MNI INTERVIEW: AI Boom Doesn't Justify Lower Rates - Haskel

An AI-driven productivity boom will not provide support for looser monetary policy, because central banks models assume forward-looking consumers will increase consumption in anticipation of greater wealth, former Bank of England Monetary Policy Committee member and professor at Imperial College Jonathan Haskel, told MNI.

"This question, which is now big in America, with Kevin Warsh as the new boss, is if there is an AI productivity miracle, does that mean that you can loosen monetary policy? Because you've got lots and lots more potential output, as it were, so to the extent that inflation is the gap between output and potential ... surely you can have looser monetary policy?" Haskel said.

"In the Bank of England models at least, I bet you in the Fed model as well, the ultimate answer is no. The reason the answer is no is that in the Bank of England's models, consumers, when they see an increase in productivity due to AI or due to anything else… they foresee an increase in the wealth of the economy, and therefore they start consuming more. So, in fact, what happens is, although potential output goes up, actual output goes up as well.”

He noted that the relationship between AI-driven productivity and rates depends on the supply of capital. While something like hi-tech chips may be in short supply in the near term, capital supply is more elastic, he said, so it is doubtful there will be any medium-term rise in necessary returns as demand for capital rises.

INFLATION

Big weights in inflation baskets for items less exposed to AI, such as food and rents, are likely to reduce its effect on headline inflation, Haskel said. 

"My expectation would be in terms of its immediate effects, no, it'd be small," he said, adding that it would also be hard to measure.

"I am quite bullish about AI spreading out throughout the economy, firms becoming more efficient, our banking services getting better, discovering better drugs," he said.

"Is it actually going to be the case that AI starts really improving people's lives, but we just don't measure it very well? That strikes me as being a very real possibility.”

Rather than capturing, say, enhanced flows of medical services, official statistics may just capture changes in prices of individual pills, he said.

FIRMS OPTIMISTIC

The BOE’s February Monetary Policy Report saw relatively subdued overall potential productivity growth, which rises from 0.5% in 2024-25 to 0.8% by 2028, despite upside risks from AI. However, firms surveyed by the BOE expected AI to boost their productivity by 0.6% per year over the next three years.

"I think putting in some upside for AI is the right thing to do," Haskel said, but expecting 0.6 "is really a lot." (See MNI INTERVIEW: Budget Shouldn't Prompt BOE Cuts-OBR's Miles)

RAPID SPREAD OF AI

Haskel has changed his mind on the probable speed of AI adoption since leaving the Bank in August 2024. He now considers that AI is an "innovation in the method of innovation” as well as a new general-purpose technology, and notes that IT infrastructure needed to it is already widespread.

"First, it's going to have a bigger effect than previous industrial revolutions. And second …  we've got the rails upon which it will run, [which] tells you it will come a bit quicker.”

DIVERGENCE BETWEEN ECONOMIES

Recent work by Haskel estimated that AI will boost annual productivity growth by one percentage point in the U.S., and about 0.3 percentage point in Europe.

The U.S. has a lot of software development, whereas "in Europe, that production of software is just much smaller. So, in Europe, we don't get a very big production effect, and so all that's left for us is the use effect," Haskel said.

Britain poses a puzzle, he said, as its software sector is between the size of America’s and Europe’s.

"We've got, actually, quite a big software producing sector, but it does seem ... we're not getting much use out of it," Haskel said. (See MNI INTERVIEW: Employment Law Hit To UK Productivity - Haskel)