What inflation is and why it moves markets

The same basket, month after month 

The list being priced is not a list of everything sold. It is a basket: a fixed selection of goods and services chosen to stand for what households in that economy actually buy. Rice, coffee, a litre of petrol, a bus fare, a haircut, a month's rent, a mobile tariff, a school fee. Each entry carries a weight, and the weights come from a household expenditure survey, so a category households spend a large share of their money on moves the index far more than one they buy rarely. Housing and rents carry heavy weights in most baskets, which is why an index can rise noticeably in a month when supermarket prices did not move at all.

Two disciplines make the number comparable across time. The first is matched pricing: the same item is priced in the same outlet in successive months, so a change in the index reflects a change in price rather than a change in what happened to be sampled. The second is quality adjustment. When the laptop in the basket is discontinued and replaced by a faster one at the same price, the agency decides how much of that is a price change and how much is a quality change, by a documented technique called a hedonic adjustment. The technique is published, but the split it produces is a judgment, which is why statisticians argue about index construction instead of treating it as arithmetic.

What the agency publishes first is an index level, not a percentage. The level means nothing on its own: it is set to a round value in a chosen base period, and every later reading is a statement about that base and nothing else. The rate that gets reported is the change in the level, either against the previous month or against the same month a year earlier. The annual comparison leads most releases, because comparing like months cancels the seasonal pattern that repeats every year: holiday travel, school terms, harvests, the timing of religious calendars.

That arithmetic is the whole of it. An inflation rate is a weighted average of price changes, and the weights decide what the number is mostly about. A basket weighted towards rents is largely a statement about the housing market; one weighted towards imported food and fuel is largely a statement about world prices and the exchange rate they are paid at. Two economies can publish the same rate for entirely different reasons, which is why the composition of a release is read alongside it.

A falling rate is not falling prices 

This is the most common misreading of an inflation release, and it survives in commentary written by people who know better. When the annual rate falls from a higher figure to a lower one, prices are still rising. They are rising more slowly. The word for that is disinflation. Prices only fall when the rate itself goes below zero, and that is deflation, a different condition with different causes and a different policy response. A headline announcing that inflation has come down describes a change in the speed of the price level, never a reversal of it.

The annual rate also moves for reasons that have nothing to do with the present. Because it compares this month's level with the level twelve months earlier, it changes when either end of the comparison changes. A sharp one off increase a year ago eventually drops out of the window, and the annual rate falls on the day it does, even if prices this month rose at exactly the pace they rose at last month. Economists call that a base effect, and it is a routine feature of the series rather than an anomaly in it. The monthly change describes what happened recently; the annual change carries a year of history inside it.

Headline and core 

Headline inflation is the whole basket, weighted as published. Core inflation is the same basket with the most volatile categories removed, conventionally food and energy, though the exact exclusions differ by agency and some also strip out alcohol, tobacco or administered prices. The reasoning is that a harvest failure, a shipping disruption or a decision by an oil producing group can swing the headline in either direction within a single month without saying anything about the pace at which prices are being set across the rest of the economy. Removing those categories is an attempt to separate a signal from noise, not an attempt to make the number look better.

Both are published because they answer different questions. Headline is the measure that describes what households actually pay, and it is the measure that feeds wage negotiations, contract indexation and the general expectation of what prices will do next. Core is a diagnostic, used to judge whether a move in the headline is likely to persist once the volatile categories settle. A central bank whose mandate is written in terms of the headline index still reads core closely, because the question a policy decision turns on is persistence rather than the value of the last print.

Economists disagree about core, and the disagreement is worth holding onto. The objection is that excluding food and energy excludes precisely what a lower income household spends most of its money on, so treating core as the real inflation rate turns an analytical convenience into a claim about people's lives. Agencies have answered by publishing further measures rather than choosing between them. A trimmed mean discards the largest moves in both directions each month, and a weighted median takes the middle of the distribution, both arguing that whether an item is volatile should be decided by that month's data rather than by a label fixed in advance. No measure has settled the question, which is why several are published side by side.

Why a central bank treats this as its central problem 

Almost every central bank has a statutory objective phrased around price stability, and most express it as a numerical target for a named index over the medium term. Mandates differ in what else they contain. Some are single mandates, in which price stability is primary and everything else is subordinate to it. Others are dual mandates, in which employment stands alongside prices and the two can pull in opposite directions, which is where most of the difficulty in a rate decision lives. The objective is a predictable price level in every case, because contracts, wages, savings and lending are all written in money, and a currency whose purchasing power moves unpredictably makes every one of those agreements harder to write.

The instrument is the policy rate: the rate at which the central bank lends to and borrows from the banking system. It propagates outward into what banks charge for credit and pay on deposits, and from there into the cost of financing spending. A higher policy rate makes borrowing more expensive and holding cash more rewarding, which slows credit financed demand relative to supply, which reduces the pressure on prices. A lower rate works the same chain in reverse. Every link is uncertain, each varies in strength with how indebted households and firms are, and the whole chain operates with a lag conventionally described as running from several quarters to around two years. That lag is why decisions are taken on a forecast of where inflation is heading rather than on the reading published this morning.

That constraint is not identical everywhere, and this region is the clearest illustration. A currency peg is a commitment by an authority to hold its exchange rate at or near a stated level against another currency, and holding it ties the domestic policy rate closely to the anchor economy's, because a large gap between the two would put pressure on the peg through capital flows. A central bank in that position addresses domestic price pressure with other instruments: reserve requirements, macroprudential rules on lending, fiscal measures, and in several economies the direct administration of fuel and utility prices. The Gulf central banks, the CBUAE and SAMA among them, operate under exactly that arrangement, so a Gulf inflation release stands in a different relationship to domestic policy than one published in a floating rate economy.

It also changes where inflation comes from. When most food and manufactured goods are imported, their domestic price is the foreign price multiplied by the exchange rate they are bought at, plus freight and margin, so price pressure arrives from outside through world commodity prices, shipping costs and the anchor currency's own inflation rather than being generated at home by domestic demand. A domestic interest rate has limited purchase on any of that. It is a structural fact about small open economies, not a criticism of anybody's policy.

The reading and the expected reading 

An inflation release is scheduled months ahead, and it does not arrive into an empty room. Economists at banks and research houses publish estimates in advance, and data providers aggregate those estimates into a consensus figure. Whatever sits in that consensus has already been acted on by everyone who intended to act on it, so it is already reflected in prices before the release exists. What is genuinely new at the moment of publication is the difference between the number published and the number expected. That difference is the surprise, and it is a different quantity from the inflation rate itself.

The two can point in opposite directions in the same release, which is the point at which the distinction stops being pedantic. A rate can fall from the previous month, so the level is lower, while still landing above what economists expected, so the surprise is positive. A very high rate that arrives exactly where the consensus put it carries no new information at all. This is why commentary describing a release as good or bad on the basis of the level alone is describing only half of what was published.

The idea that markets price the surprise rather than the level is a convention, and it has real limits. A consensus is a median of estimates gathered at different moments from a self selected set of forecasters, so it approximates expectation rather than measuring it, and positioning ahead of a release is not observable at all. Practitioners also read which components drove the change and whether core moved with the headline, so two prints carrying an identical surprise are not treated as the same event. The relationship between a data surprise and any subsequent market move is neither fixed in size nor stable in sign across regimes, which is why nobody who works with this data treats it as a formula.

Where the measure itself is contested 

A fixed basket answers what a fixed set of purchases costs today against what it cost before. It does not answer what it costs to maintain a standard of living, and the two diverge because households substitute: when beef becomes expensive some buy chicken, and an index still pricing beef at last year's weight records more of an increase than the household experiences. Chained indices update the weights more often and are criticised in turn for assuming substitution is costless. Neither approach is wrong. They measure different things, and an argument about the true inflation rate is usually an argument about which question is being asked.

Housing is the other long running dispute, and it matters more than most because housing carries a heavy weight. Rent is straightforward, being a price paid for a service in a period. Owner occupied housing is not, because a house is partly a durable good and partly an asset, and no consensus exists on how to put an asset into a consumer price index. Some agencies impute a rent the owner notionally pays themselves, some use the cost of acquiring a dwelling net of the land, and the euro area's harmonised index has historically left it out altogether. Indices also differ over rural areas, informal markets and the spending of visitors. Two national rates set side by side are therefore often two definitions rather than two outcomes.

None of this makes an index untrustworthy. It makes it a measurement with a method, and every agency that produces one publishes that method in full. What it rules out is treating a decimal place as a fact of nature. An inflation figure is the output of documented choices about what to price, how to weight it and how to handle change.