Four Research and Policy Implications from an RCT on Women Business Owners in Indonesia

Women’s profits grew over time while a rise in women’s agency was short-lived, according to new analysis.

Allowing for the effect of economic empowerment interventions to unfold over two years or more, and tracking and analyzing how time changes results of randomized control trials (RCTs), is important and worthwhile.

We learned this lesson from an analysis we conducted with Firman Witoelar, Australian National University, of what happens over time to Indonesian women owners of very small businesses who were exposed to a supply- and demand-side intervention.  The results, in a nutshell: the estimated positive effect of the intervention on women’s business profits starts low, increases over time and is even projected to continue growing well beyond the last 24-month follow-up measure.  In contrast, the estimated positive effect on women’s agency (e.g., their role in family decision-making, access to savings) appears to be mainly limited to the first six months.

What the Indonesia trial did

The Indonesia research was a collaboration between the Center for Global Development, J-PAL Indonesia, the World Bank’s East Asia and Pacific Gender Innovation Lab, and two academic researchers.  Other research papers using data from the trial are here and here.  Rural and peri-urban villages in East Java, Indonesia, were randomly assigned to a condition where a large bank offered bank agents (one per village) high or low incentives to increase uptake and usage of a new mobile banking product among business owners (this is the supply-side intervention).  Next, within each village, women business owners were randomly assigned to either a short financial literacy/business training followed by mentoring, designed and implemented by Mercy Corps Indonesia, or to no intervention (the demand-side intervention). 

The data for this study

The analysis uses data from the baseline survey and two follow up surveys (with a total of 4,240 observations).  It explores individual level variation in the time elapsed between the businesswoman’s first exposure to the randomly assigned intervention and the month in which she was re-interviewed in a follow-up survey--analyzed over four intervals: the first six months, 7-12 months, 13-18 months, and 19-24 months.

How women’s economic empowerment was measured

The RCT and this analysis follow the growing consensus in conceptualizing women’s economic empowerment as the process of empowerment (the expression of agency) as well as more tangible and easily measurable empowerment outcomes and included questions about women’s agency (their role in family decision-making) and their economic achievements (measured through savings, business capital and profits). 

Research and policy implications

There are at least four implications from our analysis of the Indonesia trial, the first two for knowledge on what works to empower women economically, the last two for the impact evaluation literature more broadly.

1. Women’s economic empowerment should be measured on a longer timeline

Some interventions designed to improve women’s economic lives need to be tracked long enough for women to manifest new and beneficial behaviors. Sometimes, business training programs have delayed effects on businesswomen’s profits (i.e. in Kenya and Peru). These delays may be partly the result of women having to practice new business-related behaviors and/or negotiate with others in traditionally male-dominated household or business environments over time. 

2. We may need better interventions

One would assume that practicing new business behaviors and increased business profits would strengthen women’s agency over time, but this was not the case here. It could be that the self-reported measures of decision-making are not capturing agency well. It could instead be that the type of agency the household decision making questions picked up is unrelated to the agency women enact in their business environments. But, if the finding truly conveys that the intervention is not transformative enough, we need to rethink and recalibrate these interventions so that they indeed empower women economically—both in terms of agency and economic achievements.

3. Time matters and should be factored into research designs

Researchers are usually under time pressures to produce results quickly while our study suggests that slowing down and allowing for time to unfold may be the smart thing to do. The same message goes to funders of this research: the impact evaluation literature that studies the effects of interventions that seek to empower women economically (and also perhaps other disempowered populations) needs sufficiently long-time frames to capture results.

4. Economic empowerment is not linear

The unfolding of economic empowerment outcomes takes time and is also not necessarily linear. This is the fourth implication of this work. In cases where estimated impacts vary over time, traditional constant impact estimation models may provide misleading results for these outcomes and for the economic analysis of projects.