TRANSCRIPT: Fed Governor Stephen Miran talks rates, stablecoins and AI
In an exclusive interview with the Peg, Stephen Miran outlines his views on inflation, AI labour displacement, balance sheet policy, and much more.
Below is the transcript of The Peg’s interview with Fed Governor Stephen Miran, which took place on Wednesday, February 18.
The Peg: How do you view the appointment of Kevin Warsh as chair, and to what degree does a more hawkish chair allow the Fed to push through larger rate cuts this year?
Stephen Miran: I’m really excited for Kevin Warsh. He comes to the Fed with an enormous reputation as a very deep and serious thinker about monetary policy, about the balance sheet, about the banking system, and about the interaction of these various things. He’s got a great reputation in financial markets and policy circles generally, so I’m very excited about what he’s going to bring to the Fed.
There is this famous Rogoff ‘85 result that if you have a central banker with a reputation for being hawkish, he gets more benefit of the doubt from financial markets, from investors, and agents in the economy, and therefore is better able to pursue appropriate policy and achieve a better trade-off for monetary policy. So I think Chairman-designate Warsh brings an enormous amount to the table, and I’m really excited to see it unfold.
The Peg: Which inflation measures best capture the underlying trend right now, and which do you find less informative?
Stephen Miran: It’s important to ask the question whether the data really say what we think they say. It’s very straightforward to tell you what the price of a pound of coffee is or what the price of a barrel of oil is. Those are concrete objects that you can touch and measure and have markets for. But inflation is the change in the general price level, and the general price level doesn’t exist. It’s an abstraction. It gets constructed by government statisticians and economists who are undertaking what’s really an impossible task.
There are 1000 methodological choices that have to be made when you’re constructing a general price level. And, for a lot of them, there’s no obvious answer to them. People make the best decisions they can, but at the end of the day, there’s no obvious answer to a lot of these really thorny philosophical and economic questions. And sometimes people can’t come to a solution at all. And so that’s why, if you look at the way the European Union measures housing inflation. They don’t have any measure of owner equivalent rents whatsoever, because the different countries that comprise the European Union couldn’t decide on one means of measuring it. And so, because they couldn’t decide, they just left it out, which I just bring up to emphasize that there’s a lot of really thorny issues in constructing the price level. You really have to think carefully about whether the data really say what we think they say.
One thing I’ve been trying to do in recent months is draw attention to a couple of anomalies that have distorted our view of inflation. There’s always biases in how you measure inflation. What matters is that the biases grow or shrink over time, because if they grow or shrink over time, then you’re implicitly moving the inflation target.
A certain level of bias is built into the inflation target, but if the measurement bias grows or shrinks, then you’re moving the inflation target around with it. And there’s been a couple of ones that I’ve been drawing attention to recently, and I’m sure there are more, but the ones that I’ve been drawing attention to, because I think they’re very prominent right now are portfolio management services, which — the way that they’re constructed — just naively track the stock market, and in my view, can conflate a quantity with a price.
If the value of your assets or management goes up, the BEA records that as an increase in the price of the assets managed, when, in my mind it’s an increase in the quantities. I understand there’s some thorny measurement issues here. The BEA may very well be doing the right thing for a national accounting purpose, which is what the BEA’s mission is. The BEA’s mission is national accounting. It’s not monetary policy. The Fed’s mission is monetary policy. And so the BEA may be making the right decision entirely for national accounting purposes, but it’s improper for us, for the Fed, to just port that concept over and apply it to monetary policy uncritically. I think that doing so has resulted in some biases in how we think of inflation. These portfolio management services have contributed 36 basis points to core inflation in the most recent data release over the last year. Historically, it’s normally more like six. So, there’s significant growth in this bias, which, in my mind, has grown over time, and we’ve implicitly moved the inflation target down by not correcting for that bias. And so I think it’s important for us to think about making corrections like that.
Another one is the housing market. This has been discussed a lot because of the way that housing inflation is measured. It’s very lagged in how it’s measured. And you know, basically, because of these very, very long lags, the measured shelter inflation is picking up the rental market of three or four years ago, it’s not picking up the rental market of today, in which market rents are growing at a 1% rate for a couple of years.
The key thing that changed about a year ago is that renewal rents caught up to new rents. And the fact that if you’re renewing your lease, the rent is now roughly where it would be if you move to a new lease for a similar home means there’s now a restriction on how much landlords can jack up renewal leases. You can no longer increase your renewal lease more than the market rent of the new lease is going up, because then the tenant would move. A few years ago, that constraint didn’t exist. And so the catch up period of renewal rents to new rents ahead of us might still conceivably have been very long, measured inflation ahead of us may conceivably still have been very large, but the fact that the renewal leases have caught up to the new leases means that there’s now a constraint in how fast renewal leases can grow, which, in my mind, is again, going to lead to a significant slowdown in measured shelter inflation going forward.
So these are some things that I’ve been thinking about, and I’ve been talking about, because I think they’ve been biasing our read of inflation. And I think that if you make those corrections, I think that inflation is not doing so bad.
The Peg: What is the biggest misconception markets currently have about your policy framework?
Stephen Miran: I don’t know what markets think about my reaction function. I would say maybe it’s that I’m unresponsive to economic developments. I don’t think that’s the case. If you look at my end of 2026 dot back in September versus December in the summary of economic projections, I moved my dot 50 basis points down from September to December for a couple of reasons. One, inflation data had come in better than I expected, and the labour market had come in worse than I expected over that period.
Now, when I look forward to what I expect to do with current data in hand — of course, there’ll be a lot more data between now and then — but with current data in hand, with what I know right now, if I had to put down a dot again today, as I will have to in March if I’m still on the Federal Reserve, which probably will be the case by the March meeting. But if I had to put down the dot today, I probably would have moved it back to where it was in September, because the labour market came in a little bit better than I came to expect over the last few months.
There’s been some signs of even more firming in goods inflation. And so those two things combined would make me undo what I did in December. My view is that monetary policy works on the economy with lags. And because it takes 12 to 18, months for monetary policy to really hit the economy, you have to make policy based on a forecast, based on your outlook for the economy. If you are excessively data dependent, you’re very backward-looking, because the data cover what happened in the past, and so there’s a time to be data dependent, and that’s when you have not a lot of confidence in your forecast. When the forecast is extremely cloudy. Now, I think the forecast is pretty clear because market rents are a really great window into the future of shelter inflation, all these other things that I’ve been talking about and putting numbers on in the speeches that I do. But just because I want to be forecast dependent and not data dependent doesn’t mean that I ignore data, right?
I still very much obviously respond to data because the data make me change my forecast. And when the data come out, you can change your forecast for the future. And so being forecast dependent doesn’t mean that you ignore data altogether. It means that the data that you’re using to make your forecast can move your forecast and it’s important to respond to those. And I responded to those data in December, and if I had no more data in hand right now and had to put a dot down at the March meeting, I would respond to the data again and probably end up moving my dot back to where it was in September.
The Peg: Governor Barr argued that the AI boom is unlikely to justify lower interest rates since technological change is structural. Do you agree?
Stephen Miran: You know, I don’t agree. I think there’s a few things that you have to think through when you think through the consequences of a productivity shock or technology shock for monetary policy.
The three major inputs into monetary policy in most frameworks are what we call the neutral rate, what we call the output gap — i.e. the difference between potential GDP and actual GDP, or supply and demand — and inflation. And so when I think about what AI does to those three counts, I think inflation seems like the easiest one. It seems quite deflationary to me. It’s making things cheaper to make. And when you make goods and services cheaper to produce, that flows through into prices. You also get more goods and services.
And you know, if you just write down a very simple supply and demand shock, and you think of a productivity shock as increasing the potential capacity of the economy, increasing supply. And you move supply out, then you know you’re going to get lower prices.
While we’re on the subject of prices, let me say one more thing, which is that the Federal Reserve has emphasized in recent years the inflationary consequences of the supply chain bottlenecks that developed in 2021/2022. If we thought that constraining supply had an effect on pushing inflation up, I think we have to acknowledge that the reverse is also true, and that expanding supply in the economy through technological advances, through things like easing of regulations, that expanding supply and easing the supply side of the economy, producing more with fewer inputs, that doing so has the opposite effect and pushes prices down.
If we rely on that as an explanation on the way up, I think we have to be honest and acknowledge that it’s a good explanation on the way down.
The second piece is the output gap, which, of course, is whether potential output, or actual output, increases by more or less. And that I think is a much more nuanced question. There are good arguments in both directions on this one. If you think back to the 90s, a pretty classic view is that the IT technology shock was what we call gap neutral. It could have moved actual output and potential output by about the same amount. So you had new technology, all this internet stuff, but it created new demand too because adopting this technology required investment. You had to hire people to dig up the streets, lay the telecom wires, put the streets back together. That’s very labour-intensive. It was very investment-intensive. And so actual output increased also, and in line with potential output, and the output gap didn’t really move very much.
I think the different types of productivity shocks will hit the economy in different ways. To me, that’s a plausible description of what happened with the IT technology shock in the 90s. That’s not at all in my mind, a plausible description of how regulatory shock hits the economy. I gave a big speech on this in January, because regulation constrains existing capital from producing and when you lift those constraints, you’ve already got the capital in place. And so you don’t need to invest in the capital the way you had to invest in laying the telecom wires. If it’s the case the economy wants more capital, because net capital is more productive — so there can be incremental investment — but in my mind, it’s absolutely going to be less than the increase in potential. And so a regulatory shock will open up an output gap, because potential GDP will increase by more than actual GDP. I talked a lot about that in January. An AI technology shock could very well be like the IT technology shock, because we’re investing in data centers, right? We’re investing in training models, we’re investing in GPUs. All that’s true.
But then there’s another question of, are those investments as labour-intensive? Do those investments create as much demand in the United States as laying telecom wires did? And if a lot of investments are going to semiconductor fab plants in Taiwan, then that might not be demand that accrues to the United States. Or if the investments are going to a small number of IP owners that aren’t spending a lot of the investments, then that might not be going into the United States. To the extent we’re building data centers that absolutely seems like real demand, just like laying telecom wires was. But if a lot of the products in those data centers are imported goods, because they’re electronics that are manufactured in another country, then not all of that activity is going into the United States. Some of it is leaking internationally. So the question of, is AI gap neutral in the way that telecoms arguably was gap neutral, is an open question. I could imagine that it is, but it’s also easy for me to imagine that it’s not.
Then finally, there’s the question of the neutral rate. I have no doubt in my mind that if you raise productivity growth for a long period of time because of technological invention that does raise the neutral rate. However, I think it’s also important to know that a lot of things have changed to move the neutral rate around in recent years. I’ve talked about that a lot. I’ve emphasized the changing population growth, which I view as much more primary, as an input into where the neutral rate is.
We just went through years of endlessly discussing whether everyone was becoming Japanese because of sheer population dynamics. Then also national savings. We’ve had these big swings in national savings, primarily driven by fiscal deficits in recent years, which again raise neutral rates or lower neutral rates, if national savings is going down, and those in my mind, are much more primary drivers of where the neutral rate has gone than a technology shock that we cannot quantify.
I have not seen any plausible quantification of what AI will do to the long-term productivity growth rate for the economy that would feed into a neutral rate estimate, certainly not along the lines of the types of estimates that we have for population growth and national savings. And if anybody presented me with a number saying, I know exactly what AI is going to do the long-term productivity growth in the United States, this is my forecast, and you can take it to the bank. I think I would probably laugh at that. Nobody’s got any right to a lot of confidence in a claim like that. So I think there’s a lot of moving parts here. Some of them argue, in my mind, for looser policy. Some of them argue for weaker policy. On balance, I think they argue for looser policy, but I also think the arguments are much clearer for regulatory-driven productivity shocks for which there’s a large existing literature, which is why I’ve focused on that.
Oh, and of course, we didn’t discuss AI potentially causing layoffs as well because of the potential destruction of some job categories. That’s another absolute key thing. So it’s always been the case that technology has destroyed jobs. Always been the case, but it’s also the case that technology creates new jobs too, and technology has been advancing for 1000s and 1000s of years, and we always manage to create new jobs. I have a really hard time believing that AI is different than the thousands and thousands of years of human history that we have: that jobs get created and jobs get destroyed and then new jobs get created. And my instinct is just to think that we probably aren’t creative enough to imagine the totally new job categories that don’t exist because we haven’t imagined it. And one example I like to give is that social media destroyed a lot of jobs in the journalism industry, but then it created new jobs too, like social media advertising strategist, which is a job category that just did not exist before social media. And nobody could have imagined that job category in 2000 and by the time there were thousands of them in 2006, 2007, 2008, it was obvious. Everybody saw it. But nobody imagined it ex ante. And right now, I think we just have a failure of imagination that we can’t imagine the new jobs that AI will create, because it’s just hard to be a really successful futurist. It’s a difficult it’s difficult task. You know, not, not all of us are Isaac Asimov.
So AI will certainly eliminate some jobs. It will also create new job opportunities. I don’t think that we can conclude that there’s going to be some structural increase in unemployment, that’s a long term increase in unemployment, because that hasn’t really happened for previous technological bouts. Why would we conclude that this, alone, in all of human history, is a technology that we will never recover from economically? That’s a bold claim that I don’t really see a lot of reason for accepting.
I also think it’s the job of the central bank to accommodate that transition. As the economy is reallocating jobs across sectors because of transformative technologies, as the economy is creating new categories of jobs because of transformative technologies, that traditionally is exactly the type of unemployment that the central bank would accommodate. So I see some reasons why AI would lead me to think monetary policy should be looser. I see some reasons why it should lead me to think monetary policy should be a little bit tighter. I have a hard time putting numbers on many of these things. I think everyone else should be honest and have a hard time putting numbers on many of these things as well. But, you know, I can’t control what people talk about.
The Peg: Moving on to stablecoins. Dollar stablecoins are currently dominant, but in the last few weeks, European officials have said they are looking to push euro-pegged stablecoins and euro liquidity backstops to promote the international role of the euro. They say that these policies could complement the introduction of a digital Euro. Do you see the dollar being challenged by such architecture, or will the dollar and its respective swap line system remain structurally dominant, even against those sorts of challengers?
Stephen Miran: I think the dollar will remain structurally dominant against those challengers. I don’t see any realistic competitors to the dollar. If you think about being an international currency, you want to be convertible, you want to be deep, you want to be liquid, and you also want to be a growing share of global GDP. And I see a lot of reasons for expecting U.S. economic growth to continue diverging to the upside versus global GDP growth. I’m pretty optimistic about that. So I don’t see a lot of credible threats to dollar dominance. And, you know, I expect the dollar to remain the dominant international currency.
The Peg: You don’t think that euro-pegged stablecoins might make a difference, or that they may be a game changer in any shape or form for the euro?
Stephen Miran: Many Europeans seem to have no problem opining about U.S. economic policy. But I don’t want to tell them what would be good or bad policy for them. But like I said, you’d want to see a deep and liquid market for a currency that was going to have a large international presence, and taking highly fragmented markets and putting a wrapper around them doesn’t necessarily create the type of infrastructure that would be a very significant rival to the dollar.
The Peg: Do you think any developments with eurobonds could change that?
Stephen Miran: I don’t want to tell the Europeans what policies they should pursue. They should decide that for themselves. The question of whether or not to adopt eurobonds gets deep into a set of questions about sovereignty and democracy, and the role of the EU and the role of the national governments. And it’s not my place to tell the Europeans what policies they should pursue. I trust them to choose the policies for themselves that they think are best, and they should do the same for us.
The Peg: If, as you’ve argued, convertibility and ease of cross-border use are central to reserve currency status. Where do AML enforcement, sanctions, and increasingly, bank-like regulation of stablecoin issuers sit on that spectrum? Can you have your cake and eat it on that front?
Stephen Miran: Yes, because those types of tools are deployed on people who are violating the law or undermining the system that the United States is underwriting. The United States is underwriting global financial system and global trading system, and those types of tools that you’re describing are deployed on people who are trying to undermine that system and so the use of those tools ends up reinforcing the system itself rather than undermining it.
The Peg: If, stablecoins continue to grow and are allowed to offer returns that compete with bank deposits, how might that impact monetary policy transmission or financial stability?
Stephen Miran: I don’t think it would really necessarily impact monetary policy transmission so much, because stablecoins are just wrappers, right? So if you own a treasury bill or money market fund or bank deposit directly, or you own a stablecoin, and the stablecoin provider then owns the treasury bill or the money market fund or the bank deposit, instead, economically, there’s not really that much of a difference, right?
Those assets are still held by the public, there’s just a wrapper around them, held by a different part of the public. And the people who are borrowing are still borrowing at the rates that are ultimately keyed off of the monetary policy rate. So I don’t think that there’s a reason for supposing this is really going to affect monetary policy transition.
Financial stability is an issue that I’m still trying — as a result of if there were yield offered on stablecoins — is an issue that I’m studying and still trying to get my head around. I’ve seen very strong arguments made by groups with all sorts of perspectives on it. It’s not an area where I’ve reached a very firm conclusion yet, and it’s something that I’m still exploring and trying to get my head around.
The Peg: You’ve spoken about the interaction between the balance sheet, monetary transmission, and regulatory dominance. Cecchetti and Schoenholtz argued this week that balance sheet shrinkage poses financial stability risk and is impractical. What’s your view on that position?
Stephen Miran: I took a look at their piece. When you say they said that balance sheet shrinkage poses risks, were they arguing about any shrinkage at all, or about a return to a scarce regimes, a scarce reserve regime?
The Peg: I think they were arguing about a return to a scarce regime specifically.
Stephen Miran: I think this is an important distinction, because I think you can reduce the Fed’s balance sheet without returning to a scarce reserve regime. And the point that I was trying to make, or maybe I should have made it more explicitly, in the November speech on regulatory dominance, is that balance sheet policy in the current era is defined by the boundaries between scarce reserves, ample reserves, and abundant reserves.
And those boundaries are determined not entirely, but in large part, by the regulatory environment. And I think that it’s possible to alter the regulatory environment and lower those thresholds. And therefore, you could lower the line that separates the boundaries from ample and abundant or ample and scarce, without actually returning to scarce. My preference would be to make improvements to the regulatory system, to lower those thresholds, and then to shrink the balance sheet, rather than saying we’re going to keep the regulatory system as it is and then go to scarce with the current lines as they are.
You might imagine that a pre-regulatory reform “scarce” threshold might actually be significantly above a post-regulatory reform “ample threshold”. And going back to scarce reserves today may require more reserves than keeping ample reserves next year, after we’ve done a lot more bank regulations. And so my preference would be not to make an immediate attempt to return to scarce reserves. I don’t know if that’s necessary. I think that’s a much larger undertaking that requires a much broader rethink of the regulatory apparatus. For all I know, it requires legislation, amending or repealing Dodd Frank. It’s a very significant undertaking, and not the type of undertaking that should be taken very lightly. However, it does strike me as very plausible to start making the types of incremental regulatory reforms in the context of the current structure that would reduce thresholds between abundant, ample, and scarce, and allow you to shrink the balance sheet without returning to scarce reserves.
The Peg: You don’t think that would have any impact on repo rates?
Stephen Miran: Let’s just suppose that one of these important regulations is the liquidity coverage ratio. And if you had a very onerous liquidity coverage ratio, then the threshold between ample and scarce is going to be very high. And if you relax that liquidity coverage ratio, then the threshold between ample and scarce is going to be lower, right? And so if you keep ample, but you relax the ratio, then you have a smaller balance sheet, and so the divergence between repo rates and other rates will probably not be that reactive, but the difference is that banks will be incentivized to hold fewer reserves, and so there will be some more financial volatility risk. Because you’ve forced the banks to be single, double, triple, quadruple insured, and then you’re going from quadruple insured to only triple insured. And so you’ve incrementally increased a little bit of risk there, but you’ve allowed yourself to shrink the balance sheet as a result, and to reduce the regulatory burden on the banks and allow them to better send credit into the economy.
The Peg: With administered rates increasingly anchoring secured funding markets and repo rates, and doing most of the work in the operating framework, do you think there’s a diminishing role for the federal funds market itself? How do you see that evolving?
Stephen Miran: That’s very clearly happened, right? The Fed funds market very clearly has diminished in importance relative to the administered rates. I don’t think there’s a lot of doubt about that. I’m not quite sure why we hold on to the Fed funds rate, but we do. Maybe someday someone will explain it to me. I would imagine that if you want to, if you want to aggressively reduce the balance sheet, and doing so requires doing some serious work on the bank regulatory front as well — and my my prediction is that Vice Chairman Bowman’s deregulatory agenda is going to get turbocharged. If you want to reduce the balance sheet, there’s no other way to do it. You’ve got to turbocharge Miki. We’re going to have “super Miki.” And so if you’re going to engage in shrinking the balance sheet and transform Miki into super Miki, I would imagine that maybe you also want to rethink the implementation mechanism at the same time. I’m not at the point yet where I have really strong views on what that should look like, but I think you’re asking a super important question.
The Peg: Markets have been increasingly focused on the interaction between Treasury issuance patterns, reserves, and money market conditions. How tightly linked are those in practice, and does the current environment require closer coordination, or at least awareness, between debt management and monetary policy?
Stephen Miran: The answer is yes, no, yes. They’re tightly linked. And you can tell they’re tightly linked because our reserve management purchases have to be responsive to the demand from reserves that waxes and wanes as a result of the payments into and out of the general account. So they are mechanically tightly linked. We respond to that. There’s no question about it. They don’t require coordination, because the Fed should be pursuing monetary policy for monetary purposes, and Treasury should be pursuing fiscal policy for fiscal policy purposes. And we should each be living in our own little fiefdom, and staying in our lane. So I don’t think they require coordination in the ordinary course of business. However, they do require awareness, because what the one does can depend on what the other does, because the choices from the one can affect conditions that matter for the choices of the other. So it does require awareness.
In theory, you could run up against a time in which there are constraints. For example, in line with the notion of reducing the footprint of the Federal Reserve in the markets, I support the idea of holding more bills and less coupon debt, less duration on the Fed’s balance sheet. That’s something I support. But, completely hypothetically, if we were to move as aggressively as we could out of coupons and into bills, we might actually have more demand for bills than Treasury is providing. And so, in theory, there’s a world in which we eat up all the available supply on a flow basis, right? And then we result in making collateral scarce, and there are repo fails and all this stuff getting in the way. And so in the ordinary course of business, I don’t think it requires coordination, but we have to be aware that the actions of the one affect the actions of the other.
And then you could imagine some extreme circumstances in which it did require coordination. And you know, we’re not going all in on replacing coupon debt on our balance sheet with with with T bills, right? We’re replacing maturing mortgage debt with T-bills. And, I would like to continue replacing more of our securities holdings with bills. But if we went all in on this in a maximalist way, it would run into supply constraints that would have consequences for issuance, consequences for markets, that then would require, I would think, that the two agencies talk to each other a little bit more about these things. But that’s sort of an extreme situation that is, in theory, possible, but I don’t see it as likely, and certainly not a situation that I want to arrive at. But I’m just sort of fleshing out the answer to your question. But they definitely do affect each other.
The Peg: Everybody is obviously talking about the gold price. Does it have any meaningful bearing on monetary policy or how you’re thinking about transmission?
Stephen Miran: I don’t think so. It doesn’t, for me. The gold price is a function of a lot of things. U.S. interest rates are one of the things that go into the gold price, but lots and lots of other things do too. And I think it would be a mistake to infer too much of relevance for Federal Reserve policy from gold prices. It’s important to understand it’s not the most liquid market in the world, also, right? It doesn’t take huge flows to generate large price moves in a market like gold.
You’ve seen that it can have very strong moves, like, what’s the move today, two and a half percent? These days, that’s normal, right?
I find it difficult to think about taking very significant inferences from the gold market. It serves a function in commodities markets. That’s important. I’m not too serious about it. I think it shows aspirations. There are some countries that really would love to get away from the dollar for their own geopolitical purposes, and they’ve wanted that for the longest time, and they still want it, and they want it even more now than they did in the past. And so maybe gold is sort of displaying an intensification of those aspirations, but at the end of the day, there’s no escape. They’re just aspirations. And maybe the gold price is a signal of how bad they want it, but it’s just not something that I spend that much time thinking about.
In the long run, capital follows economic growth. And even when you look at the worst EM offenders, they default, and then they’re back in capital markets a few years later. Investors are never really able to stay away for long. In the long run money follows opportunities. If you think that the U.S. economic growth story is a good story, whatever the reason is — the full expensing in the One Big Beautiful Bill Act, AI, deregulation, whatever it is that you think that makes the U.S. economic growth story attractive. In the long run, capital is going to follow that. And we can have lots of pearl-clutching over people not liking this policy or that policy, or they’re not liking the tone, or they’re not liking the style. But at the end of the day, money is going to follow the substance, not the style. To the extent that I continue to see U.S. economic growth on a pretty good trajectory relative to the rest of the world, I think that’s going to just underline the dollar system and reinforce it enormously.