Gross domestic product is the number that decides whether an economy is described as growing or shrinking, whether a recession has technically occurred, and whether a central bank has room to move. It is quoted constantly and understood loosely.
Most of the confusion comes from three places: which of the three ways of calculating it is being used, whether the figure is nominal or real, and how much of it is a genuine measurement rather than an estimate that will be revised.
The definition, word by word
GDP is the monetary value of all final goods and services produced within a country's borders in a given period.
Each phrase does work.
Monetary value means everything is measured in money, which is why unpaid activity is invisible to it.
Final excludes intermediate goods, to avoid double counting. The steel that goes into a car is not counted separately from the car; only the finished vehicle is.
Within a country's borders distinguishes GDP from gross national income, which counts what a country's residents earn wherever they earn it. A foreign-owned factory's output counts in the GDP of the country it sits in, and its profits count in the GNI of the country that owns it. For most economies the two are similar; for Ireland, Luxembourg and a handful of others they differ enormously, which is why comparing those countries on GDP produces strange results.
Three ways to the same number
GDP can be calculated three ways, and in theory they give the same answer, because every transaction has two sides: what one party spends is what another party earns, and both equal the value of what was produced.
Production approach. Sum the value added at every stage of production across every industry.
Income approach. Sum all incomes earned in production: wages, profits, rents, interest, plus taxes less subsidies.
Expenditure approach. Sum everything spent on final output. This is the version that gets quoted.
In practice the three never match exactly, and the gap is published as a statistical discrepancy. Its size is a rough measure of how much confidence the number deserves.
C + I + G + (X − M)
The expenditure formula is worth understanding component by component, because each one behaves differently and each tells you something the headline does not.
C — household consumption. Everything people spend on goods and services. In developed economies this is typically 55% to 70% of GDP, and around 68% in the United States. It is the largest component and the most stable, which is why consumer data is watched so closely.
I — investment. Business spending on plant, equipment, software and buildings, plus residential construction and changes in inventories. Usually 15% to 25% of GDP and by far the most volatile component. Investment collapses in recessions and rebounds hard, which is why it drives most of the cycle despite its modest share.
The inventory line inside investment causes a great deal of confusion. Goods produced but not sold count as investment, so a quarter in which firms build unsold stock shows stronger GDP than the underlying demand justifies — and the following quarter shows weaker, as inventories are run down. A large inventory contribution is a warning that the headline overstates the trend.
G — government spending. Public consumption and investment. Transfer payments — pensions, unemployment benefits — are not included, because they are not payment for production. They appear later, when the recipient spends them, inside C.
X − M — net exports. Exports minus imports. Imports are subtracted because they were already counted in C, I or G, and they were not produced domestically. This is why a widening trade deficit mechanically subtracts from GDP even when it reflects strong domestic demand.
Nominal against real, and why it matters
This distinction separates people who read GDP correctly from people who do not.
Nominal GDP is measured at the prices of the period being measured.
Real GDP adjusts for price changes, so it measures the quantity of output rather than its money value.
The difference is not small. An economy where prices rise 5% and output does not change at all reports nominal growth of 5% and real growth of zero. The headline says expansion; nothing was produced.
Growth figures quoted in the media are almost always real. The ratio between the two — the GDP deflator — is itself a useful inflation measure, and broader than the consumer price index because it covers everything produced rather than a consumer basket.
Per capita, and the limits of averages
Total GDP measures the size of an economy. It says nothing about living standards, because a large population produces a large number regardless.
GDP per capita divides by population and is a much better proxy for average prosperity. It changes rankings dramatically: economies with enormous total output can sit far down the per-capita list, and small wealthy countries dominate the top of it.
It remains an average, and averages conceal distribution. A country where all the growth accrues to a small share of the population reports the same per-capita figure as one where it is spread evenly. Two economies with identical GDP per capita can feel completely different to live in.
What GDP does not measure
The list is long and the omissions are not accidental — GDP was designed to measure production, and it does that job well. The problem is that it gets used as a proxy for wellbeing, which it was never intended to be.
Unpaid work. Childcare, eldercare, housework and volunteering are enormous in value and entirely absent. Paying someone to look after your parent increases GDP; doing it yourself does not.
The informal economy. Cash work and undeclared activity are estimated rather than measured, and in some countries the estimate is a substantial share of the total.
Distribution. Growth that goes entirely to the top is indistinguishable from growth that is broadly shared.
Environmental cost. Depleting a resource or polluting a river adds to GDP through the activity and never subtracts through the damage. Cleaning it up adds to GDP again.
Quality improvements. Statisticians attempt to adjust for products that get better at the same price, but the adjustment is imperfect and the direction of the error is not obvious.
Destruction as activity. A natural disaster reduces wealth and increases GDP through reconstruction. The number goes up while the country is worse off.
Revisions: the part that catches traders
GDP is not a measurement, it is an estimate that improves.
The United States publishes three versions of each quarter — advance, second and third — each incorporating more complete data, and then revises again in annual and benchmark exercises. Revisions of several tenths of a percentage point are routine, and revisions large enough to change whether a quarter was positive or negative are not rare.
This matters more than it sounds. Decisions get made on the advance estimate, headlines are written about it, and the number that eventually enters the historical record is frequently different. Anyone comparing a current print to history is comparing an estimate to a revised figure.
Why GDP rarely moves currencies much
For a data release of such importance, GDP is a surprisingly quiet event in foreign exchange, and the reason is worth understanding.
GDP is backward-looking and already known. By the time a quarter's GDP is published, the monthly data that goes into it — retail sales, industrial production, trade balance, employment — has already been released and analysed. Economists build the estimate from those components, so the published figure is usually close to the consensus, and the consensus is close to right.
Compare that with the inflation report or the monthly jobs figures, which are genuinely new information about a month nobody has data on yet. Those move markets. GDP mostly confirms what was already priced.
When GDP does move markets, it is usually because of a component rather than the headline: a collapse in business investment, an unexpected inventory swing, or a consumption figure that contradicts the retail data. The composition carries more information than the total.
It also matters at turning points, where the definition of a technical recession — two consecutive quarters of negative real growth — creates a headline threshold that is arbitrary economically but real in terms of sentiment and politics.
The complements
Because of GDP's limits, alternative measures exist. The Human Development Index combines income with life expectancy and education. Various wellbeing and inclusive growth measures attempt to account for distribution, environment and unpaid work. Gross national income corrects for the foreign-ownership distortion.
None of them replaces GDP, and none is trying to. They answer different questions, and the mistake is asking GDP to answer theirs.
Reading it properly
GDP is the best single summary of economic activity that exists, which is a real achievement and a limited claim.
Read it real, not nominal. Read the components, because investment and inventories explain most of the movement. Treat the first estimate as provisional. Use per capita for anything about living standards, and do not use it for anything about how those standards are distributed.
And remember what it counts: production, in money, inside a border. Everything it does not count — the unpaid, the undeclared, the environmental, the distributional — is not a flaw in the number. It is a flaw in using one number for a question it was not built to answer.
