The study is a two phased project that is testing the impact of account labeling on the savings behavior of rural customers serviced by Susu agents (savings collectors). The original labeled account study, referred to as Phase I, is an RCT comparing treatment customers who were offered a labeled Susu account to control customers who were not offered an account. The follow up study, referred to as Phase II, is an RCT that again provided a labeled Susu account to treatment customers but, this time offered an unlabeled Susu account to control customers.
External Link(s)
Citation
Karlan, Dean. 2020. "Savings Account Labeling for Susu Customers in Ghana." AEA RCT Registry. November 15. https://doi.org/10.1257/rct.6598-2.0.
The study offered labeled accounts to customers serviced by Susu agents (savings collectors) in Mumuadu Rural Bank to test the impact of account labeling on their saving behavior. The original study, phase I compared customers who were offered a labeled account to those who were not offered. The follow up study, referred to as Phase 2, compared customers who were offered a labeled account to those who were offered an additional, unlabeled account.
Intervention Start Date
2009-10-23
Intervention End Date
2013-09-27
Primary Outcomes (end points)
Monthly deposits, withdrawals, and balance of Susu accounts.
Primary Outcomes (explanation)
Secondary Outcomes (end points)
Secondary Outcomes (explanation)
Experimental Design
A census survey was administered to customers with active Susu accounts and of those who responded, half of the customers were randomly selected for treatment. Randomization for phase 1 was done for each branch of the bank and was stratified on Susu agent and Savings Balance Strata. Randomization for phase 2 was also done for each branch and stratified on Susu agent and if they had more than 1 Susu account with the bank. The presence of pure control group in phase 1 was replaced with customers who received offers for the additional unlabeled Susu accounts.
Experimental Design Details
Randomization Method
Done in office by a computer
Randomization Unit
Individual
Was the treatment clustered?
No
Sample size: planned number of clusters
2163 in phase 1, 2019 in phase 2
Sample size: planned number of observations
2163 in phase 1, 2019 in phase 2
Sample size (or number of clusters) by treatment arms
1091 in control and 1072 in treatment for phase 1
1009 in control and 1010 in treatment for phase 2
Minimum detectable effect size for main outcomes (accounting for sample design and clustering)