Planning to Reduce Violence: Understanding the Impact of Cash Transfers and Facilitated Financial Planning on Intimate Partner Violence

The first intervention is a cash transfer provided by the Government of Liberia (implemented by the Ministry of Gender, Families and Social Protection) and implemented by GiveDirectly. Households will receive a series of 6 cash transfers, each valued at roughly US$90-100 and paid through mobile money over a period of 12-15 months. The precise transfer schedule will depend on project implementation timelines. Transfers are preferentially targeted to women, who account for about 80 percent of direct beneficiaries.

The second intervention is a facilitated planning session for beneficiary households headed by married or cohabiting partners. The session will be facilitated by a GiveDirectly field officer at the time of transfer enrollment. Participants will first discuss their spending priorities and agree on a set of goals. These goals will be written on a “planning contract”, which participants can keep and update as they achieve their goals. The contract is designed to provide both a reminder of spending goals and soft commitment to follow through on these goals. Participants will also receive a poster, which contains a less detailed summary of the planning contract -- based on piloting we expect
most beneficiaries to keep the contract in a private place, while the poster is designed for display in a public place.

Given the prevalence and promise of cash transfers, and the fact that these programs generally already entail brief touchpoints with households, building on cash transfer programming to layer additional interventions specifically targeting intrahousehold gender dynamics may be a promising strategy. In this respect our planning intervention is innovative in two ways. First, it is designed to reduce decision stress and male status threat (by promoting spousal consensus and early planning at a low stress time), addressing some of the most important outstanding concerns about cash transfers. Second, it is easily implemented in limited resource settings, at minimal marginal cost. (For example, the main modification GiveDirectly needs to make to implement the intervention in this setting is to have field officers sit with both spouses when performing transfer registration as opposed to one spouse).

2022-03-01

2023-06-30

Intimate partner violence, controlling behaviors and non-violent conflict, couples' communication, intrahousehold decision-making

The evaluation includes 211 rural communities in Bomi and Maryland counties, identified by the government of Liberia as targets for the Social Cash Transfer program based on their high rate of extreme poverty. These communities will be randomly assigned to one of the three study arms with equal probability: cash transfer only, “TO”, cash transfer plus financial planning, “TP”, and control, “C”. Eligibility for transfers will not be tied to study participation; rather, all households in TO and TP communities will be eligible for transfers. GiveDirectly will be responsible for identifying households, informing them of their eligibility, and obtaining consent to participate in the interventions.

A baseline survey will be launched in February 2022. We aim to enroll up to 15 eligible couples per community, for a total sample of approximately 2,500 couples (the exact sample size will depend on the share of targeted households that have an eligible couple). We will conduct separate interviews of the male and female partner in each couple. Both surveys will collect data on socioeconomic characteristics (including individual control over assets and labor force participation), intrahousehold decision-making, financial decision-making power and use of financial services, couples’ communication, and social norms around IPV. The female survey will collect additional data on reported experience of IPV and controlling behaviors, while the male survey will collect additional data on household composition and socioeconomic status. The surveys will be conducted prior to the rollout of the cash transfer and planning interventions.

A short-run phone survey follow-up is planned after the second transfer (~5 months after the baseline, ~3 months after enrollment) and will target all 2,500 enrolled women. Since GiveDirectly offers phones to households when needed, we expect completion rates will be higher in the TO and TP arms -- thus this survey will be most useful for identifying the short-term effects of planning conditional on receiving a transfer.

A subsequent in-person endline survey will (contingent on funding) revisit all enrolled households around one year after program enrollment and financial planning intervention. Key modules in the endline include experience of IPV (we propose to use the DHS module, which is very similar to the WHO module); controlling behaviors, intrahousehold conflict, and relationship satisfaction; and decision-making dynamics.

We will assess impacts by comparing differences in IPV and related outcomes across the three treatment groups using OLS regression, controlling for randomization strata and clustering at the community level. We will also study impacts on the above-mentioned precursors to IPV, which will help us shed light on mechanisms that drive our interventions’ impact on IPV and test our hypothesis that planning ameliorates decision stress triggered by a sudden injection of new economic resources. Finally, we will examine heterogeneous treatment effects with respect to baseline conflict and decision making to shed light on mechanisms and identify groups most impacted by cash and planning.

The randomization was conducted in Stata by the research team and program assignments were then communicated to the government partners and intervention team.

Community / village

Yes

211 village

2500 households (all including a cohabiting couple; both husband and wife will be surveyed at baseline)

71 control, 71 cash transfer only (TO), 70 cash transfer + facilitated planning (TP)

Our power calculations (α=0.05, β=0.8, intra-community correlation of 0.05 based on DHS data) indicate that we are powered -- assuming a 91% retention rate at endline -- to detect a reduction in IPV of .19 standard deviations (around 9 percentage points using the DHS mean of IPV in the past year for our sample counties), in pair-wise comparisons between arms.