The benefits of improving women’s economic opportunities are clear: when women control income, they invest it in their families, particularly in the health and education of their children, helping to break cycles of poverty. Women also contribute to the economic growth of their communities. Research from McKinsey shows that since 1970, as women’s greater labor participation took off, their productivity has accounted for a quarter of U.S. GDP. In the developing world, women are estimated to own 40 to 50 percent of businesses.

Yet, as the World Bank’s World Development Report 2012 notes, “pervasive and persistent gender differences remain in productivity and earnings across different sectors and jobs. Indeed, many women around the world appear to be caught in a productivity trap—one that imposes significant costs on women’s welfare and economic empowerment today and serious disincentives to invest in the women of tomorrow.” As the World Bank’s report explains, it’s not that women are worse farmers, workers, or entrepreneurs than men, it’s just that they face particular constraints–notably in time (due to care responsibilities), skills, and access to land and credit.

Figuring out how best to address these constraints has become a high priority in development circles. But the reality is that we still know very little about what works and what doesn’t work in improving women’s economic productivity.

To address this knowledge gap, ExxonMobil and the United Nations Foundation (UNF) have joined forces to evaluate interventions promoting women’s economic opportunity. As Mayra Buvinic from the UN Foundation said last week at a meeting of the ExxonMobil roundtable series at the Council on Foreign Relations, “we know a great deal about the determinants of inequality, but little about what interventions actually succeed.”

Panelist Markus Goldstein, a senior economist at the World Bank, echoed the importance of this research and the need to “raise the bar on gender data.” Strong evidence exists on the barriers women face, but we know less about how to overcome them. Goldstein explained that while evaluations of economic development interventions do occur, they rarely have a gender dimension. Moreover, randomized controlled trials – which are costly to do – often do not disaggregate gender.

Gender programs are therefore usually run on “well-informed hypotheses,” noted Goldstein, and that can be frustrating. In order for investments in gender equity to be sustainable, it’s important to demonstrate their effectiveness through hard data. As panelist Agnes Quisumbing from the International Food Policy Research Institute argued, the design of research and the type of data collected is critical. Longitudinal studies and experimentation matter.

In some areas, we do have relatively strong evidence about what works: for example, there is robust data on job training programs for women implemented in Latin America. Research shows that job skills programs are more likely to be successful if extra subsidies are offered for transport and childcare, and on-site job trainings are more successful in increasing women’s economic earnings than in-classroom trainings. Still, it is not clear how well such results translate across regions.

With respect to business trainings–those designed to help female entrepreneurs start, grow, and sustain businesses–the findings are less clear. Preliminary research from the UN Foundation indicates that meeting the needs of poor women at the base of the pyramid is particularly challenging. Simply providing them with quality business trainings is not sufficient to make a difference. Such interventions are more successful if paired, for example, with initiatives to increase access to credit. However, this creates a tension between implementation capacity and program effectiveness for service delivery organization.

Clearly, more research is needed to understand how best to focus limited resources on improving women’s economic potential. The opportunity is significant: for example, the Food and Agriculture Organization estimates that equalizing access to productive resources between female and male farmers could increase agricultural output in developing countries by 2.5 to 4 percent. The UNF’s research seeks to marshal information that will provide a “roadmap for action” for the private and public sectors by identifying the best interventions across countries with different socio-cultural norms and stages of economic development. I will be following their progress closely and will revisit this topic when their final report is released next spring.