Why prices never come back down

Every inflation conversation contains the same confused hope: that once the "wave" passes, prices will return to where they were. They won't. Inflation is a rate of change; the level it produces is permanent. That single distinction explains a surprising amount of economics and most of the anger about it.

Suppose inflation runs at six percent for three years and then stops entirely. Prices have risen about nineteen percent, and they stay there. The news can report "inflation is back to normal" in year four while every item in the shop remains permanently more expensive than you remember. Both statements are true, and the gap between them is where most public argument about the economy happens.

Deflation — a sustained fall in the general price level — does exist. It's just rare in modern economies, and the reason is not that anyone forgot how to do it. It's that the mechanisms that would produce it are actively avoided, because of what they do to the rest of the system.

Rate versus level

Inflation is measured as the change in a price index, usually the consumer price index, which tracks a fixed basket of goods and services weighted by what households actually buy. When people say inflation is three percent, they mean the basket is three percent more expensive than a year ago. When they say prices went up twenty percent, they mean the basket is twenty percent more expensive than at some past date.

The confusion between these is the same shape as the difference between speed and distance. A car that stops accelerating has not returned to where it started. But the intuition is stubborn, because our memory of prices is bound to specific objects — the price of the coffee, the sandwich, the rent — and those reference points feel like they should be restored when the "problem" is over.

A related artifact, the base effect, adds to the noise. If prices spiked in one month last year, the annual inflation rate this year will look low in that month even if nothing has changed, simply because the comparison point is high. This is why inflation headlines can contradict the prices you personally encounter for months at a time.

Why wages don't fall, and prices follow them

A deflationary world would require something most employers won't do: cut nominal pay. Truman Bewley's 1999 book Why Wages Don't Fall during a Recession is built on interviews with hundreds of business managers, and the answer they gave was consistent. Cutting pay damages morale, invites the best people to leave, and is remembered. Firms would rather lay people off, freeze hiring, or reduce hours than lower the number on the pay stub.

Modern behavioural economics adds a sharper finding. In a well-known 1997 study, Eldar Shafir, Peter Diamond and Amos Tversky asked people about two scenarios with the same real outcome: a five percent pay cut with no inflation, versus a two percent raise during seven percent inflation. The second is worse in real terms. Participants rated it as fairer and preferred it. People reason in nominal numbers, and a raise feels like a raise.

Money illusion: the same real outcome is judged differently depending on which number is written down.

Since wages are the largest cost for most firms, sticky wages put a floor under prices. A business that has committed to roughly stable nominal payroll cannot cut prices without cutting margins and then jobs. So the general price level ratchets up and stays up.

There's a second ratchet on the price side, called menu costs: the real expense of changing prices — printing menus, updating systems, retraining staff, and the customer reaction to a visible increase. Firms therefore change prices in occasional jumps rather than continuously. When a jump happens, it's sized to cover expected future costs so it doesn't have to happen again soon. Over time, that behaviour produces exactly the pattern people find maddening: prices rise in steps and never come down.

Shrinkflation is the same thing in disguise

When a firm doesn't want to raise the visible price, it reduces the quantity. Fewer grams in the bag, a smaller scoop, one less item in the pack. The sticker price holds; the unit price rises. It looks like a trick and is often experienced as one, but it's a rational response to menu costs: the number that customers notice stays fixed while the cost is recovered invisibly.

Statistically, this is captured, if imperfectly. Price indices are adjusted for package size and quality where agencies can measure it. But measurement lags, and consumers notice the change in the bag immediately while the index adjusts later — a perception gap that makes official inflation feel understated even when it's technically correct.

Why central banks aim at two percent rather than zero

This is the part that seems most obviously wrong to anyone outside economics. If rising prices are the problem, why target any rise at all?

The answer is that deflation is worse, for two reasons. The first is mechanical. Debt is denominated in nominal terms. If prices and wages fall, the real burden of a fixed mortgage or loan grows: you owe the same number while earning less. Irving Fisher described this debt-deflation spiral in 1933, arguing that it turned the 1929 crash into the Depression — falling prices, rising real debt, forced selling, further falls. Japan's experience after 1990 is the modern case study: two decades of near-zero or negative inflation, sluggish growth, and a central bank that ran out of conventional room to respond.

The second reason is that a positive inflation target gives monetary policy room to work. Nominal interest rates can't go far below zero, because people can hold cash instead. With inflation at two percent, a central bank can cut the real interest rate several points before hitting that floor. At zero inflation, the floor is reached immediately in a downturn.

Two percent is a convention rather than a law of nature. New Zealand formalised it first, in 1990, and it spread because it was low enough that people could mostly ignore inflation when planning, and high enough to keep the buffer. It is a chosen number that became the world's default.

How a high inflation era actually ends

The 1970s and early 1980s are the reference case, and the ending is instructive about costs. US inflation peaked near fifteen percent in 1980. Paul Volcker, appointed Fed chair in 1979, raised the federal funds rate to levels not seen before or since — briefly around twenty percent between 1980 and 1981. Inflation fell to roughly three percent by 1983.

The price was severe. The economy went through two recessions in three years, and unemployment reached nearly eleven percent in 1982, the highest since the Depression. Volcker's tightening is routinely described as necessary and is still politically remembered as brutal. That is the honest shape of the trade: disinflation is achieved by suppressing demand, and suppressing demand means people lose jobs.

Which reframes the recent past. When inflation went from near-zero in 2020 to over nine percent in mid-2022 and back toward normal without a large rise in unemployment, that outcome was genuinely unusual, and economists argued publicly about whether it could last. The prices from the spike remained. That is the part of the story that never reverses.

What this changes in practice

A few implications are worth being explicit about, because they're the ones that affect decisions rather than opinions.

• Cash loses value slowly and reliably. At three percent, money in a zero-interest account loses about a quarter of its purchasing power in a decade. That's the cost of liquidity, and it's fine to pay it for an emergency fund — just know you're paying it.
• Nominal raises are not the same as real ones. A four percent raise during five percent inflation is a pay cut. Comparing the offer to last year's number is the wrong comparison; comparing it to prices is the right one.
• Long-term fixed-rate debt benefits from inflation. The nominal balance stays fixed while the currency it's repaid in is worth less. This is why inflation redistributes from lenders to borrowers, and why it's politically charged.
• Wage indexation is worth asking about, and rarely offered. Where contracts or union agreements tie pay to a price index, the adjustment is automatic. Where they don't, you renegotiate, and the burden falls on whoever has more leverage.

The perception gap is real, and it's rational

Surveys consistently find that people think inflation is higher than official statistics report. Part of this is partisanship and part of it is method, but a real component is that the index is an average of a basket you don't buy. If your rent rose twenty percent, your insurance rose fifteen, and televisions fell ten, your personal inflation rate is higher than the national one, and you are not wrong about your own life.

The useful response is not to distrust the statistic but to compute your own. Add up what you actually spend, in the categories that actually dominate your budget, and compare this year to two years ago. Most people find the number is higher than the headline and lower than their impression — and either way, it's the number that should inform what you ask for at work.