For investors, the big bet on an AI future may still seem to be the only game in town – unless markets have priced the bulk of it into leading stocks already. — Reuters
THE return of US economic data when Washington reopens may do little to clear up a macro picture that is clouded by the dash for artificial intelligence (AI) and compounded by trade distortions.
Federal Reserve (Fed) officials are doing their best at the guessing game on what comes next – but it is just that.
The more humble in the Fed’s ranks concede that it’s impossible to tell.
Cleveland Fed boss Beth Hammack, a known hawk and a voting member of the Fed’s policymaking council next year, said the AI investment frenzy and related stock price surges complicated matters by creating something of a dual economy, with higher earners and asset holders doing well even as cost-of-living pressures weigh on the rest.
In short, she reckons the Fed is walking a policy high wire and can’t lean much in either direction.
“When you see this bifurcation, it’s really difficult for monetary policy,” Hammack told the Economic Club of New York last week.
“Bifurcated economy” is the phrase of the moment.
Consumer price inflation is still well above target and rising, and financial conditions are the loosest in years – but layoffs are rising, in part due to AI adoption.
An economy creating few jobs that is still clocking annualised gross domestic product (GDP) growth of 4% – at least with the limited data the Atlanta Fed model has to go on – is a fiendishly difficult one to navigate.
How much impact AI is having on the labour market, or how much it will have in the future, remains uncertain.
But it’s conceivable that we could see job creation grind to a halt next year, while hundreds of billions of dollars in AI investment boosts top-line GDP and the mounting electricity demand puts upward pressure on prices.
Precise measurement of all that is fraught with difficulty.
For example, AI-related spending supposedly accounted for more than two-thirds of US annualised GDP growth of 1.6% in the first half of the year, leaving the rest of the economy limping along.
And yet, tariff-dodging volatility may well have distorted that picture, while AI-related stock rallies probably contributed to resilient consumption among the richer cohorts.
Morgan Stanley’s economists reckon most of the business spending this year has been in ‘intermediate’ goods such as chips, which are not included in GDP.
The bank also added that when you exclude imports of computers, servers and chips, the AI spend accounted for just 0.3 points of that first-half expansion.
More intangible?
So much for the math.
Unfortunately for policymakers, this sort of statistical fog is likely to plague economic forecasting for some time to come.
In a report last week, economists at the Institute of International Finance (IIF)in Washington zoomed in on this AI-related “bifurcation”.
It showed capital expenditure in information processing equipment and software did grow at an annualised rate of 26% in the first six months while the rest of the economy barely moved.
But it pointed out that the categories identified as “AI-related capital expenditure” are very blurry.
For example, they include things like medical equipment and office supplies but exclude data centre building, research and grid investment.
“The investment and trade data together highlight a deeper structural shift: the US economy is becoming more intangible,” the IIF said, adding that national accounts are poorly equipped to measure the value generated by software, data management and intellectual property.
At the same time, training AI language models and curating algorithms are only partly recorded as investment.
And their effects on productivity, speed and innovation remain difficult to capture.
“These gaps mean that official statistics simultaneously overstate AI’s immediate GDP contribution and understate its broader economic impact,” the IIF concluded.
And this “mismeasurement” complicates policy calibration.
It risks understating productivity, overestimating economic slack and increasing complacency regarding inflationary pressures.
Putting the number-crunching to one side, the ultimate question remains how quickly AI will diffuse through the wider economy.
“If AI adoption remains concentrated among a handful of hyperscalers and specialised firms, returns will likely plateau, leaving overall growth vulnerable once the current investment cycle peaks,” the IIF report said.
As the Fed’s Hammack said last Thursday, the AI boom could mirror what happened with the Internet build-out more than 25 years ago, leading to a structural economic change that’s not well suited to monetary policy changes.
For investors, the big bet on an AI future may still seem to be the only game in town – unless markets have priced the bulk of it into leading stocks already.
And that evokes memories of the peak of the dotcom bubble in 2000.
The real economic impact may not be clear for years. — Reuters
Mike Dolan is a columnist for Reuters. The views expressed here are the writer’s own.
