3.3% growth: too good to be true?

The US economy grew at an annualised rate of 3.3 per cent in the fourth quarter, ending 2023 well ahead of forecasts that had it slowing. Cheerleaders took the figure as proof that America has reached escape velocity. Analysts who break the number down have reason to be more careful.

A second strong quarter

The figure follows growth of 4.9 per cent in the third quarter, so the second half of the year was among the strongest since the rebound from the pandemic. Over the whole of 2023 the economy grew by about 2.5 per cent, compared with the recession that most forecasters expected at the start of the year. At the same time inflation cooled: the core price index for personal consumption, the Federal Reserve's preferred gauge, rose at an annualised rate of about 2 per cent in the fourth quarter.

Strong growth with falling inflation is the combination that central bankers hope for and rarely get. It is also the combination that invites suspicion, because in most economic models the two tend to move in opposite directions unless the economy's capacity to produce is itself growing faster.

Where the growth came from

Net exports and government spending supplied a large share of the headline, with a further lift from inventory building. Total hours worked in the private economy barely moved, which implies a sudden and remarkable jump in productivity outside farming. If productivity really has stepped up for good, growth can carry on without stoking inflation. If the figures have been flattered by quirks in the price deflators, a payback is likely in the coming quarters.

The link between hours and output is simple arithmetic. Output equals hours worked multiplied by output per hour. If output rises at an annualised rate of 3.3 per cent while hours are nearly flat, output per hour must have risen at close to the same pace, far above the average of recent decades. Productivity figures for the third quarter already showed a strong rise, and a second quarter of the same would be unusual outside the early stages of a recovery, when hours are depressed and output rebounds first.

How the deflator can flatter

Real GDP is estimated by taking spending in current dollars and dividing it by a price index. If the price index understates actual inflation, real growth is overstated by the same amount. When inflation falls quickly, as it did in the second half of the year, small errors in measuring prices for services, software or imports can make real growth look stronger than it was. Those errors are usually corrected only in later revisions as better source data arrive.

The gap between the two ways of measuring the economy is a warning sign. In principle, GDP measured from spending should equal gross domestic income, measured from wages and profits. For much of 2023 the income measure grew much more slowly than the spending measure, and the gap has been unusually large. The income figures for the fourth quarter will not be published until later in the revision cycle.

The case for a genuine boom

The strongest counterargument is that there are real reasons for productivity to have risen. Applications to start new businesses jumped after 2020 and have stayed high. The tight labour market has pushed companies to invest in software and automation, and workers who switched jobs during the pandemic may now be better matched to their roles. Immigration has expanded the labour force faster than official population estimates assumed, so hours may be understated in the data and supply larger than it looks.

Subsidies for factories and semiconductor plants have also raised business investment, which should add to capacity over time. If even part of this is right, the economy can grow faster than the usual estimates of potential without generating inflation, and the Fed can cut rates with less fear.

A precedent from the 1990s

The last time the numbers looked like this was in the mid-1990s. Growth was strong, unemployment fell and inflation stayed low, and many economists at the Fed argued for higher rates to head off overheating. Alan Greenspan resisted, arguing that computers and information technology were raising productivity in ways the statistics had not yet captured. He was proved right: productivity growth accelerated for most of a decade. The lesson is that surprising productivity figures are sometimes real, but that the confirmation takes years, not quarters.

Who should care

For bond investors the question matters directly. If the productivity gain is real, the Fed can cut rates while growth stays strong, a benign mix for both bonds and equities. If it is a statistical mirage, underlying growth is weaker than it looks, and later quarters will disappoint. Companies planning investment and hiring would also be misled by an overstated headline, and politicians in an election year will cite the figure whatever the revisions later show.

Public spending adds a second doubt. Growth that rests on government outlays, including spending by states and localities and subsidies under recent laws, depends on budget decisions rather than private demand. The federal deficit is already large, and with Congress divided, further fiscal stimulus is unlikely, so that support should fade over time.

What to watch

The Commerce Department will publish a second estimate of fourth-quarter growth at the end of February and a third at the end of March, and the annual revision later in the year often changes the picture substantially. The Labor Department's productivity report for the fourth quarter, due in early February, will show how much of the growth came from output per hour. Gross domestic income, once published, will show whether the income side confirms the spending side.

The 3.3 per cent figure is impressive, but growth built on public spending and an unexplained leap in productivity deserves scepticism. The first revisions, probably downward, will say more than the headline did. Expect the growth rate for the first half of this year to settle closer to 2 per cent, with the productivity debate unresolved until the annual revisions.