Inflation is already at target, if you squint

The October core Personal Consumption Expenditures (PCE) price index rose 3.5 per cent from a year earlier, still uncomfortably above the Fed's statutory goal. Annualised over six months, however, core inflation has slowed to 2.5 per cent, and over three months it is running at 2.0 per cent. How far disinflation has come depends on which window an analyst picks.

Why PCE is the yardstick

The Fed has aimed at 2 per cent inflation on the PCE measure since it first published a numerical target in January 2012. It prefers PCE to the better-known consumer price index because PCE covers a wider range of spending, including health care paid for by employers and the government, and because its weights shift as households substitute cheaper goods for dearer ones. The core version strips out food and energy, whose prices swing with harvests and oil markets. The result is a gauge that usually runs a little below core CPI and moves more smoothly.

In August 2020 the Fed amended its framework to seek inflation that averages 2 per cent over time, tolerating periods a little above target after periods below. That change was designed for a world of chronically low inflation. It left officials poorly placed when core PCE climbed above 5 per cent in early 2022, and it has made them wary of any reading that invites them to relax too soon.

How annualising magnifies

The short-window figures take a few months of price changes and compound them as if they persisted for a whole year. That multiplies any noise. A rise of 0.1 percentage point in a single monthly reading moves the three-month annualised rate by roughly 0.4 points, so two soft months can take the measure from worrying to comfortable. The twelve-month figure moves slowly because each month is only one of twelve in the sum.

The scale of the monthly numbers involved is small. A monthly core increase of 0.2 per cent, compounded for a year, gives about 2.4 per cent, while 0.3 per cent gives about 3.7 per cent. The difference between an economy near target and one well above it is therefore a tenth of a point a month, which is within the range that later revisions routinely shift. A three-month window rests on three such readings, each of which may yet be rewritten.

The longer measure has its own flaw. It still contains the price rises of last winter, when core inflation was running far hotter, so it describes where inflation has been more than where it is heading. That is the honest case for looking at shorter windows, and it explains why forecasters who favour them regard the job as largely done.

Pick your window

Choose one endpoint and inflation is beaten. Choose another and it is still sticky. The Federal Reserve declared victory too early in 2021 and was burned for it, so it is unlikely to set policy on three-month annualised numbers that odd seasonal patterns can easily distort. Officials will want several quarters of good data before they call the job done.

The choice has consequences for prices as well as for argument. Futures markets have moved to price several rate cuts for next year largely on the strength of the shorter measures, and long-term yields fell sharply in November as a result. If the Fed keeps anchoring on the twelve-month figure, that pricing is premature, and holders of rate-sensitive assets bought on the promise of quick cuts would give back some of their gains. Households hoping for cheaper mortgages would have to wait longer.

There is a middle route that avoids the window problem altogether. The Dallas Fed publishes a trimmed-mean PCE measure that discards the most extreme price changes each month, and the Cleveland Fed publishes a median consumer price index built on the same logic. Such measures are less vulnerable to a single category swinging the result, and officials cite them when the headline and core figures send mixed signals. If they too slow on a six-month view, the case for the optimists strengthens considerably.

Seasonal quirks since the pandemic

Seasonal adjustment is where the short windows are most fragile. Many companies reset their prices at the start of the year, and the statistical filters that strip out such regular patterns were calibrated on pre-pandemic years that no longer look typical. Since the pandemic, readings early in the year have tended to come in hot and those later in the year cooler, which flatters any three-month window that ends in the autumn. A soft October may partly be a calendar effect that unwinds in January.

Lessons from false dawns

The Fed has two reasons to distrust a good run. In the late summer of 2021 monthly price rises slowed for a couple of months, lending support to the view that inflation would prove transitory, before they accelerated again, and Jerome Powell retired that word in November of that year. In the mid-1970s inflation more than halved from its peak, the Fed eased, and prices roared back by the end of the decade. Both episodes taught officials that stopping too early costs far more than waiting a few extra months.

What the Fed will look for

Officials watch the composition as closely as the total. Goods prices have flattened as supply chains healed, and the official measure of housing costs is slowing with a long lag behind market rents. The stubborn part is services excluding housing, which moves with wages and which the Fed chair has singled out for attention since a speech at the Brookings Institution a year ago. If that component cools in the November and December figures, and the January data avoid another early-year jump, the short windows will start to look like signal rather than noise.

Wall Street can toast the three-month figures, but the Fed will wait for the longer measures to confirm them, which probably means rates stay where they are for some months yet. The likeliest sequence is a steady policy rate through the winter, with talk of cuts gathering only if the six-month figure stays near 2.5 per cent into the spring and the twelve-month rate falls below 3 per cent.