Most of Your Income Was Decided Before You Did Anything

TL;DR

The useful version of “how much of success is luck” is Roemer’s: separate circumstances nobody chose from effort and episodic luck. Milanović ran exactly that decomposition on household surveys from 118 countries covering 94% of the world’s population. Two circumstances — the average income of your country of residence and the inequality of its distribution — account for more than half of the variation in individual percentile incomes worldwide; substitute country dummies for those two and the figure runs between 66% and 73%.¹ Measured against DR Congo, the average premium for residing in the United States is 355%, Sweden 329%, Brazil 164%, Yemen 32%.¹ Assignment to a country is effectively permanent: fewer than 3% of people live in a country they were not born in.¹ The second unchosen circumstance is parents — in Italy, the UK and the US roughly half of a father’s economic advantage or disadvantage passes to his son, against under a fifth in Denmark, Norway and Finland.² The paper’s own conclusion: the role of effort “cannot play a large role in explaining the global distribution of income.”¹ And the same paper reports that this locational component is eroding — from 81% of global inequality in 1988 to 70% in 2008.¹

A wing over inhabited land. The exceptional thing in this picture is the aeroplane: 97% of people spend their lives in the country they were born in. Photo: Blake Guidry on Unsplash.³

The Question, Made Answerable

“How much of my success is luck?” cannot be answered, and the reason is not modesty. There is no measurable quantity called luck. Ask it that way and you get a temperament test: some people say most of it, some say almost none, and nobody has to produce a number.

The question becomes tractable when you change one word. Not luck but circumstances — the specific, enumerable facts about a person’s situation that they did not choose and cannot alter. Those are observable. You can put them in a regression and ask how much of the variation in outcomes they account for.

John Roemer’s framework does exactly this partition: outcomes are a function of circumstances, effort, and what he calls episodic luck.¹ The point of the partition is not to eliminate effort. It is to find out how much room is left for it once the unchosen facts have had their say.

The Biggest Unchosen Fact

Start with the one that dwarfs everything else and gets discussed least.

Branko Milanović set up the problem as a thought experiment in the shape of Rawls: imagine everyone in the world assembled, each handed a single characteristic that will shape their economic fate — country of residence — and ask how much of their income that one fact determines.¹ Then he answered it with data: household surveys from 118 countries for 2008, representing 94% of the world’s population and 96% of world dollar income.¹

The headline results are these. The elasticity of a person’s own income with respect to their country’s GDP per capita is 0.866 — nearly one for one. Together with the country’s Gini coefficient, those two circumstances explain roughly two-thirds of the variation in individual percentile incomes across the world. Replace them with country dummies, capturing every unobserved thing about a country, and the figure reaches 73%.¹ Across all six specifications the range is 48% to 73%.¹

The premiums in the chart are what those dummies mean in plain terms: the average income advantage of living somewhere other than the poorest country in the sample.¹

And there is no exit. Milanović’s flat statement: “Assignment to country is fate, decided at birth, for approximately 97% of the people in the world.”¹ Less than 3% live somewhere they were not born. For the overwhelming majority, this circumstance is not a starting condition to be overcome — it is a permanent parameter.

His conclusion is worth quoting exactly, because it is stronger than the hedged version that usually circulates: the part left over “for effort and ‘episodic luck’… is, within the worldwide context, relatively limited.”¹

The Part That Cuts the Other Way

Here is the finding in the same paper that the argument’s usual proponents skip, and it matters.

Over the twenty years to 2008, the between-country share of global interpersonal inequality fell from 81% to 70%.¹ Milanović’s own description: global inequality of opportunity due to place of residence is “huge but decreasing,” and the decrease is driven by the rapid growth of relatively poor and populous countries, China and India in particular.¹

That is a real qualification, and it points the argument somewhere less comfortable for everyone. If the dominant source of unearned advantage is which country, then the most powerful equaliser in human history is not redistribution inside rich countries — it is poor countries getting richer. The measured gains came from growth in Asia, not from transfers.

There is a second wrinkle in the same paper. Milanović finds that the country’s Gini coefficient enters with a negative sign: a one-point increase in national inequality is associated with about a 1.5% decrease in the average person’s income, because higher inequality numerically benefits fewer people than it harms.¹ But he also shows the effect is concentrated at the tails — for people around the middle of a distribution (ventiles 13 to 18), national inequality matters very little, because their income share is about the same in equal and unequal countries. What matters to them is whether the country is getting richer.¹

The Second Unchosen Fact

Country is the large circumstance. Parents are the intimate one, and the measure there is the intergenerational earnings elasticity — the share of a parent’s economic advantage or disadvantage that persists into their child’s adult earnings.

Miles Corak’s comparative estimates, covering 22 countries, put Italy, the United Kingdom and the United States at roughly 50%, against less than one-fifth in Finland, Norway and Denmark.² Same statistic, same construction, a factor of two and a half between rich democracies.

The broader relationship Corak documents is the one Alan Krueger named the Great Gatsby Curve: countries with more income inequality at a point in time also show less earnings mobility across generations.² That is a correlation across countries, not a demonstrated causal chain, and Corak is careful about it. But the direction is consistent and the implication is uncomfortable — inequality now is associated with less mobility a generation later, which means the two circumstances compound rather than offset.

Where the transmission actually lives is more specific than the headline suggests. Comparing the United States and Canada, Corak finds “a good deal of mobility in both,” with little relationship between family background and child outcomes across broad swaths of the middle. The divergence is at the extremes — a point I’ve taken up separately in the context of what North American households actually own.

What Follows, and What Doesn’t

This is where the argument usually overreaches, so it is worth being precise about which conclusions the evidence carries and which it does not.

It does not show that effort is irrelevant. A regression that leaves 27% unexplained by country leaves a great deal of room, and within a country your own decisions are most of what varies. The finding is about global variance decomposition, which is a different quantity from “how much does trying matter to me.” Confusing the two produces fatalism from a statistic that does not support it.

It does not tell you what to do. That circumstances dominate outcomes is a fact about the world; that this is unjust is a further claim requiring a moral premise. The evidence rules out one specific position — that global income distribution reflects differential effort — without establishing any particular alternative.

What it does undermine is desert in the strong sense. Not the claim that incentives work, or that skill exists, or that some people work harder. The specific claim it damages is that a person’s income is something they are intrinsically owed because they generated it. If most of the variation traces to a country you were assigned and parents you did not pick, the entitlement story is doing work the arithmetic cannot support. This is the same structure as the free-will problem: forward-looking justifications for rewarding effort survive intact, while backward-looking claims about deserving what you have do not.

There is a related trap on the other side. Treating the biological lottery as the whole story — reading every outcome as an unchosen endowment — collapses into the mirror-image error. A cause is not an alibi in either direction. The useful question is not whether you deserved your position but what the numbers say about where a marginal intervention would do the most good, and the answer that falls out of Milanović’s data is unglamorous: raising average incomes in poor and populous countries.

Where I’d Hold This Loosely

Five limits, and the first is about vintage.

The Milanović data is 2008 and the Corak elasticities are older still by construction — sons born in the early-to-mid 1960s, measured in the late 1990s. Given that the between-country share was already falling through 2008, and that China and India have grown a great deal since, today’s locational component is very likely smaller than 73%. I would treat these as establishing the shape of the answer, not the current value.

Second, R-squared is not causation, and Milanović’s regressions are not identified. Country of residence is correlated with a great many things — institutions, geography, history, culture, technology — and “country explains 73%” means that knowing someone’s country lets you predict their income well. It does not isolate a mechanism, and moving someone across a border does not automatically confer the premium.

Third, these are percentile incomes within national distributions rather than individual observations, which is what makes the cross-country comparison tractable but also means the analysis cannot see individual variation inside each percentile. The unit of analysis is doing some of the work in the headline number.

Fourth, Corak’s own warning, and it is the one most often ignored by people quoting him: the intergenerational elasticity “does not mean that it measures ‘equality of opportunity’ or the even-more-elusive ‘American Dream.’”² It summarises transmission between two generations. It is not a measure of fairness, and a country could in principle score well on it for unattractive reasons.

Fifth, and against my own framing: there is something faintly dishonest about an argument that proves circumstances dominate and then addresses itself to readers who can act on the conclusion. The people for whom the locational premium is most decisive are the least likely to be reading this, and the least able to change it. Milanović makes the practical point himself and it is sharper than mine — the question this evidence actually raises is where the efforts of people in poor countries should be directed: to work, or to migrate.¹ I do not have an answer to that. Neither, on the record, does he.

---

Footnotes

1. https://stonecenter.gc.cuny.edu/files/2015/05/milanovic-global-inequality-of-opportunity-how-much-of-our-income-is-determined-by-where-we-live-2015.pdf ↩ ↩² ↩³ ↩⁴ ↩⁵ ↩⁶ ↩⁷ ↩⁸ ↩⁹ ↩¹⁰ ↩¹¹ ↩¹² ↩¹³ ↩¹⁴ ↩¹⁵ ↩¹⁶ ↩¹⁷ ↩¹⁸

2. https://stonecenter.gc.cuny.edu/files/2013/07/corak-income-inequality-equality-of-opportunity-and-intergenerational-mobility-2013.pdf ↩ ↩² ↩³ ↩⁴

3. https://unsplash.com/photos/person-taking-picture-of-plane-wings-while-flying-during-daytime-p9vr45T2scg ↩