Focus Areas
I use data, measurement, and applied economics to support better decisions for children and vulnerable populations. Across UNICEF, the World Bank, and the Brazilian public sector, my work sits at the intersection of development economics, education policy, poverty and equity measurement, and the strengthening of national statistical systems.
My focus is operational: translating complex data into indicators, diagnostics, scalable analytics, and decision-support tools; evaluating policies and programs; and strengthening institutional capacity to produce timely, comparable, and reproducible evidence. Recent work includes creating global public goods such as the Learning Poverty measure and quantifying the impacts of COVID-19 on learning and intergenerational mobility.
Increasingly, I am leading teams to use AI- and LLM-assisted workflows to improve analytical productivity and increase the dissemination and interpretability of rigorous analysis—within clear technical and ethical guardrails and with human-in-the-loop review (e.g., validation, documentation, continuous integration and deployment, privacy/PII safeguards, and auditability).
What is Learning Poverty?
Learning Poverty is the share of children who cannot read and understand a simple text by age 10. Co-developed by João Pedro Azevedo at the World Bank, it combines schooling and learning data to capture both out-of-school children and those in school but not learning. Before COVID-19, 57% of children in low- and middle-income countries were learning poor. Post-pandemic simulations estimate this figure rose to 70%. The indicator is now used by the World Bank, UNESCO, UNICEF, FCDO, USAID, and the Bill & Melinda Gates Foundation to track SDG 4 progress and guide education investment.
Source: Azevedo et al. (2021), World Bank Policy Research Working Paper No. 9588
Core Focus Areas
1. Poverty, Equity, and Wellbeing
This agenda focuses on poverty, equity, and wellbeing—covering child poverty and child outcomes, but also broader distributional analysis, shared prosperity, and the measurement systems that translate evidence into policy.
Sub-theme: Poverty, Equity, and Distributional Analysis
Analyzing the drivers of inequality and the distributional impacts of policies and shocks—grounded in country-level diagnostics and extended to regional and cross-country benchmarking—with a focus on equity for children and vulnerable populations.
- Paper From Noise to Signal: The Successful Turnaround of Poverty Measurement in Colombia (2013)
- Report EU Regular Economic Report: Thinking CAP—Supporting Agricultural Jobs and Incomes in the EU (2018)
- Blog Going municipal: Targeting deprivation with new evidence in Croatia (2019)
- Code HOI — Human Opportunity Index (incl. Colombia and Latin America applications)
- Code ADECOMP — Shapley decomposition of poverty changes (incl. Latin America applications)
Sub-theme: Child Poverty and Wellbeing Measurement (SDGs)
Developing frameworks and indicators to monitor progress on the Sustainable Development Goals (SDGs) for children, including comparable measurement and reporting.
- Report The State of the World's Children 2025 — Statistical Compendium (2025)
- Report The State of African Children 2025 — Statistical Compendium (2025)
- Report Progress on children's well-being: SDG Country Briefs (2023)
Sub-theme: Child Health and Mortality
Investigating the determinants of child health and mortality, and how systems, services, and risk factors shape outcomes.
- Commentary Hard truths about under-5 mortality: call for urgent global action (Lancet, 2024)
Related posts
2. Education, Learning, and Human Capital
This is a central pillar of my research, with a focus on ensuring quality education for all. This work involves measuring educational outcomes, diagnosing inequalities in learning, and evaluating the impact of educational interventions and global shocks.
Sub-theme: Learning Poverty and Foundational Skills
A core contribution of my work at the World Bank was the co-creation of the global "Learning Poverty" indicator—a measure of the percentage of children unable to read and understand a simple text by age 10. This research stream involves the harmonization of learning assessments and the analysis of policies to accelerate progress in foundational learning.
- Paper Will Every Child Be Able to Read by 2030? Defining Learning Poverty (Working Paper, 2021)
- Paper Learning Poverty Updates and Revisions (Working Paper, 2021)
- Paper Learning Poverty: Measures and Simulations (Working Paper, 2020)
- Blog Will every child be able to read by 2030? (2021)
- Blog Building Back Better After COVID-19: The Importance of Tracking Learning Inequality (2021)
- Blog Learning for All: Beyond an Average Score (2020)
- Code LearningPoverty — Stata package to calculate Learning Poverty indicators
- Code EduAnalyticsToolkit — World Bank Education Global Practice analytics tools
- Code GLAD — Harmonized learning assessment microdata from 160+ countries
Sub-theme: The Educational Impacts of the COVID-19 Pandemic
In response to the global crisis, I led research to simulate and measure the impact of school closures on learning outcomes. This work quantified the scale of "learning losses," analyzed the effectiveness of remote learning strategies, and informed global education recovery efforts.
- Article COVID-19 school closures, learning losses and intergenerational mobility (Humanities and Social Sciences Communications, 2026) — Open Access
- Paper COVID-19 School Closures, Learning Losses and Intergenerational Mobility (Working Paper, 2023)
- Paper Learning Losses during COVID-19 (Working Paper, 2022)
- Paper Simulating the Potential Impacts of COVID-19 School Closures (Journal Article, 2021)
- Paper Remote Learning During COVID-19 (Report, 2021)
- Paper Learning Poverty in the Time of COVID-19 (Brief, 2020)
- Blog School closures and longer-term implications of COVID-19 (2023)
- Blog The global education crisis – even more severe than previously estimated (2022)
- Blog Five lessons from remote learning during COVID-19 (2021)
- Blog Learning losses due to COVID19 could add up to $10 trillion (2020)
- Blog Learning losses due to COVID-19 could add up to $10 trillion (2020)
- Blog We should avoid flattening the curve in education (2020)
Related posts
3. Data Systems, Methods, and Statistical Capacity
Underpinning all my research is a focus on improving the data and methods we use to measure development. This includes strengthening national statistical systems, developing new econometric tools, and making data more accessible and useful for policymakers and researchers.
Sub-theme: Data for Policy and Global Monitoring
This work focuses on building evidence-to-policy pipelines, from the global harmonization of household surveys for poverty monitoring to the development of data compacts that align countries and partners around common measurement goals.
- Blog Going municipal: Targeting deprivation with new evidence in Croatia (2019)
- Blog April 2018 global poverty update from the World Bank (2018)
- Blog Feeding the craving for precision on global poverty (2017)
Sub-theme: Global Poverty Monitoring Infrastructure
Work on building the data architecture for the World Bank's global poverty monitoring, including the harmonization of over 1,200 household surveys, development of PovcalNet/PIP, DATALIBWEB, and leadership of the Data for Goals initiative.
- Blog Is Poverty Seasonal? (2016)
- Blog How persistent is poverty in the short run? (2016)
Sub-theme: Econometric and Statistical Software Development
To support rigorous and reproducible research, I have developed over 20 open-source statistical packages, primarily for Stata. These tools have been downloaded over 25,000 times by the global research community.
- Code WBOPENDATA — Access World Bank databases from Stata
- Code SAE / FHSAE — Small area estimation: unit- and area-level models
- Code ADECOMP — Shapley decomposition of poverty changes
- Code HOI — Human Opportunity Index
- Code DFL — DiNardo-Fortin-Lemieux counterfactual density decomposition
- Blog New release of WBOPENDATA Stata module (2020)
- Blog Wbopendata Stata Module Upgrade (2014)
View all 22 software modules on the Software page →
In the Media
- Nature Statistics reach a 'crisis point': nations struggle with a critical lack of data (Nature, 2026) — News feature on the global data crisis in official statistics
Related posts
Frequently Asked Questions
What is Learning Poverty and how is it measured?
Learning Poverty measures the percentage of 10-year-old children who cannot read and understand a simple, age-appropriate text. It combines two indicators: the share of children who are out of school and the share of children in school who fall below a minimum reading proficiency threshold. The concept was introduced by the World Bank in 2019, co-developed by João Pedro Azevedo, and is used by major international organizations to track SDG 4 progress.
Who is João Pedro Azevedo?
João Pedro Azevedo is UNICEF's first Chief Statistician, appointed in 2023 to lead the organization's global data and statistical work. He holds a PhD in Economics from Newcastle University, is an elected member of the International Statistical Institute, and previously spent 16 years at the World Bank as Lead Economist. He has published in The Lancet, World Development, and the World Bank Research Observer, with over 5,500 citations and 20+ open-source statistical software packages.
What Stata packages does João Pedro Azevedo develop?
João Pedro Azevedo has developed over 22 open-source Stata modules available on SSC/RePEc, with over 25,000 downloads globally. Key packages include: WBOPENDATA (access 29,000+ World Bank indicators from Stata), ADECOMP (Shapley decomposition of poverty changes), HOI (Human Opportunity Index), SAE/FHSAE (small area estimation), DFL (DiNardo-Fortin-Lemieux decomposition), and SKDECOMP (counterfactual distribution decompositions). See the Software page for the full catalog.
How did COVID-19 affect Learning Poverty?
Simulations by Azevedo et al. (2020, 2021) estimated that COVID-19 school closures could increase Learning Poverty from 57% to 70% in low- and middle-income countries, with learning losses equivalent to 0.3–0.9 years of schooling. A 2026 peer-reviewed study by Cojocaru, Azevedo, Narayan, and Montalva Talledo, published in Humanities and Social Sciences Communications, further showed that extensive school closures significantly reduced both absolute and relative intergenerational educational mobility — in upper-middle-income countries, the share of children with more education than their parents could decline by 8 percentage points. Unequal access to continued learning across socioeconomic backgrounds made relative mobility declines even larger under optimistic assumptions about remote learning effectiveness.
For a complete and filterable list of all my work, please see the Publications Page.