Published on Development Impact

What’s the Latest Research in Development Economics? A Roundup from NEUDC 2023

This page in:

There are so many studies regarding so many aspects of development economics that it can be difficult to keep up. Last week was the North East Universities Development Consortium annual conference, often called NEUDC. Researchers presented more than 130 papers across a wide range of topics, from agriculture to education and from labor to climate; almost all of the studies are available for download. This is a great snapshot of the latest research in development economics.

Where the studies are from and what methods they use

The studies take place all over the world (Figure 1). India has more than twice as many studies (23) than the next highest country, Brazil (with 10 studies). Kenya has eight, Indonesia has six, and Bangladesh, Malawi, and Pakistan each have five. A total of 43 countries are represented (not even including regional or cross-country studies that include dozens of countries). If you examine the studies per country population, the top countries are Guinea-Bissau, Uruguay, Malawi, Chile, and Jordan. (Guinea-Bissau and Uruguay just have one study each but have smaller populations.)

Figure 1. Where are recent development economics studies focused?

Absolute number of studies per country


Studies per ten million population​​


Source: This map draws on a sample of more than 135 studies from the NEUDC 2023 conference. We categorized studies and excluded those that covered more than three countries (often broad global or regional analyses).


The research continues to draw on a wide range of empirical strategies—i.e., not just randomized controlled trials, or RCTs (Figure 2). RCTs are the single largest group, but there are still lots of studies using difference-in-differences, fixed effects, and regression discontinuity. 


Figure 2. What empirical methods do recent development economics papers use? 


Source: This figure draws on 124 studies for which we found it easy to ascertain and categorize the empirical strategy. Some studies used multiple methods, in which case we categorized the two main methods we found.


What we learned from 130+ NEUDC studies

We went through each study, and we provide a micro-summary below. Obviously these are just our quick takes. Many studies have more than one thing to teach us, so if our microsummary piques your interest, click the link to the study! Also, take these micro-summaries with a grain of salt: some of these studies are still preliminary (that’s indicated on the front page of the studies themselves), and we also largely take the studies’ claims at face value (we aren’t peer reviewers). Still, there’s a lot of exciting research here, teaching us more about both problems and solutions in low- and middle-income countries (and beyond). We hope you learn as much reading them as we did writing them!


Guide to the methodological hashtags: 




Event study


Fixed effects


Instrumental variables


Propensity score matching


Randomized controlled trials (including lab-in-the-field experiments)


Regression discontinuity


Synthetic control


Survey experiment


Households and human capital

Education and Early Childhood Development

  • Schools in Chile with more poor students tend to have lower test scores. A program that provides extra funding to those schools benefits disadvantaged students at both low and high levels of support. The results for advantaged students in the same schools are less consistent. (Cerda) #RD
  • A school voucher program in India benefits recipients, but because voucher amounts are linked to schools' tuition fees, schools have an incentive to raise their fees, which hurts non-beneficiary students. The net effect is still positive, but failing to account for the effect on non-beneficiaries would dramatically overstate the benefits. (Sahai) #DID
  • A ten-hour virtual training to help teachers in India better manage emotions, set goals, and solve problems led to a greater belief that they could boost students' learning and put more effort in their teaching. (Kaur) #RCT
  • Lots of families in western Kenya don't have books to read to their kids. When those storybooks were offered, the vast majority of families took them. The more expensive they were, the less likely people were to buy them; but even with a low subsidy, more than 90 percent of families bought them. But three months later, kids still didn't have stronger vocabularies. (Bonds, Hamory, and Ochieng) #RCT
  • In Tanzania, 10 percent more stagnant water increases diarrhea incidence among children by 30 percent and reduces test scores by 7 percent of a standard deviation. Access to improved sanitation and water sources mitigates the effect of stagnant water on these health and learning outcomes, but this effect increases with high temperatures and population density. (Berggreen and Mattisson) #DID
  • A “learning how to learn” approach in Uganda raised the pass rate in the national exam (progression from primary to secondary school) from 51 percent to 71 percent. The approach “trained teachers to teach students to learn like scientists: posing sharp questions, framing specific hypotheses, using evidence and data gathered from everyday life whenever possible.” (Ashraf, Banerjee, and Nourani) #RCT
  • In Chile, “classroom peers and older high school peers significantly shape students’ choices of college majors in male-dominated fields.” (Valdebenito) #RD
  • In Peru, a 20-minute interaction between engineering students and high school students increased students’ preferences towards engineering, especially among female students with high math aptitude. (Agurto et al.) #RCT
  • In India, primary students taught by college students scored 0.34 SD higher in math, 0.22 SD in science, and 0.15 SD in language higher than those taught by regular teachers. (Ganimian, Mbiti, and Mishra) #RCT
  • A country-wide effort to improve toilet access and other aspects of sanitation in India (the "Clean India Mission") boosted children's math—but not literacy—scores. Results are similar for boys and girls. (Karmakar and Villa) #DID #ES
  • Eleven months after introducing a program in India to encourage teachers to blend their teaching with high-quality videos, student math scores got worse, teachers taught less effectively (e.g., they gave worse feedback to students and monitored student learning less), and students had worse attitudes towards science and math. (de Barros) #RCT
  • In the Dominican Republic, all Major League Baseball teams run training academies for adolescent boys. Despite public perceptions that this leads youth to undervalue formal education, exposure to these academies has no measurable impact on school attendance. (Marein and Palsson) #DID
  • A common policy to get more underrepresented groups into college is to rank students within high schools (so that kids from poorer high schools have a shot). But in Chile, high school students switch schools to game the system, such that the current policy had a tiny impact, which would have been more than five times as large if students weren't switching schools. (Concha-Arriagada) #FE
  • With the introduction of a new bus line and a new train line in Lima (Peru), "a 17 percent reduction in commuting time to college increases enrollment rates by 6 percent," with that mostly driven by enrollment in private colleges. Boys are willing to travel 50 percent longer than girls to attend a better college. (Alba-Vivar) #DID
  • Study groups in Chinese primary schools boost student achievement. The effects are biggest for kids who were previously not doing so well in school but who are in high-achieving study groups. It also boosts those kids' level of motivation. (Gao et al.) #IV


  • An additional year of education for women in India led to "a 27% decrease in less severe physical domestic violence, a 9% decrease in sexual violence, and a 10% decrease in injuries due to domestic violence." This was likely due to women finding better partners, improved attitudes, and potentially a higher likelihood of reporting violence. (Bergonzoli, Bahure, and Agarwal) #RD
  • Higher export prices for crops grown by women in Peru reduce "domestic violence, including severe physical violence and female homicide... The effects are stronger in districts with more unequal gender norms." (Frankenthal) #FE
  • In India, short-term exposure to domestic violence doesn't seem to affect women's attitudes about violence, but over time, women who experience violence are more likely to tolerate violence—potentially as a coping mechanism. (Frezza) #RD
  • Girls' clubs for adolescents in Côte d'Ivoire alone boosted some girls' mental health but not employment outcomes. Adding separate clubs for husbands and future husbands boosted employment and sexual and reproductive health outcomes. (Boulhane et al.) #RCT
  • Women living in neighborhoods with low risk of harassment or assault on the streets are 8.5 percentage points more likely to participate in the labor market relative to women in higher risk neighborhoods—in Indonesia and India. (Cahill) #PSM
  • A successful school-expansion program in India “increased voter prioritization of leader competence over gender, boosting the share of women among candidates and state parliamentarians and the overall capability of elected officials.” (Anukriti, Calvo, and Charavarty) #RD
  • In Pakistan, providing correct information about support for women voting in society to men does not change the turnout of women, while providing the same information to women or both lowers female turnout. “This blow-back effect is caused by men discouraging women from political participation because they believe women will act according to their own (and different from men’s) preferences.” (Gulzar, Khan, and Sonnet) #FE
  • An anti-poverty program in Malawi improved households’ consumption, food security, and dietary diversity outcomes, regardless of whether men or women were targeted and whether a gender transformative training was incorporated in the program. (Bedi, King, and Vaillant) #RCT
  • A conditional cash transfer (CCT) program in Peru reinforced traditional gender-role attitudes among children, especially girls. Beneficiary mothers spent more time on home production, and this priority could be a channel for perpetuating traditional gender role attitudes. (Luong) #RD
  • Families historically exposed to higher levels of slavery in Guadeloupe and Martinique tend to be more matrifocal, with weaker fathers after emancipation. These families also face higher child mortality, stemming from poorer family environments rather than local conditions. (Beigelman) #FE
  • “On an online marriage market platform in India, ... working women are 14.5% less likely to receive interest from male suitors... Women employed in ‘masculine’ occupations are 3.2% less likely to elicit interest from suitors relative to those in ‘feminine’ occupations.” (Afridi et al.) #RCT
  • Providing "mentored girls' clubs, life skills, and vocational training" to adolescent girls in northern Nigeria reduced marriage two years after the intervention by 65 percentage points! A major reason is that it boosted girls' likelihood of staying in school. (Cohen, Abubakar, and Perlman) #RCT
  • A radio campaign against female genital cutting in Egypt reduced cutting by 13 percent. But those girls also saw their bride prices fall by nearly a quarter and child marriages rose. Across Africa, cutting is more common in settings with bride price. (Khalifa) #DID
  • Several years after the start of a cash transfer program in Malawi, those who benefited as adolescent girls were more likely to move longer distances for marriage, and those moves happened over a longer period of time, into young women's mid-twenties. (Ibrahim) #RCT
  • Parents in Pakistan prefer grooms who are relatives; they also prefer marriages where their daughters will have more freedom (e.g., to leave the house unaccompanied or to make health decisions independently). (Calvi, Farooqi, and Kandpal) #SurveyExperiment
  • In Lahore (Pakistan), women are more likely to sign up for a job search platform than they are to take a job. For less-educated job seekers, jobs with explicit gender requirements are more likely to exclude women. Women with more education are more selective about jobs, but the jobs themselves are less likely to be gender-restrictive. (Gentile et al.) #RCT
  • Offering women in West Bengal (India) the ability to multitask work with childcare and to work from home boosts labor market participation, especially for those from more traditional households. Flexible work also increases women's likelihood of accepting out-of-home work later. (Ho, Jalota, and Karandikar) #RCT
  • Increasing the minimum wage by nearly a quarter in Morocco narrowed the gender pay gap in the formal sector by about a quarter. But it also led a small fraction of women (but not men) to leave the formal labor market. (Paul-Delvaux) #DID
  • In India, horizontal communication between cisgender participants reduces discrimination against transgender people: when involved in a group discussion with two neighbors, there is no longer discrimination at all on average, even when making private choices. This effect is 1.7x larger than top-down communication that informs participants about the legal rights of transgender persons. (Webb) #RCT
  • In a lab in the field experiment in Bangladesh, the “less capable women are perceived compared to men, the less they are involved in decision-making. After the information treatment (on women’s abilities), individuals with the lowest perceptions about the wife’s skills are 20 percent more likely to make allocations in her favor.” (Nani) #RCT
  • Employers in Bangladesh discriminate paternalistically, i.e., they restrict women’s employment choices to protect them from what employers perceive as unsafe. Informing about safe worker transport at the end of the shift increases demand for female labor by 22 percent and female labor supply by 15 percent. (Buchmann, Meyer, and Sullivan) #RCT
  • Offering female entrepreneurs in Jordan access to virtual storefronts by (1) creating and managing Facebook pages for their businesses and (2) offering them digital marketing training created in collaboration with local influencers, increased business survival and revenues. Effects are driven by women with low physical mobility due to social norms at baseline. (Alhorr) #RCT

Health (including mental health)

  • Sharing information about either "the prevalence of mental health issues and the efficacy of treatment" or "the mental health struggles of a Nepali celebrity and how he benefited from treatment" boosted people's stated willingness to seek mental health treatment and to pay for counseling. The effects were biggest for those with "less severe symptoms of depression and anxiety." (Lacey et al.) #RCT
  • Sometimes people are uncertain about how risky a behavior is. In Malawi, people with greater uncertainty about how risky regular unprotected sex with an HIV-infected partner update their beliefs more drastically in response to new information. (Kerwin and Pandey) #RCT
  • Groups of friends among Syrian refugees in Jordan are good at identifying who needs mental health support. Sometimes friends don't want to share info about mental health services because of stigma, but if you nudge the sharers to disclose that they're being financially compensated for sharing the info, they're more likely to share it. (Smith) #RCT
  • Home visits inviting adults in Kenya to get vaccinated against COVID increased vaccine doses by ten percent, especially among women and people with less education or income. Announcing the home visit ahead of time (so people could either making a point of being home for the appointment or being out to avoid it) further boosted vaccinations. (Carney et al.) #RCT
  • Providing frontline health workers in Guinea-Bissau "with evidence of their program’s effectiveness in improving local health indicators" significantly boosted the effort of health workers, even six months later. (Fracchia) #RCT
  • A "mother and child health and family planning program in Bangladesh" boosted height in adulthood for those who participated as children (as well as education among the men). In the next generation, daughters of beneficiaries tend to be taller and have better cognitive development. (Barham et al.) #DID
  • Introducing “a competitive selection system for recruiting CEOs in public hospitals in Chile reduced hospital mortality by 8%”, driven by hospitals in which the new CEOs had managerial qualifications. “These CEOs improved operating room efficiency and reduced staff turnover.” (Otero and Muñoz) #DID
  • In Rajasthan (India), posting of a mid-level health care provider increased the monthly patient load by 68 percent. The number of patients diagnosed with hypertension and diabetes at treatment facilities increased as well. After one year, elderly deaths declined by 12 percent. (Agte and Soni) #DID

Migration and refugees

  • Mexican regions experiencing larger inflows of Mexican low-skilled deportees have higher rates of firm creation, firm survival, and revenue. (Osuna-Gomez and Medina-Cortina) #FE
  • In Vietnam, the Ho Khau reform, which reduced migration barriers, had more impact on reducing spatial inequality than place-based incentives. (Huynh) #DID
  • In Uganda, “landslides increase long-term displacement and migration, and affected households have substantially worse economic and mental health outcomes years afterward.” (Baseler and Hennig) #FE #DID
  • In the Philippines, typhoons increase international migration from affected municipalities, and incentivize migrants to leave for lower paying overseas jobs. (Murathanoglu) #FE
  • Municipalities from which more Moroccan soldiers were deployed to France before independence were more likely to send emigrants to France after independence. But this wasn’t true with those sent to Algeria. (Salem and Seck) #FE

Safety nets (including cash transfers and peer support)

  • How do you  target disaster aid to households? In post-2015 earthquake Nepal, the “property damage criterion excludes many liquidity-constrained households that have high demand for aid, and it includes wealthy, well-insured households that have low demand.” Dividing aid equally among all affected households has larger welfare gains. (Gordon, Hashida, and Fenichel) #RD
  • In Mumbai (India), residents who received a subsidy that could only be used for rice or wheat spent less on packaged snacks. The effect was bigger in households with children. (Aouad, Ramdas, and Sungu) #RCT
  • With access to safety nets, middle-income households in Colombia are more likely to borrow from formal lenders, and in the long run, they substitute away from predatory loans toward formal loans when experiencing severe shocks. (Álvarez et al.) #RD
  • An unconditional grant to poor kids in South Africa reduced the likelihood that girls would be underweight or obese, but it boosted the likelihood that they'd be overweight. There were no substantive effects for boys on average. (Chakraborty and Villa) #RD
  • Peer groups in Nepal where most people know each other don't necessarily choose a peer monitor in lab games. But groups where few people are immediately connected do. (Iacobelli and Singh) #RCT #DID
  • Using data from six low- and middle-income countries, a proxy means test (PMT)—i.e., using a short list of household characteristics to decide if a household is poor—fails to accurately predict eligibility for social protection programs: “the accuracy of the PMT prediction algorithm decreases steadily over time, by roughly 1.7 percentage points per year,” even though the PMT model is updated only every 5-8 years. (Aiken, Ohlenburg, and Blumenstock)


Working and saving

Banking and credit

  • Encouraging commercial banks in India to increase lending to minority borrowers in “minority concentration” districts in India increased minorities’ access to bank credit. (Khan and Ritadhi) #RD
  • In Kenya and Pakistan, equity-like contracts stimulate more profitable investments. Risk preferences play an important but nuanced role: loss-averse individuals prefer equity; however, individuals exhibiting non-linear probability weighting prefer debt. (Meki) #RCT
  • In Ghana, individual incentives increased adoption of a new technology; adding endorsement by a trusted peer doubles the impact of the individual subsidy. (Riley, Shonchoy, and Darko Osei) #RCT
  • Lots of people in urban India don't have access to credit for when there's a financial crunch or to professionals for mental health problems. Many would like to talk to the people they know, but—from a survey—68 percent underestimate others' willingness to engage on these topics. Helping people realize that boosts sign-ups for potential savings groups or for a potential program to get trained to be a volunteer to listen to other people's anxieties. (Jain and Khandelwal) #RCT

Firms and microenterprises

  • Fifteen years after the Indian Ocean tsunami, Indonesian business owners exposed to the tsunami had lower levels of capital and profits than those not exposed. These effects were biggest in rural areas. (Lombardo, Frankenberg, and Thomas) #FE
  • Providing firms in Addis Ababa (Ethiopia) with more information about candidates with college degrees led firms to hire faster, but they often hired candidates without college degrees and downgraded their expectations about college graduate productivity. (Wu and Wang) #RCT
  • For US firms, entering the whaling industry entails lots of sunk costs such that firms are slow to enter but also slow to leave the market. (Yun)
  • In Kenya, “informality is particularly prevalent in downstream economic activities and smaller regional markets.” (Wiedemann et al.) #DID #SC
  • Vendors in the Kolkata (India) vegetable market subsidized to sell additional produce earned over 60 percent higher profits, after excluding the value of the subsidy. And yet, “after the subsidy ended vendors largely stopped selling the additional produce” which may reflect “social or private preferences”. (Banerjee et al.) #RCT
  • In Tanzania, rural firms smooth both negative and positive input price shocks more than urban firms. Urban firms pass nearly 95 percent of input price increases to customers, while rural firms pass only 55 percent of input price increases. (Rudder) #FE
  • Some firms in Sri Lanka have much higher returns to inputs than others, and new econometric tools to test how much putting inputs into the wrong firms affects growth suggest that output could quadruple with better allocation of resources. (Hughes and Majerotvitz) #RCT
  • When steam power started replacing water power in the United States, water-powered mills shut down rather than transforming to steam mills, suggesting that shifting technologies is costly. (Hornbeck et al.) #FE
  • In Malawi, a survey of shopkeepers shows that they have widely varying strengths across different dimensions of productivity (like attracting customers, managing a storefront, and maintaining inventory across many products). So a one-size-fits-all management training intervention may have disappointing results. (Huntington, de la Parra, and Shenoy) 

Labor (including child labor)

  • In Uganda, an experiment with job referrals reveals that gender discrimination exists in both directions (favoring the majority gender in a given sector) but that it's worse in male-dominated sectors. But when men can make referrals anonymously, they discriminate less. (Alfonsi and Ferreira) #RCT
  • A subsidy to help people find jobs in Ethiopia seems to have smaller effects when more people participate (i.e., general equilibrium effects). The net impact on people's wellbeing still seems to be positive. (Van Vuren) #RCT
  • If the unemployment rate is 1 percent higher at the time a person first starts looking for a job in Indonesia, that person is likely to have worse long-term economic prospects: 21 percent lower income and 7 percent lower likelihood of being employed. These effects are smaller than those in high-income settings. (Marshan) #FE
  • When farmers in Burundi train their workers, they don't capture all the benefits, since the workers often take those skills and work for others. A contract that keeps workers with the same employer boosts employers' willingness to train massively (by 50 percentage points). (Cefala et al.) #RCT
  • A comprehensive training and mentoring program targeted to youth at risk in Rio de Janeiro (Brazil) boosts employment among men but it also—surprisingly—boosts fertility and subsequently welfare receipts among women. (Barros et al.) #RCT
  • How much a person is willing to work may in part just be driven by getting used to working. In Chennai (India), providing casual workers with incentives to work over a couple of months boosted how much they worked during that time (by 23 percent), and those same workers worked 16 percent in the following two months, after incentives were removed. (Cefala et al.) #RCT

Governments, institutions, and conflict

Conflict and crime

  • Which connections do people value most in times of crisis? During social unrest in Haiti, daily contacts decrease but total talk time remains constant. Checking in on close friends, family, and others remains a priority. (Putman and Lybbert) #DID
  • Historical genocide violence of the Cambodian population impacts today’s household wealth and village-level poverty rates, driven by lower human capital, limited structural transformation, and restricted public goods provision. (Mehrotra) #IV
  • In Rio de Janeiro (Brazil), installing peacemaker units of police officers reduced crime rates, which boosted available credit, which in turn further reduced crime. (Tomkowski, Monteiro, and Caluz) #DID
  • Sanctioning public acts of repression, such as beating or arrests of protesters, can encourage a regime to prevent protest through less-public means, such as obstruction or harassment of organizers (based on data from 150 countries). (Andirin et al.) #FE
  • The political leaning of Brazilian players during the 2022 World Cup affected fans’ reactions depending on their political alignment. Celebrations after goals scored by a player were more intense among politically congruent fans. (Ajzenman, Ferman and Sant'Anna) #RCT
  • In southern Ethiopia, areas with more droughts also have more violent conflict. But when herders get and use insurance to protect against drought, it reduces conflict significantly (between 17 and 50 percent). (Sakketa, Maggio, and McPeak) #RCT
  • A law in Brazil requiring firms with more than a hundred workers to reserve at least two percent of their job openings for people with disabilities increased employment of people with disabilities. When firms got fined for not meeting the quota, nearby firms—even those not covered by the law—also boosted their disability employment. (Berlinski and Gagete-Miranda) #DID
  • In Uruguay, introducing individual retirement accounts as a complement to the traditional defined benefit pension led more people to work into their fifties. It also may have reduced tax evasion. In early old age, people had similar income levels across the two systems, but when people had the chance to go back to the defined benefit system, many did. (Lauletta and Bérgolo) #RD
  • If the government doesn't enforce people sticking to contracts, how do they work? Gamblers on horse races in Pakistan provide a window, since such betting is illegal and so the government doesn't enforce the contracts. "Even in the absence of legal enforcement authority, personal relationships, and violence, more than 70% of gamblers fulfill their contractual obligations." (Mehmood and Chen) #RCT
  • South African municipalities with higher historical exposure to post-apartheid Truth and Reconciliation Commissions on media have lower levels of violence today. (Gautier, Horta-Saenz, and Russo) #FE
  • Losing a sibling during the 1994 Rwandan Genocide leads to more schooling and more siblings born after the genocide. (Gautier) #IV

Regulation, tax, and government

  • In Indonesia, “electoral defeats of the incumbent village head increase turnover in the village bureaucracy and reduce nepotism.” (Bazzi et al.) #RD
  • In India, after a bureaucrat is transferred to an important ministry with the power to make influential policies, the annual growth rate of the value of the bureaucrat’s assets is 10 percent higher, and it’s 4.4 percent higher for the number of their assets. (Chaudhury and Yuan) #DID
  • In China, “over 65% scoring auctions in public procurement show evidence of scoring rule manipulation.” (Chen) #FE
  • In Brazil, “high-ability students in (anti-corruption) audited municipalities are less likely to choose majors tailored toward public sector careers, such as business administration and law.” (Xun) #DID
  • In rural Bangladesh, the introduction of Village Courts more than doubled the share of disputes resolved in state-sanctioned courts, but the informal dispute resolution mechanism “shalish” remains most used. (Mattsson and Mobarak) #RCT
  • A comparative welfare analysis of 40 policies implemented in low- and middle-income countries since 1997 shows that the marginal value is low. (Morris) 
  • Successful democratizations lead to substantial redistribution: the size of the public sector grows, income inequality falls, and the labor share of income rises, according to data from 90 countries. (Miller) #DID
  • In India, “local democracy aligns spending more closely with citizen preferences, but these gains accrue more to men, upper castes, and other advantaged social groups.” (Arora et al.) #FE #DID
  • In Brazil, “when a young politician is in office, there is a reduction in deforestation and greenhouse gas emissions without significant effects on local GDP.” (Dahis, de las Heras, Saavedra)  #RD #DID
  • Appointing “a new city party secretary (PS), who serves as the leader of the local Chinese Communist Party (CCP) organization, is associated with a significant increase in the revealed comparative advantage in industries where the PS’s previous work location exhibited better performance.” (Lin et al.) #FE
  • Electoral turnovers improve country performance. “Electing new leaders leads to more policy change, it improves governance, and it reduces perceived corruption,” based on data from over 4,000 national elections since 1945. (Marx, Pons, and Rollet) #RD
  • In Senegal, bureaucrats with full discretion for building the lists of potential property tax payers tend to undervalue properties, and they do so even more for higher-value properties, resulting in a regressive tax profile. “In contrast, a rule-based system where bureaucrats record property characteristics (not values) that an algorithm then uses to compute values, significantly reduces this tax gap.” (Knebelmann, Pouliquen, and Sarr) #RCT
  • In Uzbekistan, mandating the use of online electronic fiscal devices, which provide regulators with real-time information on business transactions, increases company revenue reports to tax authorities by 13 percent. Adding a direct communication channel with citizens and financial rewards to act as enforcement agents increases firms’ reported revenues by an additional 34 percent. (Kobilov) #DID 
  • Research projects developed in partnership with policymakers are 17 to 20 percentage points more likely to result in observed policy change. (Bonargent)


  • Do religions codify ecological principles? In Benin, a 1 SD increase in African Traditional Religions adherence has a 0.4 SD positive impact on forest cover change. (Deopa and Rinaldo) #IV
  • When Israel reduced child allowances to large families in 2003, Jewish families substituted by enrolling their young boys in ultra-Orthodox religious schools. In the long-term, fewer boys enrolled in high schools, without affecting families’ fertility or labor supply decisions. The reform led to a 13 percent decline in completed fertility among Arab families. (Gershoni et al.) #RD
  • In Brazil, exposure to a church-affiliated TV channel increases fertility rates, lowers female labor force participation, lowers schooling for young women in the next generation, and leads to more votes for Pentecostal candidates. (Mello and Buccione) #DID
  • In Brazil, the removal of progressive Catholic leaders halted the land invasion movement, a conflict in which poor and landless peasants invaded large landholdings to force land redistribution. (Martinez-Bravo, Solá, and Tuñón) #FE
  • Hate or fear? In the ongoing conflict between Christians and Muslims in Jos (Nigeria), fear explains 76 percent, and hate 24 percent of the non-cooperative behavior. (Ortiz) #RCT 


  • Climate change leads to welfare losses of 4.8 percent of GDP across 271 regions in sub-Saharan Africa, with country-level losses as high as 43.8 percent of GDP. “Lowering trade costs can offset these losses by connecting deficit regions to surplus regions and the world market.” (Porteous)
  • In South Korea, under a better bureaucrat, exports increased by 40 percent. "In subsequent appointments, exports increase in products with greater bureaucrat experience." (Barteska and Lee) #FE
  • Giving information on tariff costs and local prices to traders (via a cell phone platform) at the Kenya-Uganda border increases switching across markets and routes, leading to large increases in traders’ profits and significant formalization of trade. (Wiseman) #RCT

Agriculture, infrastructure, and the environment

Agriculture and land

  • In India, redistribution of land ownership led to an overall increase in durable asset ownership, nonfarm employment and years of schooling, including among lower-caste descendants of households that did not receive land. (Batra) #RD
  • Households’ agricultural production in Côte d’Ivoire improves during the agricultural season overlapping with oxen delivery, and increases in land holdings and input use in the subsequent season. (Brudevold-Newman, Donald, and Rouanet) #RCT
  • In Kenya, an information campaign (training farmers to identify hybrid maize seeds that are quality-verified) improved farmers' purchasing decisions and led to gains in maize yields. While improved information caused sellers to exit the market, there are no effects on prices or quality among stayers. (Hsu and Wambugu) #RCT
  • Increasing agricultural technologies cannot rely on market prices as a mechanism for targeting high-return farmers. In Bangladesh, when farmers receive a new wheat seed variety for free, they adopt it as much as farmers that chose to buy it at a subsidized price. (Mahmoud) #RCT
  • Improving the allocation of inputs (like land, labor, and credit) across farms in Thailand could boost productivity by more than a third. But improving allocation of multiple inputs is more productive than focusing entirely on improving allocation of just one input. (Silver) 
  • A government program to subsidize the price of fertilizer encouraged farmers to specialize based on their comparative advantage: some boosted their agricultural yields; others left agriculture. (Diop) #DID
  • Social networks are key to people adopting new agricultural technology. But in Malawi (and a theoretical model predicts similar results elsewhere), using people with lots of connections to get farmers to adopt new seeds works best when lots of farmers in the network have similar farms (so that the seeds are similarly likely to work). (Chakraborty) #FE

Climate and pollution

  • Increasing water prices led to a decline in water use of richer households in Cape Town (South Africa). However, richer households substituted municipal water with privately drilled groundwater. (Cole et al.)
  • In Mozambique, “providing information on flood risk [such as physical damage and disease outbreaks] increased the implementation of suggested mitigation strategies.” (Leefers) #RCT
  • In the Philippines, large-scale tree planting reduced regional poverty and increased economic activity. (Pagel and Sileci) #DID
  • In Vietnam, rising temperatures lead more workers to move from agriculture into other industries, both in the short run and the long run. (Pham) #FE
  • Reducing localized emissions from vehicles within India’s 10 largest cities leads to a 3 times larger GDP gain than a policy controlling agricultural fires. “Further accounting for labor reallocation leads to a 6 times larger GDP gain.” (Tiwari)
  • In Colombia, “extreme temperature events increase the frequency of land sales and decrease the average farm size within municipalities.” (Arteaha et al.) #FE
  • In India, industrial water pollution does not seem to affect crop yields. Farmers respond to industrial water pollution by switching irrigation sources from surface water to (costly) groundwater and expanding irrigation. (Hagerty and Tiwari) #RD
  • To address sea level rise, the Indonesian government proposed a sea wall. Moral hazard generates severe lock-in and limits migration inland, even over the long run. (Hsiao) #RD
  • The 2009 tightening of environmental standards in the US shifted used lead-acid battery recycling, an industry that emits large amounts of lead pollution, to Mexico. Lead pollution exposure reduced students’ text scores by 0.05-0.09 standard deviations. (Litzow et al.) #DID
  • The Green Municipalities Program in the Brazilian Amazon (PMV) increased secondary forest cover (in places that have been previously deforested) by 9 percent. (Shinde et al.) #DID
  • In Nairobi (Kenya), improved stove ownership reduced high-frequency particulate matter (PM) exposure from 122 µg/m3 to 49 µg/m3, with a 0.24 SD reduction in self-reported respiratory health symptoms. (Berkouwer and Dean) #RCT
  • The beef cattle sector drives deforestation worldwide. In the Brazilian Amazon, large intermediaries drive down prices for farmer and cattle supply, but the deforestation frontier is largely competitive and thus emissions remain unaffected. (Barrozo) #IV

Infrastructure and transport

  • Chinese infrastructure projects significantly increased nighttime light in the African recipient regions, and the effects persist over time. World Bank projects, however, did not exhibit significant impacts on nighttime light. World Bank and Chinese infrastructure projects both positively influence women’s education attainment and health. (Chai and Tang) 
  • Motortaxi drivers in Uganda admire other drivers who speed. Drivers are more likely to want financial incentives to limit speeding if those incentives are visible, so they can use them to justify their reduced speed to other drivers. (Raisaro) #RCT
  • In Mexico, urban localities that spent an additional month downwind from a 0.1 degree × 0.1 degree grid cell that has adopted conservation agriculture experience a 1.3 percent reduction in the number of infant deaths. (Ferguson and Govaerts) #ES #DID
  • Discontinuous incentives around the range thresholds of Chinese driving-range-based subsidies made low-end electric vehicle manufacturers’ invest in reducing the production costs of driving ranges by 30 percentage points. (Zhang) #FE
  • “Commuters in Jakarta (Indonesia) are 2-4 times more sensitive to wait time compared to time on the bus, and inattentive to long routes.” (Kreindler et al.) #DID
  • “Winning random lotteries for the ownership of condominium houses in Ethiopia leads to significant gains in educational attainment: educational enrollment increases by 4.5-11%, secondary school completion rates by 10.5% and post-secondary attendance rates by 16%.” (Agness and Getahun) #IV

The order of authors on this blog was determined by a virtual coin flip. This blog post benefited from research assistance by Amina Mendez Acosta and editing by Jeremy Gaines. A version of this post will also appear on the CGD Blog.



Almedina Music

Senior Economist, Education Global Practice

David Evans

Senior Fellow, Center for Global Development

Join the Conversation

The content of this field is kept private and will not be shown publicly
Remaining characters: 1000