Experimental Design
In the first RCT, all of the Green Bank's 169 existing lending centers on the island of Leyte were randomly selected into control and treatment groups. The control group would continue receiving the bank's prior Grameen-style, group-liability lending model. Lending centers were made up of 15-30 entrepreneurial women, further divided into subgroups of five. At mandatory weekly meetings individuals made a repayment, subgroups were responsible to cover delinquencies, and lending center was responsible if a subgroup defaulted. Weekly deposits into individual, subgroup, and center savings accounts was mandatory. The treatment group were to switch to an individual-liability lending model. Borrowers were no longer liable if other borrowers were late or defaulted on their loan. Borrowers still met weekly for repayment and mandatory savings deposit amounts were unchanged, but all deposits were made into individual accounts. Conversion of the treatment group occurred in three waves over nine months. Administrative data was collected from prior, during, and after the intervention from the bank. A post-experiment activity survey of credit officers and three surveys of borrowers were collected as well.
In the second RCT, new areas for possible expansion of the bank's operations were identified. 124 villages, or barangays, were randomly selected. Of the 124, 68 barangays were selected as feasible for expansion. The selection criteria for the 68 chosen barangays was the same as the criteria for prior areas were lending centers had been established. The 68 new lending centers were randomly selected to offer one of three lending products: 1) group-liability loans, 2) individual-liability loans, and 3) phased-in individual-liability loans (group-liability for the first loan, individual-liability on subsequent loans on successful repayment). Administrative data from the bank was collected for the duration of the intervention in addition to the activity survey of credit officers and business census and social network surveys of borrowers.