← Maths & Science

Gross People Product

A thought experiment. GDP — PIB in French — adds up dollars, so a billionaire's next yacht counts as much as ten thousand meals. Suppose instead we added up people: score every person from −1 to +1 according to whether their needs are met, and sum. This page sets the rule out precisely, shows that it is three poverty counts in disguise, reconstructs what it would have read over the last hundred years, and finds where it breaks. A companion page asks what would happen if money itself were tied to it.

The rule

Every person is scored on three tiers of needs. Each tier is worth between −⅓ and +⅓:

  • Basic — food and a roof. Starving or without shelter scores −⅓; able to afford both scores +⅓.
  • Stable — decent housing, a good education, the ability to travel a little. Same scale.
  • Comfortable — holidays, and the rest of what makes a life feel good rather than merely secure. Same scale.

A person's score is the sum of the three, so it runs from −1 (nothing met) to +1 (everything met). The Gross People Product (GPP) is the sum over everyone alive. Its unit is a point: a village of 100 people has a GPP somewhere between −100 and +100, and the most natural way to read the world's number is as “the GPP of the world, shrunk to a village of 100”.

The property that matters most is the cap. A billionaire scores +1, exactly like a schoolteacher who takes a holiday every summer. GDP gives a dollar the same weight whoever holds it; GPP gives a person the same weight whatever they hold. Everything that follows — what it sees, what it misses, and why money tied to it would behave strangely — comes from that one choice.

From tiers to a ladder

Nobody has kept tier-by-tier records for a century, so a historical reconstruction needs two simplifications. First, the tiers stack: in practice people do not take holidays while hungry, so each person sits on one of four rungs. Second, each tier is proxied by an income line, in dollars per person per day at 2021 purchasing-power prices — the unit the World Bank uses for its poverty counts.

RungTiers metIncome proxyScore as specifiedFloor variant
Destitutenonebelow $3 a day−1−⅓
Getting bybasic$3 – $10 a day−⅓+⅓
Stablebasic + stable$10 – $30 a day+⅓+⅔
Comfortableall threeabove $30 a day+1+1

The lines are not arbitrary. $3 a day is the World Bank's extreme-poverty line since June 2025, built from the national poverty lines of the poorest countries — the cost of food and shelter and little else. $10 a day is the usual floor of the “global middle class” in the work of Homi Kharas: roughly where households can keep children in school, live in a solid home and absorb a bad month. $30 a day is roughly where rich countries draw their own poverty lines, so living above it in purchasing-power terms is a reasonable stand-in for the life of holidays and slack the third tier describes.

The last column is a variant used throughout: the same ladder, but only the basic tier can go negative; an unmet stable or comfortable tier scores 0 rather than −⅓. It answers the question “is a life with food and shelter but nothing more a minus or a plus?” differently, and you will see that this one choice flips the story.

Three poverty counts in disguise

Write Hb, Hs and Hc for the share of the population that does not meet each tier — what statisticians call a headcount ratio. A tier contributes +⅓ for the share that meets it and −⅓ for the share that does not, so it contributes ⅓(1 − 2H) to the average. Adding the three tiers:

average score = 1 − ⅔ · (Hb + Hs + Hc)

This holds whether or not the tiers stack. On the ladder it simplifies further: if r̄ is the average rung (0 to 3), the average score is ⅔·r̄ − 1. So the GPP, as specified, is nothing more exotic than three poverty headcounts at three income lines, added together and turned upside down. That is good news — headcounts are cheap, robust and easy to explain — and bad news, because it inherits every known defect of a headcount, which the last section takes in turn.

A century in four rungs

The chart reconstructs how many people stood on each rung, decade by decade, from 1925 to 2025. Switch to “Share of total” to see the mix without population growth.

Destitute Getting by Stable Comfortable
012345678919251935194519551965197519851995200520152025
Hover a year to read the breakdown.

Editorial reconstruction, rounded to whole percentage points of world population per rung; method and sources in the appendix.

Three things stand out. In 1925, about 1.5 billion of the world's 2.0 billion people were destitute by today's line. The number of destitute people kept rising for sixty years even as their share fell, peaking at about 2.1 billion around 1985, before collapsing to about 0.8 billion today — mostly through China, then India and Indonesia. And the century's real story is the orange band: the people who are fed and housed and not much more grew from 0.3 billion to 4.1 billion and are now half of humanity. Escaping destitution, at world scale, mostly meant arriving on the second rung and staying there.

The score per person

Read as a village of 100, the world's GPP as specified went from −77 in 1925 to −2 in 2025.

As specified (unmet tiers −⅓) Floor variant (only hunger counts negative)
-90-60-300306019251935194519551965197519851995200520152025
Hover a year to compare the lines.

Average score × 100, i.e. the GPP of a village of 100 people. −100 = everyone destitute; +100 = everyone comfortable.

The rise is 75 points out of a possible 200 — the world has travelled about 38% of the way from “everyone destitute” to “everyone comfortable” in a hundred years. The decade 1935–45 went backwards (war and famine); the best decade was 2005–2015, at +15 points, the decade in which extreme poverty fell faster than at any time in history; the last ten years were slower, with COVID pushing about 70 million people back below the line in 2020 and the remaining destitution concentrated in sub-Saharan Africa and fragile states.

The headline, as specified, is striking: the world has spent the whole century in negative territory and is crossing zero about now. But look at the dashed line. Under the floor variant the world crossed zero around 1963 and stands at +46. The trajectory is the same shape; the verdict on whether today's world is “net good” is entirely a matter of where zero sits.

The total: a J-curve, and a lesson in population ethics

The average says how a typical person is doing. The total — the actual Gross People Product — also counts how many people there are, and it behaves very differently.

As specified (unmet tiers −⅓) Floor variant (only hunger counts negative)
-3000-1500015003000450019251935194519551965197519851995200520152025
Hover a year to compare the lines.

Total GPP = average score × world population, in millions of points (one point = one person at +1).

As specified, the total fell from −1,503 million points in 1925 to −2,138 million in 1985, even though the average improved every peacetime decade. The reason is arithmetic: until the 1980s most people being born joined the bottom two rungs, each worth less than zero, so every additional person dragged the sum down. Only once the average got close to zero did the total turn, climbing steeply to −164 million today. Under the floor variant, where a fed and housed person counts as a plus, population growth helps instead, and the total rises almost without interruption to 3,740 million.

This is the oldest problem in population ethics, met head-on. Derek Parfit showed in 1984 that maximising a total leads to the “Repugnant Conclusion”: an enormous population of lives barely worth living beats a small, flourishing one. Maximising an average has the opposite pathology: it would rather not add a happy person if they are slightly below the mean. The GPP inherits both problems, and the zero point decides which side of the line every birth falls on. A government told to maximise the total GPP as specified would want fewer births among the poor; told to maximise the floor variant, it would want more people on every rung above destitution. The safe practice is the one used here: always report the average and the total side by side, and never let a single number decide.

GPP against GDP

Over the same hundred years, world real GDP rose roughly 25-fold, about six-fold per head. The average GPP moved 0.75 of a possible 2 points. They are not competing estimates of the same thing; they weight the world differently.

  • The value of a dollar. In GDP a dollar is worth a dollar wherever it lands. In GPP the value of a dollar is a staircase: zero almost everywhere, a jump of ⅔ of a point where it carries someone across a line. Economists usually put something smoother in between — the logarithm of income, since Daniel Bernoulli in 1738, says a dollar matters ten times more to someone ten times poorer. GPP is the extreme form of that idea: past $30 a day, nothing matters at all.
  • What GPP sees. Distribution, and the bottom. A trillion dollars added to the richest one per cent moves it by exactly nothing. A fraction of that sum, if it could be delivered as a $3-a-day floor to the 0.8 billion destitute, would add about 546 million points — ⅔ of a point each.
  • What GPP misses. Everything above the top line: the frontier of medicine, science, quality and variety. A world in which everyone lives on $31 a day scores the same as a world in which everyone lives on $300. Antibiotics or the internet register only through the people they help climb a rung, never through what they add to lives already at the top.

That last point is why GPP can complement GDP but not replace it. GDP measures the size of the economy; GPP measures how many people it reaches. See the academic critiques of GDP for the long argument about the first number, and the measures already proposed to sit beside it.

Where the rule breaks

It rewards lifting the nearly-there and ignoring the deepest poor

Each tier is a step: a person at $2.90 a day and a person at $0.50 both score −⅓ on the basic tier, and moving the first to $3.10 earns the full ⅔ of a point. The cheapest way to raise the number is to push people who are just below a line over it. Amartya Sen made exactly this objection to the poverty headcount in 1976: a good measure should get worse when a poor person gets poorer, and a headcount does not move. The cure is the one the original rule hints at when it says a tier runs “from −0.33 to +0.33”: score each tier as a ramp rather than a step, so every dollar near a line counts a little.

It counts a famine as progress

The dead leave the sum. When a destitute person dies, a −1 disappears and both the average and the total rise. The Bengal famine of 1943 or the Great Leap Forward famine of 1959–61, which killed tens of millions, would register as improvements. Ravi Kanbur and Diganta Mukherjee named this the premature-mortality problem in poverty measurement. The fix is to keep the prematurely dead on the books, at the bottom score, for the years they should have lived.

Its sign depends on where you put zero

As specified, a person who is fed and housed but has nothing more scores −⅓. Move zero down a rung — let unmet higher tiers count 0 instead of −⅓ — and the same world reads +46 per 100 people instead of −2. Nothing about the people changed; only the convention did. Because the zero decides whether one more person adds to or subtracts from the total, it is the most consequential choice in the whole design, more than any of the income lines.

Equal thirds are a convention, not a finding

Should a holiday weigh as much as a meal? The rule gives each tier a third, as the UN's Multidimensional Poverty Index gives health, education and living standards a third each. Any weighting is a value judgement; the honest course is to publish the tiers separately alongside the sum so readers can reweight.

The lines move under your feet

The thresholds are expressed in today's purchasing-power dollars, which are re-estimated every few years. When the World Bank moved its extreme-poverty line to $3 a day in June 2025, about 125 million people were added to the 2022 count overnight. And “comfortable” in 1925 meant no air travel at any price, so a fixed basket in today's money flatters the past in some ways and insults it in others — the classic index-number problem.

People are not always on a ladder

The reconstruction assumes tiers stack: nobody takes holidays while hungry. Mostly true, but there are people with degrees and travel and no secure housing — a young renter in an expensive city, say. Scoring tier by tier from surveys handles them correctly; income proxies cannot.

Its relatives

The idea has distinguished ancestors. The three tiers are Abraham Maslow's 1943 hierarchy of needs with a scoring rule attached. The UN's Human Development Index (1990, Mahbub ul Haq with Amartya Sen) blends income, schooling and life expectancy, but country by country, as an average of averages rather than a count of people. The closest relative is the Multidimensional Poverty Index built by Sabina Alkire and James Foster at Oxford and published by the UN since 2010: it scores each person on weighted deprivations and counts who is poor. The GPP is, in effect, a multidimensional poverty count with a positive side added — it gives credit for lives that go well, not only blame for lives that do not. Bhutan's Gross National Happiness belongs to the same family in spirit, built on surveys rather than income lines.

Verdict

As a way of seeing the century, the GPP earns its keep. It tells a story GDP hides: that the number of destitute people kept rising until the 1980s, that the great escape of the last forty years mostly landed people on the second rung, and that, by this rule, the world is only now reaching neutral. As a rule to steer by, it needs three repairs before anyone should trust it: ramps instead of steps, so that the deepest poor count; the dead kept on the books, so that a famine cannot read as progress; and the average and the total reported together, with the zero point argued in public rather than buried in a formula.

The next question is the provocative one: what if money itself were tied to this number? That is the People Coin.

Appendix — the numbers and how they were made

This is a reconstruction, not a dataset. No statistical office has ever scored people tier by tier, and global income distributions before the 1980s are estimates built from national accounts and a handful of surveys. The shares below are rounded to whole percentage points and should be read to within a few points either way. The shape of the century is robust; the third significant figure is not.
YearPeople (bn)DestituteGetting byStableComfort.Score /100Floor /100Total (m pts)
19251.9675%17%6%2%−77−13−1,503
19352.1973%19%6%2%−75−12−1,650
19452.3774%19%5%2%−77−13−1,817
19552.7765%23%9%3%−67−5−1,847
19653.3458%26%11%5%−58+2−1,937
19754.0752%30%11%7%−51+7−2,089
19854.8644%36%12%8%−44+13−2,138
19955.7437%41%12%10%−37+19−2,105
20056.5925%49%15%11%−25+29−1,669
20157.4313%53%20%14%−10+41−743
20258.1910%50%23%17%−2+46−164

Method

  • Bottom rung. From 1985 it follows the World Bank's extreme-poverty series, restated at the $3 line (about 10% of the world in 2025). Before 1980 it follows the shape of Bourguignon and Morrisson's historical series — about three-quarters of humanity poor in 1929, 72% in 1950 — bending toward the World Bank's level at the join. The 1945 row adds a war penalty that no series measures directly. Undernourishment (about 8% in 2024) and slum housing (about 1.1 billion people) are cross-checks: counting every slum dweller as unhoused would push today's bottom rung toward 14%.
  • Upper rungs. Anchored on Our World in Data's late-2010s estimates — about two-thirds of humanity below $10 a day, 85% below $30 — and extended backward from the population and income of the rich world, where nearly all the stable and comfortable rungs sat until the 1990s. Estimates of the “global middle class” differ by ten points or more depending on whether they start from household surveys or national accounts; this page uses the survey-based end.
  • Population is from the UN World Population Prospects (HYDE estimates before 1950). GDP growth is rounded decade averages from the Maddison Project and, for recent decades, the IMF at purchasing-power parity.
  • Everything else is arithmetic. Every score, total and chart on this page is computed from the table above by the same code, in lib/people-product.ts, so the prose and the charts cannot disagree.

Sources

  • François Bourguignon & Christian Morrisson. “Inequality Among World Citizens: 1820–1992”, American Economic Review, 2002 — the share of humanity below the poverty line from 1820, the backbone of the pre-1980 rows.
  • World Bank, Poverty and Inequality Platform. extreme-poverty headcounts since 1981, and the June 2025 move of the international line to $3.00 a day at 2021 prices, which raised the 2022 count by roughly 125 million to about 838 million.
  • Max Roser, Our World in Data. “Global poverty in an unequal world” (2021): about two-thirds of humanity below $10 a day and 85% below $30 a day, the anchors for the upper rungs.
  • Homi Kharas (Brookings). “The Unprecedented Expansion of the Global Middle Class” (2017) and, with Kristofer Hamel, “A global tipping point: Half the world is now middle class or wealthier” (2018) — the $10–11 a day middle-class floor.
  • FAO, The State of Food Security and Nutrition in the World 2025. about 8% of the world undernourished in 2024 — a cross-check on the bottom rung.
  • UN-Habitat. about 1.1 billion people living in slums or informal settlements — the housing half of the basic tier.
  • UN, World Population Prospects 2024. world population since 1950 and the projected peak of about 10.3 billion in the mid-2080s; HYDE estimates before 1950.
  • Maddison Project Database and IMF World Economic Outlook. world real GDP growth, rounded to decade averages.
  • Amartya Sen. “Poverty: An Ordinal Approach to Measurement”, Econometrica, 1976 — the axioms a headcount fails.
  • James Foster, Joel Greer & Erik Thorbecke. “A Class of Decomposable Poverty Measures”, Econometrica, 1984.
  • Sabina Alkire & James Foster. “Counting and multidimensional poverty measurement”, Journal of Public Economics, 2011 — the method behind the UN's Multidimensional Poverty Index.
  • Ravi Kanbur & Diganta Mukherjee. “Premature mortality and poverty measurement”, Bulletin of Economic Research, 2007 — why a death can lower measured poverty.
  • Derek Parfit. Reasons and Persons, 1984 — total versus average welfare and the Repugnant Conclusion.
  • Abraham Maslow. “A Theory of Human Motivation”, Psychological Review, 1943 — the original tiered ladder of needs.