Housing Stabilization Initiative – Program Evaluation

There are two treatment arms: 1) an intensive housing problem solving program with flexible financial assistance and housing focused case management, and 2) a minimal intervention strategy: housing problem solving (HPS). With excess demand, there is also a non-zero likelihood of a no intervention control group, whereby those seeking assistance are not able to be served in a timely manner (due to appointment availability) and will be referred to other continuum of care resources/organizations.

2025-11-18

2026-03-31

We measure the impact of the housing initiative on eviction, housing stability, entry into a homeless shelter, and duration of shelter stay.

Eviction is measured by observed eviction filing and eviction judgement (continuance, writ of restitutions, and request for set out) data.
Homelessness is measured by entry to shelter or unsheltered literal homelessness as observed in the local area’s HMIS (Homelessness Management Information System) data. If homelessness shelter entry is observed, the duration of shelter days will also be examined.
Housing stability is measured by address changes and self-reported results-oriented management and accountability measures of housing stability.

Secondary outcomes include pre- and post-program rent burden, location stability (same place/neighborhood), self-reported measures of wellbeing (personal health and financial health), employment (employed and hours), and even earnings.

Rent-burden is calculated as the ratio of housing costs (rent) relative to gross earnings, low income individuals with high rent burden are at greater risk of falling into arrears as they have less slack in their personal finances to weather temporary negative shocks.

Locational stability is measured by whether individuals are in the same neighborhood or place.

Self-reported measures of wellbeing are measured using a 1-10 scale where by lower scores indicate acute crises and higher score imply flourishing.

Employment and earnings are collected as part of program administration.

We also hypothesize there may be differences in the program impact depending on their individual situation, so we will seek to control for demographics and other socioeconomic characteristics such as household size, labor market participation, enrollment in benefits, area median income, perceived causes of the housing crisis, and previous experiences of literal homelessness.

This randomized research design has two treatment arms and, if/when there is excess demand, a no intervention control group. Individuals are randomly allocated to one of the two treatments, with stratification between those already displaced and those facing eviction. The two treatment arms include 1) an intensive case management and financial assistance service and 2) a low-cost light touch alternative intervention: access to housing problem solving. If/when there is excess demand, i.e., there are no timely appointments available for either treatment arm some individuals will be left untreated and comprise a no intervention control group.

Not available

Randomization done in office by a computer.

Randomization occurs at the individual level with stratification between imminent and upstream risk.

No

This is not a clustered study. There is one cluster- the full study.

Up to 2000 households/individuals requesting housing stabilization support services.

Treatment 1: an intensive case management and financial assistance service – 500 individuals/households served.
Treatment 2: low-cost light touch alternative intervention: access to housing problem solving – 500 individuals/households served.
Up to 1000 in no-treatment control

We have also conducted a statistical power calculation, given a 9% eviction filing rate (data from the Eviction Lab), with a sample size of 2000, a type I (false positive) error rate of alpha=.05 (5%), 80% power (type II error of 20%), and 1000 receiving the intervention, the smallest reduction in evictions we can detect between treatment and control is 3.27 percentage points, a 36% reduction, which would be both economically and statistically significant/meaningful. Within treatment arms, our service delivery partner has shared that historically residents who have experienced housing problem solving have had a 5% rate of eviction. Therefore, with a treated sample size of 1000, a type I (false positive) error rate of alpha=.05 (5%), 80% power (type II error of 20%), and 500 receiving the more intensive intervention, the smallest reduction in evictions we can detect between the more intensive and less intensive treatment is 3.21 percentage points, this is a 64% reduction, which would be both economically and statistically significant/economically meaningful.