Association for Public Policy Analysis & Management

In most U.S. cities, housing rents are rising much faster than incomes, particularly at the lower end of the income scale, leading to renters devoting record amounts of their income to housing costs. In many cities, it is very typical for low-income households to spend more than 50 percent of their incomes on rent. Groundbreaking research by Matthew Desmond highlights a tangible consequence of extreme housing cost burdens – the staggering rise in evictions across the country. Using data from his fieldwork in Milwaukee, Desmond reports that every year approximately 16,000 people are evicted in that city of only 105,000 renter households.

Data on evictions are sporadic and incomplete. Eviction records are not reported on a national basis or in any systematic way across localities (Hartman and Robinson 2003). Prior research often relies on data in a particular locality from examining court orders, conducting courtroom surveys, or administering neighborhood surveys (Bezdek 1992, Eldridge 2001, Desmond 2012). Although the American Housing Survey (AHS) asks questions about involuntary moves, scholars assume that these provide substantial undercounts, as they rely on open-ended questions that are likely to be misunderstood by renters in the context of eviction (Desmond and Shollenberger 2015).

Given the main obstacle to eviction research is the lack of data, we have embarked on a data collection effort that proceeds on two fronts. The first is that we have obtained a data set from American Information Research Services, Inc (AIRS), a tenant-screening and public records data warehousing company. This data set consists of publicly available U.S. eviction court records for most eviction courts in the U.S., going back as far as 1997. The second is that we are collecting additional information that AIRS does not collect on unsealed eviction court records from the Los Angeles Superior Court website using web scraping. These records are publicly available through the court's Case Summary Database, and we are augmenting the web scraping with a manual data collection effort where information exists in paper form only.

We utilize these data to first assess the spatial concentration of eviction in Southern California, and compare the spatial clustering of court-based evictions to those that arise from condo conversions (known in California as Ellis Act evictions). Second, using the data collected in Los Angeles County, we are able to identify tenant characteristics (e.g. race, gender, legal representation) associated with eviction outcomes. Third, we append eviction locations to neighborhood data during several time points, to better understand the connection between neighborhood change and eviction. Further, we zoom out at the city and county level to better understand the connections between housing markets and eviction. Through these analyses, we develop a model that describes the determinants of eviction rates in California cities and counties, as well as Southern California neighborhoods.