MNI POLICY: AI Boom Complicates Fed's Path To Lower Rates

Federal Reserve officials view a potential productivity surge driven by artificial intelligence as a reason for patience on interest-rate cuts rather than a green light for easier policy, drawing lessons from the 1990s IT boom that boosted the economy's long-run potential but also complicated real-time policy decisions. 

This could create an intellectual rift at the central bank if the new nominee for chair, Kevin Warsh, keeps pushing the idea that higher productivity can drive a burst of non-inflationary U.S. growth.

Policymakers this week cautioned against simplistic analogies to the 1990s, an era that featured not just productivity gains but also significant globalization with China's economic opening, creating favorable conditions for strong labor markets, lower unemployment and subdued inflation. 

The '90s debate centered on whether to delay rate hikes because of productivity growth -- not whether to cut rates. Among the key differences between then and now: interest rates were higher, inflation was below 2% rather than above it, and the labor market was much tighter. 

"The argument was maybe we don’t have to raise rates yet because the productivity growth rate is higher. It wasn’t ‘should we cut rates because productivity growth is high?’" Chicago Fed President Austan Goolsbee told reporters this week. "It's a cousin to the '90s but it really isn’t the same situation." 

Warsh wrote in an opinion piece in December that "the Fed should discard its forecast of stagflation in the next couple of years," adding that "AI will be a significant disinflationary force, increasing productivity and bolstering American competitiveness."

HIGHER R-STAR

The trouble for many current FOMC members is that a sustained AI productivity boost would also likely raise the neutral interest rate, or r-star, as demand for investment increases, which eventually increases supply, and a new equilibrium with a higher neutral rate is established. Fed Vice Chair Philip Jefferson said this month that that mismatch risks creating periods when demand outstrips the economy’s immediate productive capacity, keeping price pressures elevated. (See MNI INTERVIEW: AI Boom Doesn't Justify Lower Rates - Haskel)

The Laubach-Williams model, co-developed by New York Fed President John Williams in 2003 as he sought to understand what the internet boom meant for interest rates, estimated that r-star rose a full percentage point to above 3.5% in 2000 as growth topped 4%. 

The model estimate has been stuck in a much lower range over the past two decades due to aging, low birth rates and longer life expectancy, but policymakers agree AI will likely contribute to somewhat higher neutral rates. 

Alan Greenspan ultimately did raise rates starting in June 1999, taking the Fed's benchmark rate from 5% to 6.5% over the course of a year. 

OVERHEATING NOW

Goolsbee cited HVAC worker shortages in Cedar Rapids, Iowa due to data-center needs and rising farmland prices from resource competition as evidence that "in the here and now, we're facing a tighter situation," with AI demand straining scarce resources.

In the early 2000s, productivity was high, inflation was low, and the Fed kept rates low -- fueling a rapid rise in housing prices and a massive credit bubble, another cautionary tale. 

Officials have already signaled they want clearer evidence of sustainably cooling inflation before resuming rate cuts in the face of stickier inflation. The AI boom may raise questions over whether cuts remain appropriate. (See: MNI POLICY: Fed Embraces Pause As Downside Labor Risks Abate)