Estimating the Human Costs of Debtor Prisons

How to cite

Osman, Adam; Haggag, Kareem; Nagy, Abdelrahman; Ahmed, Ahmed. 2025. "Estimating the Human Costs of Debtor Prisons." AEA RCT Registry. AEARCTR-0016609. https://doi.org/10.1257/rct.16609-1.0

Download: BibTeX · RIS — for reference managers without a browser translator. Zotero and most others pick up this page's citation metadata directly.

Adam Osman* , Kareem Haggag , Abdelrahman Nagy , Ahmed Ahmed

* Primary investigator

Registry ID
AEARCTR-0016609
DOI
https://doi.org/10.1257/rct.16609-1.0
First registered
Last updated
Status
In development
Trial dates
2025-10-01 – 2028-08-30
Countries
Private
Registry record
http://www.socialscienceregistry.org/trials/16609

Abstract

We aim to provide causal evidence on the social costs of debtor prisons and on the financial market consequences of its potential abolition in Egypt. To do so, we will conduct a set of randomized controlled trials. The first randomized experiments will be with an NGO that pays off the debts of prisoners (and those facing prison), thereby releasing them. This will provide a rigorous estimate of how debtor prisons affect the lives and debtors and their families. Second, we will run an experiment with a microfinance institution to estimate the extent of moral hazard in credit markets. Together we will be able to provide the first estimates on the benefits and potential costs of abolishing debtor prisons as an institution.

Intervention

Experiment 1: Intervention 1: Debt repayment as a grant, Intervention 2: Debt repayment as a loan

Experimental design

Experiment 1: Treatment 1 will receive debt repayment as a grant, Treatment 2 will receive debt repayment as a loan. Control will receive it at a later date if funds permit.

Experiment 2: We will vary the way we will advertise to people about the opportunity to take out a new loan. For some individuals, a letter will note that for this loan the bank will not be reporting delinquent clients to the criminal authorities and instead will work with them to make sure they repay on time (group 1). Others will just receive a message about a normal loan (group 2).

We will then run a second level of randomization. Those that were told there would be no prosecution of delinquency will be split into a group that is only told that (group 1a) and another group that is told that as well as told about the reporting to a credit bureau (group 1b).

Those that were initially sent a generic letter will be split into three groups, a group informed about no prison-risk group (group 2a), and a group informed about no prison + credit reporting (group 2b), and a normal group who is not informed about the change in prison risk (group 2c). None of these clients will face prison-risk if they default, even if they are not initially informed of this.

Design details

Randomization method
Experiments 1 & 2: Randomization done in office by a computer
Randomization unit
Experiments 1 & 2: Individual level randomization
Clusters
Experiment 1: 3000 debtors Experiment 2: 2500 potential clients
Observations
Experiment 1: 3000 debtors Experiment 2: 2500 potential clients
Arms
Experiment 1: 1500 treatment debtors, 1500 control debtors Experiment 2: 500 potential clients in each treatment arm
Primary outcomes
Experiment 1: Employment, income, consumption, education of the children, and psycho-social well-being. We intend to collect this for the debtor and other members of the household. We are particularly interested in the outcomes for children in the household, who we suspect may be significantly impacted given their vulnerability. Experiment 2: Application & Default Rates

Cited by

No citation data available for this registration yet.

Keywords