Skip to content
MIT Better World

By Kathryn M. O’Neill

That’s why MIT’s Abdul Latif Jameel Poverty Action Lab (J-PAL) launched its Government Partnership Initiative—to team up with governments to advance policies that have been scientifically proven to help people living in poverty.

“Developing country governments are by far the biggest funders of poverty relief in the world. So, the biggest lever you can have in reducing poverty is getting that money spent better,” says Rachel Glennerster, who was J-PAL’s executive director until January of this year and is now on leave serving as chief economist for Britain’s aid agency, the UK Department for International Development.

J-PAL’s Government Partnership Initiative (GPI) helps governments evaluate policies, scale up successful ones, and build an internal culture of evidence-informed decision making. A competitive fund, GPI supports the work of J-PAL, which was founded in MIT’s Department of Economics in 2003 “ to reduce poverty by ensuring that policy is informed by scientific evidence.” Since then, J-PAL’s research network has conducted more than 800 randomized evaluations measuring the impact of social programs in 80 countries.

“The ‘in thing’ in development is to think about ways to bypass government and deal directly with people or via other non-state actors,” says Iqbal Dhaliwal, J-PAL’s new executive director and GPI co-chair. (GPI’s other co-chair is J-PAL Director Abhijit Banerjee, the Ford Foundation International Professor of Economics at MIT.) “We have realized in the past 12 to 13 years at J-PAL that whether you like working with a particular government or not, there’s no alternative if you want to make big changes.”

Launched in 2015, GPI has so far awarded $1.8 million for 24 partnerships in 13 countries, spanning a wide range of programmatic areas—from tackling a court backlog in Mexico to implementing educational reforms in Zambia and even creating a culture of evidence-based policy making in one of the major states of India. Average grants range from around $50,000 for a pilot to $90,000 for a full project.

Why would governments with budgets in the billions need such small sums? Dhaliwal says that it’s because most government dollars are tied to specific programs, making it surprisingly difficult to divert money to new initiatives.

“We’ve realized that small amounts of money, given for the right purpose to the right department, can have a big effect,” he says. “Small projects can change the decision-making process, then change the direction of the bigger ship.” Once the effectiveness of a new program is clear, governments can use their own resources to expand benefits to the whole population.

Low cost, big impact

“The research is demand-driven,” says J-PAL senior policy manager Claire Walsh, noting that GPI has far more requests for support than it can fulfill. To date, 85 proposals have come in from 27 countries at various levels of government requesting a combined total of $8.4 million.

While GPI often provides less support than requested due to high demand, Walsh says that even low-cost interventions can have a big impact. For example, in Mexico City, GPI provided just $50,000 for an evaluation by a J-PAL-affiliated researcher of the benefits of giving labor court claimants information about success rates of similar cases. The research showed that providing such information increased the use of alternatives (such as mediation) to resolve disputes—significantly reducing the court’s nearly four-year backlog. GPI has since approved a grant to scale up the program.

Scaling up successful programs is a central focus of GPI, which also provides funds for the technical assistance governments need to expand pilots or adapt programs already proven to work elsewhere. “You don’t always have to do new research, and you can’t use research funds to do technical assistance,” Glennerster says. “That’s the niche GPI is filling.”

In 2014, government officials of Tamil Nadu met with J-PAL’s Shobhini Mukerji (far right) and Esther Duflo and Iqbal Dhaliwal (right of center) to launch a collaboration increasing the Indian state’s use of evidence in policy making. Photo: Courtesy of Government of Tamil Nadu

In Zambia, for example, to improve chronically low learning levels, GPI has been helping educators tailor instruction to the level of the child rather than the level of the curriculum. This approach has been thoroughly tested by J-PAL with partners in India and is known to improve educational outcomes. Nevertheless, implementing the program in Zambia presented unique challenges.

“It’s not a pill you can just take off the shelf,” says Glennerster, who served as a lead advisor for the Zambia project. Glennerster notes, for example, that the school day, curriculum, and staff capabilities are very different in Zambia than in India. “You need someone who really understands the research to help governments adapt research to their context and sort through practical issues.”

GPI funded J-PAL’s Africa office to provide the technical assistance Zambia required to adapt the teaching approach to its needs, including developing a model of the program to pilot in 80 schools and conducting an independent assessment of the pilot, which proved successful. Last August the government committed to rolling out the level-teaching program to approximately 1,800 schools over the next three years.

Better decision making

Ultimately, GPI is working to institutionalize a culture of evidence-informed policy making within governments so that nations make better decisions going forward. That is exactly what is happening in Tamil Nadu, India, a state that is home to 78 million people—nearly a quarter the size of the United States population. GPI has teamed up with officials there to help the state meet its commitment to pilot and rigorously evaluate innovations in all its departments and to scale those interventions that are found to be effective.

Tamil Nadu has allotted more than $4 million to the project—an “unprecedented amount” of government support for evidence-based policy, according to Dhaliwal—and GPI is providing ongoing assistance. For example, GPI funded two J-PAL staffers to work directly with the government on its evaluations, including helping the state set up a central data analytics unit to inform its decision making.

Already, Dhaliwal says government officials in Tamil Nadu are internalizing the principles behind J-PAL’s method of assessing anti-poverty programs. He recalls a recent meeting he attended at which the head of the civil service, unprompted, pointed out the need to compare outcomes for people who received an intervention with those for a control group who did not.

“It’s incredible when such senior officials in the government understand so well the underlying science. I got goose bumps,” he says.

“These projects are changing the culture. Once you change the culture, you don’t need J-PAL and MIT to be there,”  he says. “Hopefully, governments will make more evidence-informed decisions on their own.”