Federal Reserve Chair Kevin Warsh used his Jackson Hole debut to make a clear institutional point: in normal times, forward guidance should step back, and monetary policy should return to discipline, data dependence, and judgment.
Warsh delivered the speech at the Jackson Hole global central banking conference on Aug. 28, his first appearance at the gathering since becoming Fed chair. He said the central bank should be careful about pre-signaling future policy moves and should avoid turning communication into a substitute for decision-making.
He also spent a large part of the address on artificial intelligence, describing it as a new variable for the economy and possibly even a new factor of production. Its effects on productivity, labor markets, capital intensity, and the distribution of returns remain uncertain, he said, and the Fed is watching closely.
A first Jackson Hole speech built around restraint
Warsh opened by noting that the speech came on his 100th day as Fed chair. He thanked Kansas City Fed President Jeff Schmid and his colleagues for hosting the event.
He framed the morning with a personal contrast drawn from earlier Jackson Hole hikes with former Fed officials. One, with former Vice Chair Don Kohn, felt like a punishing “death march.” Another, with former Chair Ben Bernanke, was a far easier walk through the Rockefeller Preserve. Before setting out, Warsh joked, people should ask themselves: “Is today a Kohn day or a Bernanke day?”
That setup led into the conference theme of innovation. In Warsh’s telling, innovation in how the Fed implements policy can help it pursue both full employment and price stability. But he quickly set a boundary around what he was offering. Call it an outline, he said, or a roadmap, but “don’t call it forward guidance.”
He then laid out four parts of the speech: long-term questions including AI, the practice of forward guidance and the interaction between central banks and markets, the principles that should guide policy implementation, and his reading of current economic conditions.
AI enters the policy conversation, but not today’s decision
Warsh said the policy backdrop has changed sharply from the years before and after the 2008 crisis, when economists and policymakers were still talking about “secular stagnation” and a “global savings glut.” At the time, one widely held view was that large pools of capital would remain idle because there were not enough compelling investment opportunities.
That picture looks very different now, he said. AI is the most visible example. Warsh described the technology’s pace as faster than even some of its advocates had predicted a few years ago. The economy’s potential for materially stronger growth is rising, and larger pools of capital are pouring into AI-related infrastructure.
He said something like a “super Moore’s Law” may be taking shape, with changing scaling laws altering both how innovation happens and how fast it happens. Capital and labor, in his description, are combining to produce large language models at the center of the AI buildout, while users buy tokens to gain access to those models.
According to reports cited in the speech, annualized token sales at just two leading AI labs have already exceeded $100 billion, up more than 500% from a year earlier. The Fed, he said, is following the trend closely.
Warsh posed a series of questions rather than firm conclusions. Will AI drive a meaningful and sustained rise in economy-wide productivity? If so, when? Will token usage complement labor or compete with it? Will next-generation models demand even more capital, or will the models themselves help design lower-capital solutions?
He also said the eventual market structure remains unclear. It is not yet known where returns to capital will settle, or how long that process will take. In the early phase, how much surplus value will go to owners of scarce assets such as AI labs, chipmakers, energy producers, and cloud providers? Over time, how much of that value will flow to businesses and consumers? What will that mean for workers and for the Fed’s employment mandate?
The equilibrium price of tokens is also unknown, he said. Different categories of tokens may emerge, with users willing to pay more for access to frontier models, while older-model tokens may eventually fall toward marginal cost.
Even so, Warsh drew a line between long-term study and immediate policy. Recommendations on those questions will come later, he said, and they will not affect current policy decisions. Still, doing the intellectual work now should leave the Fed better prepared for future turning points.
Why Warsh wants less forward guidance
Warsh said he has already begun changing the form and function of what is commonly described as forward guidance from the Fed chair. He has long been uncomfortable, he said, with announcing future policy decisions too early.
Transparent communication about future decisions is not a virtue by itself, in his view. Communication has to serve the Fed’s central obligation: getting monetary policy right.
He said forward guidance was adopted by him and his colleagues during the global financial crisis, when it played an important role and was rolled out prominently. But like other crisis-era inheritances, he believes it has outlived the period in which it should play a leading role. In normal times, its use should be limited and kept within clear boundaries.
Otherwise, he warned, it can create confusion in the name of clarity. Too much disclosure about policy deliberations, and too much commitment about future decisions, can mislead markets, businesses, and households. It can also narrow policymakers’ own freedom to make the right call when the moment for an actual decision arrives.
Warsh said the Fed also needs to handle its relationship with markets correctly. The central bank needs clear market signals that are as unfiltered as possible, including internal market measures, price levels and changes across markets and sectors, Treasury prices and volumes, the dollar’s exchange value, the cost and availability of credit, and broad commodity prices. Those indicators should help the Fed judge short-term activity and the inflation outlook across the cycle.
At the same time, he said, the Fed should be humble but not naive. The central bank matters enormously for both the economy and markets, and its tools are powerful. It sets the path of short-term interest rates, and market participants will always try to anticipate its next move. But the Fed should not encourage a setup in which traders rely mainly on the central bank to determine their next trade.
Warsh pointed to what the economics literature has called the “hall-of-mirrors problem.” If markets lean too heavily on Fed guidance while the Fed leans on market prices, both sides become more likely to miss new developments, get caught flat-footed at turning points, and make policy mistakes.
He argued that the biggest cost may not fall on financial market participants at all. If the Fed misreads inflation and the broader economy, the greatest damage is likely to land on working Americans without financial assets, the people who have to live with high inflation or jobs that suddenly become less secure.
He then addressed whether the Fed should at least commit to a clear reaction function if explicit forward guidance is no longer appropriate in normal conditions. Warsh said he would like the central bank’s understanding of the economy to be precise enough to produce a mechanical, well-tested answer, perhaps through a simple rule such as the Taylor rule. But the Fed does not have that level of knowledge, at least not yet, and the factors that matter most to sound implementation change over time.
Using forecasts to display a reaction function works better in theory than in practice, he said, and better in the lab than in the real world. He added that the 2021 version of forward guidance likely slowed the Fed’s policy response to high inflation. During his chairmanship, he said, he and his colleagues will work to build more reliable models and more robust rules to guide decisions.
Still, he cautioned that accurate economic forecasting remains more aspiration than certainty. Geopolitics, global supply chains, and technology are changing too quickly for false confidence. For the same reason, the Fed should hear out different views on questions that might affect monetary policy rather than shutting out competing interpretations of the economy.
Seven principles for policy implementation
Warsh then set out the principles he believes should guide the implementation of monetary policy.
First, policymakers must separate yesterday’s news from what is actually happening now. Policy should not be built on stale or inaccurate data, and it should not rest on isolated data points. Trends matter most. Because the Fed is a decision-making institution operating under uncertainty, the data it uses must be as relevant, timely, accurate, and actionable as possible.
Second, the Fed acts to keep aggregate demand broadly aligned with aggregate supply. But policymakers can directly observe only economic activity; changes on the supply side can only be inferred. That makes any assessment of the present and future balance between supply and demand inherently imprecise.
Third, there should be no ambiguity about the inflation objective. The Fed’s 2% price-stability target, measured by the Personal Consumption Expenditures price index, is fixed and firm. Price stability does not happen on its own, he said, and inflation does not naturally drift back to target without policy doing its job.
Fourth, the Fed also has a responsibility to achieve maximum employment. Over the medium term, he said, the dual mandate is not a choice of one at the expense of the other. He does not see the two goals as inherently conflicting because high inflation itself does serious damage to economic prosperity.
Fifth, short-term interest rates are the main tool for pursuing the dual mandate. Unconventional policies designed to stimulate activity may be appropriate in genuine crises, but outside those periods they should be used sparingly, or not at all if that can be avoided.
Sixth, money matters. Warsh acknowledged that saying so may not be fashionable, but he argued that money remains closely tied to monetary policy. The Fed should pay attention both to money created by the central bank and to money generated through the banking and financial systems. Financial innovation and other factors may have changed the transmission from the monetary base and velocity into the wider economy, but that is not enough reason to ignore how money ultimately affects financial conditions and prices.
Finally, a quieter Fed with more purposeful communication is better equipped to achieve its goals. Accountability, he said, should be judged by whether the institution does its job. Borrowing a line from U.S. Air Force General Chuck Yeager, he added: “At the moment of truth, there are either reasons or results.”
Warsh’s reading of the economy: jobs steady, inflation still high
Turning to the current backdrop, Warsh said the Federal Open Market Committee’s July minutes showed a shared view that the labor market has remained steady, output has been solid, and inflation is still too high.
He said he and most of his colleagues believe it is wiser to wait for new information between meetings before deciding whether rates need to change, especially given possible developments in supply chains, investment flows, and geopolitics. They also expressed a willingness to act if needed. His own impression, he said, is that the economy overall appears to have strengthened.
One test of economic strength is resilience to shocks. On that score, Warsh said, both the real economy and Wall Street have shown unusual resilience.
Business capital spending, which he called the “seed corn” of future growth, has been rising quickly. The four-quarter change in equipment and intangible investment is about 9%, the fastest pace since 2021. More than half of this year’s growth in capital spending likely reflects AI-related construction. For S&P 500 companies, profits have risen more than 20% over the past year, and margins remain high relative to history. Stock market volatility is low, and the Fed is watching internal market behavior across sectors. Expectations for both capital spending and earnings growth are elevated.
Warsh said he will keep an eye on changes in the pace of those gains, the second derivative of growth, as well as the knock-on effects for asset prices, business confidence, household income, and consumer spending. Credit spreads in corporate bonds and leveraged loans are near the low end of historical ranges, and issuance in those markets has been strong this year.
On the banking side, he pointed to the July Senior Loan Officer Opinion Survey, in which banks said standards for commercial and industrial loans were on the easier side of their historical range. That, he said, helps explain the growth in those loans this year.
His bottom line was blunt: credit and loan markets show little sign of policy restraint. Some sectors, including housing and agriculture, are under pressure, but he said it is hard to agree with the view that broad financial conditions are clearly restrictive.
Consumer spending has also held up. Real consumer outlays have grown more than 2% over the last four quarters despite repeated shocks. This year, PDFP growth has been close to 3%. Warsh said that measure often contains more useful signal than gross domestic product, and its trend is also positive.
On employment, he said the country is doing well. The labor market is fairly stable. The unemployment rate is 4.1%, low by historical standards, and has not changed much for several years. Initial jobless claims, measured as a four-week average, are near their lowest levels in decades and remain one of the stronger real-time indicators.
He said lower worker churn partly reflects the large post-pandemic rematching between employers and employees. When labor supply is barely growing, lower monthly payroll gains are not unusual. There are still pockets of concern, including recent graduates, but overall people who want work are broadly keeping jobs or finding them.
For now, Warsh said, the labor market is consistent with full employment. The inflation side of the dual mandate worries him more.
The Fed’s preferred inflation measure, the 12-month change in PCE prices, is now 3.7%, while the six-month change is 4.1%, he said. Comparable Consumer Price Index readings are also high, and so are core PCE and core CPI measures. None of them is perfect, but they tell a similar story: inflation remains above the 2% target, so the Fed’s main focus should be prices.
Policymakers, he said, need to identify underlying trend inflation, the broad movement in prices after stripping out special factors. The task is not only to judge whether that trend is rising, falling, or stalling, but also how fast it is moving. Broad inflation gauges are well below their 2022 peaks, but he remains focused on how widely price pressure is still spread through the basket.
To assess underlying inflation, Warsh said he finds it useful to break the PCE price index into 199 components. Over the past 12 months, 54% of items in the PCE basket posted price increases above 3%. That is well below the post-pandemic high of about 77%, but still noticeably above the 32% average seen in the 20 years before the pandemic. Looking only at the last six months yields a similar picture: 49% of items posted annualized increases above 3%, again well below the post-pandemic high but still quite elevated.
He also flagged the recent rise in commodity prices and said the Fed must determine whether those moves point to upside inflation risk. Just as important is whether the inflation of the past more than five years has begun to feed into expectations.
There, his tone was more constructive. Measures of medium-term inflation expectations appear broadly stable, he said, and inflation compensation in swap markets is sending a strong and consistent message. Even after recent developments, market prices still show confidence that the Fed can deliver price stability. That, he said, reflects well on the institution and fits with its best traditions. “I can assure you,” he said, “they are right.”
But he added a warning from economic history. Market-based inflation expectations often look strong and stable right up until they become unanchored. For now they remain firmly anchored, yet they must be watched closely. Keeping expectations anchored is part of the Fed’s job.
Warsh also pointed to one signal that, in his words, no one can ignore: 65 straight months of elevated inflation. Responsibility for that, he said, rests squarely with the central bank, as it should.
He then offered the standard by which he is judging the inflation outlook: policymakers must be confident that underlying inflation is moving toward target at a clear and sufficiently fast pace. If not, “we still have work to do.”
A promise of discipline, not a preset decision
Warsh closed by saying that what he was promising was discipline, not a specific policy decision. He said he and his colleagues are determined to use time well, seize the moment, and do the job with humility and resolve.
Too much depends on the choices the Fed makes, he said. Sound monetary policy can help households and businesses prosper. When implemented effectively, it can broaden and deepen the economy’s momentum and help reinforce U.S. leadership in the world.
He ended by saying the country needs the Fed to think carefully, judge prudently, and act wisely. Returning to the institution is an immense honor, he said, and he thanked his colleagues and the audience for their encouragement, advice, and attention.