And you're finding your seats, please take a moment to silence your cell phones and electronic devices, and note that the discussion this evening will be on the record. In addition to those of you joining us here in New York, we do have...
Oh, you want this? Here you go.
You ready to go, Connor? Great. Well, welcome, folks, to the C. Peter McCullough series on international economics with Minneapolis Fed President Neel Kashkari.
Thanks for being here, Neel. I think we should just kind of maybe get right into it.
So, there was a boatload of data this morning on both sides of your mandate — full employment and stable prices. Start off for me, if you would, with your take on the PCE report this morning, which had some changes of methodology and all kinds of other stuff. But what was your basic takeaway from it?
My basic takeaway on the inflation data is that inflation is still too high. It's running — you know, there are many different measures of inflation — but it's running at around a 3% rate. It's been elevated now for more than five years. I didn't think the inflation data today really changed that story for me very much. But we also saw that consumers keep spending; the economy is resilient. The bigger surprise to me today was just some of the revisions, for example in GDP, that suggest that the economic resilience is really there. It's been remarkable the last several years in the face of tariffs and trade war, in the face of the geopolitical conflict in the Middle East. It's remarkable how resilient the U.S. economy has been.
Does it surprise you? We keep seeing these numbers where oil prices are obviously high, but it doesn't seem to dent sort of consumer spending ex oil.
It is surprising. I mean, I think people are feeling a lot of pressure from the high prices that have accumulated now over the last five-plus years, and yet people keep spending. The good news is the vast majority of Americans who want to work have jobs. The national unemployment rate is 4.1%. That is quite a low number relative to history. Most people have jobs. They're not saving a lot, but they're spending it and they're making their way through it. And so consumers keep spending. You know, by the way, some of the big CEOs of the biggest banks have said that they're seeing that their consumers — their high-income consumers are spending and doing well, and their lower-income consumers are spending and doing well. That resilience continues to surprise me in a good way.
Does it give you any information about where the Fed is relative to neutral?
Yeah, that's — it's causing me — the longer this goes on, the longer the economy is proving to be resilient in the face of all these shocks, the more I am questioning how tight is monetary policy. So over the last few years, I have slowly revised up my estimate of what we call the neutral federal funds rate. And in September, I revised my estimate up to 3.25% as a neutral rate. But even that is a — it's a curious number. That number is meant to be what is the neutral federal funds rate when all the shocks have passed the economy. It's this mythical state of the world when everything is perfectly in balance — a state of the world that we'll never actually achieve because new shocks come along. But right now it may well be that the neutral rate is temporarily elevated even relative to that. So it gets quite complicated quite quickly.
Well, that's what I like — the complicated stuff. So, you're suggesting that you have in your head a short-term neutral rate and a long-term one, correct?
There's the kind of nirvana rate, but there's another rate that's like, this is the rate I need to get there.
I'm not sure. I mean, I think I look at economic conditions. As I look at the fact that so much massive investment — we all know — is going into the data center buildout and power and water and all of the infrastructure that goes with it. That's a new demand for investment capital. The neutral rate is ultimately set by the balance between demand for investment capital and supply of savings. And where do those two balance out? And if there's a lot more demand for investment capital, all else equal, that's going to push up the neutral rate. So I think the neutral rate is likely elevated temporarily, but how long is that going to persist? How long is the AI buildout going to persist? You know, there are a lot of different factors that go into it, so it's hard to know with confidence.
The bond market seems to have an idea of where the neutral rate is, and it's a lot higher than where you are right now. I have been using a metric, but you have a better one, I think. Just looking at the two-year over fed funds, and it's 100 and change. I don't remember quite being there. I understand that two years from now you ought to demand a higher yield than tomorrow, but there's a kind of normalcy to it that we're not at right now. What signal do you take from the two-year yield and where the market is suggesting you ought to be relative to where you think you ought to be?
Well, it's an interesting question that I've dug into with my staff. Today, I think the two-year ended around 4.88. And if you look at the Summary of Economic Projections, which is the dot plot that gets a lot of attention, you can compute an implied two-year rate from the expected path of interest rates in the dot plot. You look at the median dot. Our calculation is the median dot implies a two-year of a little over four — 4.05, 4.06. So 4.88 relative to 4.06 is the largest gap that it's ever been between the two-year and the median implied two-year from the dot plot. So I take note of that. I try to understand what is the market likely saying. I don't think that we should just be blindly following the market because markets can move around for lots of different reasons. There's a lot of volatility in markets. But I also don't want to dismiss it. Now, if I go back in time, you know, the dot plot was created after the financial crisis, and in the eight or so years after the financial crisis, the FOMC through the dots kept forecasting that interest rates are going to go up. And if you look at the two-year at the time, markets didn't believe the Fed. They said, 'No, you're going to end up being dovish for longer than you're telling us.' The truth was even more dovish than markets were. So markets were not right, but they were more right than the Fed. And so I don't want to blindly dismiss what markets are signaling because markets right now are signaling that policy may have to go even tighter than we expect. I want to pay attention to it, but I don't want to blindly follow it either because there are a lot of different factors that can go into some of these market judgments.
Neel, can you just walk us through what you jotted down in the last SEP when it came to rates and the outlook for the economy?
Sure. I jotted down two hikes this year — and these are jotted down in pencil. This is as of September. I had jotted down two hikes for this year: one in September and then one later on this year.
So just to be clear, you've done one already.
That would suggest one more left this year and then another one in 2027.
Just one more. But again, that is a snapshot in time based on information we had on that Wednesday of the meeting. And then we get all this new data and all these new developments, including the GDP revisions that we've now seen, including the consumer spending data. All of this needs to go into my real-time assessment of where we go from here.
All right. So, riddle me this, because I'm absolutely haunted by this one line or one paragraph in a recent UBS report I read, which might be old news to you or everybody in this room. But what they did is they did an exercise of taking out the impact of AI on the economy. And they said for the last year, the ex-AI portion of the economy has been contracting. So when I say riddle me this, I mean, how do you set an interest rate for these two things that are going on — one that's going absolute gangbusters and one that looks like it's contracting?
And you're about to slam the part that's contracting.
I'm skeptical that the rest of the economy is contracting. If you look at corporate profits, they've generally been up very strong across many sectors of the economy, not only those directly related to the AI buildout. And again, 4.1% unemployment rate, very low layoffs nationally, very low unemployment claims nationally. Most signs are — I mean, housing is clearly under a lot of pressure and housing-adjacent sectors are clearly under a lot of pressure, but many sectors of the economy I think are doing well. But AI is clearly this real engine of growth.
I probably didn't phrase it well. I mean, that was UBS said what they said. But still, there are different tracks and growth tracks for different parts of the economy. Do you set the rate for the marginal borrower, or do you set it for — you know, do you worry too much about, say, the housing sector that's going to really take it on the chin from further rate hikes?
Well, I mean, unfortunately, you all know this — monetary policy, we always say it is a blunt instrument. It affects the entire economy. We can't target it on this sector or that sector. Ultimately it will affect the economy. Another factor that we haven't talked about yet, it's been getting a lot of attention, of course, is diesel price. Diesel fuel goes into virtually everything. You know, it affects everything. And so on the margin, I know monetary policy can work. It can help stabilize inflation expectations. It can tamp down on demand. It can help get us back solidly on a trajectory down to 2% inflation. And hopefully we can do it through modest adjustments rather than something more severe.
Neel, walk us through your thinking. Was there a point in time you were on team 'look through supply shocks' and you've sort of left that team, or were you always skeptical of this idea of looking through supply shocks?
No, I wasn't. I mean, go back a little bit further than that. I was solidly on team transitory when inflation first took off. And I looked at — we had supply chain disruptions from COVID, that was the first big supply shock, and all the COVID-related. We also had a lot of fiscal spending. We had a lot of labor that was still available. There were good reasons to think that the COVID inflation would pass and monetary policy would not have to respond. It ended up that inflation got much higher than we expected and was much more persistent than we had expected. Then you had Russia invading Ukraine, which sent a commodity shockwave all the way around the world, and then a series of these different supply shocks. You know, coming into a year and a half ago, almost two years ago, it really looked like inflation was well down, going down to 2%. It was like there was a glide path for that soft landing. Then you have tariffs and trade war, which now send goods prices up that we're buying from all around the world. And then, of course, more recently, the Iran conflict. And so I'm sensitive to the fact that for five years now, five-plus years, we've had these rolling supply shocks, and eventually this gets embedded in people's decision-making and it becomes a self-fulfilling prophecy. It's our job to make sure that that doesn't happen.
Neel, one of the things you've done is look back at the '70s, and obviously there are differences with the '70s. We're not wearing bell-bottom jeans and those sorts of things.
The music was better in the '70s.
Music was pretty good. Yeah. All the best music from the '60s was done in the '70s, in my opinion. Anyway, but how does that inform the way you think now about monetary policy?
Yeah, I went back and I had different AI tools go back and read all of the transcripts from the '70s for me and summarize them and say what were policymakers looking at, what were the arguments for raising rates, against raising rates. And it's fascinating, some of the parallels. This is actually a great use of AI because it could actually go through and read all the transcripts. They're all public. It's there for them. And it was fascinating because so much of the diagnosis of the inflation of the 1970s was: these are supply shocks. Monetary policy is not well suited to address inflation from a supply shock. These will pass and then things will return back to normal. Boy, when I read this, I said, this sounds eerily familiar to deliberations that we have been having. Now, it's not all similar. As you said, there have been some important differences. In the '70s, they were experiencing a wage-price spiral. The labor market was part of the inflation story. The labor market is not part of the inflation story today. So that's an important distinction from the 1970s. But it made me realize, you know, you can be fooled year after year after year. It's now been five years. It's now been — if you'd asked me, I've been on the committee now, this is my 11th year. If you'd asked me in my first five years, would we ever endure five years of elevated inflation? I would have said, 'Not a chance. There's no way that could happen.' And yet here we are. So, you know, the inflation episode of the '70s was around a decade. Okay, we're at the halfway point. And so at some point, we have to just say enough's enough.
I feel like my economist friends would feel I would be remiss if I didn't push back on you. If you don't have the conduit of labor to transmit the inflation, where does it come from and why does it persist?
Well, it comes from many things. I mean, it's coming from — as I, for example, I said, energy goes into virtually everything that we buy. Talk to companies: shipping costs are going up all around the economy, both because of fuel prices and also for labor availability. There's a confluence of events. You know, I'm on the board of a small nonprofit, an education-oriented nonprofit, and I sat through a presentation with their CFO doing a forecast for the next few years. In one of the footnotes of their budget for this nonprofit was an expectation of a 3% annual cost-of-living adjustment. Like, that was like a gut punch for me — that was just, he wasn't making a point. That was just embedded in their forecast that 3% is what they need to start budgeting for because cost of living is going up by 3% a year. That's the mechanism by which this stuff becomes embedded in the economy.
I get it. But isn't it true, or at least theoretically true, that without the wage-price spiral, that's only a one-time supply shock — you take that price increase, you pass it along, and you're done. But it's the ever-increasing demand of wages that keeps it going.
Yeah. I mean, I hear you and I'm sympathetic with that. But when we've had these series of rolling supply shocks, I think the risk becomes this becomes embedded. If it's a one event — if it's just Russia invading Ukraine, I'm with you. If it's Russia invading Ukraine, it's on. And by the way, tariffs and trade wars — I mean, these things are going back and forth, ongoing negotiations. There's tit for tat. This stuff, what I've observed, this is taking longer to work its way through the economy and reach a new equilibrium than I would have guessed when these negotiations first started. The longer this goes on, the more risk we take.
So, I have to ask you, did you push back against this guy? Did you tell him, 'No, that's not going to be 3%. It's going to be two.'
You know, I mean, you're like...
I just quietly sat there and observed it.
Okay. You didn't tell me it was wrong. I want to go back to this — that's important. I think you didn't tell me it was wrong. I want to go back to this AI thing because I'm a little bit obsessed with it. What good does it do to raise rates on the — you know, for people in AI? I got to tell you, today I was anchoring the 2:00 show and we're interviewing a guest, an analyst about Micron, and it said 950% revenue growth and 350% earnings growth. And before we went on, I said to the producer, 'Did you not put a decimal in there?' I mean, 950% revenue growth — a quarter point is like, I'm laughing at you. The entire AI industry is just going to laugh at that and will have absolutely no effect. So what good does it do? And then you say to somebody, your mortgage is going to be seven and a half or seven and three-quarters and not seven. That matters a lot.
But I'm with you. It's a combination of factors. One is we have to use the tools that we have, and monetary policy is a blunt instrument, as I said. Second, higher interest rates are the mechanism by which capital gets reallocated in a market economy to its most productive use. So it's not comforting to the person going out and getting a seven-and-a-half percent mortgage. But what the economy is doing is taking capital that would have gone to build homes and apartment buildings and it's channeling it instead to build data centers. That is not a great position for that first-time home buyer who wants to get in. But that's what's happening in the economy. And it's not the Fed's job to stop that reallocation from happening. Our job is to recognize that that reallocation is happening and set interest rates to achieve the dual mandate that Congress has assigned us.
We'll come back to that. I do want to do a bit of a sidetrack on something else that happened today. The inspector general of the Federal Reserve came out with a report about the building renovations, and it did find management deficiencies. I'm wondering what your response is to that report. Have you had a chance to read it and react to it?
So I haven't read it. I've seen the coverage of it. I'm not surprised that the inspector general did not find any criminal liability or violations or any malfeasance or whatnot that he said, but I don't have any other insight into what went on in the building project other than what's been reported.
Is it a bad look for the Fed to have a big cost overrun like that, do you think?
Of course. I mean, one of the contributors in that report, plain as day, is inflation, right?
A little ironic, right? It's our job to keep inflation down. So I would say yes, the cost overrun is bad, but the inflation that was a big contributor to the cost overrun — you know, use the term if that's the original sin, so to speak.
Do you have concern that President Trump might use this as an excuse to fire Jay Powell?
I don't have any insight into that.
Okay. I'm being quiet for a second to see if he wants to reconsider that answer.
Definitely not. Okay, good. All right, let's move back to this. The chairman, Kevin Warsh, has talked about using financial conditions as a way to measure where the Fed is. I'm wondering about your take on that. He seems to have moved away from something that's been used over the years, which is this idea of saying we're modestly restrictive, we're neutral, we're slightly accommodative now. He doesn't want to use that framework. I'm wondering if you could tell me: is this new framework something that we can figure out in terms of that's where the Fed is going and understand the reaction function through the financial conditions index, or do you prefer the old framework?
Well, I think each policymaker is going to look at this differently. I look at both. I try to assess — they're related. So we think that monetary policy works through financial markets. So the idea that they're completely distinct, I don't think is correct. But I look at both. I look at what's happening in the real economy. Does that indicate to me that monetary policy is restraining the economy or supporting economic growth? Implied in that is a notion of what is the neutral rate and are we accommodative or contractionary relative to neutral. Financial conditions are another factor, looking at what's happening in financial markets. Financial markets, by and large, have been booming. You know, whether it's debt financings have been booming, AI companies raising huge issuances in the bond market, equities reaching new highs. None of that suggests that financial conditions are restraining the economy or that monetary policy is tight. So for me, I have to look at all of these to try to get an assessment of what's happening overall to the economy at large.
Is there a danger of a feedback loop here, where financial conditions reflect what it is they think you're going to do, and then you take a message from what they think you're going to do and do that?
It is somewhat of a looking in the mirror. But I also know at times markets can take a very different view from the Fed. And again, I go back to the post-financial-crisis era when the Fed kept forecasting that it's going to raise rates, it's going to lift off from zero, and markets weren't buying it. And markets did not have any trouble saying, 'We don't believe you. We're going to price things our own way.' So I give markets more credit than they are just blindly following the Fed. I do think we are looking at each other, but I also think we reach our own conclusions.
Neel, I first got to know you when you were working at the Treasury trying to keep the financial system from collapsing, which I think you succeeded at. I mean, it was a big group of people...
By the hair of your chinny chin chin.
You did, but it did, you know. So I think you have some interesting thoughts perhaps on this idea of not regulating AI or letting AI regulate itself. Does that bring you back to the days of the financial crisis? I've been seeing all the coverage about, you know, some people, some firms saying we should regulate ourselves, others saying no. And I just remember I have this quote in my head — I think it's an infamous quote from then-CEO Chuck Prince leading up to the financial crisis where he basically said, 'Regulators need to stop us because we can't stop ourselves. If the music is playing, we have to keep dancing, and we're still dancing.' It was a damning comment about his own reflection of what they were doing and his willingness to continue doing it. And of course, Citigroup blew up and we also had the financial crisis. And so the notion that we're just going to trust companies that they're going to be able to — with the face of public shareholders, all of the competitive pressures — that they're just going to do the right thing, it makes me skeptical.
How much do you worry about it?
I worry about it. I mean, I think there are a couple big risks from AI, at least two that I can think of. One is that AI gets out of control and does something terrible. And then, you know, it's like — think about like a nuclear reactor. If a nuclear reactor melts down, governments will spend unlimited resources to try to stabilize it. Not for the sake of the power company, but for all of our sakes, because we all suffer the consequences if that reactor melts down. And so the idea that, well, no, don't worry, the power company's going to be on the hook, they're going to be liable for any of the costs — that's just laughable in the nuclear reactor scenario. So that's a risk that I worry about. The second risk that I worry about is this massive investment ends up not being as productive as we hope. You know, I've been talking to lots of CEOs in non-tech companies and asking them, 'How are you using it?' And virtually every big company that I talk to is experimenting and experimenting and excited. And is it actually affecting your bottom line? No, not yet. A lot of it is what we hope it's going to. We hope it's going to improve productivity, but the fruits have not yet borne out. And so I hope it does come. But if this ends up being massive investment that is not nearly as productivity-enhancing as we assume, then this will have been malinvestment, and then there could be big economic consequences for the economy at large given the scale of it.
That's one way to think about it. And the other way is perhaps the other side of it, if it's massively successful. How does the Fed react in a world where AI increases unemployment, attenuates wages, especially at the bottom of the scale, and still amid massive capital investment where you would think you could react to the unemployment rate by cutting rates, but there's still this tremendous demand for capital?
Well, we have a — it's a very difficult problem. We do have an assessment where we try to judge what is the natural rate of unemployment, the neutral unemployment rate, and that is a moving target. So in that scenario that you're describing, my guess is what we would end up assessing is that the structural level of unemployment has now gone up because of these factors, and we might not actually therefore have to respond to it because we might realize there's nothing we can do about it with monetary policy.
Where do you put the probabilities? Is it more probable that all of this AI investment delivers game-changing productivity to the economy?
I think my best guess right now, just looking at the history of massive fundamental productivity, new inventions, it just takes a lot longer than the optimists think. A lot longer. And it'll show up. But, you know, sometimes I hear these Silicon Valley types talk and they're like, 'Oh, instead of 3% GDP growth, we're going to have 10% real GDP growth.' I mean, absurd numbers that have not been experienced in the last — you know, the invention of electricity didn't generate that. The invention of the engine didn't generate that. And all of a sudden they're talking about these pie-in-the-sky numbers. So I dismiss that. You know, a good scenario would be it just takes a lot longer, but it really does have transformative effects. That would be great. I hope so. I mean, I'm bullish on AI, don't get me wrong, but I'm also trying to think about what could go wrong. I mean, what people don't understand is a good scenario is one that raises productivity from one and a half to 2%.
That'd be a great scenario.
You take that all day long, right?
Absolutely. Absolutely. If it went to two and a half, it'd be a miracle.
And then, of course, that half would simply replace existing labor for prior labor force growth.
