EU Citizenship guarantees the same rights to all mobile EU citizens who move to another member state. And yet, as a recent study by Christian Adam, Xavier Fernández-i-Marín, Oliver James, Anita Manatschal, Carolin Rapp and Eva Thomann indicates, some EU citizens are more likely than others to face discrimination when interacting with their host country’s public administration. Remarkably, they find that patterns of discrimination displayed by public administrators are very similar to patterns of discriminatory behaviour displayed by the general public.
EU citizens who move to another member state have rights enshrined within the provisions of EU citizenship. They are guaranteed social rights as they are – subject to specific conditions – eligible to receive social benefits in their host country. Moreover, they enjoy political rights as they can, for example, participate in local elections. These rights have attracted substantial political conflict and controversy: they have been upheld as non-negotiable elements of EU citizenship and decried as symptoms of declining national sovereignty and uncontrollable immigration, for example, in the debates about Brexit.
Our research focuses on administrative behaviour within this strongly polarised context. Essentially, we ask two questions: are all EU citizens treated equally within this context? And are public administrators, who are socialised and who operate within legal regimes that should promote fair treatment, just as likely to discriminate against certain EU citizens as general citizens?
In the light of current controversies around migration, EU member states have tried to restrict mobile EU citizens’ access to welfare benefits by translating very restrictive interpretations of EU law into administrative practices. While EU citizens cannot count on being treated equally as national citizens in their access to social benefits, every EU citizen should expect to be treated equally to all other EU citizens. But is this the case?
Anyone who has ever moved to another country knows that interacting with the local bureaucracy can affect whether it is very cumbersome or relatively easy to get settled. Helpful, friendly, and service-oriented public servants can help you to navigate the bureaucracy. However, unfriendly and unwelcoming public servants can use their discretion to increase waiting periods and abstain from pointing out important obstacles on the administrative way to your objective.
Extant research – mostly relying on field experiments – has repeatedly shown that whether one is likely to be treated in a friendly or unfriendly manner, to receive unrequired help or not, or to receive preferential treatment or not, is not completely random. Instead, individuals’ attributes – such as their nationality, gender, language skills – influence whether people are likely to be treated favourably. Just as discrimination is an objectively observable characteristic of societal interactions, it is also omnipresent in interactions between citizens and public administrators.
Systematic comparisons between the general public and public administrators are lacking, however. Optimistically, one might expect that public servants’ public service motivation (intrinsic and/or socialised) and legal constraints would reduce the extent of discrimination within such administrative encounters. Interestingly, this proposition has never been tested explicitly through a systematic comparison between discriminatory behaviour by the general public and such behaviour by public administrators.
An empirical test
To conduct this comparison and to assess whether there are systematic differences in the treatment of EU citizens from different member states, we conducted a conjoint experiment in Germany in early March 2020. Our study draws on a general population sample of 2,974 people and includes a top-up sample of 779 people who work within the core of public administration. By subjecting both of these groups to the same bureaucratic choice experiment, we can determine whether discrimination is more or less pronounced in one group or the other, whether discrimination is triggered by the same attributes in both groups, and whether anti-discrimination interventions are similarly effective in the groups.
In our experiment, respondents perform two tasks: in one task, they handle a hypothetical application for the welfare programme colloquially dubbed ‘Hartz IV’; a means-tested benefit for (long-term) unemployed job-seekers. In a second task, respondents handled a question about missing voting documents needed to participate in municipal elections. We asked respondents to put themselves into the position of frontline workers who, due to time constraints, have to prioritise the processing of the request of a citizen from two presented to them for processing. We describe the processing task, including a statement that preferential treatment of one individual will negatively affect the other individual.
The conjoint design allowed us to simultaneously randomise five characteristics of the applicants to investigate their causal effects, with a particular interest in differences between types of mobile EU citizen. First, we varied applicants’ nationality. Specifically, we varied between Dutch and Romanian applicants as both nationalities should reflect different levels of socio-economic and cultural proximity to the German host country context (Dutch being more similar than Romanian). A second attribute that can signal varying levels of cultural proximity is proficiency in the host country’s language (broken vs fluent). Moreover, we varied applicants’ gender (male vs female), profession (nurse vs medical doctor), and age (25 vs 40 vs 55).
Prioritising one applicant over the other does not per se reflect discriminatory behaviour. Participants have to prioritise one applicant in each binary choice setting. If, however, Dutch citizens are much more often prioritised than Romanians, choices do not seem to be merely random but affected by applicants’ nationalities.
Our results indicate that language proficiency is the strongest trigger of discrimination. Whether individuals are fluent in the host country’s language or not makes the largest impact on whether they are likely to be prioritised in the respective bureaucratic choice setting or not. The second largest effect trigger of discrimination was nationality. Romanian citizens were generally much less likely to be prioritised in either bureaucratic choice setting.
While men are generally less often prioritised than women in both policy sectors, the unemployment requests of nurses are generally prioritised over those of doctors. Results for the effect of age are less consistent, but applicants in the middle age group tend to be prioritised over older and younger applicants (age 55 and 25). These patterns were remarkably similar between the two groups of respondents (general public vs public administrators) – confounding expectations that public sector workers might be less discriminatory.
To assess whether discrimination can be reduced, we introduced an anti-discrimination intervention. Specifically, we informed a treatment group that they would afterwards have to justify their decisions towards their putative manager, while the control group did not receive any such information. This was intended to introduce an element of accountability. However, this intervention proved to be ineffective – it neither reduced discrimination among the general public, nor among public administrators.
Overall, our findings support extant research on discrimination against migrants, which highlights that not all migrants face the same challenges but face differential treatment based on certain characteristics, for example, as has been found for provenance or religious orientation.
Finally, our study yields certain policy implications: if nationality is a crucial trigger of discrimination, anonymity in bureaucratic encounters could be valuable. This is, however, very difficult to implement in many bureaucratic interactions. After all, eligibility for voter registration and welfare benefits is established based on people’s nationalities.
While changes towards more anonymity in the process might not be impossible, directing efforts at, on the one hand, empowering mobile EU citizens via quick and proficient language acquisition, and, on the other hand, testing and implementing efficient anti-discrimination measures in the bureaucracy, appear to be the most fruitful approaches for overcoming discrimination.
For more information, see the authors’ accompanying study in the Journal of European Public Policy