The economic indicators behind the headlines

Why these releases dominate a news day

A handful of statistical releases move markets more than almost any corporate announcement, because they shape expectations about interest rates, which in turn affect the price of nearly every financial asset. They arrive on a published calendar, at a scheduled minute, which is why coverage appears within seconds.

They are also routinely misdescribed, in fairly predictable ways. What follows is what each measures, and the specific misreadings worth guarding against.

Inflation: CPI and PCE

The Consumer Price Index measures the change in price of a basket of goods and services bought by urban households. The PCE price index measures something similar with a different basket and different weights, and it adjusts for substitution when people switch between products. Central bank targets in the United States refer to PCE, while news coverage usually leads with CPI because it is released earlier.

Core versions of each exclude food and energy. This is not because those do not matter — they obviously do — but because they are volatile enough to obscure the underlying trend. Both figures are useful for different questions: headline inflation is closer to lived experience, core is closer to where policy attention sits.

The most consequential misreading is confusing a change in the rate with a change in the level. When inflation falls from four percent to two percent, prices have not fallen. They are still rising, more slowly, from a level that already incorporated the earlier increase. “Inflation is coming down” and “things are getting cheaper” are entirely different statements, and coverage often blurs them.

Percent versus percentage point

This distinction is small, constantly muddled, and changes the meaning of a sentence completely. If a rate moves from two percent to three percent, that is an increase of one percentage point — and an increase of fifty percent in relative terms.

Both descriptions are true and they sound wildly different. A headline choosing the relative version for a small absolute move is technically accurate and rhetorically inflated. When you see a dramatic percentage attached to any rate — inflation, unemployment, interest rates — check which of the two is being reported.

Employment: payrolls and the unemployment rate

The monthly United States jobs report contains two figures drawn from two different surveys. Nonfarm payrolls counts jobs added or lost, based on a survey of employers. The unemployment rate comes from a survey of households, and measures the share of the labour force actively looking for work but not employed.

Because they come from separate surveys, they can disagree, and reporting them as one story creates confusion. A person holding two jobs counts twice in payrolls and once in the household survey. Someone who stops looking for work leaves the labour force entirely, which lowers the unemployment rate without anyone finding a job — which is why the participation rate is a necessary companion figure rather than an optional detail.

Growth: GDP and its annualization

Gross domestic product measures the value of goods and services produced. In the United States, quarterly GDP is typically reported at an annualized rate: the quarter's growth extrapolated as though it continued for a full year. A quarter that grew about half a percent is reported as roughly two percent annualized.

This convention makes quarters comparable to annual figures, and it multiplies both real changes and noise by roughly four. Other countries report the plain quarterly change, so international comparisons in a single article are frequently not comparing like with like. Real GDP is adjusted for inflation; nominal is not, and the difference between them during a high-inflation period is large enough to reverse the apparent direction.

Surveys: PMI and sentiment indices

Purchasing managers' indices ask businesses whether conditions improved, worsened, or stayed the same. The result is a diffusion index centred on fifty: above fifty means more respondents reported improvement than deterioration.

The key property is that these measure breadth, not magnitude. A reading of fifty-two means expansion was more common than contraction, not that output grew two percent. A fall from fifty-eight to fifty-two is a slowdown in how widespread growth is, while still describing expansion — a nuance that is regularly lost in coverage describing any decline as contraction. Consumer sentiment surveys have a further limitation: they capture how people feel, which correlates imperfectly with what they subsequently do.

Revisions are the rule, not the exception

Most major statistics are published as estimates and revised as fuller data arrives — sometimes for years, and sometimes by amounts large enough to change the story entirely. A payrolls figure can be revised by tens of thousands; GDP estimates move materially between releases.

Coverage almost always treats the initial print as the event and rarely revisits the revision, so the first number is what enters public memory even after it has been superseded. If you are relying on a specific figure for anything that matters, check whether you are quoting the original estimate or the current one. They are often not the same number.

Seasonal adjustment and why raw data misleads

Retail sales rise every December; construction slows every winter; education employment drops every summer. Seasonal adjustment removes these predictable patterns so that genuine changes are visible.

It also means the adjusted figure is a model output rather than a raw count, and unusual events — a shifted holiday, an extreme weather month, a change in school calendars — can distort the adjustment itself. When a print is described as surprising, one ordinary explanation is that the seasonal pattern behaved unusually rather than that the economy did.

Reading any data release

A checklist that covers most of the failure modes above:

• Is this a level or a rate of change — and is the change relative to last month or last year?
• Percent or percentage point?
• Annualized or the plain period figure?
• Real or nominal — adjusted for inflation or not?
• Is this an initial estimate, and how large have revisions to this series been?
• Which survey does each figure in the article come from?
• How large is the move relative to the normal variation in this series?

That last question is the one most worth asking. Many prints described as surprises fall well inside the historical noise of the series, and treating routine variation as a signal is the most common error in economic coverage. This is educational material about how statistics are constructed, not investment advice.