"New Fed News Agency": 1996 or 1999? Waller's First Test is "How to View AI"

The primary challenge facing Walsh as he takes office as Federal Reserve Chair is not whether to raise or lower interest rates, but a more fundamental judgment: What kind of boom is the current AI boom? This judgment will determine the Federal Reserve's policy direction and define Walsh's historical standing.

On June 19, journalist Nick Timiraos, known as the “new Federal Reserve communications agency,” reported that the economic community holds two diametrically opposed interpretations regarding the AI construction boom:

First, the productivity dividend is about to arrive, supply will catch up with demand, and the Federal Reserve can stand pat and wait for inflation to ease naturally; second, the benefits of productivity enhancement are still distant, while demand shocks have already occurred. If the Federal Reserve waits for data confirmation, it will miss the best intervention window and be forced to raise rates by a larger margin later on.

The Federal Reserve kept rates unchanged this week, but in the latest dot plot, nearly half of officials expect further hikes this year, while others hold the opposite view. The deep internal division highlights the high level of uncertainty surrounding this core issue.

Walsh’s own inclination was faintly visible at the press conference. He repeatedly emphasized “strong productivity-driven growth is not something we fear, but something we embrace,” echoing the thinking of Greenspan in 1996.

However, the macro environment he faces—including tariff pressure, expanding fiscal deficits, and dwindling globalization dividends—is a far cry from the smooth sailing of Greenspan’s era. How to make the right judgment between these two historical playbooks will be the first real test for Walsh as Federal Reserve Chair.

Two 1990s: The Dual Legacy of Greenspan

According to Timiraos, Walsh has repeatedly referenced the 1990s as a historical parallel over the past year, but the decade itself contains two contrasting stories.

In 1996, facing rapid economic expansion, Greenspan chose to hold steady. He judged that rapid growth would not ignite inflation, and history proved him right. The economic expansion lasted for years, earning him the title of “maestro.”

In 1999, Greenspan revised his view. As the stock market surged and the labor market tightened, he began a series of rate hikes, ultimately culminating in the bursting of the internet bubble. And it was in that year that the Federal Reserve established its forward guidance mechanism for “signaling rate hikes in advance”—a practice that persists today, and one that Walsh has explicitly indicated he wants to eliminate.

The Trump administration openly endorsed the 1996 version of the Federal Reserve, and Walsh also declared before taking office that he aimed to build a central bank “confident enough to do less.” However, the current economic situation may be handing him a different script.

Walsh’s Logic: Belief in Narrative Over Waiting for Data

Before taking office, Walsh publicly expressed on Fox Business that he worried the Federal Reserve was about to make its “sixth or seventh major mistake”—prematurely tightening monetary policy during a productivity boom that should be allowed to run its course.

Timiraos notes that his core argument is: The productivity gains brought by AI will not be immediately reflected in official statistics, and may take years to appear. If the Federal Reserve insists on waiting for data confirmation, it will misjudge a benign boom as economic overheating and raise policy rates—thereby killing off the growth momentum that could otherwise subdue inflation.

The essence of this logic is to advocate using forward-looking narratives rather than lagging data as a basis for decisions. Walsh continued this tack at the press conference: when asked whether AI is currently boosting demand or expanding supply, he merely stated “demand is easier to measure than supply,” deliberately sidestepping a clear position and holding firmly to the principle of not revealing his next move in advance.

Timiraos believes that even if Walsh’s judgment is ultimately correct, the 1990s analogy is still incomplete.

When Greenspan made his famous 1996 bet, he enjoyed multiple tailwinds: imported cheap goods and labor from overseas kept inflation down, and the federal fiscal deficit was narrowing. These structural factors gave the Federal Reserve additional safety margins to “wait and see.”

Walsh faces a very different environment: tariff policy is driving up import costs, the fiscal deficit is expanding rather than shrinking, and the dividends of globalization have dissipated. This means even if the AI productivity windfall finally arrives on schedule, the inflationary pressure Walsh endures while waiting will be much greater than in Greenspan’s era.

Dissenting Voice: Chicago Fed’s “Expectation Overspending” Model

Timiraos points out that the most systematic challenge to Walsh’s logic comes from Chicago Fed President Austan Goolsbee.

According to The Wall Street Journal, last month at a Stanford conference, Goolsbee offered a key distinction: Whether a productivity boom allows the central bank to stand pat depends on whether the boom arrives unexpectedly. A boom that everyone can foresee will have the opposite effect—people will overspend future wealth in advance, significantly increasing expenditures before the productivity dividend materializes, which in turn overheats the economy.

“Ultimately, you have to raise rates by more than you would have needed had you moved sooner,” Goolsbee said.

He believes the current AI boom is exactly this kind of “obvious-to-everyone” scenario. Surveys of economists, tech workers, and the general public all indicate that the market broadly expects AI to bring about a one percentage-point annual productivity gain, with most benefits still ahead. According to his model, this expectation alone justifies a rate hike, not a cut.

Goolsbee also cited real-world “overheating signals”: AI data center construction is driving up land, electricity, and chip prices, while also raising the costs of electricians and equipment, squeezing resources from other industries. Apple’s announcement this week to increase prices due to rising costs is, in his view, evidence that this mechanism is operating.

Notably, Goolsbee’s framework is not without challengers. Federal Reserve Governor Christopher Waller pointed out at the same Stanford conference that the “expectation overspending” mechanism only works if people can borrow to spend in advance. In reality, many households’ expenditures are tightly constrained by current income, making it difficult to cash in on future wealth.

“If they can’t overspend that part in advance, this entire mechanism is cut off,” Waller said.

This rebuttal provides theoretical support for Walsh’s “do nothing” stance: If borrowing constraints are widespread enough, the front-loading effect of demand will be sharply diminished, making it more likely that a productivity boom would spur supply expansion moderately, rather than fuel inflation.

The Ultimate Paradox: Ending Forward Guidance or Being Forced to Use It

Additionally, Timiraos argues that Walsh faces a deeper paradox as he leads the Federal Reserve, a paradox that stems precisely from the thing he most wishes to change.

He has explicitly said he wants to create a Federal Reserve that “doesn’t reveal its hand in advance," rolling back forward guidance to keep the market guessing. Yet, the Federal Reserve’s current forward guidance mechanism was established in 1999—when Greenspan, to avoid catching markets off guard, began signaling rate hikes in advance.

If the economy proceeds as optimistically as the Trump administration describes, Walsh may never need to signal in advance. But if it proceeds according to the other script, he will face a dilemma:

Either continue the forward guidance practice he wishes to abolish by telling the market about rate hike plans in advance, or remain silent and let the market guess the scale and pace of hikes, risking violent financial market swings as a result.

The solution to this paradox still comes down to the same question: Is it 1996 or 1999 now?