Reducing the Fed’s Balance Sheet With Fed Board Governor Stephen Miran

So, thank you. Look, we are expanding the balance sheet now, and I voted to resume reserve management purchases in December. I supported that policy. And that's because conditional upon the current institutional framework, the regulatory framework, the implementation framework, we had run into an area approaching scarce reserves in December. And so if we were going to maintain an ample reserves framework, we needed to start reserve management purchases and increasing the quantity of reserves in the system and increasing the balance sheet. Now, what I tried to do in the paper and in the speech is to provide a set of options for reducing reserve demand so that we don't have to keep increasing our balance sheet like this. We can go back to shrinking our balance sheet, which to me is the right approach. Now, once we undertake those options, those steps, I see 1 to 2 trillion dollars of potential reserve reduction while retaining ample reserves, which sounds like a good range to me in terms of thinking of reduction, but that's the type of thing that would have to be weighed by the committee in the future in terms of what they wanted to do and how far they wanted to go.

So, as I said in the speech, it's a weak lean, and it's certainly not a hill I would die on. It's not even a hill I would take my old wounds on, I don't think. Part of the reason for that—there's two reasons for that lean. One is that we have ample reserves now, and if we have a path to remaining in ample reserves but reducing the balance sheet by up to $2 trillion, it seems if something's working reasonably well but you can reduce the balance sheet, then it doesn't necessitate an institutional change back to scarce reserves. It seems like less work, and so that seems like a positive to me. The other reason is that we still do live in a world with Dodd-Frank and Basel, and I do think that you could have scarce function in a world of Dodd-Frank and Basel, but there's some more questions in my mind. So, it's a weak lean, and I could be talked into scarce, but that's sort of my thinking at the moment, and because we're sort of still studying the subject, I could imagine my thinking on it evolving as we continue to do more research.

Sure. So I like that framing of it. Let me just repeat it a little bit so that it'll make my answer make a little bit more sense. So when we reduce our balance sheet, you know, a balance sheet has assets and liabilities, and if you reduce the balance sheet, you're reducing both assets and liabilities. Now, the bulk of the paper deals with the liability side of it, which is reserves. You know, banks own reserves, those are assets on the bank side, liabilities on the Fed side. And when we reduce reserves, you know, in the past, we get to a point where there starts to be a little bit of mayhem in short-term funding markets, and that tells you that you're approaching scarce reserves, and you need to stop. And like the 2019 repo crisis.

And those are—I listed a number of those in the speech as well without going into details because it's a 50-page paper. On the asset side, the Fed owns securities, Treasuries and MBS, and you're referring to the taper tantrum that happened in 2013, you know, big increase in long-term interest rates. And you know, on the liability side, we talked about sort of making sure that repo markets don't go haywire when you're reducing the balance sheet, but you also have to think about the asset side, too, and how do you avoid a spike in long-term interest rates that you don't necessarily want as a result of what you're doing. And so I mentioned going slowly, I think that's important. But I think another thing that you—there's a couple other steps that you can do that we mentioned in our paper, and they have to do with some regulatory implementation things. And so one of them is going to central clearing of Treasury repo. Because if you own a Treasury on repo, you've got a risk charge not for the Treasury on the risk-weighted side of the regulatory regime, but for the repo exposure, for the counterparty exposure in repo. And if you can novate across trading partners, you've got a long position with this guy and a short position with that guy, they'll net out, and then you won't have the charge. And so you reduce the regulatory burden of owning Treasuries on repo. That's one thing that you could do that we discussed in the paper. Another is the G-SIB surcharge on the risk-based side as well. And so Treasury securities in the risk-based side of the regulatory framework are supposed to have a zero capital weight because they're Treasuries, but they don't actually have a zero capital weight in calculating the G-SIB surcharge because they make banks more systemically important if you own more Treasuries. And so there's a little bit of a distortion to the way that we think about zero risk weight in the risk-based side for Treasuries. And so those are some reforms that you could take on the regulatory side that would make it a little bit easier for markets to absorb securities that are coming off of our balance sheet.

Sure. So, I still see underlying inflation as gradually moving down towards target over the next, you know, sort of 12 months, let's say. I think underlying inflation is remaining very well-behaved and with the labor market on a very, very gradually loosening trend for 3 years now, it's very difficult for me to imagine that changing. Now, the oil shock, of course, is very real. But, the way that oil shocks work when you think about inflation is they boost the price level pretty quickly. You know, gas prices move up pretty quickly, right? Or airline fares move up pretty quickly. Anything that's really tied to oil in a very powerful way tends to move up very quickly. But, monetary policy hits the economy with lags, with long and variable lags. And most people think that those are 12 to 18 months. And so, what happens with the oil price is it lifts the price level pretty immediately because it feeds through into gas prices and other stuff immediately. But, then unless there are second-round effects, then you don't have inflation as a result of oil 12 to 18 months out. Inflation 12 to 18 months out is pretty much unaffected. And you see this in inflation expectations. So, I like to look at the CPI swap market. And 1-year swaps, you know, moved up a lot. But, as you look at forward rates, if you look at the 1-year rate 1-year forward and the 1-year rate 2-years forward and the 1-year rate 3-years forward, they haven't really moved. If anything, they're actually down since the January FOMC. So, and they haven't really moved much in response to the Iran shock at all. So, if we change interest rates now, it wouldn't hit the economy for 12 to 18 months from now, but the market isn't really seeing any inflation from the oil shock 12 to 18 months from now because it all happens immediately. What you would get is a much weaker economy as a result of tighter monetary policy. And so, this is why classically central banks look through oil shocks. And I tend to think that the traditional wisdom on—I mean, I'm very, as you can tell from the speech and other things that I've said and written in the last several years, I'm very happy to challenge conventional wisdom when I think it's wrong. But, in this case, I think conventional wisdom's right.

So, the labor market, you know, was extremely tight in 2021-22. And then as the Federal Reserve started its tightening cycle, the labor market set into a very gradual cooling trend. That trend's been in place for 3 years. I think given labor markets tend to show a lot of momentum, they tend to show a lot of persistence in underlying trend. And if this trend has been going on for 3 years, it seems to me that it should be the null hypothesis when you're thinking about how the labor market has changed. I haven't seen anything that's convinced me that this 3-year-long trend has adjusted. There are some folks who think the labor market is showing signs of stabilization. I think the trend has continued and therefore merits additional support for monetary policy. Monetary policy is modestly restrictive, and I don't think that the labor market really calls for that.

Yeah, so I like model estimates. So, first of all, R-star is the neutral rate of monetary policy. It's the rate that's neither accommodative nor stimulative for the economy. If you're at neutral, you're not hitting the gas or the brakes, you're coasting. If you're above neutral, you're hitting the brakes. If you're below neutral, you're hitting the gas. And it's a very difficult concept to measure. It's an abstraction. It doesn't exist in the real world. You know, I can tell you the interest rate on a Treasury bond. I can't go out and sort of find R-star. It has to be estimated, right? And so, I tend to like some of the time series methods for estimating R-star. However, they take quite a while, I think, often to update. And the world changes more quickly than they can arrive at new estimates in some cases. And so, I like to sort of start with that and then sort of adjust it with salient things in the economy that I think are first-order importance that have changed. And so, what I'll do is I'll sort of start with that and I'll think about something like—and I've talked about this a lot—think about something like population growth, right? Population growth is something that I think is not controversial to think of it as affecting R-star. I think that's a very well-accepted concept. It wasn't that long ago that we were all talking about Japanification of the whole world because, you know, converging fertility rates coming down and demographics and aging and was that going to lead to low interest rates everywhere, right? That conversation was very common pre-COVID. And I think those pathways are valid pathways. I don't think they ever went away. Just other stuff happened in the meantime that we were paying attention to other stuff. But, I think those economic mechanisms have always been valid. And we just had the largest population growth shocks of our lifetimes in both directions in the last few years, right? We had a huge population growth spike and then a huge decline in population growth related to changing border policies in the last few years. And so, a totally 180. And so, the idea that that wouldn't sort of feed through into short-run neutral rates that had consequences for monetary policy is sort of strange to me. So, I will tend to sort of think, okay, let's take some of these models that do a good job of estimating these things, but maybe they're a little bit slow to adjust or something like that. And then adjust them based on some salient changes that I think will be relevant for it.

Yeah, I think those are great work, you know.

Sure. I mean, we may already be seeing it. You know, the labor market is weakest for new entrants and re-entrants to the labor market. And those are areas that generative AI is arguably affecting. And by the way, the fact that those labor market segments seem weak seems to me an argument against assuming that low payroll growth is a function of border policy. I think a lot of people like to dismiss low jobs numbers as a result of changing border policies being a negative labor supply shock. If that were the case, then cohorts that were sharp substitutes, that were strong substitutes for immigrant labor, would have the hottest labor markets in the country. And they'd be experiencing rising wage growth. You know, if you sort of hold demand constant and reduce supply, you get higher prices. And you don't see that in the data, especially in a period of high productivity growth. You would expect to see much stronger wage growth. And so, to me, the labor supply story isn't really consistent with it. I want to say something else about AI, which is that, you know, one criticism that I've heard people make is that we keep getting these negative supply shocks and the Fed keeps asking people to look through the negative supply shocks. Look, oil absolutely is a negative supply shock. And as I explained before, it doesn't affect the economy on a timeline—unless there's changes in inflation expectations or a wage-price spiral, which is not happening—it doesn't affect the economy on a timeline that monetary policy can respond to. But, there also are positive supply shocks. And AI is a great example of a positive supply shock. AI allows people to do more with less. Right? It boosts productivity. It reduces barriers to entry. It allows people to produce more with fewer inputs. That's a positive supply shock that really matters for economic growth, for inflation, for unemployment, and for monetary policy. And all of that is really something that's very important that I think we need to take into account. Another positive supply shock is the change in the regulatory environment. I gave a speech in January in Greece in which I looked at the modern literature, some of which is by your Hoover colleagues, Patrick McLaughlin. I looked at the modern literature on trying to quantify the regulatory code. And there's people like Patrick and Joseph Kalmenovitz who do very good work using modern AI and machine learning and quantitative methods to reduce the regulatory code to numbers that you can then study. And I use this literature and I calculated that the deregulatory shock that's been ongoing since last year would weigh on inflation by about half a point per year over the next few years. Now there was a Federal Reserve research paper that was published a couple weeks ago by two Fed staff economists, Danilo Cascaldi Garcia and Matteo Iacoviello, who using an entirely different quantification method and an entirely different empirical method came up with similar estimates of the deregulatory shock we're living through. And if you apply their model to the size of the deregulatory shock that they estimate, it causes about a 30 basis point drag on inflation, reduction in inflation per year for the next few years, right? So I estimated about a 50 basis point persistent drag. They estimated about a 30 basis point persistent drag. I think those two are within noise of each other. I wouldn't reject 30 basis points as being outside of the confidence bands given the uncertainty in these things. But this is an example of a positive supply shock. Like AI is another example of a positive supply shock that I feel is underappreciated in all of the talk about negative supply shocks. And whereas oil is generally a one-off shock, again unless there's changes to inflation expectations, unless there's a wage-price spiral, neither of which is happening, both AI and deregulation are going to have persistent effects when you look at the way that these models work. And so as a result, these are things that I think are going to be pushing out the supply side. We all know there's supply and demand. And if you hit the gas on demand while you're holding supply constant, you get inflation, right? If you push out supply as well, you're not going to get inflation.

No, so AI definitely pushes the neutral rate up. And so I gave a speech on my view on the neutral rate in September. It was the first speech that I gave as a member of the Federal Reserve Board. And in that speech I described a number of policy changes that I thought had pushed around the neutral rate. And a lot of them brought the neutral rate down. Things like population growth, things like reduced fiscal deficits due to tariff revenue. AI is something that pushes the neutral rate higher. Deregulation also pushes the neutral rate higher by improving productivity. Positive supply shocks will push the neutral rate higher because they improve the return on capital. Now in that paper I started at what I thought was a very high neutral rate and then I applied sort of changes from population growth and increased fiscal revenue, decreased national borrowing. And then I came down to a low level. So I started probably at the top end of the range of my colleagues and then I ended up at the bottom of the range of my colleagues. But in either case I was above them. But I baked AI into my starting points or I tried to.

Sure. So look, you know, in the long run we need to get our fiscal house in order. In the shorter run I think that tariffs have started raising revenue that changed some of the trajectories in the short to medium run. Now they don't—I don't see them solving the very long-term budget issues which of course matter for the interest rate market and for things that matter to monetary policy and R-star and the economy. But I think in the short term, in the medium term, those dynamics have definitely changed. So if you expect about a point and a third of GDP of additional revenue from tariffs per year over the course of a decade, I think that's a pretty powerful reduction in the deficit. Also, better productivity growth, better GDP growth also improves the deficit because revenues will go faster than outlays. And so between the two of those I think it's not far-fetched to think that the primary deficit can get better by a couple points of GDP. Which of course again does not solve our fiscal problems on a multi-decade timeline, but it will ameliorate those problems over the course of coming years, right? So I do think that this is an area for important work. But I think things have gotten somewhat better in the recent past too. And I think that's starting to be evident in the data. And so if you look at the time period after tariffs were implemented, if you look at calendar Q2 through Q4 of 2025 versus calendar Q2 through Q4 of 2024, you see the deficit starting to come in. Now not all of that is tariffs not by any means, but it's coming by hundreds of billions of dollars. And so I think that is worth acknowledging.

Thanks so much for having me.