Prize Lecture: Esther Duflo, Prize in Economic Sciences 2019

TL;DR

Field experiments influence policy through a sustained network of researchers, field staff, funders, and implementation partners, not through a single study and policy brief. Evidence becomes more persuasive when studies accumulate across settings, use comparable methods, and reveal whether results consistently support the same conclusion, as illustrated by evaluations of microcredit.

Transcript

so our next lecture will be given by Esther Duflo she was born in Paris in the early 1970s the daughter of a math professor and the pediatrician so after high school Esther went on today called normal superior with the intention to specialise in history so during a year in Moscow she was supposed to work on her undergraduate essay but she also work... Read More

Key Insights

• The economics prize is presented as recognition for a movement involving researchers, graduate students, field staff, funders, nongovernmental organizations, governments, and implementation partners. Duflo emphasizes that practical policy research depends on thousands of people rather than only the researchers publicly associated with an award.
• J-PAL's policy influence is described as extending across almost every continent through a network of about 400 associated researchers. Duflo states that policies found effective by someone in this network had affected more than 400 million people by the time of the lecture.
• The conventional policy model assumes that a small, carefully monitored experiment produces clear results, which researchers summarize in a polished brief and present to policymakers for full-scale adoption. Duflo calls this model a straw man because actual policy influence follows a more elaborate process.
• Small experiments face questions about scale and transferability because intensive monitoring or unusually strong partners may produce results that cannot be reproduced in larger programs. Evidence from one location also does not automatically establish that the same intervention will work elsewhere.
• Large-scale implementation can change an intervention's effects through factors not visible in a small experiment. Duflo specifically identifies equilibrium effects, political economy reactions, implementation challenges, and policymakers' varying priorities as reasons why experimental findings may not translate directly into policy outcomes.
• Microcredit reached about 200 million people and was initially supported by enthusiastic media coverage and success stories, but there was little hard evidence that it reduced poverty. Later reports about excessive debt and farmer suicides also remained largely anecdotal, leaving uncertainty on both sides.
• The first microcredit evaluations discussed were conducted in the Philippines and India. Researchers regarded the Indian findings cautiously because Hyderabad was already saturated with other forms of credit, making it possible that local access to alternatives influenced the intervention's measured effectiveness.
• Multiple comparable studies can provide stronger policy evidence than a single experiment. Researchers accumulated microcredit evaluations across countries, published them using the same format and template, and combined their results statistically to examine whether the individual studies told a consistent story or varied widely.

Questions & Answers

Q: How do field experiments influence public policy?

Field experiments influence public policy through an extended process involving research, replication, interpretation, partnerships, and implementation. Duflo rejects the idea that one small experiment and a polished policy brief automatically lead to full-scale adoption. Researchers must consider whether findings transfer across locations, persist at scale, interest policymakers, and remain effective after equilibrium effects, political reactions, and implementation difficulties emerge.

Q: Why is one small field experiment insufficient for policy decisions?

One small field experiment may be conducted with unusually close monitoring, carefully selected partners, and tightly controlled conditions. Its findings may therefore change when a program expands or moves elsewhere. Duflo also notes that every experiment can have limitations, while scaled implementation may introduce equilibrium effects and political economy reactions that the original study was not designed to capture.

Q: Who participates in the field experiment movement described by Esther Duflo?

The movement includes researchers, graduate students, field staff, funders, nongovernmental organizations, government partners, and other organizations involved in development and poverty research. Duflo emphasizes that policy influence requires money, institutional partners, and thousands of people working on the ground. She presents the prize as recognition shared across this broad community rather than belonging exclusively to three recipients.

Q: How much policy influence does Duflo attribute to the J-PAL network?

Duflo says the J-PAL network included about 400 associated researchers and had influenced policies affecting more than 400 million people. She explains that this estimate counts people reached by policies found effective by someone in the network. It does not count people spared from ineffective policies, even though avoiding unsuccessful programs may also represent an important form of influence.

Q: Why does Esther Duflo reject the experiment-to-policy brief model?

Duflo rejects that model because policy change does not normally proceed from one controlled experiment directly to a polished brief and then full-scale adoption. Policymakers may have different priorities or pay little attention, experimental settings may not represent other places, and implementation at scale can alter results. Actual influence requires a broader dialogue and a more elaborate accumulation of evidence.

Q: Why was rigorous evidence needed to evaluate microcredit?

Microcredit inspired strong claims in both directions without sufficient hard evidence. Early enthusiasm relied on media attention, success stories, and the possibility that a program reaching about 200 million people could reduce poverty. Later criticism emphasized debt problems and farmer suicides, but those accounts were also mostly anecdotal. Rigorous evaluations were needed to determine whether either narrative accurately represented microcredit's effects.

Q: Why were the initial microcredit results from India treated cautiously?

Researchers treated the Indian results cautiously because the evaluation took place in Hyderabad, described as a hotbed of microcredit that was saturated with other forms of credit access. The intervention's limited effectiveness might therefore have reflected the availability of alternatives rather than a general failure of microcredit. Duflo says the researchers waited for more studies instead of widely promoting an uncertain conclusion.

Q: How can multiple microcredit studies produce stronger evidence?

Multiple studies allow researchers to compare results from different countries and determine whether they support one common interpretation or vary substantially across settings. Duflo describes publishing microcredit evaluations with the same format and template, then combining the findings through statistical modeling. This approach preserves the individual results while also showing what the collected evidence indicates as a whole.

Summary

Esther Duflo's lecture focuses on field experiments and their impact on policy-making. She discusses how small experiments can lead to large-scale policy changes by providing evidence-based solutions to alleviate poverty. Duflo shares examples of successful projects, such as evaluating the effectiveness of microcredit programs, improving teaching methods, and reforming auditing systems, which have influenced policy decisions and led to significant improvements in various areas.

Questions & Answers

Q: How did Esther Duflo get into economics?

Esther Duflo initially intended to specialize in history but became interested in economics after working as a research assistant to economists in Moscow. Despite initially distrusting economists, Duflo pursued a degree in economics and eventually became an economist herself.

Q: How has Esther Duflo's career been dedicated to alleviating poverty?

Esther Duflo has devoted her career to conducting practical research on poverty alleviation strategies. She uses field experiments to study various interventions and policies, aiming to find the most effective ways to reduce poverty in its different forms.

Q: How has Duflo been successful in influencing policy?

Duflo has been successful in influencing policy by linking researchers with policymakers. Through her organization, Poverty Action Lab (J-PAL), Duflo has formed partnerships with governments, NGOs, and other stakeholders, ensuring that research findings are effectively communicated and implemented at a policy level.

Q: How does the process of policy influence differ from the traditional model of small experiments followed by policy adoption?

The traditional model of small experiments followed by policy adoption is oversimplified. Duflo explains that the process of influencing policy is much more elaborate and collaborative. It involves ongoing dialogue, co-creation, and adaptation to ensure that research findings align with the needs and realities of policymakers and institutions.

Q: What are some common criticisms of the small experiment to policy adoption model?

Critics argue that small experiments may not have the same effects when scaled up, that results may not be generalizable to other contexts, and that no experiment is perfect. They also question the relevance of research findings to policymakers and emphasize the potential for unintended consequences and political reactions when scaling up interventions.

Q: What impact did the evaluation of microcredit programs have on policy and understanding?

The evaluation of microcredit programs revealed that the average effects on poverty reduction were disappointing. However, the analysis also showed that individuals who already had businesses before accessing microcredit benefited significantly. This research shifted the discourse from microcredit being either a disaster or a miracle to exploring how to make microfinance products more effective for different target populations.

Q: How did testing teaching methods lead to policy impact at scale?

Field experiments in developing countries revealed that the prevailing curriculum was not suitable for most students, leading to low learning outcomes. By testing different teaching methods, such as teaching at the right level, researchers found that adapting the curriculum to individual student levels significantly improved learning outcomes. This research informed the scaling up of programs in India and other countries, supported by partnerships with governments and NGOs.

Q: How did the evaluation of a rice distribution program in Indonesia lead to policy change?

The evaluation of a rice distribution program showed that distributing cards to eligible individuals significantly increased program uptake and reduced leakage. The research also identified the most effective card design, with common knowledge, price information, and wide distribution having the greatest impact. Based on these findings, the government scaled up the program, implementing the recommended changes.

Q: How did Duflo's team address the conflict of interest in auditing firms?

The team identified a conflict of interest in the auditing of firms when firms themselves hired auditors, leading to biased results. They proposed a new system where auditors were randomly assigned and paid based on their accuracy, breaking the link between firms and auditors. This proposal was adopted at a state level in Gujarat, leading to more accurate audits and curbing pollution.

Q: How does Duflo envision the role of researchers in policy-making in the future?

Duflo envisions researchers making themselves less relevant over time by fostering a culture of learning within governments and institutions. She believes that if the culture of evidence-based decision-making becomes widespread, many actors will engage in learning and policy influence, making research less necessary for creating positive change.

Takeaways

Esther Duflo's lecture highlights the power of field experiments in influencing policy and alleviating poverty. She emphasizes the importance of ongoing collaboration, co-creation, and adaptation between researchers and policymakers to ensure effective policy implementation. By conducting rigorous evaluations and providing evidence-based solutions, researchers can make significant contributions to policy decisions that positively impact people's lives. The ultimate goal is to foster a culture of learning where policymakers and institutions value research and prioritize evidence-based approaches to address complex social problems.

Summary & Key Takeaways

• Esther Duflo presents the economics prize as recognition for a broad movement rather than for only three individuals. That movement includes researchers, graduate students, field staff, funders, nongovernmental organizations, governments, and other partners whose combined work connects practical research on poverty with decisions about policies and programs around the world.

• Duflo rejects the simple model in which researchers conduct one small, tightly controlled experiment, publish a polished policy brief, and secure immediate large-scale adoption. Critics reasonably question whether results survive expansion, transfer between locations, attract policymakers' attention, or withstand equilibrium effects and political reactions during implementation at a larger scale.

• Microcredit illustrates why credible policy conclusions often require multiple evaluations. Early enthusiasm and later criticism relied heavily on anecdotes, while rigorous evidence remained limited. After initial studies in the Philippines and India, researchers waited for additional results, published studies in comparable formats, and analyzed whether findings across countries supported a consistent interpretation.