Hiring discrimination on the basis of skin colour? A correspondence test in Switzerland

ABSTRACT

With the rise of the Black Lives Matter movement in 2020, the situation of Black people in many Western countries has come under closer scrutiny and ethnic discrimination has been brought to the forefront. Little is known about hiring discrimination against Blacks in many European countries. In a correspondence test in the Swiss labour market, we sent fictitious paired applications by candidates of Swiss (ostensibly White) and Cameroonian descent (ostensibly Black) in response to 354 adverts for sales assistants and electricians. We report significant discrimination against Black job seekers, who must send around 30 per cent more applications than White candidates in order to be invited to a job interview. The level of discrimination is substantively equivalent to results for applicants with a Kosovo-Albanian name that were included in previous correspondence tests in Switzerland. This suggests that in the Swiss case there is on average no additional penalty for skin colour. Explorations, however, reveal significant differences in discrimination rates between urban and rural settings, opening new avenues for understanding why ethnic and racial discrimination vary across geographical contexts.

KEYWORDS:

• Race
• Ethnicity
• Discrimination
• Labour market
• Correspondence test

1. Introduction

The Black Lives Matter protests in 2020 have rekindled interest in discrimination against Blacks and other minorities in the United States and beyond. In Western Europe, we observe public debates asking whether discrimination exists on this side of the Atlantic, too, and to what extent Blacks are affected. Recent meta-analyses of correspondence tests spanning both sides of the Atlantic (Quillian et al. 2017; Quillian et al. 2019; Zschirnt and Ruedin 2016; Thijssen 2020) and a recent comparative experiment in five European countries provide unequivocal documentation that racial and ethnic discrimination in hiring remains a concern in OECD countries (Lancee 2019). Most of the European studies focus on immigrants as ethnic minorities, and a growing number of studies also includes Black applicants with an immigrant background (e.g. Weichselbaumer 2017, for an overview see Table A1 in the Appendix).

In this study, we focus on a particular case: the discrimination against Blacks in the Swiss labour market. Earlier field experiments in this context have demonstrated discrimination against immigrant offspring (Fibbi, Kaya, and Piguet 2003; Zschirnt and Fibbi 2019; Zschirnt 2020). However, little is known about the situation of Blacks in Switzerland, with the notable exception of qualitative studies which report systematic disadvantage across all domains of everyday life (Efionayi-Mäder et al. 2017). Here we extend this work by explicitly examining labour market discrimination on the basis of skin colour. Blacks constitute a small, yet growing, minority group in Switzerland. Despite their comparatively low numbers, Black immigrants are now also being featured in the racialized anti-immigration discourse mobilised particularly by the right-wing Swiss People’s Party (SVP/UDC), mirrored in the archetypal poster of a black sheep being kicked out of the country by white sheep (Michel 2015; Boulila 2019). Public opinion polls in 2018 report that 11 per cent of the population is hostile to Blacks and 7 per cent feel disturbed in their daily lives by the presence of a person with a different skin colour (Aeberli 2019).

The correspondence tests analysing access to the labour market conducted in other European countries show Blacks at the bottom of ethnic hierarchies, e.g. in Austria (Weichselbaumer 2017), in Belgium (Baert et al. 2017), in Germany (Koopmans, Veit, and Yemane 2019), in Finland (Ahmad 2020), or in France (Cediey and Foroni 2008). To assess the role of skin colour in Switzerland empirically, we extend the study of hiring discrimination in accessing jobs requiring a vocational degree to visible minorities by focusing on children of Cameroonian descent. We find high levels of hiring discrimination against Black candidates. When comparing the results to the discrimination against White immigrant minorities, however, we find that levels of discrimination against Blacks are comparable to those of Kosovo-Albanian descendants. We cannot identify an additional penalty for skin colour at the aggregate level, in spite of significant differences between urban and rural settings.

2. Theories of discrimination in hiring on the basis of skin colour

While many studies on hiring discrimination address the theoretical question whether discrimination is taste-based (Becker 1957) or statistical (Phelps 1972; Arrow 1973), we focus on alternative explanations provided by sociologists and social psychologists which look at social closure and social distance. Both approaches are consistent with the micro-socially focused social identity theory (Tajfel and Turner 1979), according to which individuals value members of the in-group positively. Therefore, they tend to associate with people who are similar to themselves – a phenomenon known as homophily – (McPherson, Smith-Lovin, and Cook 2001); yet the two propositions predict different forms of exclusionary outcomes, by alternatively stressing majority privilege or variable intensity of prejudice/discrimination against minorities.

Concentrating on macro-social dynamics, social closure stresses that, aiming at retaining and maximising one’s own advantage, majority members establish social barriers – sometimes even legal barriers – that favour the in-group and exclude out-groups (Weber 1995). The results of some field experiments on hiring discrimination are consistent with social closure (Andriessen et al. 2012; Bovenkerk et al. 1995; Drydakis 2017; Oreopoulos 2011; Edo and Jacquemet 2014). These studies observe similar discrimination rates among the various minority groups tested. This pattern was also observed in studies that included African-origin minorities in Ireland (McGinnity and Lunn 2011) and the United States (Jacquemet and Yannelis 2012). These authors speak of ‘indiscriminate discrimination’, emphasising that the driver of discrimination is an in-group bias disadvantaging any member of minority groups.

The hypothesis based on social closure states that all minority groups are affected by hiring discrimination to a largely comparable extent and that no consistent pattern of ethnic hierarchy can be found across various contexts.

Social distance, instead, focuses on the idea that the out-group’s rejection varies according to which in- vs. out-group divide is made salient by the majority, opening the way to ethnic hierarchies among groups, particularly visible in hiring decisions. For the purposes of this study, the most relevant ones are race and culture.

Some research points to the overarching dominance of racial hierarchies in Western societies, based on skin colour engrained in deeply rooted racist ideologies and institutional racism (Feagin and Elias 2013; Bonilla-Silva 2012). As for hiring discrimination, a meta-analysis of correspondence test studies conducted on Blacks and Hispanics in the United States shows the prevalence of a racial divide largely unchanged over the last decades (Quillian et al. 2017). Black minorities are exposed to the highest levels of discrimination, especially where the penalty for being an immigrant or a direct descendant of immigrants and the penalty for skin colour accumulate. Many hiring discrimination studies consistently observe the following status order: native Whites at the top, White immigrants in an intermediate position, and Black groups at the bottom (Restifo, Roscigno, and Phillips 2019). Similarly, in their meta-analysis on UK studies, Heath and Di Stasio (2019) reveal a largely racial pattern of discrimination, with black Caribbeans facing the highest rates of discrimination. Racial hierarchies have also been documented in other countries: with indigenous people at the bottom of the hierarchy in Australia (Booth, Leigh, and Varganova 2012), or Blacks in Austria (Weichselbaumer 2017). Using a factorial survey experiment, Auer et al. (2019) show that ethnic hierarchies exist in the Swiss labour market and that they become more visible when the occupational status of a position increases.

The hypothesis centred on racially based social distance states that Blacks are targeted by hiring discrimination to a greater degree than White immigrant-based minorities.

Some research links variability of majority rejection to the cultural characteristics of minority groups. Theoretical and empirical research identifies degrees of distance based on value differences observed in various origin countries (Hraba, Hagendoorn, and Hagendoorn 1989; Hagendoorn 1995), then transposed to the corresponding immigrant group living in immigration countries. The cultural group ranking identified on this basis is: White European groups are placed at the top, followed by colonial groups, while Muslim groups are placed at the bottom. Applied to hiring discrimination, the theory predicts that the greater the difference in cultural value patterns between majority and minority groups, the higher the rate of hiring discrimination against minority applicants. Building and interpreting their findings on the basis of cultural distance theory, those studies observe a clear hierarchy in out-group rejection with Western European origin groups less affected by discrimination than non-European groups, where Muslims and Black groups either appear equally exposed because they face similar risks (Koopmans, Veit, and Yemane 2019; Heath and Di Stasio 2019) or because they are compounded in encompassing categories (e.g. ‘Middle Eastern/North African (MENA)) (Veit and Thijssen 2019). Previous research indicated that such cultural group rankings also occur in Switzerland. Auer and colleagues have demonstrated that cultural distance also plays a role in the Swiss labour market: immigrants from EU/EFTA countries have shorter periods of unemployment than immigrants from so-called third countries (non-EU/EFTA) (Auer and Fossati 2019) and signalling attachment to especially culturally different countries of origins leads to more labour market discrimination (Fossati, Liechti, and Auer 2020).

The hypothesis centred on culturally based social distance states that Blacks are targeted by hiring discrimination, although it is unclear whether to a greater extent than White immigrant-based minorities.

The above-mentioned theoretical approaches lead to the expectation of hiring discrimination against outgroups, especially immigrant-origin ones. It is unclear however, which cleavage structures the ranking: the national one discriminating all outgroups, or the racial one with the double penalty of skin colour and immigrant background, or indeed the cultural value one where skin colour is not singularised.

We further explore the variability of hiring discrimination by comparing degrees of out-group marginalisation in urban and rural settings, the former being more ethnically diverse and, hence, favouring intergroup contact. The long-standing contact theory suggests that intergroup contact between majority and minority is likely to reduce prejudice to out-groups altogether (Pettigrew 2016), especially when it occurs under favourable conditions (Allport 1954). Such intergroup contact is more common in cities than in the countryside.

The hypothesis based on contact theory stipulates that hiring discrimination is less pronounced in urban contexts than in rural ones. This variability in intensity however does not affect the structure of ethnic hierarchies.

Here we use a correspondence test in Switzerland, to study discrimination against Black applicants. Since we use the exact same research design as Zschirnt (2020) and Zschirnt and Fibbi (2019), we can draw conclusions on the relative merit of the different hypotheses. Existing correspondence tests in Switzerland show discrimination against immigrants and their descendants coming from countries outside of the European Union (Fibbi, Lerch, and Wanner 2006; Auer et al. 2019; Zschirnt and Fibbi 2019; Zschirnt 2020), but barely of descendants of immigrants from (White) neighbouring countries. A recent study based on the search behaviour of recruiters on the online platform of the Swiss public employment service (Hangartner, Kopp, and Siegenthaler 2021) confirms the disadvantage of non-European job candidates, and especially of individuals from sub-Saharan Africa and Asia. In contrast to existing country comparisons (Larsen and Di Stasio 2019; Koopmans, Veit, and Yemane 2019; Thijssen et al. 2019; Quillian et al. 2018; Quillian et al. 2019), we use a within country comparison where we can ascertain that the legal framework and public discourse on migration is the same.

3. Research design and fieldwork

Switzerland hosts a large immigrant population, with around 38 per cent of the population in 2018 being an immigrant or born to immigrant parents. Around two thirds of the immigrants come from countries of the European Union (EU), especially Italy, Germany, and Portugal. Among non-EU origin immigrants, the largest groups come from Kosovo (5.2%) and Turkey (3.2%). The modest share of Blacks in the resident population (1.4%)¹ is probably a consequence of the absence of formal colonial ties to the sub-Saharan region but is also due to the country’s traditional migrant labour policy refraining from recruiting workers from ‘distant’ countries.² In 2018, some 85,000 individuals making up 4 per cent of foreign citizens residing in Switzerland were nationals of a sub-Saharan African country.³ In the 1960s and 1970s, African migration to Switzerland consisted mainly of students, elites, and international civil servants, residing temporarily in Switzerland. The 1980s brought political refugees and economic migrants from sub-Saharan Africa (Zodogome 2017; Efionayi-Mäder, Pecoraro, and Steiner 2011). In the last three decades, the share of immigrants from sub-Saharan countries in Switzerland grew above average, mostly due to asylum-related flows from East Africa.

Despite its high share of immigrants, institutionally, Switzerland never adopted comprehensive anti-discrimination legislation, in contrast to most other Western countries (Huddleston 2015). The country is characterised by a very pronounced federal structure, which also has an impact on migrant integration: Local conditions for integration vary between cantons in spite of the common legal federal framework and quite comparable labour-market situations (Probst et al. 2019). The strong Swiss economy means low unemployment across the country, coupled with a high number of vacancies: as such a favourable context for minority access to employment. Despite this, however, the unemployment rate of non-Swiss citizens is more than double that of Swiss citizens – in 2018, the unemployment rates were 8.2 per cent and 3.4 per cent respectively (Federal Statistical Office (FSO) 2019).⁴ Even though the unemployment rate is low overall, research reports substantial levels of self-declared discrimination for Blacks and other minorities in the labour market (Auer and Ruedin 2019; Efionayi-Mäder et al. 2017).

To capture discrimination against Blacks in hiring, we use a correspondence test in which two qualitatively equivalent fictious résumés are sent to a publicly advertised job. The résumés are sent within a few hours, and discrimination is observed when the White majority candidate is invited for an interview, and the Black minority candidate is not. Since it is customary to include photographs in job applications in Switzerland, it was straightforward to signal skin colour with the combination of the photograph and name.⁵ Almost all Blacks in Switzerland are immigrants or born to immigrant parents, so we specified that they are (naturalised) Swiss citizens to isolate the effect of skin colour and to avoid potential questions regarding work permits. What is more, since we ensured that the research design and material used are the same as in recent studies by Zschirnt (2020) and Zschirnt and Fibbi (2019), we can directly compare our results to these other correspondence tests to draw conclusions on discrimination on the basis of skin colour.

In the present study, the minority group is composed of Swiss citizens of Cameroonian descent that hold dual nationality. With this choice, we ensure that the applicants are not perceived as asylum seekers, which would have confounded our measurement. Objectively, migrants from Cameroon to Switzerland are highly educated, with a third having completed tertiary education (Kamdem 2018). Around two thirds are Christian, and around one third are Muslim: There is no clear religious connotation for Cameroonians in Switzerland. The share of Cameroonians has increased noticeably in the last four decades, and Cameroonians represent the largest number of naturalised youths of sub-Saharan descent in the last 10 years. In this sense, applications from naturalised young adults of Cameroonian descent are plausible to employers.

To signal ethnicity, we followed the design by Zschirnt (2020) and Zschirnt and Fibbi (2019), who used two names for the Swiss candidates: Pascal and Fabienne Kälin.⁶ These typical Swiss names evoke an image of White skin colour. In the absence of statistics on Cameroonian names in Switzerland, we used typical Cameroonian family names according to Diop (1954, 484–489), and searched for them in the Swiss phone directory to ensure that the family names of Cameroonians actually occur in Switzerland. The search in the phone directory allowed us to identify first names associated with the family names. We had a local Cameroonian community leader verify that our selection corresponds to people in the age range targeted by our study: Blaise Ndongo (male) and Gladys Dgomo (female). In addition, in line with Zschirnt (2020) and Zschirnt and Fibbi (2019), the Black applicants indicated dual citizenship (Swiss and Cameroonian) and speaking two native languages: German or French (depending on the linguistic region) as well as Duala. As is common in Switzerland, the applicants included a photograph where skin colour is evident. For reasons of feasibility we opted for multiple markers of race (Stockstill and Carson 2021).

As in Zschirnt (2020) and Zschirnt and Fibbi (2019), we tested two occupations that require a vocational degree at the upper-secondary level: sales assistant and electrician. Applicants show having completed their vocational education and training (VET) as well as having 3–4 years of relevant work experience. Vacancies were retrieved from popular Swiss search websites. The fieldwork was carried out in the German- and French-speaking regions from January to December 2018. Overall, 708 applications were sent in pairs to 354 job advertisements for positions of sales assistants (N=180) and electricians (N=174). As soon as a candidate was invited for a job interview, the invitation was politely declined. Both male and female candidates applied for positions as sales assistant whereas only male applications were sent to open positions for the predominantly masculine profession of electrician. For each vacancy all the individual elements of the application (i.e. the cover letter, the CV, reference letter, apprenticeship diploma and layout) were randomly assigned to the two fictitious candidates, except for the photographs that had to match the expected phenotypes. By their nature, correspondence tests observe a specific moment in the hiring process: the decision to invite an applicant for a job interview. They do not provide information about actual hiring decision or conditions that were offered, thus providing a minimum rate of discrimination.

The outcome in this study is the response made by the employers. We created a dichotomous variable that distinguishes between positive and negative answers. Positive answers included invitations to an interview, questions about the application, and a request for a call-back. We considered answers as negative when we received an explicit rejection, no more than a confirmation of receipt, or no response at all. Analytically, we present descriptive statistics from the field experiment, and regression analyses run using Hamiltonian Monte Carlo (HMC) sampling in Stan (Stan Development Team 2020) with uninformative but regularising priors: student t(μ=3, σ=0, df=2.5), using the brms package (Bürkner 2017). In the logistic regression analyses, we predict whether an applicant was invited to a job interview with groups as predictors. To do so, we combine the data from the present study with that by Zschirnt and Fibbi (2019). Ethnic groups are the predictor variables, and we include interaction terms to explore discrimination against Blacks in context. Applications are clustered within individual jobs.

For considerations of context, urban versus rural locations were coded as part of the field experiment. The distinction stems from the 2012 typology of municipalities of the Federal Statistical Office used to delimit the Swiss territory according to morphological (density) and functional (commuting flows) criteria.⁷ 59 percent of the jobs were in areas classified as urban. For the vote share, we focus on the share for the Swiss People’s Party (SVP/UDC) in national elections, as provided by the Swiss Federal Statistical Office. Vote shares range between 5.3 and 83.8 percent, with a median at 22.7 percent.

4. Results: hiring discrimination against Blacks and ethnic hierarchies

We find substantive evidence of discrimination against Black candidates in hiring: a relative call back rate of 1.30 (Table 1). This means that to be invited to a job interview, they need to send 30 per cent more applications than their White competitors. As is common in correspondence tests, equal treatment dominates: Equal treatment was observed in 82 per cent of the 354 cases (N=186 neither invited, N=103 both invited). The net discrimination rate is the difference between the cases where only the White candidate was invited and the ones where only the Black applicant was invited. The 20.8 per cent means that Black applicants in the study stood about a one-in-five chance of reaching the interview stage in the hiring process.⁸

Table 1. Correspondence testing results for Black and White candidates in occupations requiring a vocational degree, Switzerland 2018.

	Number of jobs [1]	None invited [2]	Both invited [3]	Only majority invited [4]	Only minority invited [5]	Net discrimination rate % [6]	% call back majority [7]	% call back minority [8]	Relative call back rate [9]	χ² count [10]
Aggregate results	354	186	103	50	15	20.8%	43.2%	33.3%	1.30	***
Occupations
Sales	180	116	36	23	5	28.1%	32.7%	22.8%	1.44	***
Electricians	174	70	67	27	10	16.3%	54.0%	44.2%	1.22	***

Notes: Column [1] shows the total number of vacancies used. Columns 2–5 express the treatment of the pair of applications sent. Column [2] specifies in how many cases both applications were rejected or did not receive an answer. Column [3] shows the number of cases where both candidates received a positive reply. Column [4] lists the cases in which only the Swiss-origin candidate received a positive reply. Column [5] are the cases in which only the Swiss minority candidate received a positive reply. Column [6] shows the net discrimination rate: ([4] - [5]) / ([3] + [4] + [5]). Columns [7] and [8] report the success rates of the majority ([3] + [4]) / [1] and minority candidates ([3] + [5]) / [1], Column [9] presents the relative call back rate, i.e. the ratio between the success rates of the majority [7] and minority candidates [8]. Column [10] shows the statistical significance of the difference between [4] and [5] using a chi-square test (significant at the ***1% level).

By combining the data from the present study with data from Zschirnt and Fibbi (2019), we can compare the results for Blacks with other ethnic minorities defined entirely by their immigration background. This allows us to investigate ethnic hierarchies (Table 2). The relative call back rate for Blacks is of similar magnitude to that of descendants of Kosovo-Albanian immigrants, but noticeably larger than that of descendants of immigrants from Turkey and neighbouring countries (France, Germany).⁹

Table 2. Correspondence testing results for the Swiss labour market in occupations requiring a vocational degree by minority group.

Ethnic groups	Number of jobs [1]	Net discrimination rate % [2]	Relative call back rate [3]	χ² count [4]
Black: Cameroonian	354	20.8%	1.30	***
Kosovo-Albanian^†	182	28.2%	1.44	***
French + German^†	179	14.3%	1.19	**
Turkish^†	180	10.9%	1.15
Total White Minorities	541	18.4%	1.3	***

Notes: Chi-square test, significant at the ** 5% and ***1% level, ^† Source: Zschirnt and Fibbi 2019

Using multivariate regression analysis, we can better understand the discrimination against Blacks relative to other minority groups. In Figure 1, we consider the gender of the applicant, as well as the occupation – with sales jobs requiring more intensive customer contact. At the top of the figure, the base-line results resemble those in Table 2. We also note that call backs are lower for sales jobs compared to the reference of electricians. The interaction between minority group and occupation suggests that Blacks do not experience an additional penalty in sales jobs, contrary to White applicants from neighbouring countries.

Figure 1. Multivariate analysis of hiring decision for sale assistants: interaction effect of minority group and occupation, Switzerland 2018 Logistic regression, coefficient as circle and 95% credibility interval as bars. R-hat = 1.00, based on N=1,736 applications.

5. Contextual variation in discrimination

By combining the results from the correspondence test with data on the context, we can further explore the hiring discrimination against Blacks. In Table 3 we present bivariate evidence suggesting a clear urban-rural divide in the kind of ethnic discrimination. While the positive response rates for White majority applications do not vary according to contexts, aggregate data show that minority candidates altogether face significantly higher barriers in rural areas than in urban ones, both for Black and other ethnic minority applicants.

Table 3. Correspondence testing results by context and minority group, Switzerland 2018.

	Number of jobs [1]	Net discrimination rate % [2]	Relative call back rate [3]	χ² count [4]
Urban contexts: Aggregate results		14.7%	1.21	***
Black: Cameroonian	196	12.2%	1.16	**
Kosovo-Albanian	112	23.7%	1.38	**
German + French	98	26.5%	1.45	**
Turkish	110	−3.7%	0.95
Rural contexts: Aggregate results		25.2%	1.37	***
Black: Cameroonian	152	34.3%	1.56	***
Kosovo-Albanian	70	35.3%	1.57	***
German and French	81	0.0%	1.00
Turkish	70	25.0%	1.37	**

Notes: Column [1] shows the total number of vacancies used. Column [2] shows the net discrimination rate, Column [3] the relative call back rate, while Column [4] presents the statistical significance for majority and minority applicants, using a chi-square test (significant at ** 5% and ***1% level). The number of jobs in urban and rural contexts for Black Cameroonians do not add up to the total number of 354 presented in Table 2, since we lack information on the job location for 6 vacancies.

If we look at the findings presented in Table 3 in terms of ethnic hierarchies, it appears that these hierarchies differ according to context. On the one hand, extra-EU origin groups are least affected by hiring discrimination in urban settings and most concerned in rural ones. On the other hand, EU-origin groups are subject to high discrimination risks in urban contexts, while they appear unaffected by it in rural ones.

The multivariate results in Figure 2 show higher levels of discrimination in urban areas; the interaction term between urban contexts and ethnic groups highlights that Blacks fare relatively better in urban areas. Put differently, having taken into consideration the general discrimination against Blacks and the higher levels of discrimination in urban areas, we observe a relative advantage for Blacks in urban areas. A similar but weaker effect appears to exist for other ethnic minorities, while White applicants from neighbouring countries face an additional penalty in urban areas.

Figure 2. Multivariate analysis of hiring decision: interaction effects of ethnicity and type of context, Switzerland 2018 Logistic regression, coefficient and 95% credibility interval. R-hat = 1.00, based on N=1,736 applications.

Taking the results from Table 3 and Figure 2 together, we can see that the level of discrimination varies clearly by context, even though overall, we see the highest levels of discrimination against Blacks and non-neighbouring White minorities. In Figure 2, the variation in the role of the context is visible at the bottom of the figure, where the interaction effects are shown: Relative to the base-line level of discrimination shown at the top of the figure, Blacks in particular experience a relatively better treatment – the level of discrimination against Blacks is lower in urban areas. The same is probably the case for the White non-EU origin immigrants, although both the effect size and the certainty are reduced. The particular results for Turkish immigrants in Table 3 will require further study to fully understand, as will the reasons for the different ethnic hierarchies between urban and rural areas.¹⁰

To better understand the meaning of contextual effects with the data at hand, we use the vote share for the Swiss People’s Party (SVP/UDC) as a proxy for the prevalent attitudes to immigrants. The party politicises against immigration (Ruedin 2013), and tends to do better in rural areas. In Figure 3, we look at the vote share in the municipality of the enterprise. We show only the interaction effects between the invitation rate for the different groups and the vote share for the SVP/UDC to emphasise the additional role of attitudes. Similarly to the link observed by Carlsson and Eriksson (2017) between attitudes and discrimination, we can see an additional penalty for Black applicants in municipalities with greater support for the anti-immigrant Swiss People’s Party, but cannot make headway in understanding the urban-rural differences in ethnic hierarchies.

Figure 3. Multivariate analysis of hiring decision: interaction effects of ethnicity and political context (SVP/UDC vote), Switzerland 2018 Logistic regression, coefficient and 95% credibility interval. Only interaction effects are shown to focus on additional discrimination. R-hat = 1.00, based on N=1,736 applications. The full model is included in the appendix.

6. Discussion and conclusion

While hiring discrimination against the second generation of migrant-origin groups in Switzerland has been reported consistently (Zschirnt 2020; Zschirnt and Fibbi 2019; Fibbi, Kaya, and Piguet 2003), the persisting impact of negative attitudes on labour market opportunities in the case of migrants’ offspring is in line with the very limited birthplace effect identified in the literature (Veit and Thijssen 2019): Our results corroborate the taste-based character of hiring discrimination. This study focuses on skin colour and addresses the relative position Blacks occupy in ethnic hierarchies in the Swiss labour market. The study demonstrates that in Switzerland – like in other Western countries – Black applicants face hiring discrimination. They must write 30 per cent more applications than White native-origin applicants to be invited for a job interview. This disadvantage of Black jobseekers finds further confirmation in the recent, large scale study by Hangartner, Kopp, and Siegenthaler (2021): Sub-Saharan candidates are the second-least likely to be contacted by recruiters.

Thanks to a comparison with the recent findings concerning White immigrant-origin groups, we assess the relative position of Blacks in immigrant ethnic hierarchies in Switzerland. We identify different theoretical expectations regarding which cleavage structures the ranking: the national one discriminating all outgroups, or the racial one with the double penalty of skin colour and immigrant background, or indeed the cultural value one, not singularising the relevance of skin colour. The results show that in pairs where either Black or White migrant origin candidates compete with White Swiss candidates, Blacks are affected by hiring discrimination as much as other White migrant-origin minority groups, as if skin colour does not play a specific role in their marginalisation in the labour market. However, since we used this paired design that mirrors the previous Swiss field experiments, we are unable to say how these Black or White migrant origin candidates would fare if they applied to the same position in direct competition against each other. This is a general limitation of field experiments on hiring discrimination, where other applicants are unobserved.

Our findings do not support the hypothesis of a racial hierarchy distinguishing applicants because of skin colour alone. While the cultural distance hypothesis seems to explain differences between White applicants from neighbouring countries and non-European Union countries – as shown by in the study of unemployment duration of immigrants (Auer and Fossati 2019) and the signalling of cultural attachment to the country of origin (Fossati, Liechti, and Auer 2020) –, it cannot account for other results. Candidates with Turkish origins are perceived as culturally distant from the Swiss population, yet they are (at the study level) the least affected by unequal treatment among the non-European Union groups. At a first glance this seems to contradict the findings by Fossati, Liechti, and Auer (2020), where Turkish named candidates have the lowest probability to be invited for a job interview. However, the comparison groups here have Spanish and Polish origins and thus belong to the group of EU/EFTA countries that are usually perceived as culturally closer to the Swiss (Ruedin 2020).

The fact that we also observe discrimination against candidates of German or French origin might relate to the specific positions that were tested in this field experiment (i.e. electricians and sales assistant). As Auer et al. (2019) have shown, the status of an occupation might also contribute to the degree of discrimination that different minority groups experience: Auer and colleagues do not find significant differences between Swiss, Portuguese, Serbian, or Senegalese candidates for cleaning positions, while for receptionist positions they observe an ethnic penalty for immigrant-origin candidates, with Portuguese candidates still faring somewhat better than Serbians or Senegalese candidates. Here too, skin colour alone does not seem to be the reason for the position in these ethnic hierarchies. Future studies in this field should ensure to include more occupations on different status levels to examine this idea of matching ethnic and occupational hierarchies.

However, the cultural proximity of EU origin groups did not systematically lower the risk of hiring discrimination, notably in urban settings. These findings contrast with an earlier Swiss field experiment that did not account for the urban-rural divide (Fibbi, Lerch, and Wanner 2006); however, they are in line with various studies on attitudes rather than behaviour that find negative attitudes and subtle forms of discrimination towards immigrants from neighbouring countries (Helbling 2011; Binggeli, Krings, and Sczesny 2014a, b; Krings et al. 2014).

These results lend support to the hypothesis that the preference for the majority in-group, rather than prejudice against ethnic out-groups, is at the root of discrimination on the labour market: Homophily and social closure are the most pertinent mechanisms accounting for hiring discrimination. Thus, the relevant cleavage distinguishing ‘us’ from ‘them’ puts in opposition the established to the outsiders, nationals to various ‘aliens’, taking ever-changing profiles of otherness (Blacks, Kosovo-Albanians and, at times, Germans, and French). Recent findings from attitude surveys underpin the idea of a significant penalty for out-groups: Both skin colour and nationality seem to provoke a similar uneasy reaction among respondents (Aeberli 2019; Ruedin 2020). Moreover, the paramount relevance of the national cleavage in Switzerland echoes the concept of ‘over-foreignisation’ – the idea of there being too many immigrants and immigrants being too foreign –, historically rooted in the early twentieth century and a long-lasting leitmotiv in Swiss migration policy (Mahnig and Cattacin 2005; Ruedin and D'Amato 2015; Arlettaz and Arlettaz 2004).

However, the boundary between ‘us’ and ‘them’ in this study does not appear to be consistent across contexts and jobs. Regional variations of hiring discrimination mirror the well-known urban-rural cleavage in attitudes toward otherness. Its translation in diverging social and political attitudes in Switzerland is increasingly polarised (Koseki and Lévy 2018). Yet the analysis yields the somewhat unexpected result of sharp regional variability in the exposure to the risk of hiring discrimination against the tested groups in diverse cities and more homogeneous contexts, in contrast to the finding of a similar regional analysis conducted in Austria, where the preference order of minority groups is the same in Vienna and outside the capital city (Weichselbaumer 2017). The minority group ranking observed in Swiss rural settings is compatible with both the cultural distance hypothesis¹¹ and with contact theory, according to which limited contact with minorities leads to more frequent negative attitudes towards them. The importance of the SVP/UDC vote makes plausible the perspective of a greater leeway for political actors to mobilise arguments against out-groups.

The ranking observed in urban contexts, by contrast, gives a completely different picture: In contrast to the expectations of the cultural distance hypothesis, European Union origin groups are also targeted significantly. This unexpected finding hints at the critical relevance of contextual factors, which are often neglected. They may mediate the way in which employers’ attitudes translate into actions, as suggested by a Norwegian study on employers, producing variability of ethnic and racial hierarchies in time and place (Friberg and Midtbøen 2018). New research is needed to explore this line of inquiry, possibly expanding the theory of matching hierarchies (Auer et al. 2019) and perceptions of what a ‘normal’ or ‘typical’ candidate looks like (Bohnet 2016).

Despite the high levels of discrimination in hiring for Blacks, the relatively small size of the affected group means that anti-immigration politicisation in Switzerland has not focused much on skin colour in the past (Van der Brug et al. 2015). As a corollary, the discrimination against Blacks cannot be explained by anti-immigrant propaganda and politicisation as it could probably be done for other immigrant groups, though relevant anti-Black narratives and stereotypes are clearly present in the population (Aeberli 2019; Purtschert and Fischer-Tiné 2015). Only in the last decade has skin colour become an increasingly mediatised topic in Switzerland, with a stronger presence of postcolonial studies, awareness campaigns, and, most notably in 2020, with a mobilisation around the Black Lives Matter protests in the United States. The discrimination against Blacks presented in this study remains nonetheless a low-key topic in the media and public discourse, perhaps because it does not translate into very high unemployment rates in the labour market and, lacking an anti-discrimination law to speak of, only a handful of cases are brought in front of the courts each year.

While the findings of this study may have been sadly predictable on the basis of international literature, a few considerations allow us to better understand the meaning of the results. In a meta-analysis that looks at correspondence tests including Black candidates, Thijssen (2020) show that Blacks experience discrimination in all countries studied, the highest discrimination rates being reported in French experiments (average discrimination ratio 2.9). Both the high discrimination rates in France as well as the comparatively lower discrimination rate in Switzerland in our study correspond with previous findings from meta-analyses. With a relative call back rate of 1.30 in the present experimental study, the extent of discrimination against Black citizens in Switzerland is largely comparable to the average ratio for Blacks in the United States (1.36) calculated in an exhaustive cross-country meta-analysis (Quillian et al. 2019), as far as these rates can be compared: similar rates, yet different awareness of the problem and even more diverse policies toward it. Yet, with the strengthening of the Black Lives Matter movement, the emergence of stronger postcolonial movements, and a more confident second generation of immigrant descent objecting more vocally to unequal treatment based on their origin, the topic of ethnic and racial discrimination may raise awareness and generate a broader public debate. Whether this momentum will translate into political action towards the adoption of an anti-discrimination law remains to be seen.

Author contributions

RF, EZ, DR designed the study; RS and EZ collected the data; RF and DR analysed the data; RF, DR, RS, EZ wrote the paper.

Acknowledgements

This study was funded by the Swiss National Science Foundation via the NCCR on the move and obtained ethical approval from the University of Neuchâtel, on May 11th, 2016.¹²

Disclosure statement

No potential conflict of interest was reported by the author(s).

Additional information

Funding

This work was supported by Schweizerischer Nationalfonds zur Förderung der Wissenschaftlichen Forschung: [Grant Number 51NF40-142020].

Notes

1 In the absence of statistics based on skin colour, the number of black residents is estimated by FSO on the basis of birthplace in sub-Saharan Africa or the Caribbean of the permanent resident population on 31 December 2016. https://www.bfs.admin.ch/bfs/fr/home/statistiques/population/migration-integration/nationalite-etrangere.html. See also (Efionayi-Mäder and Ruedin 2018).

2 Circulaire de la Police fédérale des étrangers aux postes de douanes, du 30 nov. 1961, E 7170 (b) 1977/67/248. ‘At the beginning of the 1960s, for the purposes of migration policy, the Federal Aliens Police created the category of “distant countries”, which included not only those in Asia or Africa, but also European states such as Yugoslavia, Greece, Turkey, Malta, Cyprus and Portugal’ (Cerutti 2005, 133), see also (Clark 1983).

3 https://www.bfs.admin.ch/bfs/fr/home/statistiques/population/migration-integration/nationalite-etrangere.assetdetail.13707205.html

4 https://www.bfs.admin.ch/bfs/en/home/statistics/work-income/unemployment-underemployment-vacancies.html

5 We were able to use the pictures that Doris Weichselbaumer used for her correspondence test in Austria. A detailed description on how the photographs were produced, evaluated and selected is available in Weichselbaumer (2017). Pictures were pretested for ‘looks, likeability, intelligence, reliability, as well as in their overall score’ (p. 246). For the Black candidate, we had one male and one female picture, for the White candidate we had two male and two female pictures which were randomly assigned. Which White picture was used had no effect on the reported results.

6 The inclusion of only one group of Blacks and two occupations requiring a vocational degree limit the generalizability of the results to some extent. This is counteracted by the possibility to compare the findings from this correspondence test to previous experiments using the same research design, focusing on different ethnic groups. Templates used are available from OSF: https://osf.io/ea5n8/

7 See https://www.bfs.admin.ch/bfs/fr/home/actualites/quoi-de-neuf.assetdetail.2543324.html for detailed explanations.

8 The net discrimination rate is not affected by the level of success that the research team had in achieving the required quality of applications (Wood et al. 2009).

9 The same four White photographs (two males, two females) were randomly assigned to the Swiss, Kosovo-Albanian, Turkish, French, and German profiles. All candidates who do not portray a Cameroonian background are considered White in this study.

10 We would like to thank an anonymous reviewer to point to a possible historical pattern: In the past the settlement of Blacks in Switzerland was concentrated in the French-speaking area, and Black immigrants tended to be highly educated. The combination of high levels of education and contact could lead to relatively lower levels of discrimination in the French-speaking area. We cannot confirm difference in education, note reduced geographical concentration in recent years (Zufferey 2019), and cannot clearly attribute the French-speaking coefficient to only the purported effect. However, to spark future investigation, we show in the appendix, that the interaction effect between Blacks and the French-speaking area is indeed positive.

11 Let us bear in mind though that the link between behaviour and prejudice in real settings is not unanimously ascertained in the literature. For hiring discrimination, see (Carlsson and Eriksson 2017).

12 Participants were informed ex-post of their participation, received the results and were given the possibility to react. For a detailed discussion on research ethics in field experiments and particularly the issue of informed consent, see also Zschirnt (2019).

Appendix

Table A1. Correspondence Tests in European Countries with Fictitious Candidates of African and/or Black origin.

Austria	Nigerians (Weichselbaumer 2017)
Belgium	Ghanaian (Baert et al. 2017), Congolese (Capéau et al. 2012)
Finland	Somali (Ahmad 2020)
France	Antillean (Bovenkerk et al. 1979), African (Cediey and Foroni 2008), Senegalese Muslims and Christians (Adida, Laitin, and Valfort 2010), North African (Pierné 2013, 2018), North African (Challe et al. 2018), Northern Africa / Foreign (Edo, Jacquemet, and Yannelis 2019)
Germany	Nigerian (Koopmans, Veit, and Yemane 2019; Veit and Yemane 2018), Ugandan (Koopmans, Veit, and Yemane 2019; Veit and Yemane 2018), Ethiopian (Koopmans, Veit, and Yemane 2019; Veit and Yemane 2018)
Ireland	African (McGinnity and Lunn 2011)
Italy	North Africa (Busetta, Fiorillo, and Palomba 2021)
Netherlands	Surinamese (Bovenkerk et al. 1979), Surinamese (Bovenkerk et al. 1995), Surinamese, Antillian (Andriessen et al. 2012), Antillean, Surinamese (Thijssen, Coenders, and Lancee 2021)
Sweden	Somali (Vernby and Dancygier 2019), Horn of Africa (Bursell 2007)
United Kingdom	West Indian (Jowell and Prescott-Clarke 1970), West Indian (McIntosh and Smith 1974), West Indian (Hubbock and Carter 1980), West Indian (Bovenkerk et al. 1979), African, West Indian (Firth 1981), Black African (Wood et al. 2009), West Indian (Brown and Gay 1985)

Studies include fictitious candidates with the respective backgrounds. Countries in bold are part of Sub-Saharan Africa or the Horn of Africa region (as defined by https://openknowledge.worldbank.org/pages/focus-sub-saharan-africa). Following the meta-analysis in Thijssen (2020), Antilleans and Surinamese are also classified as Blacks.

Appendix A2. Regression Model for Figure 3.

	Estimate	Est. Error
Intercept	0.08	0.73
Cameroun	0.09	0.67
French	0.09	1.5
German	−1.09	1.21
Kosovo	−2.49	1.14
Turkish	−0.42	0.93
male	−1.61	0.54
sales	−3.41	0.54
SVP vote	0.05	0.02
Cameroun * SVP vote	−0.04	0.02
French * SVP vote	−0.05	0.07
German * SVP vote	0.02	0.04
Kosovo * SVP vote	0.03	0.04
Turkish * SVP vote	−0.02	0.03

Logistic regression, coefficient and median absolute deviation as standard error. R-hat = 1.00, based on N=1,736 applications.

Appendix A3. Interaction between Ethnicity and Language of the Advert.

	Estimate	Est. Error
Intercept	1.98	0.61
Cameroun	−1.28	0.36
French or German	−0.91	0.47
Kosovo	−1.03	0.47
Turkish	−1.08	0.48
French language advert	−1.55	0.40
male	−1.51	0.51
sales	−3.45	0.54
Cameroun * French advert	0.76	0.50
French or German * French advert	0.32	0.76
Kosovo * French advert	−1.88	0.83
Turkish * French advert	0.06	0.70

Logistic regression, coefficient and median absolute deviation as standard error. R-hat = 1.00, based on N=1,736 applications. Neighbouring countries (French and German) were combined to avoid empty cells.