Don't bet on inflation falling just because the maths says so

Equity traders are treating a slowdown in the statistics as if it were a truce in the economy. Headline consumer price inflation was 8.5 per cent year-on-year in July, down from 9.1 per cent the month before, and risk assets rallied as though price stability had been restored by decree. A flat reading for month-on-month consumer prices was taken as proof of painless disinflation. Anyone who looks at how price indices are built knows that base effects flatter the numbers. Inflation falls when demand gives way to supply that cannot expand quickly, and a bigger denominator from twelve months ago does nothing to bring that about.

Annual headline rates and short-run annualised trends are telling different stories, and the gap hides persistent momentum outside energy. A fall in wholesale petrol prices can drag down the monthly headline figure while core prices keep compounding underneath. Treating one month of flat headline prices as lasting relief confuses the temperature of the water with the state of the dam.

How base effects flatter

Year-on-year measures distort badly when prices are volatile. The large monthly increases recorded during the first supply disruptions of late 2021 drop out of the twelve-month calculation, pulling the annual rate down even if month-on-month changes stay uncomfortably positive. If core prices keep rising at 0.4 per cent a month, the annualised pace stays near 5 per cent, far above anything the Federal Reserve's mandate can accept.

The same arithmetic will keep flattering the headline for months. Consumer prices jumped by 1.3 per cent in June alone, and that single month stays inside the annual comparison until next summer, when its exit will pull the headline rate down sharply unless that month brings another jump. A falling annual rate in the first half of next year will therefore say more about the spring of 2022 than about the state of demand.

Three-month annualised figures, for their part, depend heavily on where the window starts and ends. Annualise three months that include one freak fall in energy prices and the result is a disinflation trend that vanishes once commodity prices settle. Companies do not set prices off three-month moving averages. They price contracts against what they pay today for freight and specialised labour, as well as for working capital, and those costs are still drifting up.

Petrol did the heavy lifting

The flat July reading owed almost everything to the pump. The national average price of regular petrol topped $5 a gallon in mid-June for the first time and had dropped back below $4 by August, as crude retreated from the spike that followed Russia's invasion of Ukraine. Strip out food and energy and the July index still rose by 0.3 per cent on the month, leaving core inflation at 5.9 per cent on the year, the same rate as in June. Energy can reverse as quickly as it fell, and the approach of winter gives it every opportunity.

Services refuse to budge

Price pressure has moved from traded goods into domestic services, where prices are sticky. In cars and electronics, supply bottlenecks are easing: spot container rates have fallen and wholesale inventories are building. That goods disinflation is being offset by shelter and medical services, along with commercial insurance, which respond little to price and adjust over years rather than months.

Shelter makes up nearly a third of the headline index. It enters the official figures through rental-equivalence surveys that lag actual new leases by four to six quarters, so the sharp rise in market rents through late 2021 and early 2022 has only just started to show up. Private indices of new leases already show rent growth slowing from its peak, but by the index's own construction that relief will take several quarters to arrive.

What the Fed reads instead

The Fed's target is framed in terms of a different index, the personal consumption expenditures price measure, and its core version rose by 4.6 per cent in the year to July, more than double the 2 per cent goal. Pay is running faster still: average hourly earnings were up 5.2 per cent on a year earlier in August. Wage growth of that order, with productivity growing slowly, is consistent with services inflation well above target, which is why officials keep pointing to the labour market when asked about a peak in prices.

The 1970s warning

The last time America lived with inflation of this kind, a statistical retreat misled policymakers. Headline inflation more than halved between the end of 1974 and the end of 1976 as the first oil shock passed through the index, and the Fed under Arthur Burns let policy ease. The underlying pressures had not gone away, and by 1980 inflation stood above its 1974 peak. Jerome Powell invoked that episode at Jackson Hole in August, which is one reason a single soft month is unlikely to move him.

The optimists' case rests on goods. Container rates, wholesale used-car prices and swelling retail inventories all point to falling goods prices into the winter, and goods carried much of the 2021 surge, so deep enough deflation there could drag core inflation lower even with services firm. The flaw is proportion: services, shelter included, account for well over half of the index, and they are the part moving the wrong way. The August figures, due on 13 September, land a week before the Fed meets on 20 and 21 September, and the monthly change in core services will matter far more than the headline annual rate.

The misreading has a price for those who act on it. Equity investors who bought the July rally paid for a peak in the policy rate that the Fed has not endorsed, and holders of long-dated bonds are exposed if services keep compounding. Buyers of inflation-linked securities are paid on the realised index, so they gain from exactly the stickiness that the equity market is discounting.

For investors, published statistics and policy will point in different directions for some time. The Fed will not pause simply because twelve-month headline rates come off their peaks. Until month-on-month core inflation falls towards two-tenths of a per cent, real policy rates will probably have to keep rising, and portfolios built on a statistical truce are exposed. A Fed that waits for base effects to do its work will find services inflation written into the next round of wage settlements, and equity holders who bet on an early pause will be the first to give back the summer's gains.