Online rental housing market representation and the digital reproduction of urban inequality

Abstract

As the rental housing market moves online, the Internet offers divergent possible futures: either the promise of more-equal access to information for previously marginalized homeseekers, or a reproduction of longstanding information inequalities. Biases in online listings' representativeness could impact different communities' access to housing search information, reinforcing traditional information segregation patterns through a digital divide. They could also circumscribe housing practitioners' and researchers' ability to draw broad market insights from listings to understand rental supply and affordability. This study examines millions of Craigslist rental listings across the US and finds that they spatially concentrate and over-represent whiter, wealthier, and better-educated communities. Other significant demographic differences exist in age, language, college enrollment, rent, poverty rate, and household size. Most cities' online housing markets are digitally segregated by race and class, and we discuss various implications for residential mobility, community legibility, gentrification, housing voucher utilization, and automated monitoring and analytics in the smart cities paradigm. While Craigslist contains valuable crowdsourced data to better understand affordability and available rental supply in real-time, it does not evenly represent all market segments. The Internet promises information democratization, and online listings can reduce housing search costs and increase choice sets. However, technology access/preferences and information channel segregation can concentrate such information-broadcasting benefits in already-advantaged communities, reproducing traditional inequalities and reinforcing residential sorting and segregation dynamics. Technology platforms like Craigslist construct new institutions with the power to shape spatial economies, human interactions, and planners' ability to monitor and respond to urban challenges. Prospective renters heterogeneously rely on a constellation of information sources to identify available units, including websites, newspapers, agents/brokers, property for-rent signs, and word of mouth. Krysan and Crowder (2017) theorize a two-stage search process in which homeseekers decide which neighborhoods to search and then choose which units within them to consider, emphasizing the importance of information supplies. Rae (2015) contends that the Internet has emerged as the first port of call for such searches, but online information also impacts the second stage by rendering individual units and in turn neighborhoods more legible and accessible to seekers. As the Internet constitutes an ever-increasing share of the rental housing information supply, different communities' access to information and capacity to find housing depend both on listings' representativeness and communities' abilities/interests in engaging with these platforms. This can be considered from the supplyside (i.e., the supply of information by landlords, managers, and brokers) and the demand-side (i.e...

The Unequal Availability of Rental Housing Information Across Neighborhoods

“…Following grow ing rec og ni tion of the value of data col lected online for under stand ing demo graphic pro cesses (Cesare et al 2018), we exam ine adver tise ments col lected from Craigslist. Not all rental hous ing in the United States is adver tised on Craigslist; indeed, Boeing (2020) found that adver tise ments for rental hous ing on Craigslist in Unequal Availability of Rental Housing Information 2014 were over rep re sented in neigh bor hoods with higher shares of White res i dents, dem on strat ing how offline inequalities are reproduced in the sup ply of infor ma tion online. However, our goal is to under stand what kind of infor ma tion is shared on Craigslist, the most com pre hen sive and timely source of hous ing mar ket infor ma tion in the United States (Boeing and Waddell 2017;Kuk et al 2021).…”

“…10 We exam ine these var i ables, which mea sure either the qual ity of the units in a neigh bor hood or neigh bor hood demo graph ics, because they could plau si bly affect the ways land lords list their units. For exam ple, a more highly edu cated renter pool could prompt land lords to adver tise cer tain types of ame ni ties, whereas a higher vacancy rate might cre ate more com pe ti tion for poten tial rent ers and prompt land lords to add more infor ma tion to their list ings (Boeing et al 2020).…”

“…Recent work sug gests that infor ma tion presented in online hous ing adver tise ments indeed varies by neigh bor hood. For exam ple, Craigslist adver tise ments from poorer neigh bor hoods and neigh bor hoods with more Black res i dents con tain fewer words, on aver age, and are less likely to state an exact address for the rental (Boeing et al 2020). In addi tion, the prev a lence of Craigslist list ings rel a tive to the under ly ing hous ing stock is greater in neigh bor hoods with more White, higherincome, and more edu cated res i dents (Boeing 2020).…”

“…For exam ple, Craigslist adver tise ments from poorer neigh bor hoods and neigh bor hoods with more Black res i dents con tain fewer words, on aver age, and are less likely to state an exact address for the rental (Boeing et al 2020). In addi tion, the prev a lence of Craigslist list ings rel a tive to the under ly ing hous ing stock is greater in neigh bor hoods with more White, higherincome, and more edu cated res i dents (Boeing 2020). Critically, how ever, lit tle work has exam ined the main tex tual descrip tion of list ings, with the excep tion of Kennedy et al's (2021) recent study.…”

“…Second, in ana lyz ing an increas ingly impor tant source of infor ma tion for homeseekers, we empha size that mobil ity deci sions are partly a prod uct of the sup ply of infor ma tion on avail able units. To under stand the sup ply of infor ma tion, we col lected 1.6 mil lion geocoded adver tise ments for rental hous ing from the 50 larg est met ro pol i tan areas in the United States posted on Craigslist, the dom i nant plat form for today's met ro pol i tan rental hous ing mar ket (Boeing 2020;Boeing and Waddell 2017). 1 Using com pu ta tional text anal y sis tech niques, we first iden tify com mon types of infor ma tion displayed in online hous ing adver tise ments.…”