Methodology: Eviction rates in NYC down since COVID-19
This story is written with Mariam Lobjanidze
The story lives here.
In this project, our main goal has been to understand why eviction rates are down compared to pre-pandemic levels even though the initial moratorium is not in place anymore. To deepen the analysis, we also looked at how pre and post-pandemic numbers are different for different races. We used data from NYC Open Data and Census data and by merging those, we were able to see numbers by race, poverty levels, and income levels. We talked to three experts who explained trends and the difference in numbers. However, they were not able to explain smaller details about race as many housing policy analysts do not track data by race.
Eviction rates in our analysis were calculated by dividing eviction numbers by population. However, Another point that came up during our reporting was whether we should use eviction filings or executed evictions since those are two distinct things—not all eviction filings get executed. When we talk about the fact that eviction rates are down, we are talking about executions specifically. However, we used both execution data from open data and the filing data from the eviction lab to understand how filings and executions have changed for each race.
The following is our memo regarding our initial findings and analysis:
The Tenant Safe Harbor Act, in effect since March 2020, protects New York City tenants from eviction who fail to pay their rent due to COVID-19. After the Act ended in January 2022, the number of eviction cases almost returned to pre-covid levels.
However, by comparing the data in 2019 and in 2023, we found racial disparity in the change of evictions. Linear regression shows that change number and change rate for eviction both have correlation with the proportions of Hispanic and Black tenants.
In census tracts with a majority of Hispanic tenants, the number of evictions carried out after the epidemic is the closest to pre-epidemic cases. Eviction rates fell the most in areas with the most white tenants.
For areas that had the same eviction number in 2019, the eviction number in 2023 for hispanic dominated area can be 100% higher than white, while blacks 60% higher than white, according to the linear model.
Further investigation reveals that the eviction change is unrelated to financial factors such as the median income and median rent of the census tract.
The overall trend of fewer evictions post-covid can be explained in part by the city's offer of free legal aid to tenants, according to Peter Hepburn, a statistician and quantitative analyst at Princeton University Eviction Lab.
The free assistance in housing court was first offered to low-income tenants in 2017, when New York City passed the nation's first right to counsel law. It was piloted in ten zip codes and expanded throughout the city during the pandemic.
The increased involvement of lawyers has slowed down the court process. Even if the number of eviction filing remained the same, the number of tenants executed in 2023 has been delayed or reduced for this reason, said Hepburn.
In theory, areas with more eviction filing rates would have higher execution rates. Individual filed records from the New York Housing Court are not available, but the Princeton Eviction Lab has compiled data at the zip code level for one year starting April 1, 2023. We were able to compare their filing rate with the execution rate in the same time period. Once again, differences between races became apparent.
The relationship is essentially linear in areas with the highest concentrations of Hispanic or white tenants. However, there is no obvious correlation in areas where black tenants predominate, which may suggest systemic biases in how eviction laws and practices are applied, potentially disadvantageous to black tenants.
Low-income areas tend to receive more legal assistance. Linear regression shows that the execution among the filing cases in the same year has a strong correlation with median income, even when controlling for racial factors.
