Authors: Helen X. H. Bao
Year: 2026
PDF: Download
Research line: Exclusion and spatial inequality in housing markets
One-line summary
Proposes a tripartite classification of housing-discrimination targets (colonial / graylined / foreign), anchors each category in a distinct mixture of Becker / Phelps-Arrow / Akerlof-Kranton mechanisms, and provides a meta-regression identification test on 54 correspondence-test studies across 22 countries showing that each category responds to its theoretically-paired country-level fractionalisation moderator.
Abstract
Correspondence-test studies of housing discrimination have accumulated steadily over the past two decades and consistently document that minority-named, immigrant-named or religiously-marked applicants receive fewer responses to rental inquiries than otherwise-comparable majority applicants. The economic theory of discrimination distinguishes three canonical mechanisms underlying this pattern: taste-based (Becker, 1957), statistical (Phelps, 1972; Arrow, 1973) and identity-driven (Akerlof and Kranton, 2000). We propose that the published correspondence-test record can be organised into three analytical categories of discrimination targets — colonial, graylined and foreign — each theoretically anchored in a different mixture of the three economic mechanisms. We extend the cross-study sample to fifty-four correspondence-test studies across twenty-two countries with fieldwork between 2003 and 2022 and provide a meta-regression identification test of the three-category reading. Each category responds to its theoretically-paired country-level diversity index: the colonial-by-ethnic-fractionalisation, graylined-by-religious-fractionalisation, and foreign-by-linguistic-fractionalisation interactions are each positive, statistically significant, and consistent in sign with the underlying mechanism. The pattern is robust to a FAT-PET correction for publication selection. Recent fieldwork from Singapore, Kuala Lumpur and Malaysia is consistent with the identity-economics reading and shows that the same identity can carry the discriminator role in one jurisdiction and the target role in another.