Collinson et al. (2023)

The economics of Housing & Homelessness

Overview

Motivation

Eviction Filing Rates exceed 10% in some areas (2010)

Costs of Eviction

Implicit Index of Housing Issues

Interested in the Costs of Eviction for these tenants

Eviction Filed

Causal Inference Challenge

Tenant's who are Evicted face two "Issues"

The underlying source for the eviction filing which may have lasting impacts

(1)

The Eviction Judgement

(2)

The timing of these events are likely to be relative close

Causal Inference Challenge

Judge-IV

Implicit Measure of Rental Issue of Tenant

Judge A

Judge B

\overset{\textrm{No}}{\leftarrow}
\overset{\textrm{No}}{\leftarrow}
\overset{\textrm{Yes}}{\rightarrow}
\overset{\textrm{Yes}}{\rightarrow}

Exploit Random Assignment to Judges

Results

  • Likelihood of Moving increase by 8p.p
  • Staying at Emergency Shelters increases by 3.4p.p
  • Little change in neighborhood quality
  • Earnings per quarter decrease by $324 in the first year
  • Financial Health worsens with each passing years (1-6 years post eviction) 

Background

Definition of Eviction

  • "Last recorded outcome in the case history provides legal authority for the landlord to take possession of the property via an enforcement agent"
  • "Includes instances when a tenant fails to meet the terms of an initial settlement and the judge ultimately issues an eviction order"

Eviction Timeline

New York

Cook County

Tenants

Landlords

3\%
75\%
95\%
1\%

Legal Representation Rate 

  • There are 6 Districts in Cook County
  • Cases are filed in the district of the property
  • Cases are randomly assigned to courtrooms within the district

New York

Cook County

Census

Credit Reports

40\%
68\%
61.3\%
52\%

Data Linking Match Rate

Note: I really like when this is mentioned in the introduction. I think you don't want to present the reader with any "surprises" about the data. 

Baseline Characteristics

Reflect on why Diff-in-Diff is not likely to be a credible strategy in this setting

Analysis

(\mathbb{E}[D_i \vert X_i, Z_i] - \mathbb{E}[D_i \vert X_i])

Judge IV

0.09
0.03
-0.1

Source of Variation

Tenant

Judge

Controls include lagged dependent values, court by year F.E., back rent, race, gender, poverty rate, census tract rent

Consider: Why aren't the coefficients 1?

Which one is more important/interesting?

Results

Housing Stability

Should we be surprised that the OLS estimates don't differ from the IV estimates?

Employment

You would think that estimate of OLS is biased down...

\textrm{Selection Bias} := \mathbb{E}[\tilde{Y}_i(0) \vert \textrm{Evicted}] - \mathbb{E}[\tilde{Y}_i(0) \vert \textrm{Not Evicted}]

Financial Health

Potential Concerns

I would like to see something on the appearance rate

The Monotonicity Assumption

Latent Ranking of Tenants

  • One component of the monotonicty assumption is that we can represent each Judge's decision to evict a tenant along a consistent ordering of tenants.
  • Each judge, denoted by a color line, evicts tenants to the right of their line

Doesn't directly address

the monotonicity assumption

Comments/Thoughts

These are intended to be helpful to students reading the paper

"Measuring address-level moves at an annual frequency in the United States is challenging, particularly so for our population of unstably housed tenants. We believe these administrative data sets provide the best measures available."

Comment

  • This is great writing! 

"Although it is possible to outline assumptions under which bias is not a concern or under which the bias can be signed (see Heckman and Robb 1985), we instead rely on our quasi- experimental instrumental variables research design."

Comment

  • This is great writing! 

"Despite the large number of evictions and the growing policy interest, the consequences of eviction are not well understood."

Comment

  • I go back and forth on whether I like this sentence. 

Suggestion

  • Clarify that while some consequences are well understood (it increases housing instability) other outcomes warrant greater attention...

"In New York, this indicator additionally includes applications to shelter, which cover instances where families are diverted or deemed ineligible"

Comment

  • This is a great variable to have!

"We first show that tenants in our linked housing court sample differ substantially from randomly chosen tenants who live in the same neighborhoods."

Comment

  • To what extent is this because in New York, the matched sample only contains people with a public benefit history?

Presentation of Collinson et al. (2023)

By Patrick Power

Presentation of Collinson et al. (2023)

  • 178