Bonnet, F., Lale, E., Safi, M., & Wasmer, E. (2016). Better residential
than ethnic discrimination! Reconciling audit and interview findings in
the Parisian housing market.
Urban Studies
,
53
(13), 2815-2833.
https://doi.org/10.1177/0042098015596107
Peer reviewed version
Link to published version (if available):
10.1177/0042098015596107
Link to publication record in Explore Bristol Research
PDF-document
This is the author accepted manuscript (AAM). The final published version (version of record) is available online
via Sage at http://usj.sagepub.com/content/53/13/2815. Please refer to any applicable terms of use of the
publisher.
University of Bristol - Explore Bristol Research
General rights
This document is made available in accordance with publisher policies. Please cite only the
published version using the reference above. Full terms of use are available:
http://www.bristol.ac.uk/red/research-policy/pure/user-guides/ebr-terms/
Better residential than ethnic discrimination!
Reconciling audit and interview findings
in the Parisian housing market
∗
François Bonnet
†
, Etienne Lalé
‡
, Mirna Safi
§
, Etienne Wasmer
¶
This version: January 2015
Abstract
This article investigates discrimination and the interplay of residential and ethnic stigma
on the French housing market using two different methods, paired-testing audit study of real
estate agencies and face-to-face interviews with real estate agents. The juxtaposition of their
findings leads to a paradox: interviews reveal high levels of ethnic discrimination but little to
none residential discrimination, while the audit study shows that living in deprived suburbs is
associated with a lower probability of obtaining an appointment for a housing vacancy but eth-
nic origin (signaled by the candidate’s name) has no significant discriminatory effect. We have
three priors potentially consistent with this apparent paradox and re-evaluate their likelihood
in light of these findings: (i) agents make use of any statistical information about insolvency,
including residency; (ii) there are two distinct and independent taste discriminations, one about
space and one about ethnicity; (iii) these two dimensions exist and complement each other.
Keywords: Audit, Discrimination, Neighborhood Effects, Housing
JEL codes: J71, R23
∗
This research was financed under the grants 2007 ANR-07-1210254 (DISCRI-SEGRE), ANR-10-BLANC-1819-
01(EVALPOLPUB), ANR-11-LABX-0091, ANR-11-IDEX-0005-02. We are grateful to Lorraine Pinto and Lucile
Romanello for their contribution early on in this project. We also thank Aurélien Chauvier, Augustin Célier, Fatima
Hocini, Mohamed El Dahshan, Alberic de Gayardon and Jean de Saint-Chéron for research assistance. Comments by
participants at the Sciences Po workshop on discrimination and segregation, notably David Latin and Jeffrey Zax were
most useful.
†
CNRS, UMR Pacte.
‡
Department of Economics, University of Bristol.
§
Department of Sociology, Sciences Po, OSC, CNRS.
¶
Sciences Po, Department of Economics and LIEPP.
1
Introduction
Ethnic minorities are often concentrated in underprivileged neighborhoods in Western societies (John-
ston, Poulsen and Forrest, 2007; Musterd, 2005; Peach, Robinson and Smith, 1981). Emanating from
discriminatory processes, ethnic segregation may also lead in turn to “residential traps” that affect mi-
nority populations’ socioeconomic achievements, from access to education to economic success and the
building of social networks (Crane, 1991; Sampson, Morenoff and Gannon-Rowley, 2002).
While there is a wide range of literature on the cumulative aspect of residential and ethnic/racial
inequalities in the United States (Denton and Massey, 1993; Wilson, 1978, 1987), the interaction between
ethnic and residential stigma has seldom been analyzed in audit studies on housing discrimination; to our
knowledge, the only exception is a study of the credit market by Ross and Yinger (2002).
1
Chronic urban riots in the French banlieues in 2005 once again raised the question of ethnic minorities
in France (Waddington, Jobard and King, 2013), and yet urban studies carried out in France rarely con-
sider ethnicity in general and ethnic discrimination in the housing market in particular. Some exceptions
can be found in the form of recent works providing evidence of both residential and ethnic discrimination;
these are reviewed below. By “residential discrimination” we mean discriminatory decisions precisely
oriented against the ability of individuals to choose their residential area. This practice has been explicitly
illegal in France since 14 December 2013: residential discrimination is the 20th criterion for assessing
whether discrimination against individuals has taken place.
2
France is an interesting case to study discrimination because the country ostensibly promotes a color-
blind ideal of race relations (Sabbagh and Peer, 2008; Safi, 2008; Simon, 2008). For instance, the French
Republican model forcefully rejects ethnicity, culture, and religion as a basis for political organization,
claims-making, and even as the basis of categories for official statistics. Ethnicity per se is therefore
not reported in any public-sector statistical survey in France, which makes it difficult for race-based
affirmative action to be enforced and for inequality to be documented through representative data.
In light of the above, our article aims at analyzing the relationship between ethnic origin (specifically
North African descent) and residential origin (residency in a deprived neighborhood), in a potentially dis-
criminatory interaction in the housing market in France. Our goal is to explore how ethnic and residential
stigma can be disentangled in the measurement of discrimination by analyzing practices and discourses
relating to the overlap between these two criteria. To this end, we employ two methodological designs:
• An experimental paired-testing audit study, in the tradition of statistical analyses of discrimination,
which aims to measure the interaction of ethnic and residential effects.
• A qualitative study, based on interviews with real-estate agents who were asked open-ended ques-
1
The link between racial and spatial stigma has been explored more frequently in qualitative research; see for instance
Kirschenman and Neckerman (1991).
2
It should be noted that residential location in deprived and segregated areas has also considerable legal socioeconomic
consequences, well documented in the urban literature, undermining people’s life chances in terms of education, health,
employment, etc.
2
tions about the selection process for housing applicants.
3
Each method has its own advantages and limitations. While the audit study actually measures the effect
which the two dimensions have on discrimination, it does not detail the mechanisms underlying ethnic
and residential stigma. Conversely, face-to-face interviews provide discursive evidence on discrimination
and describe its underlying micro-social processes, but do not assess the magnitude of discriminatory
practices. The complementary use of both methods helps overcome the shortcomings of each. In fact,
the juxtaposition of our audit and interview findings leads to a double paradox:
1. Although real estate agents believe ethnic discrimination to be widespread, it is not statistically
significant in an audit study that controls for residential origin but it is when residential origin is
not controlled for.
2. Although real-estate agents do not mention residential discrimination, the audit findings suggest
that it is statistically significant.
This article analyzes this discrepancy highlighted in many studies of discrimination between discourses
and practices, and attempts to provide interpretations of it. We first review the research background and
present our two sets of findings, and then we attempt to decode the double paradox that emerges from
them. We then review the various hypotheses consistent with these findings that may help us explain the
paradox.
1 The relationship between residence-based and ethnic discrimina-
tion in housing: Background and hypotheses
1.1 Space and ethnicity in discrimination studies
There is now extensive sociological and economic literature showing, through survey results, that residen-
tial location has a significant effect on employment, education and crime, among other things (Brueckner
and Zenou, 2003; Fernandez and Su, 2004; Sampson and Sharkey, 2008). Despite the well-documented
role played by segregation and residential location in the production and perpetuation of inequality, the
concept of “residential discrimination” has seldom been used and its effect is rarely measured in au-
dit studies on discrimination. Conversely, ethnic and racial discrimination in American cities has been
measured with audit studies for more than three decades (Committee on National Statistics, 2002, 2004;
Fix and Struyk, 1993). These studies regularly document the unequal treatment disadvantaging ethnic
and racial minorities at various stages in the housing search on both the rental and ownership markets
(Turner et al., 2013). The 2012 study by the U.S. Department of Housing and Urban Development (HUD)
3
The agents audited are not necessarily the same as those interviewed, and we will explain why this is not of consequence
below.
3
showed in particular that, of all ethnic groups, Hispanics face the highest levels of discrimination on the
rental market (followed by Blacks and Asians); for example, they become aware of roughly 12% fewer of
the housing units available than white housing applicants when they contact real-estate agents to inquire
about recently advertised properties.
Comparisons of the HUD’s findings over time show that although “blatant” discrimination has de-
clined, with minority applicants now less likely to have a door slammed in their faces, overall unequal
treatment remains high because of more subtle forms of discrimination. For example, minority applicants
are more likely to be told that they must talk to a lender before being shown an advertised home for sale,
whereas a white tester is more likely to meet with the agents without being asked about prequalification.
This changing framing of discriminatory practices challenges the methods which pair-testing studies have
traditionally used to measure discrimination and requires more attention to several details in the nature
and quality of the interactions between the testers and the audited agencies or landlords.
Finally, in the majority of studies, evidence on ethnic and racial discrimination is interpreted as being
related to conscious motivations and taste-based mechanisms.
4
This is corroborated by some findings on
ethnic and racial neighborhood preferences highlighting whites’ unwillingness to live in neighborhoods
with a high proportion of ethnic minorities, particularly African Americans (Charles, 2009). Finally, in
an audit similar to ours, Ahmed and Hammarstedt (2008) find that male applicants on the rental market
with a Swedish name are much more likely to be called back than those with Arab names; evidence for
female applicants is less categorical.
1.2 Review of discrimination and immigration studies in the French context
Studies such as those cited above have rarely been conducted in France, and research on ethnic segre-
gation and discrimination in the area of housing has emerged only recently there. Most scholars have
analyzed ethnic segregation as being directly linked to “color-blind” market mechanisms of social strat-
ification. Recent studies challenge this, however, showing that ethnic segregation is more prevalent than
socioeconomic segregation and that it decreases very slowly (Préteceille, 2009; Rathelot and Safi, 2014;
Safi, 2009; Verdugo, 2011). French government statistics on housing also document considerable inequal-
ity between natives and immigrants. Such inequality occurs with regard to not only housing access and
tenure, but also housing quality, in terms of factors such as amenities and apartment size (Barou, 2002;
Breem, 2009). The immigrants’ housing situation is particularly disadvantaged for non-Europeans and
is resistant to standard socioeconomic controls, suggesting underlying ethnic discrimination. In a recent
comprehensive survey on immigration and discrimination, first- and second-generation immigrants were
shown to report twice as much discrimination in housing access as natives of non-immigrant background
(Pan Ké Shon and Scodellaro, 2011; Safi and Simon, 2014). Controlling for socioeconomic variables
suggests that ethnic or racial discrimination may, at least partly, be at play. In the same survey (TeO), the
4
Taste-based discrimination still seems to be a major mechanism explaining persisting unequal treatment toward African
Americans in the US labor market (Bertrand and Mullainathan, 2004).
4
