How do we measure whether a country is doing well?

The one number three questions lean on

Three of the questions that bring readers to this site turn out to lean on a single number, and none of the three is about accounting. The first is the suspicion behind a headline: the news reports that the economy grew, and the raise never came, so what was actually growing (Volume IV, on why a growing economy can still feel like standing still). The second is the largest question in this volume: was the shape of a life mostly settled by where it began, rather than by anything the person did (the next chapter). The third shadows every figure a page like this one prints: why trust an economist, or this site, at all (Volume V, on why anyone should trust an economic number).

Each of the three is read off the same instrument, the number the news calls “the economy” and economists call gross domestic product. Before any of them can be answered, the instrument itself has to be understood: what it counts, what it was built to do, and what it was never designed to see. This chapter calibrates the gauge that every later chapter in this book, and most of the headlines in between, quietly rely on. It is the one piece of machinery the rest of the volume presupposes.

A gauge with a date of manufacture

Gross domestic product feels like a fact of nature, a thing an economy simply has, the way a body has a temperature. It is not. It is a measuring convention with an author and a date. The modern national accounts were built in the United States in the early 1930s, when the Depression had erased perhaps a third of output and nobody could say so with a number, because no number existed. Congress commissioned the young economist Simon Kuznets to build one, and in 1934 he delivered a report, National Income, 1929–1932, that gave the government its first comprehensive measure of the size of the economy.

Kuznets was careful about what the measure was, and careful in the same breath about what it was not. National income, he wrote, describes “the total activity of the national economy under one aspect, viz., the size of the final net product.” One aspect: the size of the output, not its worth to the people who produced it. He set down the warning in the report itself, in a sentence that has been quoted against the number ever since.

The instrument the news reads today is not quite the one Kuznets built. His measure was a net figure, oriented toward the income reaching people. During the Second World War the accounts were rebuilt into a gross measure of output, because the wartime question was not how well citizens were living but how much the economy could produce, and whether guns could be added without starving the supply of butter. Two shifts did most of that work: the accounts stopped netting out the capital used up in production, so the headline became a gross figure rather than a net one, and government spending was now counted as final product rather than as a cost subtracted from it, a treatment Kuznets objected to at the time. The gauge that resulted was superbly suited to its purpose, planning maximum output under mobilization, and that purpose is stamped into what it still counts and still ignores. The number was not invented for the war. It was born in the Depression to measure collapse, then rebuilt during the war into the output gauge that survived.

The boundary is drawn, not found

Everything the gauge counts sits inside a line that statisticians call the production boundary. The line is not given by nature. It is a convention, written down and periodically revised, most recently in the international rulebook known as the System of National Accounts. The rulebook is explicit that it draws a narrower line than it could. It defines a broad, general idea of production, any activity that uses labour and materials to make something a person could in principle have paid someone else to make, and then it deliberately measures a smaller circle inside that idea.

The strangest thing the boundary does is cut through a single household. If a farmer grows maize and eats it, that maize is counted, because it is a good and someone could have bought it. If a parent cooks the maize, minds the children, nurses a sick grandparent, and cleans the house, none of that is counted, because those are services performed for one’s own family. Hire someone to do the identical work and it enters the gauge at once, as wages. Kuznets saw this at the outset and named it: his estimates covered “primarily only efforts whose results appear on the market place,” so anyone interested in the total product of the nation, “including those efforts which, like housewives’ services, do not appear on the market,” could use the figure only with qualifications. That unpaid work is real economic activity, and the question of what it is worth and who does it has its own home later in this series (Volume IV, on the work GDP doesn’t count). What matters here is narrower and structural: the boundary that hides it is a line the accountants drew, and could have drawn elsewhere.

What the gauge cannot see

Two blind spots follow from the way the line is drawn, and both are the kind a headline routinely forgets. The first is the unpaid and unrecorded work just described. The second is distribution. A national total is a sum, and a sum keeps no memory of how it was shared. An economy can post a healthy growth rate while the median household’s income stalls, if the growth accrued to the top, and the aggregate would report none of it. This is exactly the gap between the headline and the raise that sends readers to the felt-stagnation question in a later volume. It is also the second thing Kuznets warned of in the same paragraph as the first: welfare cannot be judged, he wrote, without knowing how income is distributed. The gauge answers “how much,” and is silent on “for whom.”

None of this makes the number worthless. A measure of total output is one of the most useful instruments the social sciences ever built, and much of this volume depends on it. What it measures is output, and output is not the same as wellbeing, safety, fairness, or a life going well. When those are what a reader wants to know, other gauges exist, and they do not agree with the first.

When the gauges disagree

Set three ways of ranking countries side by side and watch the order change. The first is gross domestic product per person, adjusted for what money actually buys in each country. The second is the Human Development Index, which folds in life expectancy and schooling alongside income. The third is average life satisfaction, from a global survey that simply asks people to rate their own lives on a ladder from zero to ten. Each is a defensible answer to “is this country doing well.” They rank the same countries differently, and the differences are the argument.

The reorder is real, not an artifact of a small sample. Costa Rica sits outside the richest fifty countries by income and rates as the sixth-happiest on earth in the 2025 survey, ahead of every economy above it here. Singapore, among the two or three richest large economies in the world by output per head, ranks 34th for life satisfaction. Income clearly matters for a good life, and the poorest countries cluster near the bottom of every one of these gauges. Yet above a certain level the three measures come apart, and which one is called “doing well” is a choice, not a reading. The gross domestic product line answers one specific question, the size of marketed output per person, and answers it well. It was never the same question as whether the people are healthy, educated, secure, or content.

The economy that leaves no record

There is a further gap, and for much of the world it is the largest. The gauge can only count what leaves a record: a wage on a payroll, a sale with a receipt, a firm that files accounts. A great deal of real economic life leaves no such trace. A street vendor, a day labourer paid in cash, a workshop that never registered, a family farm selling at the roadside, all of them produce, trade, and earn, and much of it passes below the level at which any statistical office can see it. Economists call this the informal economy, and it is not a rounding error.

More than three in five of the world’s workers, some two billion people, make their living in the informal economy, and in the poorest regions the share is higher still. Statistical offices know this and build estimates to fill the gap, so the informal economy is not simply missing from a modern GDP figure. But the estimates are exactly that, estimates, and their quality falls just where informality is greatest. A gauge that reads cleanest in rich countries with dense records reads roughest in poor countries where most work leaves none, which is an awkward property for a number used above all to compare rich countries with poor ones. Why some states can see and tax their economies while others cannot is machinery this volume takes up in its own right (the chapter on why some states can tax and enforce while others cannot).

Reading the gauge for what it is

None of this argues for throwing the number away, and the verdict below is not that GDP lies. A single figure that lets a government know within weeks whether output is rising or falling is a genuine achievement, and the chapters that follow lean on it constantly, because for the questions it was built to answer nothing else comes close. The discipline the number demands is only to read it as what it is: a measure of the size of marketed output, built in a particular decade for a particular purpose, blind by construction to unpaid work, to distribution, and to much of what the poorest half of the world does all day. Kuznets said as much on the first pages of the document that created it. The gauge answers “how big.” Whether a country is doing well is a larger question, and it takes more than one instrument to read.

Where the argument goes next

With the gauge calibrated, the volume can ask the question that most readers came for. The single largest fact about the number just described is how far it varies from one country to the next: output per person differs across the world by something like a hundredfold, a gap so large it dwarfs almost anything that happens inside a single country over a lifetime. Why the gap is that size, and what actually explains it, is the spine of this book and the subject of the next chapter.