London Macro Policy Forum 2026 - Fireside Chat with John Williams

Yes.

So we have, what you said, two hours for this one question. Okay. So first of all, as you mentioned, this was a unanimous decision of the Federal Open Market Committee. You know, we had a very robust discussion of all the things that are happening, both in terms of our maximum employment and our price stability goals. And I think, from my perspective — and I should state right now that everything I say reflects my own views and not necessarily anyone else in the Federal Reserve — but to me it's really about the evolution of the data, and I'll get to your question at the end. We've seen a remarkable resilience in the US economy despite some significant shocks: whether last year with geopolitical trade issues, this year with conflict in the East and the broader rise in energy prices due to constraints on refining capacity. So we've seen some very significant global shocks, but at the same time our economy continues to grow around 2%. We've seen resilience around the world, quite honestly, in growth in many, many countries around the world. And so from the dual mandate point of view, the downside risks to achieving maximum employment definitely seem to have receded. And, you know, the labour market continues to actually, I think, improve on the margin. But on the inflation front, that's where the big challenge is. That's the job — we still have a lot of work to do. Inflation's been above target for over five years.

And, you know, again it's kind of assessing where are the risks to achieving our goals. And with the conflict in the Middle East clearly not coming to an end right now, and we're seeing higher energy prices, and also importantly — I know we're going to come back to this — some pretty strong demand coming from AI at large, which is all the aspects of AI investment that's going on. And so to my mind it was really about assessing the outlook for the economy: what's the likely path of employment and inflation, what are the risks to achieving our goals, and thinking about how do we appropriately calibrate or set monetary policy to balance those risks. And as Chairman Walsh said in Jackson Hole, and the committee also said, we really want to see not only inflation get back to 2% — which is absolutely essential to achieve that — but also we want to see that happen, in his words, in a timely manner. And in the decision of the committee in the statement, I think it clearly said, with this increase in interest rates, we see that this is going to help support getting inflation back to 2% in a timely manner. So I think that, again, I don't think it's that the data have suddenly shifted on us or things have changed dramatically. It's more of an assessment of the evolution of the data and what it's telling us about the inflation outlook, especially, and the risk to achieving our goal in a timely manner.

So when I think about the SCP — so I fill out the economic projections, as do most of my colleagues — first of all, it's a very uncertain environment. The geopolitical developments, the conflicts in the world, other factors just make it very uncertain. And I think you look at the central — the median of the SCP — people see that it's likely that another rate hike may be appropriate by the end of the year. That seems to me a reasonable kind of way of thinking about it, but we're going to have to see. We have to collect the data and do what we did between July and September: we took all that data, assessed what's it telling us about achieving our goals and the risks to that, and how do we best adjust policy as we learn more along the way.

In terms of our financial conditions, accommodative or restrictive and things. I mean, I think this is a hugely challenging question and you can look at it different ways. And as Chairman Walsh has pointed out, if you look at equity markets, credit markets, these are very strong markets, positive markets. I mean, there's always a question of why are financial conditions changing, and you have to not just look at what the numbers are, but kind of why they're strong in the US. I think the economy again has proven its resilience. We're in a pretty good growth path. I think with AI investment I expect productivity growth to actually pick up over future years. And so it's a pretty positive outlook both for the economy, but especially for AI-related companies. And so the fact that the equity markets, credit markets are viewing this economy in a positive way and pricing accordingly is not that much of a surprise. But again, let me get to why it's a little more challenging to think about how to read this. I mean, it could be that one of the reasons the equity markets are very strong is that investors think that the AI-related investments are going to have a very high return, but maybe that isn't associated with very high profits and high markups through market power of some of these players as we've seen in the past. That may be a positive for those companies, and those investments may not be as big a positive for the overall economy. So you always have to look at these kind of credit, financial conditions and think through what are the drivers.

So to my mind, it's very hard to say how to assess financial and monetary conditions. As you said, from my perspective, the way I look at it is: is monetary policy where it needs to be in order to achieve our maximum employment and price stability goals? Is inflation likely to be on the right track to get us to 2% on a sustained basis and in a timely manner? And is it doing so in a way that reflects supporting our maximum employment goal? So it's really not about looking at financial or monetary conditions alone, but looking at it in the broader, holistic, totality of the data: is monetary policy positioned to get us to where we need to be, or does it need to be adjusted somewhat in order to achieve that? So to me, that's what we have to collect the data on — all the data, including financial conditions and including economic data — assess that and come to a judgment about really where are we likely to be going and how do we make sure that we get inflation back to 2%.

So that seems like a trick question, because you're trying to give me forward guidance on the future of the dot plot. So Charlie, very clever, reminds me of some exams from the past. So, no — first of all, we have a task force, as everybody knows, Chairman Walsh has established this task force. One of them is looking at FOMC communications. Importantly, I haven't seen that work. I'm looking forward to seeing that, hopefully later this year. But then it's really the FOMC, getting in the room, analyzing, thinking through these issues and thinking about how do we most effectively communicate monetary policy. I prefer not to think of it as 'are you for or against the dots', but more — I think we're all for more effective, better communication that supports the effectiveness of monetary policy. That's what we're about. It's not about, you know, do you like this or not like this? It's how do we more effectively communicate monetary policy so that it has the greatest effect on the economy and in achieving our goals.

So in terms of forward guidance generally, I mean, I agree that the time for explicit, very directive forward guidance was over. The committee very much agrees with that, and in the last few meetings we got away from FOMC statements and other communication saying where we think policy is going to be going in the near term. So I think that's because of the uncertainty and everything. I do think that it's helpful for the general public and market participants to understand our strategy and a long-run goal. So we have that anchor of our long-run goals and strategy: absolutely committed to the 2% inflation goal in the longer run and achieving both of our dual mandate goals. I do think that we have other communication tools that are very effective — they're not forward guidance, but like the minutes and other things. So I think you have to think about this holistically in terms of when we look at all the communication tools.

One historical thing I will bring up, and this has been studied quite a bit, is that forward guidance of the Fed really started under Alan Greenspan. So the forward guidance we often talk about is very explicit — you know, 'we're not going to raise rates' kind of forward guidance. But Chairman Greenspan had experienced in 1994 a significant bond market reaction to the Fed's tightening cycle back then, and perhaps an overreaction, or at least a lot of volatility in the bond market in '94. And so Chairman Greenspan, over the years, his thinking about this evolved quite a bit, and by 1999 and then 2000 and the early 2000s the Fed started putting its toe into forward guidance — very near-term forward guidance. And then in 2003 and '04 that became very, very explicit with the 'considerable period', 'measured pace' kind of language like that. So, not the post-crisis kind, but something that says we're giving statements about what we're doing, but we also want you to understand how we're thinking about the near future. So forward guidance, again, has many different flavors. I think that the guidance that we used in 2011 and after, the guidance we used during the pandemic, those really are, in a way, the tools you use with the zero lower bound. But at the same time, I think one thing we've learned from going back to Chairman Greenspan and since then is that there are times when market participants are really struggling to understand where things are going, and there can be a much gentler version of that — but we'll see. I think the reason I bring this up is there's lots of different flavors of this. I think the most important thing is that people are clear about what we're trying to achieve, how we're seeing the economy, and then how we're thinking about our policy decisions. And like you said, that does not necessarily involve being very explicit about where we think the next meeting's decisions will be.

So the honest answer is: we don't know, and I don't think any of us know. I think you're absolutely right. If we were having this conversation in the past, we would talk about low r-star and what does it mean. So I guess a couple of things. So, you know, if you think about the economics of what a neutral real interest rate or a natural rate of interest is, it's really about what's the price of either savings or credit or borrowing that equates supply and demand — investment and saving — not only nationally but across the globe. And that's why some of the stories, like Ben Bernanke's famous analysis on the global savings glut, was about the savings side. Today we're really focused on the investment side, and I think that is the right framing of this. And you look at the model of r-star that I've worked on or other people have worked on, behind the scenes is this notion that those big drivers in supply or demand for savings or investment are going to drive the price of that and drive neutral rates. And I think those are the big factors that drove r-star down: lower productivity growth back then, lower — you know — demographics, people living longer, lower birth rates and other factors.

So today, okay, so let me give you my assessment. So first of all, these longer-run drivers like demographics, I think are still fully in play, if not even stronger, globally. On the investment side, I definitely see very strong investment. You see it from AI, but not just in the US but around the world. So I definitely see a shift outward in investment demand. And then on the fiscal side, of course, governments have much larger amounts of debt relative to GDP and deficits in many cases. So I think again that's kind of a driver that pushes the neutral rate up, all else equal. But what is interesting is — so in my own model, with Catherine Holston and Thomas Laubach — and my model I was just looking at this morning, thinking you might ask about r-star — that model shows the growth rate estimate in that model has gone up quite a bit since the pre-pandemic period. So it's picking up, through statistical methods, that the trend growth rate of the economy, and presumably that's productivity, not labour force growth, because we don't have much labour force growth at all in the US, that's in the model, and that's pushing up r-star. It's less so for Europe and I think other countries, but probably a positive there. But in the model the other factors are still pulling r-star down. So this is a tug of war in a way between higher growth pulling it up, which is showing up — I think the estimates are 2.4% for trend growth of the US economy, that's very high actually, that's what the model estimate is — and then other things pulling it down, I think it's demographics and some other factors. So I think at the end of the day these model estimates haven't moved up a lot. They moved up a little bit. Other model estimates have moved up more. So I think again we don't know. My view is if we really do get this productivity boom that we're hoping for, maybe something like the 90s and early 2000s, that will push up the natural rate of interest. I think it's still a bit early days. What's really interesting to me is you look at the rest of the world and you're not seeing this kind of evidence, at least from the models, of higher neutral rates. Our estimate of the Euro area neutral rate is essentially zero, and that's incredibly low.

Okay, what about the bond market? So I'll just say what I think — I've been studying this and I know others have looked at this: the bond market is not a great predictor of r-star in the future. So it is hard to infer from the five-year, five-year forwards or other longer-term rates from the bond market. You have to infer what's the real rate component, what's the inflation component, what's the term premium, and all of those factors can be moving. I think definitely real rates are a big part of the story — expectations of real rates. But I think so that part, I understand. But I don't think that, at least from my point of view, it's really important to understand what's happening in the markets and what it means for the economy. I'm just — I think other factors move long rates besides the neutral rate that matters for monetary policy, and so we have to kind of be cognizant of that.

So obviously at the Fed and other central banks around the world, the independence of our actions on monetary policy is something we take extremely seriously. And we're able to make our own decisions for the reasons that we think is best to achieve our mandate goals, and that continues today. So that's crucial. But you're asking a more subtle point. If, for the US, the bond market is not functioning well, if it's dislocated, that could be extremely disruptive to the economy, that could be extremely disruptive to the global economy, since the US Treasury market is in a way the heart of the global financial system. So of course, I think my answer to that is it's very important to separate out market functioning and dislocation from just volatility in markets. So prices in financial markets can be volatile because the world is changing and market participants are uncertain about how that plays out. So my view is volatility is not the issue, and we've seen a lot of volatility in markets, but the markets have been functioning extremely well, looking at various measures of market liquidity and functioning. So I think it's important to say: let the markets do what they need to do. And then when it's really because the market's not functioning, which will then spill over into our maximum employment and price stability goals, or financial stability, then you're thinking about it. And your example from the UK is very good: is there a way to do this that is very targeted at a specific market dysfunction, and you're only — in a way — addressing an acute thing.

I mean, a lot of criticism we definitely heard in the Fed is that our interventions in 2020 and 2021 — balance sheet, QE interventions — it was harder for people to say, is this market functioning or is this monetary policy? And the honest answer, since I was there for this, was both. And we did try our best to communicate: well, markets are now functioning, we're pulling back from that, and now we're just doing QE. But that was — you know — a global pandemic, a severe crisis for a while in markets. So it's hard to communicate perfectly about exactly the way academics would like us to do — exactly, on what day did this go from being market functioning to QE. But that was that situation. The UK example, some other examples, I think can be used to say, well, if it is a very discreet kind of thing, you can do that. I mean, I go back to some of the other emergency things we did in the pandemic, where we did — in the corporate bond market facility, markets at the time in spring of 2020 were not functioning well. We introduced in the Fed a corporate bond facility, which was very novel. When those markets recovered, which was very quick and very effective, then we stopped the purchases and eventually sold all those assets. So that's a nice little.

For example, it's not the treasury market, but it is an example where, in a case of market dysfunction, you can move in and then move out cleanly. And nobody really is bringing that up as a, you know, is that QE, but it is a challenge because, as you know, the situation you're dealing with the next time is going to be very different than last time.

So on the fiscal, you know, my first answer, which you hinted at already, was I'm not going to comment on, you know, the decisions or comments of other elected officials. But, you know, the real point is that, you know, we take the world as it is, you know, whether it's fiscal policy or actions of other governments, and then assess what does it mean for the US economy, what does it mean for maximum employment, price stability, the risks of that. So to me, it was just part of that environment that we're in, and then we just focus on making the decisions that we make. You know, clearly, markets are responding to changes in information and things, so we're getting the information we need from the markets.

The second question was on AI, no, productivity. Yeah, productivity. How much is AI? Yeah, so I think, you know, so US productivity growth since the fourth quarter of 2019 has been about a little above 2%. Okay, that happens to be the long-run average since World War II. But it's better than it was before the pandemic, which was running, you know, 1.5% or something like that. So why is that? I think one thing that happened in the US economy was a remarkable shifting of people between jobs and a better matching of, you know, people with their skills and the jobs and things. So that was just enormous scale. I mean, I felt it at the New York Fed. I mean, our turnover was the highest we've ever seen. And, you know, we're hiring, you know, we're hiring a lot of people. So there's just a lot of, I think, improved matching.

The second thing is the whole topic of, I can't remember, years ago at Jackson Hole, there was a topic about the dynamism in the US economy seemed to have been disappearing in terms of new business formation and things. That's completely reversed. Part of that's pandemic related, but it's still going on and ever stronger. I've talked to people in that space. There's a lot of startups. So I'm getting to your AI question. A lot of one-person, two-person startups that they think they basically have this agentic staff that's going to do a lot of their work. So I think that is part of this creation of new businesses and jobs and the amazing dynamism of the US economy. So I think that's another part of it. Part of this is also influencers. So I'm not sure how that gets into national income accounts, and I'm way too old to actually try to understand what that means, but actually that is another part of the changing labor force. A lot of these people who are starting doing startups are actually influencers. So I think it's a set of issues that has boosted productivity.

If you ask me, is that a level effect or a growth rate effect? It's probably a level effect, but it's stretched out over a number of years, and that's good. I mean, I love positive supply shocks, so that's good. Is it AI? There's some research that says maybe a little bit of the recent productivity could be AI because businesses are starting to use it more, and businesses who use AI more, I think, show more productivity, at least in terms of revenue per person or something. So maybe a little bit, but I think most of that is before AI. My view on AI is it could easily be something like the '96 to 2005 productivity boom where productivity went from that 1.5%, 1.25% to 3% for a decade. 3% productivity growth for a decade is like unbelievably a big positive shock to the economy. It far outweighs some of the uncertainty issues that we've talked about, some of these other issues. I'm not dismissing some of these, but if you think about what does it mean to get productivity growth for a decade that's 2.5%, 3.5%, 3%, 3.5%. I mean, this is just enormous. And I know we'll talk about, I'm sure people ask about it. So my view is we could be getting that in the future. I think it's still early days. I mean, you look at the amount of investment in AI, the amount of work we're doing using AI. I think there's going to be a boom in productivity. How big and how long-lasting, we'll see.

The third is on real wages, wages and prices. I mean, I think one of the real challenges for understanding these issues, and I think it's a really important set of questions, is that historically, maybe, you know, we even talked about this when I was a grad student, historically the labor share of income in the United States was relatively constant. That wasn't true of some other countries, but it was true of the United States. It's not true at all anymore. Labor share of income, so that's, you know, basically wage times hours divided by, you know, GDP kind of thing. Labor share of income has fallen dramatically. So part of the story of prices and wages haven't fully adjusted. I mean, real wages have grown over the last five, six years. They've grown in the US, but the fact is they haven't grown as much as productivity because now labor share is at an all-time low. So I think one of the big policy questions for all of us is, you know, what is driving that change in labor share? Is it structural? Is it related to the productivity? Daron Acemoglu and co-authors have emphasized that, you know, changes in technology are basically making this economy more, making capital share higher. There's other different views. So this is something I think is very important to understand because historically you kind of think of these things should, you know, I'll be very nerdy, error-correct in a way together, but they're clearly not in the US, and I think we're seeing that in other places. And maybe AI, this is one of my maybe concerns around AI, is that AI, if it ends up being dominated by a few firms with very high markups, we'll see the higher productivity growth, we'll see some of those benefits, but it won't be shared in...