Patrick Power PRO
Economics PhD @ Boston University
Hungerman et al. (2024)
The economics of Housing & Homelessness
THe Essence of the Paper
"Our results indicate that any income effect of temporary
transfers for those in crisis is minimal and that these transfers may convey labor market benefits for the poorest of the poor"
Framing
Note
Please note, I am highly opinionated and many reasonable people will disagree with me.
The point of this presentation is not to criticize a paper so to illustrate how I read/analyze a paper so that you can better learn to analyze papers
Put simply - it's about calling a spade a spade
I would push back on the framing of the paper.
"A simple labor supply model predicts that a pure non labor income transfer such as emergency financial assistance would reduce labor supply if leisure is a normal good"
This is a "Strawman Model" for this context
No one would expect this to happen in this context. We're not worried about decreases in the labor supply from the provision of essentially one months worth of rent for those in need
"Generally, we can reject that earnings fall by
more than 3 percent of their pre-transfer average."
I don't personally find the following a meaningful result
Background
Emergency Financial Assistance (in this context)
Facing a housing crisis
Crisis can be solved with limited financial assistance
Caller has sufficient future income to pay additional expenses
If an individual is eligible and funds exist, then they are referred
Emergency Financial Assistance (in this context)
"Not all callers who are initially deemed eligible and offered assistance ultimately receive assistance"
Caller does not follow through with the application process
The delegate agency determines that the caller is not eligible.
1)
2)
They receive assistance from some entity besides the HPCC.
"Callers who are initially denied funds may subsequently receive assistance"
They call the HPCC back and ultimately receive assistance
1)
2)
Baseline Comparisons
Timeseries
Data
Take-up
A 2011 descriptive study showed that 71% of HPCC callers received or expected to receive assistance within 7 days of the initial call
The authors observe that 5.4% of people call back and are later referred
The 2011 study found that 8.4% of callers who are turned away are assisted elsewhere
Suggests that the effect of referral on assistance is 0.59
Analysis
Identification Concerns
Selection into the sample of callers
Selection into the sample of callers who get referred
Baseline Differences of Callers
by Referral Status
Baseline Differences of Callers
by Referral Status
Regression Specification
Y_i = \alpha _i + \textrm{Funds}_i \beta + X_i \Gamma + Z_i \Pi + \varepsilon_i
Y_i = \beta \big(\textrm{Funds}_i - \mathbb{E}[\textrm{Funds}_i \vert X_i, Z_i] \big) + \eta_i
What they estimate
How to interpret it
Z_i + X_i := \textrm{Caller Characteristics}
Results
The apparent pre-trend effects make this difficult to interpret
The apparent pre-trend effects make this difficult to interpret
We should applaud the authors for not cutting the data at two years post treatment. If they did so, the effects would seem larger than they appear to be over four years
Final Thoughts
I disagree with the framing of the paper
I think the authors should be applauded for their transparent analysis
I think some people in their position wouldn't have been so forthright -
" In the overall sample, whether being referred to assistance increases earnings is sensitive to the specification."
1)
2)
" We can reject the hypothesis that being referred to funds leads to a decline in earnings, which is an important result."
Presentation of Hungerman et al. (2024)
By Patrick Power
