![Sample our Social Sciences journals, sign in here to start your FREE access for 14 days]({"type":"image" , "src" : "/sda/110822/TOC-Subject-Sample-2021-Social-Sciences.jpg"})
ABSTRACT
This article focusses on the scope and causes of ethnic discrimination against applicants with different ethnic backgrounds in the German housing market. Whereas ethnic discrimination is a well-documented phenomenon in the commercial housing market, little is known about the shared housing market.
I conducted a field experiment in the German shared housing market for students, sending out fictitious enquiries with names signalling a Turkish, a Syrian or an US background to advertisers of shared livings in Berlin, Germany.
The results show that ethnic discrimination against Turks and Syrians is widespread, whereas the US applicant is not discriminated against substantially. Discrimination against the Syrian applicant is furthermore mediated by the language used in the enquiries. The results show that patterns of discrimination differ between permanent and temporary contracts.
Disclosure statement
No potential conflict of interest was reported by the author.
Notes
1 The #metwo channel on Twitter is a platform for Germans with a migration background to exchange their discrimination experiences.
2 I would like to thank the anonymous JEMS reviewer for their constructive and helpful comments.
3 For a more general overview of discrimination in Germany see Klink and Wagner (Citation1999) and Auspurg, Hinz, and Schmid (Citation2017).
4 With ‘Americans’ I am referring to white US Americans.
5 In 1961, 1.2% of the German population were foreigners (Federal Statistical Office Citation2017a).
6 About 11.6% were Germans but had an indirect or direct migration history and 10.9% were non-Germans.
7 With ‘Turks’ I am referring to persons in Germany who have a direct or indirect immigration history from Turkey.
8 See Said (Citation1978) for a critique on Orientalism.
9 The evaluation of the ethnic background of the advertisers was based on the surname of the contact person.
10 For a description of causal inference and the preconditions for causal inference, see Pearl (Citation2010).
11 See e.g. Hall (Citation1992) and Said (Citation1978) for an analysis of the discursive creation of ‘the west’/‘occident’ vs. ‘the rest’/‘orient’ and Huntington (Citation1993, Citation1997, Citation2004) as example for the (scientific) reproduction of this division.
12 Between 2006 and 2015 the absolute and the relative number of non-native students has increased in Berlin (2006: 15%, 2017: 20%, Amt für Statistik Berlin-Brandenburg Citation2018). Hence, the prevalence of English as language of communication is increasingly natural in this market segment. However, the results of a pre-test indicate that the use of English is still recognised as unusual by the students (see section 4.2 for more information on the pre-test).
13 For a discussion on detecting ethnic discrimination in field experimental settings see Neumark (Citation2012)
14 The platform itself states that they have ‘up to 8 million visitors a month’ (https://www.wg-gesucht.de/).
15 The setting of the experiment was based on the expertise of the Federal Anti-Discrimination Office (Citation2011) and the vote of the university’s ethical committee.
16 See e.g. the appearance of and the public attention given to the far-right party ‘Alternative für Deutschland’ (AfD; Alternative for Germany).
17 I decided for male identities as most points of reference in the empirical literature concentrate on discrimination against men, and men are overrepresented among the refugees from Syria (DIW Citation2017).
18 In the pre-test 100 enquiries from the Syrian applicant where sent to advertisements, varying the amount of information given in the enquiries. The share of (positive) responses was increased when the enquiries included more details about the applicants. However, as the main aim was to measure the effect of ethnicity on invitations, giving more information would have jeopardised the ability to measure the direct causal effect of ethnicity on invitation rates. Afterwards the advertisers where invited for a qualitative interview. 14 interviews where realised discussing if the names where perceived as distinct for the respective ethnicity and if the length of the enquiry was perceived as appropriate.
21 e.g. Following Carlsson and Eriksson (Citation2015), the first idea was to name the Syrian immigrant Mohamed Hussain. However, although Mohamed Hussain is an Arabic name, the students did not unanimously agree that Mohamed Hussain clearly indicates an Arabic instead of a Turkish migration background.
22 As there was no specific theoretical expectation about the influence of the size of the apartment/the room, the amount of the rent and the location of the flat these variables where only included as robustness check but not part of the final model. The main effect was not substantially affected by the inclusion of these variables. Furthermore, although there was an initial expectation about the impact of the number of persons living in an apartment, this assumption was excluded from the empirical model as the interpretation of this variable was inconclusive. There was no effect of the variable on patterns of discrimination.
23 A complete enquiry looked like this (for the Syrian applicant the information that he came a year ago from Syria was included):
Hi [name of the advertiser],My name is [name of the applicant]. I’m 22 years old and just finished my Bachelor’s. Next semester I will start my Master’s in Berlin. I just saw your ad on WG-Gesucht and the room sounds perfect for me. Is it still available? Ím in Berlin from tomorrow onwards and would love to come around to see the room and get to know you.Best regards,Burak
24 The period was restricted to two weeks for ethical reasons. I had to inform the advertisers that the enquiries where sent in the scope of a research project within two weeks to ensure that I will be able to contact them. However, the positive responses I received where all sent to me within the first five days after the enquiries where sent.
25 The results are similar if these 3 cases are included as well.
26 The dominance of a no-answer behaviour is in line with the findings of Carlsson and Eriksson (Citation2015) in the London Market for Shared Housing.
27 From each of the resulting six treatment conditions 100 enquiries where sent out. The sample size for the field-experiment was calculated using the R package pwr. Calculations where performed under the assumption of one-sided tests at α = 0.05.
28 It should be noted that most of the persons with a Turkish migration background in Germany could be considered natives as well because they were born in Germany. Nevertheless, in the following they are referred to as ‘Turks’ or ‘Turkish’.
29 Tests for statistical significance between the different ethnic groups have been conducted separately by changing the reference category in the regression model.
30 When comparing these two conditions it has to be considered that the observed difference in invitation rates is at least to some degree explainable because Americans might be ascribed as native English speakers whereas Syrians writing in English are not. In this interpretation living together with an American might be considered an investment in one’s own English skills by learning native English through day-to-day communication.