The same cart, a bigger bill.

Every month the Bureau of Economic Analysis adds up what Americans spent, and every month it performs a second, quieter calculation: how much of that spending bought more things, and how much merely paid higher prices for the same things. In September the two calculations told different stories. Consumers spent $113.1 billion more than in August, a rise of 0.5 percent. Once the effect of higher prices was removed, the extra goods and services they actually took home came to $2.6 billion, a change so small it rounds to zero. The country walked out of the store with the same cart and a bigger bill.

That is not how inflation is supposed to look. In the picture most of us carry around, prices rise because people want more than the economy can make: too much money chasing too few goods, shoppers elbowing each other for the last television on the shelf. September showed nothing of the kind. People were paying more without getting more. If there is a frenzy somewhere, it is not in the shopping aisles.

My argument in this essay is that today's inflation is best understood not as one force but as four ordinary mechanisms acting at once. Costs are being passed along. Demand is being propped up at the top of the income ladder. A wave of spending is coming from buyers who are not households at all. And expectations have learned, after five years, to assume the worst. None of the four requires a buying frenzy. Together they produce something that resembles a mild version of the 1970s, and they explain why the Federal Reserve, armed with an instrument that works mostly on household demand, is divided about what to do.

Two numbers on every receipt

Begin with a distinction that economists consider elementary and almost everyone else finds slippery: the difference between nominal and real. A nominal figure is counted in dollars as they come. A real figure is counted in dollars of fixed purchasing power, so that a change in it reflects a change in the amount of stuff rather than in the price tags.

Suppose you buy ten apples for $10 in August and the same ten apples for $11 in September. Your nominal spending rose 10 percent. Your real spending did not change at all, because you ate the same ten apples. The government performs this subtraction for the entire economy using an index of prices called the PCE price index, and in September that index rose 0.5 percent, exactly as much as spending did. Every cent of the increase went to price.

The same subtraction applied to paychecks yields a darker number. Disposable income rose 0.2 percent in dollars but fell 0.3 percent in real terms, because prices outran pay. How, then, did households keep their real spending flat? They saved less. The personal saving rate slipped to 3.8 percent, thin by historical standards. Families are holding the cart steady by dipping into the cushion, and a cushion is a finite thing.

Two ways to raise a price

Economists distinguish two routes by which the overall price level can rise, and the distinction is easier to see in a picture than in words. Draw two lines on a sheet of graph paper. The first slopes down. It is demand: the lower the price level, the more output buyers want. The second slopes up. It is supply: the higher the price level, the more producers are willing to make. Where the lines cross, the economy settles.

There are two ways to push that crossing point higher. In demand-pull inflation, buyers want more at every price, the demand line shifts outward, and prices and output rise together. That is the frenzy. In cost-push inflation, producing anything becomes more expensive, the supply line shifts upward, and prices rise while output stalls or falls. Every economics student learns the two cases. Fewer people notice that they leave different fingerprints in the data. Demand pull raises real spending. Cost push does not.

September's fingerprints belong mostly to cost push, with a twist. If costs alone had risen, real output would have fallen, because a supply shock with nothing else going on shrinks the cart. Instead it held steady, which means something was also nudging demand up, just enough to offset the squeeze. Much of the rest of this essay is about what that something is.

Why higher costs stick

Start with the costs. In late February, war between Iran and the United States and Israel shut most traffic through the Strait of Hormuz, which normally carries about a fifth of the world's oil trade. The International Energy Agency called it the largest supply disruption in the history of the global oil market. Brent crude peaked at $118 a barrel at the end of March. By September, gasoline was 34.2 percent more expensive than a year earlier, energy prices overall were up 20.4 percent, and airline fares were up 22.4 percent. Electricity, pushed by forces of its own that we will come to, was up 4.4 percent.

A cost increase becomes a price increase through a process called pass-through, and how much passes through depends on what economists call elasticity: how sharply buyers cut back when a price goes up. Gasoline is the classic inelastic good. People still have to drive to work, so a refiner that raises prices loses few customers. When every competitor faces the same cost shock at the same moment, no single firm fears being undercut, because its rivals are raising prices too. Some economists, notably Isabella Weber, have argued that shocks like this give firms cover to raise prices by more than their costs; others find that markups have mostly tracked costs. Either way, the mechanism needs one condition, which is that demand stays flat rather than collapsing. A firm cannot pass a cost increase on to customers who have stopped coming. It can pass it on to customers who keep coming but buy no more.

Notice what the headline hides, though. Core inflation, which excludes food and energy, was 3.1 percent by the PCE measure. That is lower than the headline 3.6 percent, consistent with energy doing the heavy lifting, but it is still well above the Fed's 2 percent goal, so energy is not the whole story. Some of the shock has come in through the side door. Airline fares count as a core service, not as energy, yet they run on jet fuel. Shipping companies have added fuel surcharges to everything they carry. An oil shock does not stay in the energy column for long.

An average with two arms

If households as a whole are barely holding their real spending steady, who is keeping demand alive? Part of the answer is that "households as a whole" is a statistical fiction. An average of one family spending freely and one family cutting back describes neither of them.

Researchers at the Federal Reserve Bank of New York split retail spending by income. Between January 2023 and March 2026, real spending by households earning more than $125,000 a year grew about 7.6 percent. Middle-income households managed about 3 percent. Households earning under $40,000 grew just over 1 percent. Before the pandemic, the pattern ran the other way. Economists call this a K-shaped economy, after the letter's two arms: one climbing, the other flat or sinking.

The upper arm has a simple explanation. Higher-income households own most of the stocks, much of the home equity, and most of the savings that now earn real interest. When those assets rise, their owners feel richer and spend more, a regularity economists call the wealth effect, and they spend it disproportionately on services such as travel, restaurants, and hotels. Then comes the subtle part. Prices are set at the margin, by the customers still willing to pay. A hotel that is full at $300 a night has no reason to charge $250, even if most of the town could no longer afford a room. The buyers on the upper arm of the K set the prices that everyone faces.

The buyers who are not shoppers

The other part of the answer lies outside the household altogether. Every introductory macroeconomics course teaches an accounting identity: total output equals consumer spending, plus business investment, plus government purchases, plus net exports. When the first term is flat, the others can still be booming.

Investment is booming. The five largest technology and cloud companies are on track to spend roughly $725 billion on capital in 2026, by one mid-range estimate, most of it on data centers for artificial intelligence. One analysis estimates that AI-related investment supplied about 1.5 percentage points of the economy's 2.0 percent annualized growth in the first quarter, while consumers supplied about 0.3. (The government publishes no separate line for AI, so figures like these are estimates.) A data center does not buy groceries, but it buys what households also need: electricity, electricians, concrete, transformers, and a place in the utility's queue. When a new buyer with very deep pockets enters those markets, prices rise for the old buyers too. That is part of why your electric bill is climbing.

Government is spending well beyond its revenue as well. The federal deficit for fiscal 2026 came to $2.0 trillion, or 6.2 percent of GDP, according to the Congressional Budget Office. A deficit that size with unemployment near 4 percent is historically unusual; it puts dollars into the economy that taxes do not take back out. This is how the economy-wide numbers can look strong while households feel squeezed. In October, the Fed upgraded its description of growth from "solid" to "strong." Both impressions are accurate, because they describe different terms of the same identity.

The sticker gun

The fourth mechanism lives in people's heads. In the late 1960s Milton Friedman and Edmund Phelps argued that inflation depends partly on what people expect it to be. If workers expect prices to rise 4 percent, they bargain for 4 percent raises, and if firms expect their costs to rise 4 percent, they raise prices to match. Expectations, once formed, tend to fulfill themselves.

For most of the two decades before 2021, ordinary people hardly thought about inflation, and that inattention was a kind of national asset: prices rose about 2 percent a year because everyone quietly assumed they would. Inflation has now run above the Fed's target for more than five years. In October, households surveyed by the University of Michigan expected prices to rise 4.7 percent over the coming year and 3.5 percent a year over the next five. A business that raises its prices every spring no longer needs a reason, because its customers have stopped asking for one. Each new price goes on top of the last, like a sticker from a pricing gun, and nobody bothers to peel off the old one.

Stagflation lite

Put the four mechanisms together and you get prices rising mostly from the supply side while ordinary households feel poorer. Economists have a word for the full-strength version. Stagflation was coined in Britain in the 1960s and made famous in America in the 1970s, when the oil embargo of 1973 and the Iranian revolution of 1979 sent energy prices soaring, inflation reached double digits, and unemployment climbed along with it.

The parallels are uncomfortable, right down to the map. But the differences matter just as much. Inflation today is 3.6 percent, not 13. Unemployment is low, and the Fed expects it to end the year around 4.1 percent. The economy is growing, not shrinking. What we have is the 1970s with the volume turned down: a supply shock, a squeezed consumer, and expectations that have begun to drift, without the mass unemployment that made the original so painful. Call it stagflation lite. The "lite" is real, and so is the risk that it will not stay that way.

The case that it is demand after all

A fair argument has to meet its strongest opponents, and on this question some of them sit on the Federal Open Market Committee. In September the Fed raised its benchmark rate for the first time since 2023, to a range of 3.75 to 4 percent. In October it held, but two members, Governor Lisa Cook and Cleveland Fed President Beth Hammack, dissented in favor of another increase.

Their case is a serious one. Growth is strong. Unemployment is low. Core inflation, which strips out the energy shock, is 3.1 percent and has been above target for years. Expectations are rising, and a central bank that waits for proof that expectations have come loose usually waits too long. From this angle, a $2 trillion deficit and a $700 billion investment boom are exactly the kind of demand that monetary policy exists to restrain, whether or not shoppers are the ones doing the spending. The disagreement is not really over the facts in this essay. It is over which risk is worse: tightening into a supply shock, or letting another year of high inflation teach people to expect it.

One lever, the wrong lane

Here is the Fed's predicament in a sentence. Its main tool, the short-term interest rate, does most of its work on the buyers who are already pulling back.

Higher rates work by making borrowing more expensive. The borrowers who feel it fastest are those with mortgages to refinance, credit card balances, car loans, and small businesses running on bank lines of credit, who sit disproportionately on the lower arm of the K. The largest technology companies pay for much of their building out of their own enormous cash flows, and they are betting on returns large enough that another quarter point barely changes the arithmetic. The federal deficit does not respond to interest rates at all. If anything, higher rates enlarge it: the government's interest payments rose $115 billion in fiscal 2026, and those payments flow to bondholders and savers, who sit mostly on the upper arm. The medicine, in other words, can feed some of the very spending it is meant to cool.

None of this means the Fed is helpless. It means its choices are worse than usual. It can raise rates further and accept that the cost will fall mainly on households that did not cause the problem. Or it can hold, hope that energy prices ease as the 1970s shocks eventually did, and risk letting expectations harden. Arthur Burns, who led the Fed through most of the 1970s, is remembered for waiting. Paul Volcker, who took over in 1979, is remembered for pushing rates to around 20 percent and breaking inflation at the cost of a deep recession. Every Fed chair since has kept both portraits on the wall.

Interest rates press hardest on the households already buying less, and lightest on the buyers driving prices up.

What the cart is telling us

The September receipt is a small document, but it carries a large lesson. Inflation is not one thing. Prices can rise because buyers want more or because producers must charge more, and the cure for one can be the wrong treatment for the other. Today's version is mostly the second kind, with enough of the first, on the upper arm of the income ladder and in corporate and government budgets, to keep it from fading on its own.

That combination explains the national mood better than any single statistic. The economy is growing, stock prices are high, and data centers are rising across the countryside, yet a large share of households are paying more without getting more. Both pictures are accurate. They are the two arms of the same letter, and any policy that treats the economy as a single average will keep surprising the people who live on the lower one.

Four things to watch

• Core against headline. If core inflation drifts down while energy stays high, the shock is passing through and fading. If core climbs, it is spreading.
• The five-year expectation. A one-year reading above 4 percent is a warning; a rising five-year reading is a louder one, because it means people think the problem is permanent.
• Real disposable income. While it falls, households can keep their carts full only by saving less, and the saving rate cannot fall forever.
• The Strait of Hormuz. Energy is the one cost the Fed cannot touch, and the one that could reverse fastest.