Labor markets and immigration concerns: a brief overview

One of the main theories studying opposition to immigration is group conflict theory, which states that individuals who perceive themselves as competing with immigrants, for example in the labor market, are more likely to oppose immigration (Dehdari Reference Dehdari2022; Malhotra, Margalit and Mo Reference Malhotra, Margalit and Mo2013). However, recent research has challenged this assertion and argued that individuals’ values, rather than labor market competition, are the primary drivers of immigration concerns (Bansak, Hainmueller and Hangartner Reference Bansak, Hainmueller and Hangartner2016; Hainmueller and Hopkins Reference Hainmueller and Hopkins2014).

Partly in response to claims that labor markets are not related to anti-immigration sentiment, the literature has identified two nonexclusive arguments that could explain this relationship: fear of prospective competition with immigrants in the labor market and threats of status loss. The prospective competition framework holds that potential competition with immigrants in the labor market, rather than direct competition itself, fuels opposition to immigration (Pardos-Prado and Xena Reference Pardos-Prado and Xena2019). In particular, the authors argue that individuals in occupations with hard-to-transfer skills may fear competing with immigrants in the future, which could lead to an economic backlash against immigration (see also Kunovich (Reference Kunovich2013) and Polavieja (Reference Polavieja2016)).

Status loss arguments, on the other hand, emphasize how economic risks are related to status threats and demands for cultural homogeneity. In assessing the recent cultural backlash in Western countries, Inglehart states that ‘economic and physical insecurity are conducive to xenophobia, strong in-group solidarity, authoritarian politics, and rigid adherence to their group’s traditional cultural norms’ (Reference Inglehart2020: 8; see also Ballard-Rosa, Malik, Rickard et al., Reference Ballard-Rosa, Malik, Rickard and Scheve2021). Instead of analyzing labor market competition, this line of research argues that perceived subjective status is closely linked to anti-immigration attitudes (Gidron and Hall Reference Gidron and Hall2017), and that individuals perceive their societal status as connected to their occupations (Breyer Reference Breyer2025; Kurer Reference Kurer2020).

The argument

The argument developed here builds on and extends the literature examining the relationship between labor market risks and anti-immigration attitudes. Following sociological accounts, it is argued that possessing specific skills provides workers with income and job stability (Emmenegger Reference Emmenegger2009; Goldthorpe Reference Goldthorpe2000). However, these returns are contingent on occupational labor markets: when occupational unemployment rates (OUR) are high, firms can more easily replace workers without incurring training costs, reducing the protective effect of OSS against job loss (Jäger and Heining Reference Jäger and Heining2022; Serrano-Serrat Reference Serrano-Serrat2026). Additionally, workers may fear changing occupations because it entails income losses, as they cannot transfer OSS between occupations (Huckfeldt Reference Huckfeldt2022; Serrano-Serrat Reference Serrano-Serrat2026; Zangelidis Reference Zangelidis2008). A key assumption of the argument is that changing occupations involves substantial costs for both workers and firms, a fact recently documented in the literature (Huckfeldt Reference Huckfeldt2022; Jäger and Heining Reference Jäger and Heining2022).

Given the increased sense of insecurity, a reaction against immigration is to be expected (Anelli, Colantone and Stanig Reference Anelli, Colantone and Stanig2021; Kaihovaara and Im Reference Kaihovaara and Im2020; Pardos-Prado and Xena Reference Pardos-Prado and Xena2019). Drawing on previous discussion, it is argued that this can occur through increased perception of prospective labor market competition and/or through threats of status loss.Footnote ² First, as indicated by the prospect competition framework, occupational hazards make workers fear potential competition with immigrants (Kunovich Reference Kunovich2013; Pardos-Prado and Xena Reference Pardos-Prado and Xena2019; Polavieja Reference Polavieja2016). When the labor market context deteriorates, people feel that the probability of job loss increases, especially for those with high levels of OSS. If they become unemployed, they would fear having to compete with immigrants in the general labor market, where OSS are not valued. This, in turn, increases concerns about immigration.

Second, formerly protected groups facing an anxious economic situation may become more concerned about their status in the social hierarchy (Baccini and Weymouth Reference Baccini and Weymouth2021; Kurer Reference Kurer2020; Petersen, Osmundsen and Arceneaux Reference Petersen, Osmundsen and Arceneaux2023), prompting a shift toward cultural and group-identity politics (Ansell, Hjorth, Nyrup et al., Reference Ansell, Hjorth, Nyrup and Larsen2022; Kurer Reference Kurer2020). According to the proposed argument, social recognition derived from OSS is expected to decline when the labor market deteriorates (Hall and Lamont Reference Hall and Lamont2013; Sørensen Reference Sørensen2000). As the prestige of an occupation falls, people compare their current unsatisfactory situation to an idealized past, reinforcing nostalgia and a willingness to regain status at the expense of immigrants through identity-related issues (Anelli, Colantone and Stanig Reference Anelli, Colantone and Stanig2021; Baccini and Weymouth Reference Baccini and Weymouth2021; Gest, Reny and Mayer Reference Gest, Reny and Mayer2018; Gidron and Hall Reference Gidron and Hall2017). These complementary mechanisms generate the following testable hypothesis:

The argument presented is analytically distinct from accounts that emphasize task specificity. In that line of research, greater specificity means fewer jobs performing comparable tasks (Lazear Reference Lazear2009). These accounts claim that as the number of available jobs requiring similar tasks increases, the costs associated with changing occupations decrease (Gathmann and Schonberg Reference Gathmann and Schönberg2010). This, in turn, may reduce the potential negative political reactions associated with labor market deterioration (Bisbee and Rosendorff Reference Bisbee and Rosendorff2025). However, occupational skills depend not only on tasks but also on specialized knowledge, and only by considering both can one account for the costs of mobility.Footnote ³ Indeed, Cortes and Gallipoli (Reference Cortes and Gallipoli2018) show that task similarity explains only a small share of the costs associated with occupational switching (about 6%). In contrast to accounts that analyze task specificity, OSS are specific because they are primarily usable within a given occupation – that is, their specificity does not depend on the broader labor market context (Lazear Reference Lazear2009; Streeck Reference Streeck, Busemeyer and Trampusch2011) – and greater OSS indicate higher skill levels.

Furthermore, the proposed relationship operates independently of other skill dimensions, such as general skills, firm-specific skills, or possession of a vocational degree. First, general skills, by definition, are transferable across occupations and are therefore unlikely to moderate the effects of occupational labor markets. Thus, the proposed theory concerns not skill levels in general, but in particular the amount of OSS. Second, the argument also differs from perspectives that emphasize the role of firm-specific skills. Firm-specific skills, unlike OSS, may provide some protection to workers during economic downturns due to the reciprocal relationship and mutual investment between workers and firms (Pahontu Reference Pahontu2022; Tåhlin Reference Tåhlin2007). Finally, it is important to clarify that the focus is not on the type of training – whether general or vocational – but rather on the level of OSS required to effectively perform a job.

Data

A summary of the operationalization of the main variables used in the paper is provided in Table 1.Footnote ⁴ Importantly, the GSOEP includes occupational identifiers, allowing occupational variables – obtained from external sources – to be linked to individuals.

Table 1.

Summary operationalization of variables

Table 1. Long description

A table with three columns labeled Variables, Operationalization, and Source, and three rows detailing the operationalization of variables. Row 1: Occupation unemployment rates, main explanatory variable. Operationalization: For employed workers, information of the current occupation. For the unemployed, information on previous occupation. For each year, calculate % of unemployed at the ISCO 2-digit. Source: German LFS. Row 2: Occupation-specific skills, moderator. Operationalization: Supposing that someone with this level of qualification was applying today, how much related work experience would they need to get this job? Answers coded in months and aggregated at ISCO 2-digits for European Union countries. Only 30+ years old. Source: PIAAC. Row 3: Immigration concerns, dependent variable. Operationalization: How concerned are you about immigration to Germany? Scale: Very, (1) Somewhat, (1) and Not at all (0). Source: GSOEP.

Occupational context is operationalized as OUR, a well-established measure in the literature (Pardos-Prado and Xena Reference Pardos-Prado and Xena2019; Rehm Reference Rehm2009; Rovny and Rovny Reference Rovny and Rovny2017). Data are drawn from the German Labor Force Survey (LFS) provided by Eurostat, which includes information on the previous occupation of unemployed workers. Unemployment rates are calculated for each occupation-year, aggregated at the International Standard Classification of Occupations (ISCO) 2-digit level.Footnote ⁵ Following Rehm (Reference Rehm2009), the variable is smoothed and winsorized at the 1 percent level.Footnote ⁶ Online Appendix A.1 details this transformation, presents the descriptive statistics – including the variation exploited in empirical analyses – and examines potential measurement error.

Occupation-specific skills have been theoretically defined as the amount of skills required to work in a particular occupation that cannot be transferred to other occupations. To measure this, scholars have used survey-based measures in which individuals self-report the difficulty or cost of learning their job (Polavieja Reference Polavieja2016; Serrano-Serrat Reference Serrano-Serrat2026). This research note uses the Survey of Adult Skills questionnaire from the Programme for the International Assessment of Adult Competencies (PIAAC), which includes a question about the amount of related experience required to work in a given job (Serrano-Serrat Reference Serrano-Serrat2026).Footnote ⁷ By asking about requirements in terms of related experience, this variable taps at the two analytical dimensions considered: the experience loss workers face if they change occupations and the costs firms incur when hiring someone without the required skills. A crucial element of the question is that it asks about required experience, assuming the individual already holds the formal qualifications to perform the job, which limits concerns that the variable simply captures education-based skill.

Individuals’ responses are coded in months and aggregated at the ISCO 2-digit level for European Union countries. Managers and Professionals are the occupations with the highest OSS, while Elementary Occupations have the lowest levels (see Appendix A.2). One concern with this variable is that it may simply reflect other aspects of occupations or skills. To address this, it will be shown that the observed relationship differs from task specificity measures and from other skill dimensions, such as years of education and firm tenure. Moreover, Appendix A.2 demonstrates that the variable captures the vertical dimension of social class (Goldthorpe Reference Goldthorpe2000), and Appendix D shows that the results hold when operationalizing OSS as the occupational mean wage.

Immigration concerns are drawn from the GSOEP (Ersanilli and Prag Reference Ersanilli and Präg2023; Pardos-Prado and Xena Reference Pardos-Prado and Xena2019). From 1999 to 2019, respondents were asked whether they are (somewhat or very) concerned about immigration. Answers indicating that respondents are not concerned at all about immigration are coded as 0, while those showing some degree of concern are coded as 1.Footnote ⁸ This variable has been shown to primarily reflect anti-immigrant sentiment rather than just saliency. Appendix A.3 shows that forty percent of individuals changed their preferences one to three times, and about five percent changed them more than three times during the period studied.

Method

It has been hypothesized that OUR are positively associated with anti-immigration attitudes, especially among individuals with high levels of OSS. This is tested using the following linear model:

(1) ${\rm Imm}{_{i,t}} = \alpha {\rm OUR}_{{o\left( {i,t} \right),t}} + \beta {\rm OUR}_{{o\left( {i,t} \right),t}} \times {\rm OSS}_{{o\left( {i,t} \right)}} + \delta {Z_{i,t}} + {\theta _{i,o\left( {i,t} \right)}} + {e_{i,t}}$

where Imm_i,t is immigration concerns of individual i in year t. The subindex o(i,t is a function that yields the occupation hold by individual i at time t. Thus, OUR _o(i,t),t is the occupational unemployment rate faced by individual i in period t, who is employed in occupation o(t). This is similar for $OS{S_{o\left( {i,t} \right)}}$ ; however, for a given occupation, OSS do not vary over time.Footnote ⁹ $\beta $ is the parameter of interest, which is expected to be positive, indicating that increases in OUR leads to greater immigration concerns the higher the amount of OSS. ${Z_{i,t}}$ are controls, interacted with OSS (Blackwell and Olson Reference Blackwell and Olson2022), and ${\theta _{i,o\left( {i,t} \right)}}$ are individual-occupation FEs. Finally, ${e_{i,t}}$ are the individual errors. Since the variation of the interaction occurs at the occupational level, standard errors are clustered at the ISCO 2-digits (Abadie, Athey, Imbens et al., Reference Abadie, Athey, Imbens and Wooldridge2023).

In this regard, Equation (1) examines different sources of variation for OUR and OSS. For OUR, by including individual-occupation FEs, the analysis exploits aggregate changes in occupational labor market conditions over time. Each time an individual changes occupations, it is treated as a separate employment spell, so the analysis does not leverage variation from individuals endogenously changing occupations.Footnote ¹⁰ In contrast, OSS variation mainly reflects cross-sectional differences between workers rather than within-worker changes over time, so the estimates partly reflect selection into occupations (Keele and Stevenson Reference Keele and Stevenson2021). Importantly, the main sources of selection likely work against finding a positive interaction: if more able individuals – who are generally more sheltered from occupational unemployment – sort into high-OSS occupations, the estimates would be downwardly biased. Appendix D.2 explores narrower variation in OSS, including within ISCO 1-digit variation, finding similar results.

Finally, it should be noted that although the dependent variable is binary, the model is regressed via OLS. This is done because nonlinear models (eg logit) cannot estimate individual-occupation FEs due to the incidental parameters problem (Wooldridge Reference Wooldridge2010). Appendix D presents comparable results from (ordered) logit specifications using a Mundlak specification (Mundlak Reference Mundlak1978).