Experimental Design
156 distinct clusters of houses in remote areas not previously served by POS programs were identified. A listing exercise was conducted to gather data on employment and use of banking services for 3,657 households in the 156 clusters. Inclusion criteria for households were: (1) no member had any transaction with a bank in the previous month, and (2) a primary income earner received income with a frequency of at least once per week. 795 individuals were identified. For the first 6 months, 78 randomly assigned clusters received weekly home treatment (N=389), and 78 clusters were in the control group (N=406). For the last 6 months, the 78 treatment clusters were reassigned as follows: 40 remained in weekly home treatment (N=197), 19 switched to biweekly home treatment (N=85), and 19 switched to weekly box pickup (N=107). The 78 control clusters were reassigned as follows: 52 clusters remained as pure control (N=256), 13 switched to weekly box pickup (N=89), and 13 switched to biweekly box pickup (N=61).
In order to measure the effect of more frequent surveys on behavior, individuals were either surveyed monthly (18 rounds) or quarterly (8 rounds). The survey consisted of a detailed expenditure and cash flow survey to identify changes in both formal and informal savings behavior. 498 individuals were surveyed monthly and 297 were surveyed quarterly.