Tackling discrimination on the housing market through correspondence tests
City/Region : Ghent , Belgium

The City of Ghent asked the University of Ghent to investigate the extent of discrimination on the housing market.

The aim of this study was twofold. Firstly, we aimed to investigate the extent of discrimination against ethnic minorities, disabled people and people on social welfare on the private rental housing market in Ghent. Secondly, we wanted to examine whether discrimination could be reduced by monitoring real estate agents and landlords through correspondence testing.

The study was carried out in 2014-2015.

Afterwards the city and the real estate sector created and signed the charter titled "Equal acess to housing".

Since the study has proven that situation testing reduce discrimination, we have started a new round of testing this year.

Agendas addressed
HousingSocial inclusion and integration
Pathways followed
  • Ensure equal access to municipal services

Ghent is a city that is home to more than 160 nationalities. 30% of all Ghent's citizens have a migration background. The Ghent population is very diverse.

Earlier studies had already proven that discrimination on the housing market does exist. But these studies where not focused on Ghent and/or examined only one or two discrimination grounds (ethnic background, disablity) .

Decent housing is human right. We cannot tolerate that people are beeing discriminated on the housing market.

That is why we asked the University of Ghent to carry out this study.

We also investigated whether situation testing has an effect on the behaviour of landlords.

In Action

The extent of discrimination was measured through a first wave of pairwise-matched correspondence tests (between November 2014 and March 2015). In this research method two candidate renters apply to the same rental advertisement on Immoweb by contacting the real estate agent or landlord via e-mail. The candidates ask whether the dwelling is still available and whether it is possible to arrange a meeting to view the dwelling. Both candidates should not differ from each other on most relevant characteristics, except with respect to the potential ground of discrimination (e.g. ethnicity, disability or income source). Afterwards, we examine whether one of the candidates is adversely treated by the realtor or landlord in comparison with the other candidate. An adverse treatment is assumed to be due to discrimination. As a consequence, it is truly important that both candidates are as similar as possible, apart from the ground of discrimination.

"To be clear, with this research method, no discriminatory questions were asked. We examined only the extent to which two similar candidate-renters are unequally treated." says Pieter-Paul Verhaeghe, researcher at the University of Ghent.

Afterwards, we calculated net discrimination rates, which show the levels of structural discrimination against a group, given the random unequal treatment of candidates on the rental market.

The following correspondence tests were performed:

• Ethnic discrimination was measured by comparing a rental candidate with a typically Belgian sounding name with a Dutch-speaking candidate with a typically Arabic, Turkish or Eastern-European sounding name. We performed pairwise-matched correspondence tests on ethnic discrimination for 754 rental dwellings in Ghent.
• Discrimination against disabled people was measured by comparing a non-blind candidate with a blind candidate with an assistance dog. Here, we executed 268 pairwise-matched correspondence tests.
• We performed 305 pairwise-matched correspondence tests on discrimination against people on social welfare, in which we compared a rental candidate with a wage with a candidate with a benefit for disability or a social benefit, supplemented with a rental subsidy. We only selected rental vacancies with a rent that could be afforded by the wage or the (supplemented) social or invalidity benefit. In Belgium, landlords and real estate agents may select tenants based on the level of their income, but it is not allowed to select on the basis of the source of their income. Disadvantaging candidates with social benefits versus candidates with wages is considered as discrimination.
• We also examined discrimination on the combined grounds of ethnicity and income source. Here, we compared a rental candidate with a wage and a typically Belgian sounding name with a candidate with a social or disability benefit and a typically Turkish sounding name. For this comparison, we performed pairwise-matched correspondence tests for 195 rental dwellings in Ghent.

In total, 1753 rental dwellings in Ghent have been tested in the first wave through pairwise-matched correspondence tests.


From the correspondence tests, it appears that there is structural discrimination on the private rental housing market in Ghent. In 37% of the rental advertisements, candidates with an Arabic, Turkish or Eastern-European name were systematically adversely treated in comparison with candidates with a Belgian name. There were no significant differences between Arabic, Turkish and Eastern-European names. English-speaking candidates with a Turkish name were discriminated against in 49% of the cases. In other words, there seems to be an additional penalty on not speaking the local language. Landlords and real estate agents also discriminate against blind candidate renters with an assistance dog in 36% of the rental advertisements. The discrimination of blind people is probably due to the apparent reluctance to (assistance) dogs. Moreover, we found that realtors and landlords in Ghent discriminate on the basis of the source of income. In 48% of the rental advertisements, the rental candidates with a social or dsiability benefit were systematically adversely treated in comparison with rental candidates with a wage. Finally, candidates with a Turkish name and a social or disability benefit were discriminated against in 70% of the cases. In general, we found that private landlords discriminate more than real estate agents.

We also discussed the results of this study with the real estate sector. This led to the charter "equal access to housing" in which the city and the sector both engage themselves to battle discrimination on the housing market. Cfr. attachment.


After the first wave of correspondence tests, we prepared a press text of their results and announced that we would organise a second wave of correspondence tests to evaluate discrimination further. In addition, each real estate agent received a letter which contained the same message: that discrimination has been observed during correspondence tests and that they would be subjected to further tests during the months to come. The news was widely covered in the mainstream public media. One week after the announcement via public media and the personal letter to real estate agents, we repeated the correspondence tests in a second wave (from April to June 2015). Here, we measured discrimination again with the same methodology as in the first wave. However, due to time limitations we could not test discrimination against blind people with an assistance dog in the second wave. All other discrimination grounds were replicated. During the second wave, we performed in total 1666 pairwise-matched correspondence tests in Ghent. By comparing both waves of correspondence tests, we can assess whether discriminatory behavior is adapted when landlords and real estate agents are aware that they are monitored through correspondence tests.

It appears that the results are different for real estate agents and private landlords. Among landlords, we did not find any change in their discriminatory behavior between the two waves. This means that landlords do not stop discriminating when they know that they are being monitored through correspondence tests. Among real estate agents, we did find a significant reduction in ethnic discrimination from 24% in the first wave to 10% in the second wave. In other words, a substantial part of the real estate agents stopped discriminating against ethnic minorities when they were aware that they were being monitored through correspondence tests. However, a minority of 10% kept on discriminating. Moreover, we found only a significant reduction in ethnic discrimination,

This year the city and university have started further situation testing for ethnicity and visual impairment. This time we are using real people to test the individual real estate agents. Every agent will be tested 10 times. If there is a suspicion of discrimination, the antidiscrimination center (Unia) will take legal action against the real estate agent.      

Challenges and lessons learned

Discrimination on the private rental housing market in Ghent is widespread. Ethnic minorities, blind people with an assistance dog and people on social welfare are systematically adversely treated by landlords and real estate agents. Ethnic discrimination can, however, be substantially reduced among real estate agents when you monitor them through correspondence tests. Unfortunately, this seems not to be the case among private landlords and for other discrimination grounds.



ICLEI – Local Governments for Sustainability
European Secretariat

Leopoldring 3
79098 Freiburg

Tel.: +49 (0) 761 – 368 92 0
Fax: +49 (0) 761 – 368 92 19



Legal Notice / Impressum


Data Protection Policy / Datenschutzerklärung