Discrimination and segregation in the rental housing market in Delhi

How to cite

Syed, Hamza Mohammad. 2026. "Discrimination and segregation in the rental housing market in Delhi." AEA RCT Registry. AEARCTR-0017735. https://doi.org/10.1257/rct.17735-1.0

Download: BibTeX · RIS — for reference managers without a browser translator. Zotero and most others pick up this page's citation metadata directly.

Hamza Mohammad Syed*

* Primary investigator

Registry ID
AEARCTR-0017735
DOI
https://doi.org/10.1257/rct.17735-1.0
First registered
Last updated
Status
On going
Trial dates
2026-01-21 – 2026-03-31
Countries
Private
Registry record
http://www.socialscienceregistry.org/trials/17735

Abstract

How do discrimination and segregation have an effect on how urban spaces are developed? Does residential segregation affect belief polarisation? The segregation that we look at, when it comes to cities in developing countries, are the movements in and out of slums as people’s incomes grow (Marx, Stoker and Suri, 2013). If we add discrimination to the equation, even a well-off but discriminated-against population might be forced to live in a ghetto-like situation because they cannot get accommodation elsewhere. This segregation, especially for minorities, affects their access to public services and imposes several direct and indirect costs on them. We can view discrimination and segregation as a cycle, whereby if people perceive discrimination to be high and if they are denied rental housing because of discrimination, they might end up living in segregated localities. And if segregation increases, people will interact less and this will lead to more polarisation and discrimination.
The study uses a combination of theory, survey experiments, novel data on locality wise amenities to understand the link between discrimination and segregation.

Intervention

In this project, I'm using a survey experiment to parse out taste-based vs statistical discrimination against Muslims in the housing market in Delhi. Respondents will be shown randomized profiles of prospective tenants. There will be three randomized groups of respondents: A - profiles with predominantly Hindu sounding names; B - profiles with same characteristics as A but more Muslim sounding names; C - same profiles as B but with additional information about tenant reliability and police verification, etc. The difference in acceptance rates between groups A and B will give us a measure of total discrimination. The difference between groups B and C will give us a measure of statistical discrimination. Deducting statistical discrimination from total discrimination will give us a measure of taste-based discrimination. My expectation for Delhi is that taste-based discrimination against Muslims is higher than statistical discrimination. And total discrimination against Muslimsis higher than Hindus.
I will also be using a public goods game and an area choice game to get stated preferences for area types both as a planned and as a tenant. We are also testing these results for areas which have a history of conflict vs areas which don't. We are expecting to find higher levels of discrimination in the areas that have a history of conflict.
Another component is a broker survey, treating them as experts, to map out the probability of finding a house for different identities in localities across Delhi.

Experimental design

Stratified random sampling of localities in Delhi is done based on the religious mix of the locality and geographical spread. Within these localities, random starting locations are selected within the residential areas and then every 5th house is surveyed. The profiles are randomized across areas and respondents.

Design details

Randomization method
Randomization done in office by a computer
Randomization unit
Stratified randomisation at locality-resident level for resident survey; stratified randomisation at pincode-broker level for broker survey
Clusters
72 localities and 74 pincodes
Observations
3200 residents and 600 brokers
Arms
18 localities in each strata - only temples, only mosques, mix of both, neither 1066 respondents in benchmark/control group, 1066 respondents in treatment 1, 1066 respondents in treatment 2
Primary outcomes
total discrimination, taste-based discrimination, statistical discrimination
Secondary outcomes
locality preferences, amenity preferences, probability of finding a house

Cited by

No citation data available for this registration yet.

Keywords