Identifying what actually works to reduce poverty and improve population well-being is a key challenge in development economics. When something is thought to work, the next challenge is determining why it works and the conditions under which it works; that is, assessing the extent to which conclusions are generalizable. These are key research themes in the Development Economics Program.
One exciting source of new results on these questions arises from a multifaceted, focused initiative known as the "Graduation" Program. This program, developed by BRAC, a large NGO formerly known as the Bangladesh Rural Advancement Committee, was designed to provide the poorest people with a sustainable pathway out of extreme poverty. The program provides resources to address participants' immediate needs and longer-term investments, with the goal of building sustainable livelihoods. The Graduation Program has three central planks designed to provide a holistic set of resources and services to increase the productivity of the ultra-poor: a grant to acquire productive assets, access to a savings account, and two years of training and support, including life skills coaching. (1)
To investigate how well the program works, Abhijit Banerjee, Esther Duflo, Nathanael Goldberg, Dean Karlan, Robert Osei, William Pariente, Jeremy Shapiro, Bram Thuysbaert, and Christopher R. Udry conducted an ambitious set of coordinated randomized controlled trials (RCTs) in villages in Ethiopia, Ghana, Honduras, India, Pakistan, and Peru. They identified the poorest households in each study village and randomly offered about half of them the BRAC program, with the other households serving as controls. The program was a stunning success, as measured by a very large and broad set of markers of well-being.
At the end of the intervention, which lasted two years, relative to controls, Graduation Program households reported higher levels of per capita consumption, more income, greater savings, more assets and improved mental health. Effects were not only large and statistically significant, but also long-lived, persisting for at least a year after the intervention ended in all the study settings (2) and in India for at least another four years. (3)
Figure 1 illustrates estimates of the magnitude of some of the average standardized treatment effects two years after the start of the program in the six countries. Based on this evidence, many countries are currently experimenting with this type of multifaceted package as they endeavor to reduce persistent poverty.
New Thinking about Poverty Alleviation Strategies
The study clearly establishes that the Graduation Program has transformed the lives of the poorest not just in one small area but across vastly different settings over three continents. This is important because many of the most promising anti-poverty programs that have documented successful poverty reduction in some contexts have not been successful in other settings.
Microcredit is one example of an anti-poverty strategy that has been extensively analyzed. In 2006, the Nobel Peace Prize was awarded to Muhammad Yunus and Grameen Bank for leading the microcredit revolution that brought small loans first to the poor in Bangladesh, and then to the poor more broadly. Micro-loans, which are made mostly to women, involve some form of group liability and report excellent repayment rates. The number of people who have received the loans has grown rapidly and microfinance for a time was heralded as the magic bullet that would end poverty as we know it. However, results from rigorous studies investigating the impacts of microcredit have not been encouraging. Duflo, Banerjee, Rachel Glennerster, and Cynthia Kinnan conducted a randomized evaluation of the impact of a micro-finance firm entering markets in Hyderabad, India, and found that, while loans were made and recipients invested in their new businesses, the effects were transitory, with no discernible improvements in consumption or well-being for any but a small fraction of recipients at the top of the income distribution. (4)
Non-experimental studies have drawn similar conclusions. This includes research that has taken a structural approach to modelling credit constraints, income uncertainty, and lumpy investments, exploiting quasi-experimental variation in microcredit programs in Southeast Asia. Overall, estimated impacts of microcredit have been mixed, at best, and the outcomes that are affected vary substantially across studies and contexts. Francisco Buera, Joseph Kaboski, and Yongseok Shin summarize the evidence on micro-credit as indicating that, while it can help segments of the population increase both income and consumption, there is little reason to believe that it has had a transformative impact on the lives of the poorest. (5)
It is possible that microcredit loans are too small and too short-term to have a sustained impact on the lives of the recipients. This was investigated in a clever RCT conducted in rural Mali by Lori Beaman, Karlan, Thuysbaert, and Udry that provided capital to farmers at the beginning of the planting season to be repaid as a lump sum after the harvest. (6) About half the study villages were assigned to participate in a loan program. Women in those villages were able to form associations and apply for loans. After all loan decisions had been made, a random sub-sample of the women living in the same villages who did not borrow were given cash grants. In the other villages, randomly selected households were given a cash grant--the first plank of the Graduation Program. Those who received cash grants significantly increased investments and net revenue on their farms. In contrast, in the villages where loans were available, the farmers who borrowed increased their investments and revenue even more, whereas there were no significant increases among the cash grant recipients in these villages. The researchers conclude that when borrowers are self-selected, or selected by a loan officer, returns to credit are large, significant, and sustained, but when borrowers are not self-selected, average returns are effectively zero. A similar point is made by Pushkar Maitra, Sandip Mitra, Dilip Mookherjee, Alberto Motta, and Sujata Visaria, who show that crop yields and income of potato farmers increased when trader-lender agents were given authority to select borrowers. (7)
An important point of this literature is that while credit is a powerful tool for poverty alleviation for a selected group of people, credit alone is not sufficient to combat widespread poverty at the population level.
Other work on microcredit highlights the importance of liquidity constraints after a shock. Emily Breza and Kinnan show that after the 2010 crackdown on microfinance in Andhra Pradesh, India, there were declines in wages and consumption. (8)
Elizabeth Frankenberg and I reported similar evidence during the Indonesian financial crisis, a large-scale, unanticipated shock, although they also note that families draw on all of their resources, including savings and even their own health and human capital, to mitigate deleterious impacts of the negative shock on the well-being of family members. (9)
Access to a savings account is the second important component of the Graduation Program. Research at the macro level has established the importance of the banking sector for growth and development, and many national and international agencies have invested substantial resources in an effort to shift people from informal savings structures into formal institutions. In a series of RCTs, Pascaline Dupas, Karlan, Jonathan Robinson, and Diego Ubfal find that expanding access to basic bank accounts to a population-representative sample of unbanked households in...