Experimental Design
This research builds on a RCT, conducted by BIGD in 2016. Randomization was done at the branch office. The research covers 88 branch offices (66 treatment and 22 control branches) in 11 districts of Bangladesh. After their selection following a rigorous process, we completed a baseline survey in July 2016, covering 8,973 eligible households (1,931 control and 7,042 treatment).
The baseline survey collected detailed information from the main female member of the household on the demographic and socioeconomic characteristics of the household, including household composition, labor supply of each household member, occupation, income, assets, financial market participation, expenditures, food security, health outcomes, women's mental health (anxiety), women empowerment and women and children’s anthropometric data. This survey also collected data on IPV; however, with a limited set of questions, focusing on physical abuse and abuse for dowry. Since IPV in South Asia is often related to dowry disputes (Ambrus, Field and Torero, 2010), it included specific questions to capture this. Moreover, this survey collected data on risk (i.e., choosing lotteries) and time preferences (i.e., choosing between today's and future payoffs). Thereafter, BRAC implemented the intervention described above in the treatment branches until December 2017.
An endline survey (hereafter referred to as endline1) was conducted in January 2019, a year after the completion of the programme, in which all households in the control group and a random subsample of two-thirds of the households in the treatment group were re-interviewed. This resulted in 4,212 treated and 1,739 control households, with around 10% attrition in both treatment and control groups. The endline1 used the same modules as the baseline survey. In addition, it has an extended version of the IPV module, using five established indicators covering physical and emotional abuse.