Introduction

The report examines how extreme poverty declined from 36% of the global population in 1990 to 9% in 2015. It focuses on China, India, Indonesia, Mexico and South Africa, which together accounted for 75% of global poverty reduction between 1990 and 2015. The study analyses how poverty fell across cohorts and how household livelihoods changed as people moved into and out of poverty.

Data

The study uses repeated cross-sectional and panel survey data covering income and consumption-based poverty measures. Extreme poverty is defined as living on less than US$2.15 per day in 2017 purchasing power parity terms.

The datasets track household livelihoods, including sector of employment, occupation type, migration and female labour force participation. The report notes the surveys are high quality relative to most poverty research and tests results against alternative poverty lines, household definitions and measurement assumptions. Findings were generally robust across methodologies.

Poverty decline within and between cohorts

The report finds poverty reduction occurred mainly within cohorts rather than through younger, less-poor generations replacing older cohorts. Poverty rates were broadly similar across age groups at any point in time, with declines occurring through parallel downward shifts across all ages.

Changes in population age structures therefore played little role in aggregate poverty reduction. The pace of poverty decline within cohorts roughly matched the pace between cohorts, implying most poverty reduction occurred during people’s lifetimes.

The results were consistent across both consumption and income measures. Using a higher poverty threshold of US$3.65 per day strengthened evidence that poverty decline primarily reflected within-cohort improvements.

Within-cohort poverty dynamics

Panel data showed substantial household mobility both into and out of poverty. Initially poor households frequently escaped poverty, but many non-poor households later fell into poverty. Among initially poor households, the share remaining poor at endline ranged from 55% in rural India to 33% in China. Poverty entry rates among initially non-poor households ranged from 13% in Indonesia to 34% in Mexico.

No single pathway out of poverty dominated. Most households that exited poverty did so without changing sector, occupation, migration status or female labour force participation.

Sectoral shifts from agriculture to non-agricultural activities contributed to poverty reduction in some countries. Among households exiting poverty, 37% in China and 21% in India moved out of agriculture. Occupational patterns differed by development stage. In China, India and Indonesia, self-employment was associated with poverty reduction, while wage employment drove most poverty reduction in Mexico and South Africa.

Migration generally played a limited role. Rural-to-urban migration contributed modestly in South Africa, while rural-to-rural migration accounted for nearly one-third of poverty decline in Indonesia.

Female labour force participation produced mixed results. Households where women entered the workforce generally experienced strong poverty reduction outcomes. In South Africa, these households accounted for 43% of net poverty exits. China differed, with nearly half of poverty reduction linked to households where women left the labour force.

Conclusion

The report concludes that poverty reduction reflected broad-based improvements experienced simultaneously across cohorts, alongside significant household-level mobility. Structural change, occupational shifts, migration and female labour force participation all contributed to poverty reduction, but no single factor explained most poverty exits. The authors note the findings are descriptive rather than causal and suggest future research should focus on barriers limiting household economic decisions.