What Inflation Is, and Why Prices Never Go Back

Inflation is the speed at which prices rise. When it comes down, the car slows; the odometer never runs backward.

In one sentence

Inflation is how fast the average price of what households buy is rising, measured each month by the Bureau of Labor Statistics against the same month a year earlier.

The headline figure, the Consumer Price Index, is built from roughly 80,000 prices collected around the country and weighted by how much a typical household spends on each thing. Shelter alone is about a third of the basket. Food is about an eighth and energy less than a tenth, which is why a 30% jump in gasoline moves the total by a point or so and not by 30.

New articles and charts arrive in one short email on Monday mornings. Get the weekly digest

The analogy

Inflation is a speedometer. The price level, the thing you feel at the register, is the odometer. The speedometer says how fast prices are rising right now; the odometer says how far they have come.

Most arguments about inflation are arguments between the two gauges. "Inflation is down to 3%" is a speedometer reading. "Everything costs 30% more than in 2019" is the odometer. Both are true at once. When the Fed says it beat inflation, it means the car slowed from 9 to 3, not that it backed up; outside the Depression and a few months of 2009 the odometer has only ever run one way. Rate vs. level applies the same trick to the deficit, interest rates and jobs.

How it reaches your kitchen table

In order of how fast you feel it:

The pump and the grocery store: this week. Energy and food prices are set in world markets and change daily. They are the loudest part of inflation and the part the Fed can do least about, which is why economists strip them out to get "core" inflation. Core is not a claim that food and gas don't matter. It is a claim that they are a poor guide to where the rest of prices are going.

Rent: at your next lease. Rents on new leases move first. The index catches up over about a year, because it samples all renters, most of whom are mid-lease. So the shelter line in the CPI runs roughly twelve months behind apartment listings, and since shelter is a third of the index, that lag alone held the headline up through 2023 after new rents had cooled.

Your paycheck: at your next raise. Wages follow prices with a lag of six months to a year, and only partly. Whether you are gaining ground is the real wage: your raise minus inflation. Pay for production and nonsupervisory workers rose 3.4% in the year to September 2026 against inflation of 3.7%, so typical paychecks lost ground over the year.

Social Security: next January. The cost-of-living adjustment is set from the CPI for urban wage earners (CPI-W) averaged over July–September, announced in October and paid from January. The 2027 adjustment is 3.6%. The lag runs the same direction every time: a price jump in the spring is not made up until the following January, and a shock in the fall waits fifteen months.

Your mortgage and your savings: through the Fed. Inflation doesn't change a fixed mortgage payment; it shrinks it in real terms, the one way a homeowner comes out ahead. Savers come out behind unless their yield beats the rate. And inflation above target moves the Fed, which moves every other rate in the country; the Fed funds page has that chain.

What moves it

Two families of causes, and the right response to each is different. Demand: too much spending chasing what the economy can make, as after the 2020–21 relief payments. Supply: the economy makes less, or it costs more to make, as with the 2021 ports, the 2022 oil and wheat shocks, and the 2026 disruption in the Strait of Hormuz. Then expectations: once workers and businesses expect 5%, they ask for 5%, and a one-time shock becomes self-sustaining. Most of what the Fed does is aimed at keeping the third from happening.

How to read this month's number

Three questions before reacting to a headline.

Year over year, or the month? The headline "3.7%" compares September 2026 with September 2025. The monthly change, 0.4% in that report, is what is happening now; at that pace for a year it would be about 5%. One hot month is noise. Three in a row is a trend. The yearly figure also carries "base effects": a bad month a year ago dropping out of the window lowers the rate without anything happening now.

Headline or core? Headline includes everything. Core strips food and energy. When they diverge, as in September 2026 (3.7% headline, 2.6% core), the gap says the story is a commodity shock, not a broad one. When core climbs too, it has spread.

Which index? The Fed's 2% target is on a different index, PCE, which weights health care more and shelter less and usually runs a few tenths below the CPI. The almanac shows the CPI because your COLA and most contracts use it; the PCE companion on the chart is what the Fed is aiming at.

The speedometer and the odometer. Inflation (top) is the yearly rate; the price level underneath, with 2019 set to 100, is where prices stand. PCE inflation is the index the 2% target actually applies to. Full chart and context →

Where the analogy breaks

A speedometer measures one car. The CPI measures an average household that doesn't exist. If you rent, drive a lot and eat beef, your inflation since 2019 ran well above the headline; if you own with a fixed mortgage and don't commute, it ran below. The index also can't see everything you feel: it adjusts for quality (a phone that costs the same and does more is scored as a price cut), it assumes some substitution, and it counts your rent even if you haven't moved. And a speedometer reads the present. A third of the CPI is shelter, which reads rents from about a year ago. The Fed is driving by a gauge that lags.

What people argue about

What caused 2021–24. One side: supply shocks (shipping, chips, energy, food) started it and would have faded on their own; the rate hikes mainly made housing worse. The other: $5 trillion in relief and two years of zero rates turned a spike into a plateau, and the hikes are why it ended. Bernanke and Blanchard's 2023 study puts the initial surge on supply and the persistence on an overheated labor market, which hands each side half. The data can't say what would have happened with smaller relief or earlier hikes, because that economy didn't run.

Whether 2% is the right target. Some economists want 3%, for more room to cut in a downturn and because 2% was never derived from anything in particular. Others say a target you move after missing it isn't a target. The Fed reaffirmed 2% in its 2025 framework review.

Shelter. Most of the housing third is "owners' equivalent rent," an estimate of what homeowners would pay to rent their own homes; nobody writes that check. One camp says it overstates inflation for the two-thirds who own; the other says the year-long lag means the Fed is always reacting to old rents. Both are right about the mechanics; what to do about it is a policy question, not a data one.

Rates against an oil shock, again. With oil driving the 2026 numbers, the 2022 argument is running in real time: raise rates to stop a second round, or hold because rates can't produce oil and the job market is already soft. The chart shows headline against core; it won't show which side was right until later.

See also

Sources

The weekly digest

One email on Monday mornings: the week's new articles and charts, the numbers that moved most, and what comes out in the week ahead. No ads, and no tracking of opens or clicks.

Past issuesHow addresses are kept