For whom does the phone (not) ring? Discrimination in the rental housing market in Delhi, India

Using an audit experiment carried out on of India's largest real estate websites, we document striking variations between landlords' treatment of upper-caste Hindus, Other Backward Castes, Scheduled Castes, and Muslims. We find strong evidence of discrimination against Muslim applicants, both in terms of probability of being contacted and the number of contacts, relative to upper-caste Hindu (UC) applicants, in the rental housing market in Delhi and its largest suburbs. While the probability that a landlord responds to an upper-caste applicant is 0.35, this is only 0.22 for a Muslim applicant. We also find suggestive evidence that when landlords respond to both UC and Muslim applicants, they call back the UC applicant sooner. Muslim applicants are especially disadvantaged when applying to rent one-bedroom houses; there is an additional 20 percentage points reduction in the probability of a callback. In contrast, we find no clear evidence that landlords are less likely to respond to Scheduled Castes and Other Backward Classes. However, our estimates may understate the true differentials in callback ratios as a result of our failure to perfectly link all callbacks to a listing.