Abstract
We provide guaranteed (unconditional) income (monthly transfers) to disadvantaged individuals living in rural and semi-rural areas of Anderson City, South Carolina. Individuals in a treatment group receive 200 dollars a month for 24 months, beginning on August 25, 2023. Individuals in a control group receive 10 dollars in the same time frame.
We enrolled individuals in the summer of 2023. They were all beneficiaries of different programs implemented by Anderson Interfaith Ministries (AIM). We first canvassed the participants in person while they were at AIM to participate in the different programs provided there (e.g., food pantry and financial literary classes). We made appointments with them so they could answer a baseline survey. At the time of the baseline survey, Spero Financial, a local credit union that partnered with us for this project, opened checking accounts for the participants of the baseline survey.
We collected an initial sample of 100 individuals. We required that their total annual household income be at most 60,000 dollars. All 100 participants agreed to participate, opened a checking account with Spero and signed several agreements so that we could use their data. These agreements included their permission to obtain their data for ten years starting on August 25, 2023, from different agencies in South Carolina (data are collected and linked by the South Carolina Revenue and Fiscal Affairs Office). They agree that their data be linked through their social security numbers and other demographic information. They also agreed to be linked manually to real estate, credit score, and criminal records sources using social security numbers and demographic information.
The monthly transfers are deposited to the treatment and control individuals in the checking accounts opened for them by Spero. The participants agreed that Spero share with us data on their usage of the debit card linked to their checking accounts.
The South Carolina Revenue and Fiscal Affairs Office accepted our proposal to link the participants to the data provided by them, which include quarterly wages (provided by the Department of Employment and Workforce), benefits received from SNAP and TANF (provided by the Department of Social Services), data on hospitalization and corresponding medical expenditures (provided by the state), and state criminal records provided (provided by the South Carolina Department of Corrections).
All 100 participants agreed to disclose all of the data described above and be contacted for future surveys. They all opened an account with Spero. However, 7 participants were impossible to locate, and we were not able to confirm they received the debit card Spero created for them when they opened the checking account. Our final pre-randomization sample was thus composed of 93 individuals.
For budgetary reasons, we could only assign 40 individuals to the treatment group. We randomized the 40 individuals to the treatment group and 53 individuals to the control group using the procedure in Bai (2022):
1. We first matched individuals based on the baseline registered demographics and the baseline registered outcomes from the survey. Outcomes from the administrative data from the South Carolina Revenue and Fiscal Affairs Office were not used in the randomization protocol. Based on the matching, we formed 22 "matched groups" of 4 individuals and 1 "matched group" of 5 individuals.
2. Within the 23 groups, we assigned two individuals to the treatment group randomly, and the other individuals to the control group. This totaled 46 individuals initially assigned to the treatment group.
3. We randomly chose 6 of the matched groups and then, within each group, randomly picked one of the individuals initially assigned to the treatment group and switched them to the control group.
The final randomization assignment has six matched groups, with one individual assigned to the treatment group and three individuals assigned to the control group; 1 matched group, with two individuals assigned to the treatment group and three individuals assigned to the control group; and 16 matched groups, with two individuals assigned to the treatment group and two individuals assigned to the control group. This assignment totals 40 treatment-group individuals and 53 control-group individuals.
The baseline registered demographics collected during the baseline survey are: indicators for being a woman, being white, Black, other race, Hispanic, having a high school degree, and living in a house owned by them or a family member, as well as age the number of individuals living in the household.
The outcomes collected during the baseline survey: a labor force participation indicator, a variable containing income from all sources during the last month multiplied by 12, an indicator of being in poor or fair health, an indicator of smoking, and an indicator of being poor (constructed from the income and number of individuals in the household variables).
These two sets of variables were used in the matching process for the randomization.
We plan to follow up with individuals to collect updates on the baseline outcomes from the baseline survey during and after the implementation of the intervention. The first follow-up will take place during September and October of 2024.
The registered outcomes from the administrative data, which were not used for the randomization protocol, are:
1. Quarterly wages from all sources.
2. Monthly benefit amounts from SNAP and TANF (as well as indicators for receiving either of these benefits).
3. Total expenditure (out of pocket and covered by any form of insurance) in out-patient and in-patient hospitalizations.
4. Days spent in the hospital.
5. Engagement in the correctional system of South Carolina due to any crime.
We plan to implement our main tests on the follow-up of the outcomes collected during the baseline survey and the administrative outcomes. Our plan is to test whether these outcomes change at all as a result of the intervention.