Inflation and CPI Explained: Headline, Core and Reading the Release

Every consumer price index release opens with the same two sentences, built the same way, month after month, something close to this: the Consumer Price Index for All Urban Consumers rose 0.2 percent on a seasonally adjusted basis in the month, the Bureau of Labor Statistics reported, and over the last twelve months the all items index increased 3.1 percent before seasonal adjustment.

Two numbers. One describes a month, the other a year, the second is not derived from the first in any way an ordinary reader would guess, and the rest of the release runs to a dozen tables that exist to explain where those two figures came from. Learning to read the tables is most of the work.

Where the numbers come from

The index has a wartime origin. The Bureau of Labor Statistics began collecting family expenditure data during the First World War so that wages in shipbuilding yards could be adjusted for the cost of living, and regular publication of national indexes followed in the early 1920s, with the series extended back to 1913. The purpose has never really changed. A price index exists so that contracts, benefits and wages can be written in terms of what money buys.

The modern index tracks a basket of goods and services bought by urban consumers. Prices are collected each month from retailers, service providers and landlords across the country, weighted according to a survey of what households actually spend, and published as an index level. The level has no natural units. When the index reads 311.1, the basket costs 3.111 times what it cost in the 1982 to 1984 base period, and the number means nothing on its own. Only the change matters.

Component group	What it covers	Relative size
Shelter	Rent paid by tenants and owners’ equivalent rent	Largest single group by a wide margin
Food	Groceries and food away from home	Large, split into two sub-indexes
Transportation	New and used vehicles, fuel, insurance, repairs	Large and highly volatile
Medical care	Services, drugs, insurance	Moderate, and measured differently from PCE
Energy	Gasoline, electricity, piped gas, fuel oil	Small weight, outsized swings
Apparel, recreation, education, other	Everything remaining	Small individually, meaningful together

The BLS publishes the exact relative importance of every category alongside each release, and updates those weights on a regular schedule, so reading that one table will tell you more about why a print came in hot than any amount of commentary about it.

From an index level to an inflation rate

The whole release rests on this arithmetic, and it is worth doing by hand once. Take the following index levels as an illustration.

Last month the index stood at 310.326. This month it is 311.100. The monthly change is 311.100 / 310.326 - 1 = 0.00249. That is 0.2 percent, rounded to the tenth the BLS reports.

Twelve months earlier the index stood at 301.836. The annual rate is 311.100 / 301.836 - 1 = 0.0307, or 3.1 percent.

To see what the latest month implies if it continued, compound it twelve times: 1.002 ^ 12 - 1 = 0.0243, or 2.4 percent annualized.

That last line is why a single monthly print can matter so much. A run of 0.2 percent months annualizes to something the committee can live with. A run of 0.4 percent months annualizes to close to 5 percent, and no committee with a 2 percent target can watch that for long without acting.

Headline, core, and the slices the desks read

Headline CPI is the whole basket. Core CPI removes food and energy. The distinction came into general use during the 1970s, when oil embargoes and harvest failures were moving the headline so violently that it had stopped describing anything a central bank could influence, and stripping the two most shock-prone categories out left a series that behaved more like a trend.

Analysts have gone on slicing it since. Core goods separates physical items, which were the main source of the surge that followed the pandemic and then largely reversed, and core services covers everything from haircuts to insurance premiums. Core services excluding shelter, sometimes called supercore, gets attention because it tracks the part of the basket most closely tied to wage growth, which connects this release directly to the jobs report.

None of these is more real than the others, since they are different windows onto the same collected prices, and the one a commentator chooses usually tells you which argument is being made.

Shelter, and the year of old leases inside it

Shelter is the largest group in the index and the slowest moving, and the reason is worth understanding, because it distorts the number for months at a time.

Tenant rent is measured from actual leases, and each unit in the sample is revisited roughly twice a year. Owners’ equivalent rent estimates what a homeowner would pay to rent the home they already live in, which is how the index captures the cost of housing services without treating a house purchase as consumption, since a house is an asset. Both series therefore reflect contracts signed over the preceding year. This week’s asking rents are somewhere in next year’s index.

When the market for new leases turns, the index goes on reporting the old trend for a long while. That lag has often run close to a year. Anyone reading the release in real time needs to know which part of the number is current information and which part is last year’s leases still working their way through the sample, because the two lead to opposite conclusions about where inflation is going.

CPI against the index the Fed actually targets

The 2 percent target is written on the personal consumption expenditures price index, produced by the Bureau of Economic Analysis and published later in the month than CPI. Two differences drive most of the gap. PCE includes spending made on households’ behalf, notably employer-paid and government-paid medical care, so health costs carry a considerably larger weight, and PCE also updates its weights every period, which captures consumers substituting toward whatever has become cheaper, while a fixed basket by construction cannot.

The combined effect is that PCE usually prints a little below CPI. Markets nonetheless react far harder to CPI, because it arrives first and because the ingredients of the later PCE figure can be estimated from it once the producer price data is out.

Accuracy is an old argument here. In 1996 a congressionally appointed panel led by Michael Boskin concluded that the CPI at that time overstated increases in the cost of living by roughly 1.1 percentage points a year, largely through substitution and quality adjustment, and a series of methodological changes followed. The argument has never fully closed. Remember that whenever someone treats the index as a fact.

What inflation does to a return

Inflation is the reason a nominal return overstates what you gained. The exact relationship is:

(1 + nominal) / (1 + inflation) - 1

A portfolio that returned 5 percent in a year when inflation ran 3 percent earned 1.05 / 1.03 - 1 = 0.0194, a real return of 1.94 percent.

Simple subtraction gives 2 percent. That is close enough for conversation. It drifts further from the truth as both numbers get larger. Run your own figures through the real return calculator and the gap becomes obvious at higher inflation rates, and the same index sits underneath Treasury inflation-protected securities, whose principal is adjusted by the CPI and which are covered in the Treasury securities guide.

Where the measure falls down

The CPI averages a national basket. Nobody buys the national basket. A renter in a tight housing market and a homeowner with an old fixed mortgage face different inflation rates for years on end, and neither of them is being measured incorrectly.

Quality adjustment is the second complaint, and the most contested. When a new car carries features the old model lacked, the BLS treats part of the higher price as payment for more product, which lowers the measured increase. That is defensible in theory and infuriating to the person writing the check, and a fixed basket also cannot fully capture households trading down, which is a real reduction in living standards that never appears as inflation.

Knowing those limits is the whole difference between reading the release and reading the headline about it, and the terms used here are collected in the markets dictionary.

Work through the other half of the mandate in the jobs report explained, or see where CPI sits in the month in the economic calendar guide. The Fed and the economy quiz tests the arithmetic above.