Discrimination, Information Treatment, and the Screening Margin in Malaysia's Rental Housing Market: A Correspondence Experiment in Kuala Lumpur

Last registered on June 22, 2026
View Trial History

Pre-Trial

Trial Information

General Information

Title
Discrimination, Information Treatment, and the Screening Margin in Malaysia's Rental Housing Market: A Correspondence Experiment in Kuala Lumpur
RCT ID
AEARCTR-0018948
Initial registration date
June 16, 2026

Initial registration date is when the trial was registered.

It corresponds to when the registration was submitted to the Registry to be reviewed for publication.

First published
June 22, 2026, 6:59 AM EDT

First published corresponds to when the trial was first made public on the Registry after being reviewed.

Locations

There is information in this trial unavailable to the public. Use the button below to request access.

Request Information

Primary Investigator

Name
Helen Bao
Affiliation
University of Cambridge
Contact Primary Investigator

Other Primary Investigator(s)

PI Name
Nur Hafizah Binti Mohammad Ismail
PI Affiliation
Universiti Utara Malaysia & University of Cambridge
Contact Investigator

Additional Trial Information

Status
On going
Start date
2025-01-15
End date
2026-07-15
Keywords
Behavior, Other
Additional Keywords
correspondence test; rental discrimination; ethnic discrimination; nationality; behavioural intervention; information disclosure; screening; Malaysia; Kuala Lumpur; multi-ethnic society
JEL code(s)
J15, R21, C93, D83, J16
Secondary IDs
Prior work
This trial does not extend or rely on any prior RCTs.
Abstract
We test for ethnic and national-origin discrimination in the rental housing market of Wilayah Persekutuan Kuala Lumpur, Malaysia, using a paired correspondence design with a behavioural-intervention layer. Synthetic applicant inquiries varying ethnicity (Bumiputera, Chinese, Indian), nationality (Malaysian, Singaporean, Chinese, British, American), gender, and age are sent to 2,400 listings on Mudah.com
(private-landlord-dominant) and PropertyGuru (agent-dominant). A randomised information-disclosure treatment varies whether the inquiry includes additional applicant characteristics (occupation, length-of-stay intent, etc.).

The design tests three contributions:

1. Response-margin discrimination by ethnicity and nationality in a multi-ethnic, post-colonial federal-governance setting.
2. Behavioural-intervention efficacy: whether information disclosure narrows the discriminatory gap.
3. Screening-margin discrimination: using a novel Viewing-invitation-plus-screening outcome category to separate discrimination at the response margin from discrimination at the subsequent tenant-screening margin.

We also test whether discrimination travels through the agent-intermediary channel (statistical / customer-prejudice) or the landlord-principal channel (taste-based), exploiting the Mudah vs PropertyGuru platform split.
External Link(s)

Registration Citation

Citation
Bao, Helen and Nur Hafizah Binti Mohammad Ismail. 2026. "Discrimination, Information Treatment, and the Screening Margin in Malaysia's Rental Housing Market: A Correspondence Experiment in Kuala Lumpur." AEA RCT Registry. June 22. https://doi.org/10.1257/rct.18948-1.0
Sponsors & Partners
Experimental Details

Interventions

Intervention(s)
Each property listing on Mudah.com or PropertyGuru.com receives one or more inquiries from synthetic applicant profiles. Each inquiry manipulates applicant identity (ethnicity, nationality, gender, age)
and the information-disclosure treatment. Profiles are drawn from a stratified design.
Intervention Start Date
2025-01-15
Intervention End Date
2026-07-15

Primary Outcomes

Primary Outcomes (end points)
Response_received (binary; main response-margin outcome)
Viewing_unconditional (binary; main screening-margin outcome).
Primary Outcomes (explanation)
Response_received: Constructed from the categorical response field. Equal to 0 if response == "No response", else 1.
Viewing_unconditional: Defined only within the engaged sub-sample (rows where response ∈ {Viewing invitation, Viewing invitation + screening}). Equal to 1 if response == "Viewing invitation", 0 if response == "Viewing invitation + screening".

Secondary Outcomes

Secondary Outcomes (end points)
Response_time_hours (continuous).
Response (7-category)
LM_response
Secondary Outcomes (explanation)
Response_time_hours (continuous): (respond_1_time − application_sent_time) expressed in fractional hours. Defined only when both timestamps are present; missing otherwise.

Response (7-category): Used as the dependent variable in a multinomial-logit robustness specification (seven outcome categories: the six original manual codes plus the derived Viewing invitation + screening, minus the dropped Listing not available any more, plus No response).

LM_response (categorical; robustness cross-check): The prediction of the same character-n-gram TF-IDF + Logistic Regression classifier applied to respond_1_content, irrespective of whether respond_1_type is populated. Used in the robustness pass to confirm that conclusions do not depend on the human coding scheme. Missing when respond_1_content is empty.

Experimental Design

Experimental Design
Stratified factorial design. The Sampling Scheme sheet specifies the targeted number of observations per stratum. Within each stratum, the information-disclosure treatment is randomised.

Experimental Design Details
Not available
Randomization Method
Block-randomised within stratum at the property level. Each property is randomly assigned a small set of applicant profiles such that the information-treatment cells are balanced across applicant identity.
Randomization Unit
Property listing × applicant profile.
Was the treatment clustered?
Yes

Experiment Characteristics

Sample size: planned number of clusters
2400 property listings
Sample size: planned number of observations
4000 inquiries
Sample size (or number of clusters) by treatment arms
Information-disclosure treatment: ~50% Information, ~50% No information, balanced within applicant-identity stratum.
Minimum detectable effect size for main outcomes (accounting for sample design and clustering)
The wave-1 response-rate gap between Bumiputera-Malaysian and Indian-Malaysian applicants was 13.1 pp (p = 0.003). Power calculations on the full-sample target (~3,400 retained observations) indicate the design detects a 5-pp gap at α = 0.05 with power > 0.80 for the main ethnicity comparisons and the information-treatment heterogeneity terms.
Supporting Documents and Materials
IRB

Institutional Review Boards (IRBs)

IRB Name
Land Economy Research Committee
IRB Approval Date
2025-03-05
IRB Approval Number
DC05032025
Analysis Plan