Financial Education and Intimate Partner Violence: Evidence from Colombia during Covid-19

The proposed intervention is designed to be scalable and cost-effective. Built on content from LISTA, a proven tablet-based financial education program targeting women (Attanasio et al., 2019), it will consist of two complementary WhatsApp components - an intelligent bot for individual users and a group chat. The bot will push information to individual users based on their selection of a menu of options including Covid-19, financial education, and partner communication. In parallel, the same users will be enrolled in moderated support groups based on gender, geography, and company membership. Besides encouraging use of the bot, the support group moderator will share interactive content (e.g., images, audio, links to videos and ATM and banking simulators, group games and quizzes) on six topics (e.g., A=ahorro (savings), B=bien asegurados (insurance), C=cuentas claras (budgeting), D=deudas sanas (debts), and F=familia (family). The communication module includes behavioral strategies to be more assertive, tips to manage partner disagreement, calming methods, balanced time use within the household, and discussion about gender norms. The primary user will be encouraged to share the content with his or her partner and even other household members.

2020-06-24

2020-08-01

Intimate Partner Violence

Women’s empowerment may reduce or increase IPV depending on the degree of male backlash. To measure this relationship we will capture data on partner relationship quality, including type and degree of communication, and will rely on WHO measures for male controlling behavior and incidences of psychological, emotional, physical and sexual abuse.

Financial capability
Women’s empowerment

Secondary outcomes will focus on financial capability and women’s empowerment. Financial indicators include financial knowledge, attitudes, practices, and performance. Empowerment will consist of measures of psychological and economic empowerment, including self-efficacy; beliefs and attitudes about gender norms; shared household tasks; decision-making, especially in relation to consumption; and relative and absolute distribution of household income.

This project will conduct a cluster randomized controlled trial comparing a control group to a treatment receiving the LISTA WhatsApp intervention. To avoid risk of spillover randomization will be conducting at the firm level.

In Colombia, every individual in the formal labour market (or that has a labour contract and may receive social security either as a net contributor or beneficiary) must be affiliated to a family compensation fund such as Comfama. These funds operate at the state (or ‘departmento’) level and in some cases compete with other funds operating in the same geography, although in the majority of departments only one compensation fund operates. The employer or hiring firm is responsible for affiliating its employees to a compensation fund and the general outcome of this process is that all employees in a firm are affiliated to the same compensation fund.

Power calculations sought to identify the sample size necessary to detect 0.2 SD shift between groups, taking into account take-up and attrition assumptions, as well as three waves of data collection (baseline and two follow ups), intraclass correlation among firms, and adjustment for multiple hypothesis testing. The sample size needed to get a minimum detectable effect of 0.2 SD, using 174 clusters an intraclass correlation of 0.05, and assuming a take up rate of 60% is 2,733 (1367 treated individuals and 1,367 control individuals).The sample size of baseline is roughly 2,300 (800 males and 1,500 females) selected from around 350 companies, with half of the firms and individuals assigned to the treatment group. An additional sample of about 1800 eligible individuals that were not in baseline are considered for the intervention and follow up surveys, assigning roughly half to the treatment group.

Fundación Capital is providing technical support to Comfama for the intervention, including the hiring, training, and supervision of facilitators with experience in social interventions, thus leaving Comfama with the capacity to further scale the intervention after the evaluation, while paying Fundación Capital a small licensing fee. Facilitators will establish contact with firms through Comfama’s existing channels. During the intervention an intelligent bot diffused or embedded in WhatsApp will be accessed by participants who will also belong to support groups. Group facilitators also send content and provide support during the interaction with the content and with one another.

To better characterize the population and to improve statistical power we will control by baseline covariates, ensuring balance across key variables. A baseline phone survey was implemented in June and November 2019. Randomization was conducted across roughly 2.300 participants interviewed at baseline and 1800 participants not included at baseline. After completing the intervention, a follow-up survey is planned to take place in August 2020, with a second follow-up some months later at a still undetermined time.

Randomization was made using the randtreat Stata command.

Treatment was randomized at firm level.

Yes

350 firms

2,300 from baseline and 1,800 additional eligible individuals. Total 4,100 individuals.

Around 175 firms assigned to the treatment group and approximately 175 assigned to the control group.

The smallest minimum detectable effect in IPV is 0.2 standard deviations.