STRATEGYOngoing practice92% confidence

Life Chances Matrix

Predict someone's future from postcode and parental education — not income.

Problem it solves

Misidentifying inequality as income-driven when the real lever is geography and parental education

Best for

Policymakers, urban planners, and researchers diagnosing where inequality is structural rather than behavioural

Not ideal for

Individuals seeking a personal action plan — this is a systemic diagnostic lens, not a self-help tool

Overview

Why this framework exists

The Life Chances Matrix is an analytical lens Collier developed with the IFS to measure what someone's future is likely to look like based on two inputs: where they were born and whether their parents hold a university degree. The metric is deliberately not income — it is the probability of reaching the top 40% of earners by age 28, starting from measurements taken at 16. This reframes inequality from a question of wealth gaps to a question of opportunity gaps.

The results are stark. Children born to affluent, degree-holding parents in London have roughly a 70% chance of landing in that top 40%. Children born in the bottom fifth of the income distribution to non-graduate parents face a 17% chance. But the sharpest cut is by GCSE performance: the top third of GCSE performers have roughly a 70% success rate; the bottom third face a 4% chance. Collier argues this is not a talent gap but a memory-testing artefact — GCSEs measure recall at 16, not the skills, judgment, and intuition that the Sheffield crucible steelworkers demonstrated for generations.

The most disadvantaged person-group his research identified is a white woman from South Yorkshire with non-graduate parents who did not attend university herself. The 6:1 ratio between her life chances and those of the most advantaged group — a white London professional — is the widest such gap in any high-income country. The framework's power is that it makes the structural nature of this gap undeniable: luck of birth geography and family education, not character or effort, is the dominant variable.

Core principles

5 total
  1. Life chances — the probability of a decent outcome — are a more honest inequality metric than income comparisons alone.
  2. Geography and parental education together predict future outcomes more reliably than any measure of individual talent.
  3. A country where luck of birth determines success at a 6:1 ratio is not a meritocracy, regardless of rhetoric.
  4. GCSE results at 16 are a memory test, not a talent screen — they systematically exclude non-academic intelligence.
  5. Data suppression on life outcomes is itself a policy choice that lets structural inequality remain invisible.

Steps

5 steps
  1. Define the outcome metric clearly
    Fix a concrete, modest benchmark — Collier uses being in the top 40% of earners at age 28. This avoids conflating 'success' with wealth and keeps the analysis honest about what a decent life requires.
    Pro tipTop 40% is deliberately humble — roughly a higher-rate taxpayer. The framework loses power if you chase top-decile outcomes because confounding variables multiply.
  2. Measure at two ages, same cohort
    Capture baseline inputs at 16 (GCSE tier, parental education, postcode) and outcomes at 28. The gap between input and outcome is the life-chance ratio for that group.
    WarningData quality is the binding constraint in the UK — lobby for longitudinal linkage between school records and tax/NI data before attempting this at national scale.
  3. Segment by place and parental education, not income
    Run the numbers by region × graduate-parents vs non-graduate-parents. Income enters as the outcome variable, not the segmentation variable — this is what reveals the structural driver.
    Pro tipAdd ethnicity as a third axis only after place and education are established; Collier found place dominates ethnicity as a predictor for the bottom group.
  4. Report the ratio, not just the percentages
    Express results as the ratio between the most and least advantaged group (Collier's is 6:1). Ratios are harder to explain away than percentages and make international comparisons legible.
    WarningAvoid framing the ratio as a race or gender gap before checking whether geography dominates those variables in your dataset.
  5. Compare to peer nations
    Benchmark your ratio against equivalent countries — Spain, Germany, France. Collier's comparison shows the UK is an outlier: equivalent social groups in Spain have higher upward mobility than their UK counterparts.
    Pro tipThe international comparison shifts the frame from 'this is inevitable' to 'this is a policy choice' — essential for generating political will.

Checklist

Saved in your browser

Examples

3 cases
The Sheffield crucible steelworker

Nineteenth-century Sheffield steelworkers read the hue of furnace flames to adjust metallurgy in real time — high-skill, judgment-intensive work that built the first industrial factories on Earth. Their descendants today score in the bottom GCSE tier and are statistically written off at 16.

OutcomeA 4% chance of decent earnings for the bottom GCSE third in the same region whose grandparents mastered the crucible process — illustrating how the measurement of talent, not talent itself, has changed.
The most disadvantaged person in Britain

Collier's IFS data identified the worst life-chance profile as a white woman in South Yorkshire with non-graduate parents who did not attend university. Her probability of reaching the top 40% of earners is so low as to be statistically negligible.

OutcomeA 6:1 gap in life chances versus a white London professional — the widest such ratio in any high-income country, making the UK an outlier by its own peer-group standards.
Spain comparison

Collier uses Spain as a direct comparator: a person born in the bottom fifth in Spain has a higher probability of reaching the top 40% of earners than an equivalent person in the UK.

OutcomeDemonstrates that low social mobility in the UK is a policy choice, not an economic law — comparable countries achieve materially better life-chance ratios.

Common mistakes

5 traps
Using income quintiles as the segmentation variable
Sorting by income hides the causal mechanism — geography and parental education are the upstream drivers. Income sorting produces the symptom, not the diagnosis.
Treating GCSE tier as a proxy for talent
GCSEs are a memory test at 16. Using them as a talent screen writes off a third of young people based on a narrow, culturally loaded metric that doesn't capture the judgment, craft, or interpersonal skill that real economies need.
Focusing only on ethnic or gender inequality
Collier's data shows the most disadvantaged group is white and female but defined by place — South Yorkshire. Diversity-first framing misses the geographic dimension entirely, leaving the bulk of inequality unaddressed.
Stopping the measurement at 16
Without the age-28 follow-up, you only see inputs. The life-chances insight requires linking school-age inputs to early-career outcomes — the gap is invisible without that second measurement point.
Accepting data gaps as neutral
Poor longitudinal data in the UK is, in Collier's reading, a form of policy — what is not measured cannot embarrass the Treasury. Treating data gaps as technical rather than political misses the accountability dimension.

Origin story

How this framework came to be

Collier conducted this research with the Institute for Fiscal Studies (IFS), overcoming a significant data problem: Britain has unusually poor longitudinal data on life outcomes because, in his words, 'everything's been covered up.' The IFS study tracked cohorts from age 16 to 28 — too early to see a full life but long enough to see the trajectory gap open unmistakably between the fortunate and the left-behind. The framework crystallised for Collier when he found that the most disadvantaged group was not defined by ethnicity but by place and parental education, cutting against dominant narratives about who inequality really hits.

Source

Traced to primary
Source · PODCAST
The UK Is The Most Unequal High-Income Country In The World
Paul Collier · 2024
Open source →

Related frameworks

Browse all Strategy →