Experimental Design
Our research design offers subsidies to small-scale Fairtrade farmers to cover pecuniary costs involved in the opening of formal saving accounts in rural Ethiopia. These transactions costs include the pecuniary costs associate to obtaining two passport-size pictures, transportation to and from the bank branch, and the required minimum account balance. These subsidies are distributed through vouchers that are tailored according to the randomly assigned saving group: we set a one-time, small subsidy of 70 Birr (US$ 3.20) for single saving accounts and slightly higher subsidies of 100 Birr (US$4.50) for joint saving accounts. We also offered a marketing and financial education workshop through public events in which farmers received a half-day financial training on basic bank operations and bank regulations on single and joint deposit accounts, as well as messages reinforcing social capital (trust) on formal banking institutions.
The sampling framework follows from a population of 5100 agricultural households who belong to four Fairtrade coffee cooperatives located in Jimma (2) and Sidama (2) in rural Ethiopia. This sampling frame comes from the cooperatives’ 2014 administrative records. This population framework entails geographic, cultural, and agricultural output variation. Based on this population we select a representative sample based on a 2x2 stratified design applied independently to each Fairtrade cooperative. Stratification is based on two variables of interest, level of coffee production and the gender of the heads of households. In doing so, we split the population of Fairtrade members into high- and low-production groups according to whether households’ production is above or below the 2014 per-household median of coffee production. The resulting stratified sample is composed of 1,200 households. The allocation of household units to treatment and control groups is implemented through complete randomization at the household level by simple random methods. As a result, 450 households were assigned to the single-account treatment group, 450 to the joint-account treatment group, and the remaining 300 to a pure control group. The setting of this study covers two remote, agricultural areas of Ethiopia within the preeminent coffee-producing regions of the country. Our financial intervention covers 12 rural districts (Kebeles) across two provinces, Jimma and Sidama, in the west- and south-central parts of the country. These two areas entail an important cultural variation. Orthodox Christian households (55 percent of our sample) mainly populate Sidama, while Jimma is a predominantly Muslim area (45 percent).
A cross-randomized, complementary design was implemented to investigate the role of commitment incentives on take-up rates and short-term (6 months) account usage. Specifically, we randomly split the treatment group households into three equal-sized sub-samples and offered them marginal variations in some salient features of the account. The first randomly assigned group is the ‘standard’, single- or joint-account treatment households, the default treatment group. The second randomly assigned group was offered an additional reward equivalent to 2.5% on the saving balance if account holders actively used the account within the next six months from setting up the account. This short-term ‘active’ account usage was defined as having at least two monetary deposits and a positive saving balance in the account as of July 1st, 2016, six months following the intervention. The third randomly assigned group was offered an additional prize equivalent to 2.5% on the saving balance if an account holder did not withdraw her initial deposit until July 1st, 2016. Information about these additional incentives and their corresponding vouchers were provided during the individual household visits implemented as part of the outreach campaign in December 2015/January 2016. This cross-randomized allocation design means that we have the same number of household units (300) in each of four independent farmers groups, including the pure control group