From the Director: Understanding inequality in the age of artificial intelligence
When I joined Oxford and took over directorship of INET's Economics, Inequality & Opportunity Programme in October 2025, I benefited from the kind of luck that every new director hopes for: inheriting an already stellar research programme from Brian Nolan. In the past year, we have expanded to 20 core researchers from across the social sciences, and we now face a daunting challenge: how should we understand, measure, and prepare for changes in inequality amidst rapid advancements in artificial intelligence?
Thus far, we have embraced that challenge with creativity and experimentation. We formed an LLMs Subgroup, meeting regularly to ask how we could use generative AI to advance the study of inequality itself. We have experimented with new forms of collaboration, such as an open, unstructured, 'contribute-at-will' model for producing academic research. And we have opted for a virtual, outward-facing seminar series, the INET Oxford Inequality Working Group, that has brought together nearly 500 researchers from outside Oxford to engage with Dani Rodrik on 'productivism', Anthropic economist Maxim Massenkoff on the effects of AI on jobs, and representatives of the European Commission on the EU Anti-Poverty Strategy.
More broadly, we have organised our work around four priority areas: (1) poverty and income inequality, (2) social mobility and economic opportunity, (3) labour markets, and (4) the use of LLMs in inequality research. You will find examples of our team's work in these areas in the pages to follow.
Our first year closes as we host the Atkinson Conference on Economic and Social Inequality. On September 10th and 11th, 100 academics will gather at Nuffield to deliberate on the measurement, causes, and consequences of inequality. The conference's namesake, the late Sir Anthony B. (Tony) Atkinson, combined careful measurement and rigorous analysis with serious policy consideration. Tony was fixated not only on measuring and understanding inequality, but also on What Can Be Done to reduce it. The Atkinson Conference, and our team's work more broadly, aims to carry that focus forward.
Prof. Zachary Parolin - Director, INET Oxford programme on Economics, Inequality and Opportunity; Professor - Department of Social Policy & Intervention; Professorial Fellow - Nuffield College
Read this Report as a PDF
Our mission
We pursue policy-relevant evidence on the causes and consequences of economic inequality.
We are an interdisciplinary group of social scientists based at the Institute for New Economic Thinking at the Oxford Martin School. We combine administrative and survey data, advanced quantitative research methods, and close engagement with policymakers to understand why inequality persists and what can be done to widen opportunity.
We have 36 researchers, affiliates and fellows, who you can find at the Economics, Inequality & Opportunity programme page.
Our research: Four priority streams for understanding economic and social inequality
1. Poverty & Income Inequality
How can we improve the measurement of poverty and income inequality in high-income countries? Which policies are effective at generating a more equal income distribution, and at what cost? Our team studies these questions with the aim of providing evidence-based guidance on how state policies can make life better for low-income households. Recent work has revealed the role of an expanded Child Tax Credit in the U.S. in reducing poverty and hardship, and why the persistence of poverty in the UK has increased over time.
2. Social Mobility & Opportunity
Whether a child born to a disadvantaged family can achieve prosperity in adulthood is a central indicator of a healthy and inclusive society. Our team produces new evidence on how and why social mobility varies across high-income countries, with particular attention to upward mobility from poverty. Using a mix of administrative and survey data, we study how childhood exposure to poverty shapes later-life outcomes, and which policies are most effective at breaking the intergenerational cycle of poverty.
3. Labour Markets
Labour markets are a central engine of economic inequality: the wages, bargaining power, and job opportunities available to workers shape who prospers and who is left behind. Our team studies how labour market institutions and policies — from minimum wages to collective bargaining — influence the distribution of earnings, and how the power of employers affects the fortunes of low-wage workers. Recent work examines how minimum wage policies interact with worker power to shape wage growth, how the expansion of dominant employers such as Walmart Supercenters affects local labour markets, and how labour market institutions structure social and economic stratification.
4. LLMs for Inequality Research
Large language models and artificial intelligence stand to reshape social and economic inequality in the years to come, but they also offer powerful new tools for studying it. Our team pursues both strands: examining how the rise of AI affects employment patterns, income inequality, and wealth inequality, and how these new tools can reveal fresh insights into inequality itself. Recent work includes using text-as-data methods to track how the content of collective bargaining agreements has evolved, and constructing and evaluating a large database of UK housing planning applications.
The sections that follow highlight a selection of projects drawn from across these four streams.
"More than one in five British children born after 2013 spend at least half of their childhood in poverty." Professor Zachary Parolin, University of Oxford
The persistence of poverty
Poverty is often treated as a temporary setback. Two strands of our research (Poverty & Income Inequality and Social Mobility & Opportunity) show how it often lasts much longer, stretching across childhoods and repeating from one generation to the next.
Part I - Growing up poor in Britain
For nearly a quarter of British children born today, poverty lasts for most of childhood. Tracking every birth cohort from 1991 to 2017, Selçuk Bedük and Anna Yong find that long-term childhood poverty (spending at least half of childhood poor) fell from 25% for children born in the early 1990s to around 13% by the end of that decade, as reforms after 1997 boosted support for low income families.
Then the trend reversed. Among children born after the austerity reforms of 2013, long-term poverty climbed back to 23%, close to where Britain started three decades ago. Policy changes since the crisis have led to a rise in long-term poverty in the UK.
Part II - The persistence of poverty in America
What happens to children who grow up in poverty in the U.S.? Zachary Parolin follows American cohorts born from 1960 to 1988 and finds that, despite half a century of policy change and falling poverty rates, the odds that a poor child becomes a poor adult have not improved. His work finds that gaps in employment, not educational attainment, explain the largest share of poverty's persistence.
Can policy break the cycle? A companion study of the Earned Income Tax Credit, one of America's largest antipoverty programs, suggests it is harder than hoped. The credit lifts long-run earnings, but shows little evidence of improving a poor child's chances of escaping poverty as an adult.
Further reading
In 21 of 24 EU countries, the new measure tracks material deprivation more closely than the official poverty rate over two decades of data. Selçuk Bedük, University of Oxford
A New Measure of Poverty for Europe?
Europe's 'at-risk-of-poverty' measure does not track other indicators of material and social deprivation in many countries. Our team proposes a revised poverty measure anchored in what families typically spend in each country.
Tracking trends in poverty rates is essential for understanding the economic security of nations' most financially vulnerable residents and assessing how policy changes could improve well-being. Across the European Union, that tracking rests on one primary measure adopted in 2001: a person is poor if their household income falls below 60% of the national median. The poverty measure is transparent and easy to replicate, but it measures where people sit in the income distribution rather than whether they can afford to participate adequately in society.
Lewis Anderson, Selçuk Bedük, Solhee Han, and Zachary Parolin propose an alternative measure: the European Supplemental Poverty Measure (EU-SPM). Modeled on the U.S. Supplemental Poverty Measure, it keeps the EU's income data but sets the poverty line by what typical families actually spend on food, clothing, shelter and utilities, adjusted for housing costs within each country.
Applied to twenty years of data across the EU, the EU-SPM outperforms the standard measure in tracking other measures of destitution. In 21 of 24 countries examined, the new measure better identifies which individuals report being unable to make ends meet. The advantage is largest in downturns; whereas the official rate fell while hardship was rising in many countries, the EU-SPM instead tracks hardship outcomes even during recessions.
The study in brief
- The new measure: The European Supplemental Poverty
Measure (EU-SPM) compares post-tax/ transfer household income to a poverty line set at what typical families spend on food, clothing, shelter and utilities, adjusted for housing costs within
each country. - Validation: The authors compare the performance of the EU-SPM to the EU's official at-risk-of poverty rate (AROP) across two decades
of EU-SILC data. The validation tests assess how either measure tracks material deprivation and households' own reports of being unable to make ends meet across time and place. - Results: In 21 of 24 countries, the EU-SPM tracks deprivation more closely than AROP, and it better identifies disadvantaged households. Its advantage is largest in economic downturns and in periods of high inflation.
"Cost-of-living escalators, once a standard feature of American union contracts, now appear in only about one agreement in ten." Matthew Bone
Large Language Models in Inequality Research
Our team has been experimenting with use cases of large language models to advance inequality research. One output: turning the text of thousands of union contracts into data.
Empirical research on unions has long relied on a single binary — covered by a collective bargaining agreement or not — even though one agreement may secure employer-paid family healthcare and binding arbitration while another secures little beyond union recognition. Research led by INET Oxford members Matthew Bone, Prashant Garg, and Chenxi Li assembles 8,546 U.S. collective bargaining agreements spanning 1907 to 2025 and uses large language models to score each on nine areas of worker entitlement, from compensation and healthcare to job security and dispute resolution.
The result is a more comprehensive portrait of what it means to be unionized in the United States. The authors find that collective bargaining agreements that pay well also tend to protect well: compensation and non-wage protections such as healthcare, job security, and grievance rights are strongly, positively correlated.
Over the century, contract generosity rose through the 1970s and has drifted down since. Some individual provisions follow the same trend: automatic cost-of living adjustments, once a standard feature of American union contracts, now appear in about one in every 10 collective bargaining agreements, a decline from past decades.
Further reading
When Walmart Comes to Town
Our researchers have made advances in measuring the impact of monopsony - the dominance of one or a few large employers - in local labour markets, quantifying changes in poverty, public finances, and worker power in the decade after one arrives.
Twin papers published this summer in Labour Economics and American Sociological Review piece together the impact of Walmart Supercenter openings on local communities across the US. INET Oxford's Zachary Parolin and Lukas Lehner collaborated on both, joined by Clemente Pignatti, Rafael Pintro Schmitt, and Joshua Choper. Walmart is the largest private employer in the US — and as it expanded across rural America, outcompeting independent grocers, it came to dominate local labour markets.
Monopsony power describes a labour market where one or a few employers dominate, leaving workers with few outside options. Few American firms exhibit monopsony power quite like Walmart. The authors of these recent studies find that a stronger presence of organized labour can deter the entry of a Walmart Supercenter and the rise of monopsony power. However, when Walmart does succeed in opening, local communities tend to face adverse economic consequences.
The authors find that in the decade after a Supercenter opened, household incomes dropped by about 4%; poverty rose by 2.2 percentage points relative to counties without an opening; and union membership fell by up to 3.5 percentage points, weakening worker power across the local labour market. Using a restricted dataset that follows the same workers across their careers, the study is the first to measure Walmart's effects with the most robust poverty indicator the U.S. Census Bureau offers.
Further reading
Where to find us next
- The Atkinson Conference on Economic and Social Inequality - We are thrilled to host the Atkinson Conference at Nuffield College in September. Named in honour of Sir Tony Atkinson, Warden of Nuffield College from 1994 to 2005 and one of the most influential scholars of inequality, the Atkinson Conference brings together 100 researchers from economics, sociology, politics, and social policy for two days of deliberation. Co-organised by INET Oxford, the Department of Social Policy and Intervention, and Nuffield College, the conference features a keynote from Janet Gornick (Graduate Center, City University of New York), a closing keynote from Nobel laureate Philippe Aghion, and an opening panel on inequality in the age of artificial intelligence moderated by Sarah O'Connor of the Financial Times.
- The Wage Standard, with Prof. Arin Dube - Our INET Oxford team will host Arin Dube, Provost Professor of Economics at the University of Massachusetts Amherst and one of the world's most influential researchers on wage inequality and the minimum wage, for a presentation of his new book, The Wage Standard: What's Wrong in the Labour Market and How to Fix It. Drawing on two decades of research, Dube asks why pay at the bottom and middle of the distribution failed to keep up with a growing economy, and lays out the policy levers that could reverse the divergence. The talk can be attended at the Oxford Martin School or followed online.
- INET Oxford Inequality Working Group - Throughout term, our working group hosts researchers and practitioners working at the frontier of inequality research, followed by open discussion with attendees. Recent sessions have featured Dani Rodrik of the Harvard Kennedy School, economists from the European Commission presenting on poverty measurement in Europe, and an economist from Anthropic on how artificial intelligence is reshaping work. The series is open to the Oxford research community and visitors alike; the full schedule is at inet.ox.ac.uk/events.
Work with us
If your work connects to any of our four priority areas — poverty and income inequality, social mobility and opportunity, labour markets, or LLMs for inequality research — we would like to hear from you. We welcome visiting researchers each term, who join our seminars, present work in progress, and collaborate with the team during their stay.
Research highlights at a glace
More than one in five British children born after 2013 spend at least half of their childhood in poverty Bedük & Yong in “Long-term Childhood Poverty in Britain”
"When a Walmart Supercentre enters a locla labour market, the local poverty rate rises by 18%, and stays elevated for a decade." Lehner et al. in “Power and Poverty”
"Once a Walmart Supercenter opens, union membership falls by 3.5 percentage points - and even worker who keep their union cards see earnings decline." Choper, Lehner & Parolin in “Countervailing Powers”
"The share of causal claims within economics papers rose from about 4% in 1990 to 28% in 2020." Garg & Fetzer in “Causal Claims in Economics”
"Higher local authority spending on housing repairs is associated with better resident wellbeing in England - with tenants in the poorest housing conditions benefiting the most." Neftalem Emanuel in “Damp Places, Unhappy Faces”
"Generosity of U.S. collective bargaining agreements increased through the 1970s but has declined ever since." Bone et al. in “The Substance of Union Contracts”
"The odds that a child who grows up in poverty in the U.S. will escape poverty in adulthood have not improved in fifty years." Parolin in “The Intergenerational Persistence of Poverty in the United States”
