What Landlords Say They Do: Algorithmic Recommendations and Categorical Screening Criteria

This study examines landlord screening behavior, namely landlords’ use of algorithmic recommendations and their own categorical screening criteria, using surveys of 591 landlords in California, Georgia, and North Carolina. We find that 62.5% of landlords receive algorithmic recommendations and about one in five defer to them entirely, while most retain some discretion. More than 40% report blanket denial rules, and thresholds such as no rental debt and no felony conviction are each endorsed by a third or more of landlords. Blanket denial is most common among smaller landlords and in majority white neighborhoods, while mid-size landlords are most likely to adopt algorithmic services. These patterns suggest that these services add a lower-friction infrastructure that makes it easier to apply the rules landlords already favored, rather than new or more consequential information they could not have assembled alone. Because these categorical rules recur across criterion domains and appear to operate largely independently of what screening services deliver, we argue that policy attention belongs on the categorical rules themselves and on the conditions that lead smaller landlords to adopt them.