The inflation number the Fed secretly cares about

US consumer prices rose 6.5 per cent in the year to December, confirming that headline inflation is coming down from its summer peak. Equity markets reacted with their usual enthusiasm, bidding up tech valuations and pricing in Federal Reserve rate cuts before autumn. Anyone listening closely to officials at the Eccles Building knows the Fed has already moved past headline CPI and even conventional core CPI. The number that now drives US monetary policy is core services excluding housing, which Fed staff call "supercore" inflation.

The reasoning is sound. Headline inflation is dominated by volatile food and energy prices, and core goods inflation is fading fast as supply chains heal and retailers clear their stock.

Where the headline fell

December's figures showed how much of the improvement comes from the volatile parts. Headline prices fell slightly on the month, the first monthly decline since the spring of 2020, largely because petrol became cheaper. Core prices, which exclude food and energy, still rose 0.3 per cent in the month and 5.7 per cent over the year. The annual headline rate has dropped from 9.1 per cent in June, and most of that fall reflects energy and used cars.

That is the easy part of disinflation, and markets have treated it as the whole story. Petrol prices can reverse as quickly as they fell, and used-car prices have already corrected much of their pandemic surge, so neither can keep pulling the headline down for long.

Why supercore matters

Even housing inflation, high as it is in the official figures, is a known laggard. Private-sector spot rents have already levelled off, which all but guarantees that official shelter numbers will turn down by mid-2023. Shelter accounts for about a third of the consumer price index, and the official measure averages rents across existing leases, most of which reset only once a year, so it trails market rents by many months.

That leaves core services excluding housing, which covers healthcare, transport, hospitality and professional services, as the clearest read on the domestic labour market. In these industries wages make up most of the costs, and prices move closely with pay. It is also the largest component of the core index the Fed targets, the personal consumption expenditures measure, which gives it more weight in the Fed's thinking than its prominence in the CPI suggests.

Jerome Powell made the case explicitly in a speech at the Brookings Institution at the end of November, singling out this category as the most important for understanding where core inflation goes next. The minutes of December's policy meeting, published last week, showed that no official expected it would be appropriate to cut rates this year.

Wages set the pace

Supercore services, unlike manufactured goods, cannot be imported from cheap factories abroad. As long as average hourly earnings grow at 4.5 to 5 per cent a year, supercore inflation is likely to stay stuck near 4 per cent, and a return to the Fed's 2 per cent target will be arithmetically out of reach.

The arithmetic is simple enough. Pay growth consistent with 2 per cent inflation is roughly the target plus the trend rate of productivity growth, and American productivity has been weak. Wage growth near 5 per cent is therefore well above what the Fed can tolerate, and in labour-heavy services the excess passes into prices with little to offset it. The Federal Reserve cannot declare victory while the measure it has chosen keeps accelerating.

Officials also have a history they are determined not to repeat. In the mid-1970s, under Arthur Burns, the Fed eased as headline inflation fell back after the first oil shock, and inflation returned later in the decade at higher rates than before. Paul Volcker then had to push rates to painful heights to finish the job. Powell has cited that episode repeatedly in arguing that the Fed must keep policy tight until inflation is clearly beaten, which makes a slowdown in petrol prices a weak reason for officials to relax.

A noisy measure

The strongest objection is that supercore is a volatile series that can swing on airfares, car insurance and other erratic items. One quirk will flatter it for months: the CPI measures health insurance using insurers' profits, updates the method once a year, and began subtracting from services inflation in October. A few soft readings driven by that quirk would say little about wages.

The labour market is also cooling at the edges. Job openings have come down from their peak, the rate at which workers quit has eased, and some measures of wage growth have slowed from their highs. If that continues, supercore inflation will follow with a lag. The disagreement is over how fast, and the Fed has no incentive to bet on the optimistic answer after being wrong about inflation in 2021.

What the Fed will watch

The fourth-quarter Employment Cost Index, due at the end of January, is the cleanest reading of labour costs, and it arrives a day before the Fed's decision on 1 February. The January consumer price report in mid-February will show whether supercore is slowing on a monthly basis once the health-insurance distortion is set aside. Monthly personal consumption data will show the same category in the Fed's preferred index.

Investors cheering cheaper petrol and used cars are watching the wrong numbers, and until supercore inflation gives way to a looser labour market, the Fed will probably keep rates high through 2023 and disappoint anyone betting on cuts before autumn.