Development Research Group

This study tests the effects of large language model (LLM) decision support on patient care at two outpatient clinics in Nigeria. Health workers were given the option to make revisions to their initial care plan based on LLM feedback... (Read more).

Can algorithms enhance bureaucrats’ work in developing countries? In data-poor environments, bureaucrats often exercise discretion over key decisions, such as audit selection. Exploiting newly digitized micro-data, this study conducted an at-scale field experiment... (Read more).

Rice is a staple crop for over half the world’s population, and accurate, timely information on its planted area and production is crucial for food security and agricultural policy, particularly in developing nations like Sri Lanka. However, reliable rice monitoring in regions like Sri Lanka faces significant challenges due to frequent cloud cover... (Read more).

Worldwide, account ownership increased by 50 percent in the 10 years spanning 2011 to 2021, to reach 76 percent of the global adult population. The goal of financial inclusion is not just for more adults to have accounts but for account owners to benefit from using them... (Read more).

An online platform—entitled NaijaCare—offered digital ordering, business development, peer exchange and business skills training to informal medicine vendors in Lagos, Nigeria... (Read more).

This paper develops an economic evaluation framework to assist in establishing the economic value of Digital Health Interventions(DHIs) to inform policy... (Read more).

We explore prompt designs for a lightweight integration of a large language model (LLM) into clinical decision support in primary care in Nigeria. The LLM integration is designed to give immediate, actionable "second opinions" to frontline healthworkers... (Read more).

Digital technologies offer the possibility of a new era in development. The April 2024 edition of the World Bank research newsletter makes the case that investing in digital infrastructure and skills will be critical for countries... (Read more).

Large Language Models (LLMs) are quickly becoming ubiquitous, but the implications for social science research are not yet well understood. This paper asks whether LLMs can help us analyse large-N qualitative data from open-ended interviews... (Read more)

Ravi is the chief innovation officer at EHA Clinics, a startup in Nigeria that blends technology solutions with skilled clinicians to deliver high-quality primary care in the private sector. EHA Clinics is collaborating with Anja Sautmann on an AI project in Nigeria. 

Daniel leads the AI and Development initiative at CDEP which incubates projects using technology to better serve the world's poor. 

Ezequiel serves as the regional focal point for Teachers and Educational Technology (EdTech) at the Global Education Department of the World Bank.

Since 2011, the Global Findex Database has been the definitive source of data on global access to financial services from payments to savings and borrowing.

DRELM ensures the provision of quality education with equity; it oversees and supervises the implementation of educational policies and regulations, promoting the holistic development of students.

Drawing on the development experience and advances in economic analysis since the 1950s, World Development Report 2024 identifies what developing economies can do to avoid the “middle-income trap.” Lower-middle-income countries must go beyond investment-driven strategies—they must also adopt modern technologies and successful business practices from abroad and infuse them across their economies. Upper-middle-income countries need to accelerate the shift to innovation by pushing the global frontiers of technology. This requires reconfiguring economic structures governing enterprises, labor, and energy use to enable greater economic freedom, social mobility, and political contestability.

This chapter discusses and quantifies the role of entry barriers and productivity-constraining distortions in South Asian countries’ formal sectors to rationalize the pervasiveness of informality 

The role of firm productivity in enhancing living standards and driving sustained economic growth is widely acknowledged. Extensive academic research, newspaper articles, and influential reports have highlighted the various channels through which firm productivity contributes to prosperity. However, the factors driving firm and aggregate productivity still need to be understood.

A “missing middle” is often interpreted as having too few “mid-sized firms” and sometimes raise calls for policies focusing on small and medium enterprises. While these policies can be useful, the issue of the “missing middle” should be interpreted more broadly – a preponderance of small informal firms and larger firms that fail to absorb the workforce. A “missing middle” indicates that there is scope for shifting workers from small enterprises into medium-sized and larger ones.  Given that larger firms tend to be more productive due to economies of scale, this shift would imply a productivity gain.

This paper develops a scalable method to estimate local greenhouse gas emission changes using satellite-based CO₂ and CH₄ data from NASA’s OCO-2 and ESA’s Sentinel-5P. By analyzing monthly anomalies and grid-level trends, the study identifies significant emission patterns across 242 countries and subnational units. Results show declining long-term CO₂ and CH₄ trends, though CO₂ increased recently, highlighting regional disparities and supporting data-driven climate mitigation strategies worldwide.

Climate modeling suggests that achieving international climate goals requires a reduction in current CO2 emissions by over 90%, with any remaining emissions to be addressed through carbon dioxide removal (CDR) solutions. Sixteen CDR strategies are evaluated by integrating technical potential, environmental, social, and governance (ESG) criteria, along with sequestration permanence. This evaluation, conducted by ENGIE’s scientific council using an interdisciplinary Delphi panel methodology, proposes a “quality” measure for each technology. This measure combines ESG scores and sequestration timescales to rank and select the most promising solutions. The findings highlight the necessity for further research to understand and mitigate ESG impacts, aiming to inform both future research and current decision-making to support the effective and legitimate use of CDR strategies.

This paper evaluates alternative options for rebating revenues from a unilateral emissions price, focusing on energy-intensive and trade-exposed industries. A theoretical model is developed to demonstrate that conditional rebating policies---which would be distortionary in a first-best world---may be welfare-improving. The findings indicate that from a domestic perspective, rebating emissions revenues proportionately to firm output is typically superior to other rebating options when the emissions price is set close to the social cost of emissions.

This paper's analytical model reveals how outcomes differ depending on the nature of the emissions trading systems (ETS) and the types of policies with which it overlaps. It highlights when overlapping policies may have counterproductive impacts with rate-based ETSs, and when they can enhance effectiveness in ways that differ from the effects with mass-based trading systems. The insights are complemented by results from a numerical general equilibrium model representing important characteristics of the recently implemented Chinese nationwide ETS context. 

While countries increasingly commit to pricing greenhouse gases directly through carbon taxes or emissions trading systems, indirect forms of carbon pricing—such as fuel excise taxes and fuel subsidy reforms—remain important factors affecting the mitigation incentives in an economy. Taken together, how can policy makers think about the overall price signal for carbon emissions and the incentive it creates? We develop a methodology for calculating a total carbon price applied to carbon emissions in a sector, a fuel, or the whole economy.

Carbon-pricing policies worldwide are increasingly coupled with direct or indirect subsidies where emissions pricing revenues are rebated to the regulated entities. This study analyzes the incentives created by two novel forms of rebating that reward additional emission intensity reductions: one given in proportion to output (intensity-based output rebating) and another that rebates a share of emission payments (intensity-based emission rebating). These forms are contrasted with output-based rebating, abatement-based rebating, and lump-sum rebating. Given the same emission price, intensity-based output rebating incentivizes the most intensity reductions, while abatement-based rebating causes the most output reductions, and output-based rebating puts the least pressure on output (and emissions); intensity-based emissions rebating lies in between these, by implicitly subsidizing emissions while incentivizing intensity reductions. The study supplements partial equilibrium theoretical analysis with numerical simulations to assess the performance of different mechanisms in a multisector general equilibrium model that accounts for economy-wide market interactions.

This study investigates the impact of information provision on household behavior in connection with indoor and outdoor air pollution in Tbilisi, Georgia. The study implemented a randomized controlled trial to assess whether providing households with different levels of pollution information changes their knowledge and behavior around air pollution, and improves their health outcomes. The study evaluates three treatments: a pamphlet with general information, the pamphlet combined with daily text messages about local outdoor pollution, and the pamphlet with messages about both indoor and outdoor pollution levels, supplemented with an indoor air pollution monitor. The findings show that while the pamphlet alone did not lead to behavioral change, daily text messages significantly enhanced knowledge about pollution, led to increased avoidance behaviors, and improved health outcomes. 

Air pollution is a major threat to health and the dangers are particularly acute in low- and middle-income countries where levels of exposure tend to be high and adaptation resources are often limited. However, little is known about how the burden of pollution is spread across different income groups within these countries. Understanding who is impacted by air pollution is important for designing equitable policy solutions. In this study, we used data providing high-resolution wealth estimates for more than 100 countries, combined with high-resolution estimates of air pollution, to estimate how wealth is correlated with ambient air pollution in low- and middle-income countries around the world. 

This paper develops a global CO₂ emissions model using satellite data at 25 km resolution, incorporating industrial sources, fires, and population-related factors linked to household incomes and energy use. It evaluates the impact of subways in 192 cities, finding they reduce population-related CO₂ emissions by 50% in those cities and 11% globally. Future subway expansion is analyzed under varying cost assumptions, showing substantial climate benefits. Even under pessimistic scenarios, many cities achieve significant emissions reductions, justifying subway investments on both climate and traditional grounds, such as traffic and air pollution reduction.

This paper identifies high-priority areas for methane emissions reduction and assesses recent emissions changes using Sentinel-5P satellite data. It examines 775 key areas, with case studies on landfills in Madrid, rice production in India, and oil production in Iraq. Two fixed-effects models estimate monthly trends (2019–2022) and annual differences. Findings show increasing emissions outnumber decreases four to one, except in gas and oil sectors. High-income economies see more reductions, while other groups experience widespread increases. The study provides emissions trend reports and maps, serving as a template for tracking progress toward the Global Methane Pledge.

Exposure to fine particulate pollution (PM2.5) increases mortality and morbidity and reduces human capital formation and worker productivity. As a consequence, high levels of particulate pollution may adversely affect economic activity. Using a novel dataset of changes in the annual gross domestic product of Indian districts, this paper investigates the impact of changes in the level of ambient PM2.5 on district-level gross domestic product. Using daily temperature inversions as an instrument for pollution exposure, this paper finds that higher levels of particulate pollution reduce gross domestic product.

This paper uses an exogenous shock to wages from the world’s largest anti-poverty program to show that higher wages can lead to increased air pollution, likely by inducing farmers to shift into a labor-saving and mechanized production process. The results suggest that absent policies to correct for environmental externalities of mechanization at all stages of development, labor market shocks may lead to inefficient levels of mechanization

This paper presents a satellite-based CO₂ tracking model that enables monitoring of local and regional concentration changes using NASA’s OCO-2 data. By comparing trend estimation models, the study finds that a simple model using pre-filtered satellite data effectively isolates local anomalies. Applications include long-term trend analysis and short-term yearly changes. The paper also estimates CO₂ emissions for thousands of cities, identifying those with unexpectedly high or low emissions. Given computational constraints, the World Bank has established an open web facility to provide pre-filtered CO₂ data, annual tracking reports, and urban-to-administrative unit linkages.

This paper highlights the critical role of methane reduction in near-term climate mitigation and the data challenges facing the Global Methane Pledge. Current emissions estimates rely on bottom-up inventories like EDGAR, which lack direct measurements. To address this, the World Bank has developed a new database using Sentinel-5P satellite data to track monthly methane concentrations at high spatial resolution. A case study on irrigated rice production demonstrates its potential for refining emissions estimates and supporting mitigation efforts. This approach could enhance monitoring and verification across other methane-emitting sectors.

This study estimates an urban CO2 emissions model using satellite-measured CO2 concentrations for 1,236 cities in 138 countries from 2014 to 2020. The model accounts for global CO2 trends, seasonal variations, industry structure, agricultural and forest fires, demographics, income effects, and geographic conditions. The study also examines regional variations, decomposes emissions sources, and explores the role of subway investments in emissions reduction. The research supports policy discussions on Pigouvian pricing and urban planning for low-carbon transitions.

Cities worldwide can now track their carbon and methane emissions from space, thanks to a new dataset from the World Bank that puts high-resolution, real-time urban data at everyone’s fingertips.

A low-emission growth path across key sectors — including agriculture, mining, forestry, manufacturing, infrastructure, and utilities — is critical for sustainable development, poverty reduction, and a livable planet. Integrating emissions mitigation into development planning is vital for long-term environmental and economic resilience. 

As the world grapples with the climate crisis, subway expansions present a viable and urgent opportunity for meaningful climate action. World Bank research results to date show that rapid subway expansion provides an excellent opportunity for significant reduction of carbon emissions in our efforts to address climate change. 

The lack of accurate and consistent carbon dioxide (CO₂) and methane (CH₄) emissions data has hindered the efforts to rapidly reduce emissions. Recognizing the urgent need for a comprehensive and reliable GHG monitoring scheme, a multi-disciplinary team at the World Bank launched an initiative to leverage advancements in satellite based GHG measurements to fill the global information gap. 

Clearly, subway systems can be beneficial in reducing congestion and improving health — but the upfront costs may be beyond the reach of most fiscally stressed city governments. But subways don’t simply reduce congestion locally, they cut carbon emissions globally.

World Bank researchers developed a new database and open data service for tracking CH₄ emissions building on earlier carbon emissions work. The database responds to the SDG Decade of Action, which calls for broadening global access to policy-critical data and demonstrating their use. 

To better track carbon pledges and support mitigation finance, World Bank researchers have developed a new database and web facility that place carbon data at the user’s fingertips.  The database uses satellite data from NASA’s OCO (Orbiting Carbon Observatory)-2 satellite, which provides reliable information about global CO2 emissions at high levels of spatial resolution.

The World Bank’s research suggests that satellite-based CO2 measures can provide valuable support for analyzing and assessing urban emissions performance.  While the study has focused on large cities, the same modeling approach can be used in geographic settings as varied as large and small cities within regions or countries, regions within countries, or specific project areas.

Sentinel satellites can be powerful allies, with high-resolution greenhouse gas emissions measures now available from many platforms.

Worldwide, account ownership increased by 50 percent in the 10 years spanning 2011 to 2021, to reach 76 percent of the global adult population. The goal of financial inclusion is not just for more adults to have accounts but for account owners to benefit from using them, such as for digital payments, which provide a range of positive benefits that extend far beyond convenience. This paper reviews the evidence demonstrating how digital payments can expand financial inclusion among recipients and encourage the use of additional formal financial services, such as savings, credit, and insurance, and it explores how digital transactions offer greater security and privacy, especially for women, as well as opportunities to build a digital credit history for credit risk assessments.

This paper studies the welfare impacts of agricultural shocks on households with detailed heterogeneity, by taking consumption, land, and labor allocation choices into account. Using household survey data from 51 developing countries, the paper analyzes the welfare consequences of the food price hikes induced by the Russian Federation's invasion of Ukraine and future climate change. Both repress income and exacerbate inequality. War-induced food inflation reduced real household incomes across developing countries by 2.90 percent on average, while changes in yields due to climate change will reduce real incomes by 11.99 percent. The welfare impacts of both shocks vary enormously across the income distribution, with already vulnerable and poorest households bearing the brunt of the costs.

Since 2000, China has been upgrading its infrastructure, exemplified by the expansion of the high-speed railways (HSR), while simultaneously moving up the global value chains, evidenced by the rising domestic value-added ratio (DVAR) in exports. To analyze the impacts of the HSR on China’s DVAR, this paper develops a new methodology to estimate firm-level DVAR using only customs transaction data, without relying on the industry input-output tables or matching firm-level industrial census data. This paper also proposes a novel way to capture firm-level input-output linkages by matching the custom product codes of importers and exporters. The results confirm that the HSR increases the DVAR through firm linkages by connecting downstream exporters with upstream domestic suppliers. A two-sector model shows that, by improving the probability of exporters connecting with low-communication cost domestic suppliers, the HSR decreases domestic material prices and increases the variety of accessible domestic materials, hence pushing up DVAR.

Firms’ import response to tariffs is particularly adverse for products associated with green value chains relative to average imports, driven by the solar value chain and downstream segments across all green value chains. This effect is pervasive for both the values and quantities imported by firms as well as for the probability of firms importing these products. Moreover, the effect is even more negative for undiversified firms. In contrast, import regulations have a smaller and more varied impact on firms’ imports of products associated with green value chains. The findings suggest that governments in emerging markets should avoid adopting protectionist policies that are increasingly used in high-income countries, as their local firms rely on imports for the short-term diffusion of green technologies. 

This study explores how multinational firms and global value chain linkages drive the cross-country diffusion of emerging technologies. Using online job posting data from 17 countries (2014–2022), combined with multinational and firm-to-firm network data, the analysis shows that emerging tech jobs are heavily concentrated within large firms and their supply chains. About one-third of these postings originate from Fortune 500 firms or their affiliates, buyers, suppliers, or innovation partners. While tech hubs initially dominate, diffusion accelerates over time in wealthier, nearby regions. Crucially, firm-to-firm linkages—especially with prior buyers or innovation partners—boost tech job growth. Cross-firm relationships beyond corporate boundaries significantly shape diffusion patterns, often more than geography alone.

This paper examines the evolution and resilience of global value chains (GVCs) in the agrifood sector, which intensified since the 1994 Uruguay Round Agreement. Using data from FactSet and Fortune 500 lists, the study analyzes 17,500 agribusiness firms and over 150,000 supplier–customer links from 2014 to 2022. Findings show that large corporations have become more central in GVCs by diversifying geographically while maintaining a concentrated supply strategy. Greater network complexity and connectivity are associated with higher resilience—firms with deeper linkages are less likely to exit the industry. The paper offers new insights into agribusiness network dynamics and underscores the importance of firm-to-firm ties in navigating disruptions such as COVID-19, extreme weather episodes, and geopolitical tensions.

Restrictive NTMs coexist with lower tariffs, particularly for high-income importing countries, low-income exporting countries, country pairs with deep trade agreements, and products with consumption externalities. Designed model is further used to shed light on the recent Sino-EU battery electric vehicle (BEV) disputes, whereby the EU imposed NTMs on top of the tariffs on China’s BEVs.

The US-China trade war created net export opportunities rather than simply shifting trade across destinations. Many "bystander" countries grew their exports of taxed products into the rest of the world (excluding the United States and China). Country-specific components of tariff elasticities, rather than specialization patterns, drove large cross-country variation in export growth of tariff-exposed products. The elasticities of exports to US-Chinese tariffs identify whether a country's exports complement or substitute the United States or China and its supply curve's slope. Countries that operate along downward-sloping supplies whose exports substitute (complement) the United States and China are among the larger (smaller) beneficiaries of the trade war.

In the face of rising geopolitical fragmentation and the rise of protectionist measures, it is becoming increasingly important to understand how easy it is for firms to switch suppliers in the short term.  Fujiy et al., 2025 finds that in the short term it is very hard for buyers to switch suppliers even for the same product due to reasons as varied as long-term buyer-supplier relationships, customized products, and institutional capacities in terms of contract enforcement. The costs of disrupting supply chains, especially severing links for highly connected firms, can be large.

Brexit’s trade impacts were smaller than anticipated, suggesting that highly interconnected economies may absorb some barriers through reorganization.

Geoeconomic fragmentation can also manifest through policies ostensibly enacted for environmental or health reasons, but which function as protectionist measures. Ghose et al. (2023) provide an example of such a policy in their analysis of Sri Lanka’s fertilizer import ban, a policy introduced on environmental grounds. This restriction led to a drastic decline in agricultural production and exports of fertilizer-dependent crops, resulting in a 4.35%  average income reduction and concentrated welfare losses among landowners tied to agriculture. This example illustrates how policies framed for non-economic goals can act as de facto trade barriers, impacting global trade flows and domestic welfare. The findings also quantify the costs of lack of fertilizer access and the role of trade and industrial policy in determining such access.

Globalization has slowed—but not reversed—since the 2008–09 financial crisis. While trade remained resilient during COVID-19, recent shifts in policy and public sentiment mark a potential turning point. Concerns about labor market impacts from trade with low-wage countries like China set the stage, but didn’t drive the shift. The pandemic raised supply chain concerns, yet trade rebounded quickly in 2021. Russia’s invasion of Ukraine intensified fears about geopolitical risk, prompting calls to “de-risk” supply chains and restrict exports to China. These trends suggest a move toward more fragmented trade, with uncertain long-term effects. While some industries may gain, global growth, innovation, and poverty reduction could slow. Measuring resilience remains challenging without clear policy benchmarks.

Cali et al. (2024) shows that fixed costs of non-tariff measures affect a firm’s ability to respond to global shocks by changing sources of imported inputs, thereby negatively affecting their export performance. The costs of protectionism and trade fragmentation would thus be underestimated if we measure trade barriers by solely converting them into iceberg costs as reflected in import values.

This paper introduces a new approach to measuring Global Value Chain (GVC) participation, crucial for policy-making. It classifies GVC involvement into backward, forward, and two-sided linkages, incorporating production data beyond traditional trade metrics. The findings reveal that existing trade-based GVC measures underestimate global activity, misrepresent key sectors like services and manufacturing, and exaggerate risks in early trade liberalization. The study also refines backward-forward classifications, showing they overestimate backward linkages. Applying these new metrics to economic models improves predictions of trade disruptions, highlighting that while GVC participation increases exposure to external shocks, it enhances output stability. The dataset is available on the World Bank’s WITS Platform and is regularly updated for further analysis.