support for immigration into high-tech/IT industries and skilled labor in general is fairly weak. This reluctance to open doors to skilled workers may reflect the more acute shortage in unskilled than in skilled labor in Japan at the time of the survey (Ministry of Health, Labor and Welfare 2016b), or it may reflect a desire on the part of businesses to protect managerial authority.
All four types, including Diversifiers, markedly prefer female over male immigrants, as is typically the case among developed economies. Females may be thought more pliant, less violent, and less of a threat to standard male jobs. In Japan, females are also expected to become care workers, addressing acute labor shortages in that field.
As expected, Resisters want immigrants to stay for a shorter duration. Exploiters and Integrators also prefer immigrants to go home after a short stay, but not to the extent as Resisters, for reasons that are unclear. The difference between Exploiters and Integrators here misses statistical significance. Only Diversifiers are indifferent as to immigrants’ length of stay.
On the desired number of immigrants, Resisters favored the fewest, as expected, while Integrators favored the most, and Exploiters and Diversifiers fell in between. Figure D in the online appendix shows that this is one of the few areas in which the preferences of Integrators and Exploiters diverge. Although the two types agree about the industry, skill level, and duration of stay, Integrators will countenance a large (50% or 100%) increase in immigration, whereas Exploiters will not.
To test the robustness of our two-dimensional typology, including whether or not it is appropriate to include crime in the cultural dimension, we reran the conjoint analysis using an alternative measure that excludes crime. The results are shown in Figure E in the online appendix. The figure should be interpreted with caution, because it drops quite a few respondents (N = 9,770), but overall, the results remain largely unchanged. On country of origin, once the crime item is removed, Exploiters are less favorable toward those of Western European origin, but otherwise the results are largely similar to those shown in Figure 1. Diversifiers remain indifferent as to which industry should admit foreign workers, whereas Exploiters and Integrators favor foreign workers in the care industry. As with earlier results, Integrators, Exploiters, and Resisters are reluctant to accept foreign workers in the high-tech/IT industries. Removing the crime item weakens Resisters’ support for female foreign workers and opposition to male workers. Resisters prefer short-term workers and dislike long-term immigration. Integrators favor a larger increase in foreign workers compared with all other “types,” including Exploiters, and are the least likely to want a drop in the number of foreign workers. Overall, the results remain largely similar.
Discussion
Existing studies have missed an important feature of citizens’ attitudes toward immigration: many of the same people who disfavor immigrants on one dimension may favor them on another. Some people, to be sure, dislike the idea of increasing the number of immigrants for both cultural and economic reasons, but a greater number appears to appreciate immigrants on either or both dimensions. Attitudes toward immigrants, when broken into cultural and economic dimensions, present a more complex picture than has previously been understood. This paper moves the debate from asking whether the public concerns about immigration are rooted in economic or cultural considerations, to probing which segments of the population are concerned about which, and why.
We found that factors such as job position, education, and contact with foreigners predict attitudes toward immigration. These findings show both how attitudinal types emerge and how they might change. For example, in Japan, college enrollment rates have been rising steadily over the last few decades, which could produce more Integrators. At the same time, continued labor shortages across different industries in Japan could expand the number of Exploiters.
Our empirical analyses reveal systematic differences in the immigration policy preferences across the four types of respondents. Integrators welcome larger numbers of immigrants than Resisters and Exploiters, and Diversifiers lie in between. Diversifiers favor opening the door to immigrants irrespective of immigrants’ economic position and from any geographical origin (except for the rest of East Asia, reflecting Japan’s current touchy diplomatic environment in the region). Exploiters express largely similar preferences as Integrators but support a smaller number of immigrants, again, especially of East Asian origin.
The typology of citizens offered in this paper may be applied fruitfully to contexts outside of Japan. The European Social Survey asks about economic and cultural concerns in language slightly different from our own, but nevertheless shows a similar distribution across the four types in the Czech Republic and Hungary as in Japan: a relatively low percentage of Integrators and a commensurately large group of Resisters (see Figure G in the online appendix). In all three countries, Integrators comprised 40 percent or fewer of all respondents and Resisters made up more than 25 percent. Slovenia is close to Japan in having a small proportion of Integrators (roughly 40%), but it has a significantly smaller percentage of Resisters (just over 10%) than Japan. France is also similar to Japan in having a high percentage of Resisters (over 30%), but France has a much smaller proportion of Exploiters (less than 10%). The determinants of cross-national distributions across the four types bear future exploration.
Consider the policy implications. Support for greater immigration among Integrators and Exploiters is driven to a considerable degree by individuals’ labor market positions, although noneconomic factors matter for the other types. Somewhat ominously, one implication of this finding is that alleviating labor shortages could produce a groundswell of anti-immigrant sentiment. Within the Japanese context, the government has attempted to address the labor shortage by encouraging more women to stay in the workforce after childbirth. Greater female labor force participation may dampen enthusiasm for immigration among Integrators and Exploiters.
Our findings also suggest that opening the door to a large number of immigrants could provoke a backlash from Resisters. Even as Integrators and Exploiters might cheer the prioritizing of immigrants from developing countries in Asia, Resisters and Diversifiers could object.
Finally, our research suggests the possibility of policy coalitions across the different types of voters. Integrators and Exploiters, though perhaps belonging to different political and social circles, are likely to favor immigrants in labor-scarce industries. A coalition of Integrators and Diversifiers would welcome an even larger opening of doors to immigrants, regardless of geographic origin or economic need. These are important insights for policymakers piecing together plausible coalitions net of backlash.
A two-dimensional typology could form the basis of a more dynamic account of immigrant aversion across the globe than any current theory on offer. Given the xenophobic rhetoric that surrounds the politics of immigration today, the possibilities of tapping into positive attitudes toward immigration are ripe for more systematic study.
Supplemental Material
sj-pdf-1-prq-10.1177_1065912921993552 – Supplemental material for Varieties of Public Attitudes toward Immigration: Evidence from Survey Experiments in Japan
Supplemental material, sj-pdf-1-prq-10.1177_1065912921993552 for Varieties of Public Attitudes toward Immigration: Evidence from Survey Experiments in Japan by Rieko Kage, Frances M. Rosenbluth and Seiki Tanaka in Political Research Quarterly
Footnotes
Acknowledgements
We thank Daniel Aldrich, Kiichiro Arai, Jack Citrin, Hilary Hilbrow, Jacques Hymans, Roland Kappe, Yuko Kasuya, Hiroto Katsumata, Taeku Lee, Kenneth McElwain, Hirofumi Miwa, Hugh Patrick, Katharina Pfaff, Susan Pharr, Alex Quiroz Flores, Mark Ramseyer, Dan Smith, Michael Thies, Michio Umeda, and audiences at the 2016 Keio University Comparative Politics Workshop, the 2016 UCLA Global Japan Forum, the 2016 UTokyo-UC Berkeley Strategic Partnership Workshop Panel Discussion, the 2017 European Political Science Association Annual Conference, the 2017 and 2018 American Political Science Association Annual Conference, and the Harvard University Contemporary Japanese Politics Study Group (March 2018) for their comments and suggestions. Naosuke Mukoyama and especially Keita Ando provided excellent research assistance.
Authors’ Note
The human subject protocol of the research was approved by the Institutional Review Board, Graduate School of Arts and Sciences, University of Tokyo (Research Project 444, approved January 21, 2016), and the research was conducted following the approval. The authors are listed in alphabetical order.
Declaration of Conflicting Interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: We thank the Mitsubishi Foundation and the Murata Science Foundation for generous support of this project.
Notes
References
Supplementary Material
Please find the following supplemental material available below.
For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.
For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.
