Back to History Current Version

Debt after Death: Randomized Evidence on Emergency Financial Assistance for the Urban Poor

Last registered on March 13, 2023


Trial Information

General Information

Debt after Death: Randomized Evidence on Emergency Financial Assistance for the Urban Poor
Initial registration date
March 08, 2023

Initial registration date is when the trial was registered.

It corresponds to when the registration was submitted to the Registry to be reviewed for publication.

First published
March 13, 2023, 3:07 PM EDT

First published corresponds to when the trial was first made public on the Registry after being reviewed.


Primary Investigator

University of Notre Dame

Other Primary Investigator(s)

PI Affiliation
University of Notre Dame

Additional Trial Information

In development
Start date
End date
Secondary IDs
Prior work
This trial does not extend or rely on any prior RCTs.
Widespread financial fragility in the United States, in which one in three adults report they could not pay a $400 emergency expense with cash, could leave low-income households vulnerable to a cycle of debt that prevents them from being able to escape poverty. There is little causal evidence on the implications of unanticipated financial shocks for low-income individuals, and even less evidence on emergency financial assistance provided at the time of such shock. In this paper, we partner with a non-profit that provides financial assistance for burial and funeral (BaF) services to individuals with household incomes below 40% AMI in a large U.S. city. Nationwide the median cost of a funeral with viewing and burial is about $8,000. Even for low-income families, these expenses are hard to avoid for cultural and religious reasons. We estimate that 1 in 50 households experience a death each year. Using a randomized controlled trail, two-thirds of study participants will randomly receive $1000 towards BaF expenses, while the other third will receive a higher amount of financial support (to be revealed publicly at study completion). Using administrative records, we will passively track impacts on credit outcomes, housing stability (moves, evictions, and interactions with homelessness services), and earnings and employment. We anticipate that funding will support a sample size of 672 participants, enabling us to detect a 0.20 standard deviation change in a standardized credit index. We will also explore an event-study analysis and (potentially) a matched-sample event-study approach, to estimate the impact of the shock itself with little or no financial assistance.
External Link(s)

Registration Citation

Batistich, Mary Kate and Jonathan Tebes. 2023. "Debt after Death: Randomized Evidence on Emergency Financial Assistance for the Urban Poor." AEA RCT Registry. March 13.
Sponsors & Partners

There is information in this trial unavailable to the public. Use the button below to request access.

Request Information
Experimental Details


Intervention Start Date
Intervention End Date

Primary Outcomes

Primary Outcomes (end points)
- Standardized financial distress index that combines a "credit access" index, a delinquency index, and (if available) a payday borrowing index.
- An indicator for having a higher financial distress index than in the year prior to receiving assistance.
Primary Outcomes (explanation)
We are primarily interested in understanding how emergency financial assistance may protect low-income households from experiencing financial distress and the downstream consequences of that financial distress. Our main outcome will therefore aggregate rich consumer credit data into a singular measure of financial distress. Our main index will combine three separate indices -- a credit index, a delinquency index, and a payday borrowing index (if available). These indices are adapted from Miller and Soo (2021) and Collinson et al. (2022). Together, they provide a holistic measure of financial distress. The exact variables that enter these indices may also change as we learn more about data availability, quality, and improvements in financial distress measurement.

Each component of the index will be standardized based on the control mean and standard deviation in the year prior to receipt of financial assistance. Each component will then be summed, assigning directions to components such that they appropriately point in the direction of financial distress, and then we will re-standardize the aggregated index using the control group mean and standard deviation in the year prior to receipt of financial assistance. All other standardized indices described in below will be constructed in a similar manner. We denote the sign of how each index will enter the financial distress index in parentheses.

Credit Access Index (-)
- Credit score estimated using VantageScore
- Indicator for any positive balance on an auto loan or lease, a proxy for durable good consumption
- No open source of revolving credit, a proxy for having limited access to credit (-)
- Total amount of credit available on all accounts (or only across all active credit cards, if these data are more reliable)

Delinquency Index
- Amount held 30 days past due or more on all open accounts (e.g., overdue credit bill)
- Amount in tax liens or overdue taxes
- Amount ordered to be paid by a court judgment (e.g., unpaid rent cases or child support)
- Debt past due held by third-party collection agencies

We would like to collect information on payday loans from Clarity. Clarity maintains the largest database of subprime borrowers, containing over 62 million unique consumers and covering about 70% of payday loans. It is unclear whether these data will become available for this research. If they are, we will construct an index using some or all of the following variables (depending on data availability and quality).

Payday Borrowing Index
- Amount borrowed across all payday loans
- Amount borrowed across payday loans taken out online
- Payday amounts borrowed in person at physical storefronts
- Any payday loan inquiry
- Number of payday loan inquiries
- Any payday loan
- Number of payday loans


Collinson, R., J. E. Humphries, N. S. Mader, D. K. Reed, D. I. Tannenbaum, and W. Van Dijk (2022). Eviction and poverty in american cities. Technical report, National Bureau of Economic Research.

Dobkin, C., A. Finkelstein, R. Kluender, and M. J. Notowidigdo (2018). The economic consequences of hospital admissions. American Economic Review 108 (2), 308–52.

Miller, S. and C. K. Soo (2021). Do neighborhoods affect the credit market decisions of low-income borrowers? evidence from the moving to opportunity experiment. The Review of Financial Studies 34 (2), 827–863.

Secondary Outcomes

Secondary Outcomes (end points)
- Housing instability index
- Labor market index (if available)
- Index summarizing total impact across all three indices (financial distress, housing instability, and labor market outcomes)
- Indicators for increases in each of the above indices relative to levels observed in the year prior to assistance receipt.
Secondary Outcomes (explanation)
In addition to our financial distress, we will explore two other outcome domains -- housing instability and labor market performance. To reduce the concern of multiple hypothesis testing, we will combine all of our three outcome indices -- financial distress index, housing instability index, and labor market index -- into a central standardized index, applying equal weight to each index. (Our weighting scheme may change if future methods suggest a more efficient way to weight outcomes that adjusts for differential noise/power across outcomes.) To understand what drives aggregate effects, we will also report effects for each index and its sub-components, although point estimates on sub-components should be viewed as suggestive. All of the below measures will be captured over the two years after the caller receives financial assistance. Data sources are denoted in brackets.

Housing Instability Index
- Any eviction filing [Eviction Court Records]
- Was ever evicted [Eviction Court Records]
- Number of address changes in Infutor records [Infutor]
- Any move to a lower-quality neighborhood, as measured by a composite neighborhood quality index (We will generate a composite neighborhood quality index using five-year ACS measures of median household income, poverty rate, % receiving SSI, % with HS degree or lower, and % single-parent headed households, as well as publicly-available information on neighborhood crime.) [Infutor, ACS, and Crime Reports]
- Any interaction with other homelessness prevention services [HMIS]
- Any stay in emergency shelter [HMIS]
- Number of interactions with homelessness services (or fraction of days spent in shelters if data are available) [HMIS]

Labor Market Index
- Quarterly earnings [State UI Records]
- Indicator for having any positive earnings. [State UI Records]

Planned Sub-Group Analyses:
We will also examine heterogeneity of effects on the listed primary and secondary outcomes by baseline levels of reported household income (or earnings if data become available), financial distress, and the labor force status/age of the deceased (as a proxy for additional hardship imposed on the family by the death outside of the funeral bill).

Experimental Design

Experimental Design
Individuals seeking financial assistance with a funeral or burial will call into the CCAC hotline and will be directed to the funeral and burial assistance team. The client will be told about the burial assistance program, the research study, and what documents must be provided for participation. These include finalized invoices for burial and funeral expenses, a document confirming household income, a signed consent form to participate in the program, and a voluntary consent form to participate in the study if they are deemed eligible. With this paperwork complete, the intake staff will complete a short Qualtrics survey over the phone. This survey consists of two sections: eligibility and intake.
There will be two separate components to the eligibility screen: eligibility for the program and eligibility for the study.

Program Eligibility
- Callers will be eligible for the burial assistance program if they meet the following criteria:
- Have household income at 40% AMI or below
- Deceased is not an honorably discharged veteran (honorably discharged veterans will be referred to the Veterans Funeral Assistance program)
- Have an invoice of $12,000 or below

CCAC will confirm callers’ income by asking them to provide proof of income (for example, pay stub, benefit letters, letters from employer, etc.), or, if they do not have proof of income, asking them to complete a self-attestation form. CCAC will ensure that they have a copy of a caller’s invoice and that the invoice is within the program’s eligibility range before they begin intake with the caller. In addition to these forms, callers will complete CCAC program consent forms and consent forms for participation in the research study.

Implementing this income threshold will allow us to target the program to those who are most in need. Past research has shown that people with income in the lowest quartile see the most significant impacts of financial shocks. However, because this financial shock is larger than shocks in previous research, we expect there to be a wider population that experiences prolonged negative financial outcomes. In particular, we have chosen the cutoff of 40% AMI because past program experience has shown that 30% AMI cutoffs make it difficult to recruit enough people into programming, while 50% AMI cutoffs screen out very few people. We have chosen 40% AMI as a midpoint between these two, and we will closely monitor intake data at the beginning of the study to determine if this threshold needs to be adjusted.

The goal of the $12,000 upper bound is to prevent very large bills, and if a caller has a bill above this threshold, CCAC will work with the caller to determine if there is anything that can be done to reduce the bill. Placing this limit on the size of invoices that are eligible for the program will also serve as a second way of ensuring that the program is targeting those most in need, in addition to the income requirement. $12,000 was chosen as the cutoff as it was approximately equal to the largest historical requests.

Clients who are eligible for the program will complete the intake portion of the Qualtrics form with a CCAC staff member by phone. This form will collect baseline demographic information on callers, their household, and the deceased, including their race, gender, age, and income.

Study Eligibility
- Callers will be eligible for the study if they meet the following criteria:
- Are eligible for the burial assistance program
- Consent to participate in the research study
- Have an invoice of $3,000 or more

The requirement that invoices must be $3,000 or more for the client to be eligible for the study serves two purposes. First, it ensures that the program is targeted towards those who are experiencing a more significant financial shock (and thus have higher need). Second, the cutoff increases statistical power by ensuring that there is a substantial difference between the financial assistance provided to treatment and control groups.

Those who did not consent to the study, including those who are ineligible for the study, will automatically receive $750 of assistance, while those who did consent will be randomized to receive either $1,000 (control group) or a higher amount of financial assistance (to be revealed at study completion).

All program-eligible callers will receive a minimum of $750, including those who do not consent to the study. This amount is large enough to provide substantive benefit to those who do not consent to participate, but also is small enough that the majority of limited resources go towards those in the study. To prevent coercion, we set the difference between the payments of the control group and non-consenting group at $250. Importantly, all enrollment and consenting materials do not mention the potential of the higher treatment group payment, and instead informs potential participants that they may receive an additional $250 or more if they participate in the study. This design was approved by the Institutional Review Board at the University of Notre Dame. This design allows us to test a high “dosage” of financial support that eliminates debt for the overwhelming majority of the treatment group. Inclusion of an upper bound prevents very large individual payments and allows limited financial resources to assist as many people as possible.
Experimental Design Details
Not available
Randomization Method
Computer random number generator
Randomization Unit
Was the treatment clustered?

Experiment Characteristics

Sample size: planned number of clusters
Sample size: planned number of observations
672 individuals in panel data; observation count will depend on longitudinal nature of data.
Sample size (or number of clusters) by treatment arms
We anticipate 224 individuals to be randomized to Treatment and 448 individuals to Control. The exact number will depend upon the fraction of callers who consent to participate in the research (assumed 85%), the fraction who take-up the treatment when offered (assumed 90%), and the average bill amount among the treatment group (to be revealed afters study completion). We will enroll participants until June 30th, 2023 or until funding for the program is exhausted.
Minimum detectable effect size for main outcomes (accounting for sample design and clustering)
We expect that 85 percent of callers will meet the study eligibility criteria and consent to participate, and that 15 percent of callers will not be in the study and will receive $750. The control group will receive $1,000 and we project, based on historical payment data, that the treatment group will have a higher average payment (to be disclosed after study completion). Since total funding for this project is fixed, we improve power by increasing the size of the less-expensive control group; in particular, we will randomize one-third of participants into the treatment group and two-thirds into the control group. Given these assumptions, we anticipate having 224 households in the treatment group and 448 households in the control group for a total sample size of 672 households. Assuming 80% statistical power and a 90% post-randomization take-up rate, this sample size yields a minimal detectable effect of 0.204 standard deviations on a standardized credit index. In terms of a binary outcome (for example, the likelihood that an individual experiences an increase in their financial distress index relative to baseline) the minimum detectable effect size is 11.4 percentage points, assuming a base rate of 50%. These ranges are similar to effects found in related economic shock literature (Collinson, 2022; Mello, 2021; Dobkin, 2018). The shock that we are studying is a greater magnitude than some of the prior literature (e.g., Mello (2021) saw an average of $190 vs. a significantly larger shock in this study (to be revealed at study completion)), so we expect to see larger effect sizes. While we do have concerns about our power given this sample size, we believe that this payment structure gives us the best chance of having statistical power, as it will maximize the intensity of the treatment within the budget constraints of the program.

Institutional Review Boards (IRBs)

IRB Name
University of Notre Dame Institutional Review Board
IRB Approval Date
IRB Approval Number