GDP Explained: What It Measures and the Three Estimates

Simon Kuznets built the first national accounts for the United States in the early 1930s, because Congress had discovered during the Depression that nobody could say with any confidence how much the economy had shrunk. When he delivered the work in 1934 he attached a warning to it, observing that the welfare of a nation can scarcely be inferred from a measurement of national income. The warning has been ignored for ninety years. He designed a production ledger. It is read as a verdict on whether a country is doing well.

Carry that history into the release. Most of the ways people misuse gross domestic product are ways Kuznets predicted, and what follows is the contents of the accounts, the arithmetic behind the growth rate, and the short list of things the quarterly print is genuinely good for.

What the accounts add up

The expenditure approach counts output by who bought it:

GDP = C + I + G + (X - M)

Component	What it is	Notes
C, personal consumption	Household spending on goods and services	Close to two thirds of the total, and the steadiest part
I, gross private investment	Business equipment, structures, software, housing, inventories	Small share, enormous swings, drives most cycles
G, government consumption and investment	Federal, state and local purchases	Transfer payments are excluded, since they buy no output
X - M, net exports	Exports minus imports	Imports are subtracted because they were produced elsewhere

Inventories can swing a quarter by a full percentage point in either direction while telling you nothing about demand, because a warehouse filling up counts as investment whether the goods were wanted or not, and that single entry does much of the damage at turning points. Net exports can do the same. A surge of imports arriving ahead of a policy change reads as a subtraction from growth, even though it usually signals buyers who expect to sell the goods on later.

Real, nominal and the annualized convention

Nominal GDP uses the dollars of the day. Real GDP removes the effect of prices using the GDP deflator, which is built from the mix of things the country actually produced in the period, so it differs from the consumer price index in construction and in coverage.

The relationship is the same real-return arithmetic that applies to a portfolio:

real growth = (1 + nominal growth) / (1 + deflator) - 1

If nominal GDP grew 5.0 percent over a year while the deflator rose 3.0 percent:

1.05 / 1.03 - 1 = 0.0194, so real growth was 1.9 percent. The identical calculation applied to your own returns is in the real return calculator.

Annualization is the second convention, and it causes international confusion every quarter, because an American quarterly growth figure is reported at an annual rate, meaning the quarter’s growth compounded as though it continued for four quarters. If output rose 0.5 percent during the quarter:

1.005 ^ 4 - 1 = 0.0202, reported as 2.0 percent.

European statistical agencies quote the quarter-on-quarter change itself. American figures therefore look roughly four times larger for the same performance, and comparing a US annualized rate against a euro area quarterly rate is a mistake that appears in print with some regularity.

Three estimates, and then the revisions

Release	Timing	What changes
Advance estimate	About a month after the quarter ends	First print, partly built on assumptions for the final month
Second estimate	A month later	Better trade, inventory and services data
Third estimate	A month after that	Fuller source data, plus corporate profits
Annual update	Mid-year	Revises several years using tax and census records
Comprehensive update	Every five years or so	Redefines methods and restates decades

Revisions between the advance and third estimates have exceeded half a percentage point on occasion, which is larger than the gap between a quarter the commentary called strong and one it called weak, so the figure that eventually enters the historical record is frequently not the one that was reported on the day. Comprehensive updates have redrawn whole cycles.

Why the release rarely moves markets

The number is old. The quarter being measured ended weeks before it appears, and its ingredients, retail sales, trade balances, construction spending, factory orders and inventories, were published month by month along the way. Forecasters, including the nowcasting models several regional Federal Reserve banks publish openly, update a running estimate after each of those releases, so by the time the advance print lands the market already holds a tight distribution around it.

Compare that with the jobs report, which delivers genuinely new information about the month just ended and reprices the front of the Treasury curve inside a second. Information value depends on surprise. By publication, GDP has almost none.

The exceptions are worth naming, since they do occur. GDP moves markets when a component breaks sharply from what the monthly data implied, when the deflator suggests an inflation path different from what CPI had suggested, and when the headline crosses zero and revives the recession conversation.

Two measures of the same thing

There is a second way to count. Gross domestic income adds up what was earned in production: wages, profits, rents and interest. In a perfectly measured world GDP and GDI would be identical, since every dollar spent is a dollar earned by somebody, and in practice they are assembled from different sources and differ by what the BEA calls the statistical discrepancy.

When the two diverge for several quarters running, the average of them has historically been a better guide to what was actually happening than either alone, and the BEA publishes that average for exactly this reason. It repays the attention. The discrepancy has tended to widen around turning points.

Who decides that a recession happened

Two consecutive quarters of falling real GDP is the popular definition. It has no official standing in the United States. The call is made by the business cycle dating committee of the National Bureau of Economic Research, which defines a recession as a significant decline in economic activity spread across the economy and lasting more than a few months, and which examines employment, real income, consumer spending and industrial production at monthly frequency.

The committee announces its decisions long after the turning point, sometimes more than a year later, and it has occasionally dated a recession that the two-quarter rule would have missed entirely. That delay is why forward-looking measures attract so much attention. They are collected in recession indicators.

What the accounts leave out

The measure counts transactions. Anything produced outside a market is absent, so unpaid household work, childcare done by a parent and volunteering all raise living standards without appearing anywhere in the accounts. Rebuilding after a hurricane adds to GDP while the destruction itself subtracts nothing, which is the clearest illustration of why the number says nothing about whether a country is better off. Depletion of natural resources is treated the same way.

Distribution is absent too. A quarter of solid aggregate growth is entirely consistent with most households experiencing no improvement, and output per person is a separate series that can fall while the headline rises. If an economy grew 2.0 percent while its population grew 1.2 percent, output per person grew roughly 0.8 percent, which is closer to what an individual would notice.

Quality change is the remaining weak point. Statisticians adjust for improvements in goods, those adjustments are estimates, and reasonable people argue about them constantly, particularly for software and services where defining the output at all is genuinely difficult.

What to do with the number

For a long-term investor the useful content is the trend, and real growth running near potential with contained inflation is the background that supports earnings without forcing the committee described in how the Federal Reserve works to act. Growth well above potential tends to pull policy tighter. Growth stalling while unemployment turns up is the combination that has ended cycles, and the framework for thinking about those combinations across asset classes is in macro regimes and asset allocation.

See where GDP sits among the month’s releases in the economic calendar guide, or test the annualization arithmetic in the Fed and the economy quiz.