Darrell Duffie On How to Fix the World’s Most Important Market

0:00You're being sold an AI future where you're obsolete or irrelevant. That vision is wrong. At Palantir, they're building AI that helps workers and unlocks their full potential. American workers are our nation's greatest strength. AI shouldn't eliminate them. It should elevate them. Palantir is here to tell their stories. From factories to hospitals, AI is freeing people from drudgery, letting them do what humans do best. Create. Solve. Build. Palantir, making Americans irreplaceable.

1:01Acrobat Studio. Learn more at adobe.com slash do that with Acrobat.

1:17Hello and welcome to another episode of the Odd Lots podcast. I'm Tracy Alloway. And I'm Joe Weisenthal. Joe, have you been watching Treasury Yields lately? They're up. They've gone. I'm aware. That's it. That's the show. I'm aware that the line has gone up and to the right lately. But not just that. They've moved quite quickly up. And I think volatility in the treasury market has once again become a talking point. And I always get a little bit of a sense of deja vu because whenever things start happening in the market for US government bonds, extreme things, these extreme moves, it feels like everyone says, oh, this shouldn't be happening in the world's most liquid market, the world's safest market.

1:56We shouldn't be seeing these types of dramatic shifts. Right. The expectation is this is like a very extremely liquid, slow moving market, but it has been a very fast move. And people don't really know why. There's a lot of debate. You have people talk about, well, look, the economy is proving to be more resilient than people might have guessed six months ago, even a month ago. And I think there's this expectation, there's this sense that things are moving faster. There's also a lot of talk about treasury supply, supply demand imbalances, and so forth. So things are on the move, to say the least.

2:29And you see it in mortgage spreads are very wide. You see another risk assets not seeming to like this treasury volatility. So lots going on there. So you mentioned a couple of things there, which is, yes, the economy seems to be doing relatively well. And yes, the amount of treasury supply is exploding and has been going up for a very long time. There's another factor here, which is the actual inner workings of the treasury market. So how treasuries are actually traded and whether or not that is potentially contributing to some of the volatility that we've seen. And again, I feel like this topic keeps coming up.

3:06Every year there's some massive move in treasury markets and everyone starts talking about liquidity issues. And yet you don't really see any big solutions being proposed to it. You see talk about, you know, maybe loosening some of the regulation around the supplementary leverage ratio or something like that. But it feels like we need to talk about this and maybe try to come up with a solution. Let's do it. And we're in a good place to talk about it. Yes, we are at Jackson Hole. I thought that was going to be the first. That is like, we're at Jackson Hole. I like how you like, we eased into it a little bit.

3:37Oh, I want this to be an evergreen episode for an evergreen problem, it feels like. But the people here care about this topic. Yes, absolutely. The way U.S. Treasury debt is traded has a lot of implications, particularly for the cost of financing. the US deficit. That's an obvious one. And so we should talk about it. Let's talk about it. I am very pleased to say that we have the perfect guest. We are going to be speaking with Daryl Duffy. He is, of course, a professor of finance at Stanford's Graduate School of Business. And he is presenting a paper at Jackson Hole in front of the world's top central bankers all about this issue.

4:14It's called Resilience Redux in the US Treasury Market. So Daryl, Thank you so much for coming on OddLots. Thanks for having me, Tracy, Joe. It's terrific to be here. You know it's the perfect guest when it's the guest. It's like not just like we want to have him on OddLots, that the top central bankers around the world want to hear from Professor Duffy. So yes, I'm very excited about this episode. When treasuries are moving, they call up Daryl Duffy. All right. Well, Professor, thank you again for coming on. Why don't we just start with the basics? We sort of alluded to this in the intro, but it feels like this is a subject that just keeps coming up and never really seems to go away.

4:54What's going on here? Well, over the past century, there have been, as you suggested, many episodes of increased volatility and liquidity problems in the treasury market. But I do think these are happening more frequently. Just recently, I think you summarized pretty well some of the stresses in the treasury market coming from the fiscal side. The US is issuing more than people expected. There was a recent downgrade by Fitch. And the Fed is struggling with what to do about inflation. That additional monetary policy uncertainty also contributes to volatility. Let me back up a minute. I just spent most of the last year on a sabbatical at the New York Fed and working with some terrific economists there.

5:36Michael Fleming, Frank Keen, Orr Shekhar, Peter Van Tassel, Claire Nelson, we decided we needed to look into the relationship between the volatility that you two discussed and the liquidity in the market. They're closely intertwined. So we dug deep and went into a lot of data. And yeah, volatility seems to be the main determinant of liquidity in the market. So when fiscal uncertainty or debt ceiling debate or a COVID crisis or monetary policy uncertainty start to get volatility higher and higher, market becomes less and less liquid. It's an extremely regular relationship. About 80 % of illiquidity is explained simply by variation in yield volatility.

6:25So just to press on this relationship between rates volatility or yield volatility and illiquidity, which way does the direction run? Because we can come up with these fundamental stories, right? Like, okay, maybe the economy is better than expected. Maybe we're going to have more rate hikes than previously anticipated, in which case you say, okay, well, that's a fundamental story, a fundamental driver. And then you could look at certain nature of the structure, like the size of dealer balance sheets, et cetera. It's like, okay, well, this is something technical that contributes to, also could be a contributor to volatility.

7:01How do you think about the directions of causality when you look into a problem like this? Yeah, well, both. The direct kind of fundamentals, fiscal monetary fundamentals, and the global economy and geopolitics recently all play a direct fundamental role. And then as you alluded, there's also kind of a feedback effect. When volatility rises for fundamental reasons, dealers are going to struggle with providing sufficient liquidity to the market. And to the extent that dealer balance sheets are not sufficiently flexible to accommodate the provision of liquidity to the market, that in and of itself increases illiquidity, increases volatility, and they kind of feed back on themselves.

7:43And you can get an episode like we had when COVID hit in March 2020, where liquidity becomes even worse than would be suggested by volatility alone, much worse. And that's just a sign that the market is not capable of intermediating those extreme demands for liquidity. Right. So this seems to be a distinct characteristic of treasuries in particular, which is when stuff starts to go really wrong, like it did in March of 2020, people often sell the safest stuff first, which means they sell treasuries. So you get this big wave of selling at precisely the moment that a lot of dealers want to wind down or back away from risk.

8:28Is that just a fundamental tension in the market? Is that always going to be the case? As long as US treasuries are the world's most important safe haven, which is clearly the case by miles, that's always going to be the result for basically two reasons. Number one, a whole lot of major investors like foreign exchange reserve managers, firms that are storing a safe liquid asset just in case. Well, the just in case happened and they are going to liquidate those positions. The other channel for this is as the volatility grows and uncertainty grows, a lot of investors are kind of finding it too hot to handle and they have to unload some risk.

9:11And treasuries are the easiest security to unload in the world. The market's got a good reputation for being the deepest and most liquid market in the world. So I know we've talked about this a little bit in the past with a few different guests, including Josh Younger. He made the point on one of the episodes that we did, which is like, if you're thinking about like, what do we want to do to have better treasury market structure, that it doesn't necessarily make sense to optimize for, well, we never want to have a March 2020 again. Because you don't necessarily want to have optimized for the one out of every 100-year pandemic.

9:47But what would you say is the goal? If you're like, OK, there does seem to be this relationship between volatility and illiquidity. It does seem like some of these bouts of volatility and illiquidity become more frequent. If you think about designing sort of an optimal market structure for treasuries, what would you say we're trying to achieve? Josh, I've known him since he worked at JP Morgan. And now that he's moved to the Fed, we get to talk a lot more. This is a terrific insight that he has. Do you really want to design a market for the worst day in 1 ,000? Isn't that very expensive and maybe overdoing it because 999 days out of 1 ,000, you didn't really need that kind of a market structure.

10:28I'm going to be a little provocative here. I think you do want to build a market for the worst day in 1 ,000 for the following reason. If I'm, let's say, managing the foreign exchange reserves of an emerging market central bank, when do I need to actually take advantage of the depth and liquidity of the US treasury market? It's that one day in a thousand when all the other safe haven investors are trying to do the same thing. In the paper that Tracy mentioned, I'm giving here at Jackson Hole, I talk about this wrong way risk from the viewpoint of illiquidity. You don't want the market to be great except on that very singular day on which everybody needs the liquidity.

11:10Why not? Well, because A, this is the linchpin of global financial market stability. You want it to work day in, day out. And B, if you discourage safe haven investors from believing that even though everybody else is liquidating that day, they could also liquidate at low cost with ease, then they won't use the US treasury security as much as their safe haven. They'll diversify. And that's what we've been seeing somewhat over the last couple of decades, a degree of diversification away from the US treasury, still by far the dominant safe haven, something like 59 % of foreign exchange reserves are held in US treasuries.

11:49But from the viewpoint of US taxpayer, You want everyone to believe that on the worst day in a thousand, that market is going to be there for them. Wait, I'm going to be provocative now or try to. OK, so on this note, we did see in March 2020, the Fed unveiled all these different emergency programs aimed at supporting the U.S. Treasury market. So, you know, we have the new standing repo facility. I think, you know, they were buying U.S. Treasuries in exchange for reserves. And then they exempted all of those from the supplementary leverage ratios for banks. So it seems like the backstop is in place.

12:26If something bad were to happen again, I would presume that the Fed would unveil either those exact measures again or something very similar. So is the issue solved? Yeah. Tracy, that is provocative. So I would definitely say - That's a polite way of saying you're so wrong. No. The Fed came out guns blazing. Unlimited financing in the repo market for anyone that had access to the Fed. A trillion dollars of purchases in the first three weeks, nearly a trillion, of US treasuries relieving dealer balance sheets of their overloading. Getting the supplementary leverage ratio dialed back and it was causing problems took a little And it took, I think, some backroom negotiations with the other bank regulators to come on board.

13:14So that got delayed and that was a problem. But the Fed did a terrific job at crisis management during those weeks. And I say weeks because they didn't solve the problem. They only made it less bad than it otherwise would have been. It took five, six, seven, eight weeks before market liquidity was restored. And again, going back to my wrong way risk point, if I'm looking for a market that's going to work for me in a crisis, I don't want to have to wait weeks in order to get liquidity or to pay a low cost for liquidity. I want it to be working all the time. Now, of course, it's unrealistic that it should work every single day.

13:55But if we rely only on central banks, and I speak more broadly, to bail out their government securities market when they get into trouble, it's not going to be 100 % effective. And it raises moral hazard. It says to the rest of the world, we'll use the central bank balance sheet to bail you out. You don't need to focus on improving market structure, reducing undue leverage. We have your backs. That message, while it needs to be there, is not a substitute for improving the market structure. Yeah, you sort of anticipated my next question, but okay, it does seem like in an emergency, the Fed can say, we're going to buy, do QE at a scale that we've never seen before.

14:37And as Tracy mentioned, unveil these new facilities kind of on the fly. And it seems like basically since 2008, 2009, the Fed has gotten really good at standing up new facilities very quickly. So that's like a skill that they've developed. But can you talk a little bit more about what you perceive as to be the cost of a sort of stability regime that sort of presumes that, yeah, we know there's some frailties, but our solution is that in that seven sigma or 12 sigma a day that the Fed is there. And talk more about why that's not a good system. Okay. Well, just I want to reemphasize that the Fed - I don't know how many sigmas it really is.

15:16Is that once every billion years? I don't know. I just throw out a number. The Fed does need to be there. It's not as though one should say, let's take the Fed's balance sheet out of the equation and try to do without it. It needs to be there. It's a backstop. It's the last resort. The Fed is the buyer of last resort. After it's become the lender of last resort, it can't do anything else but bail out the market by buying securities. But relying on that has several problems. I already mentioned, it's not 100 % effective on the first day. And there's also the size of the Fed's balance sheet. That's controversial.

15:51I mean, even if you think it's innocuous, it raises political concerns. There are those that say, well, maybe the Fed's balance sheet is too big and we need to curtail the ability of central banks, including the Fed, to expand their balance sheets to the extent that they have been. And they've been using them very, very liberally over the past couple of decades. The other concern is once the balance sheet is large, it eventually is going to come back down and those treasuries are going to be adding to the stock of securities that other investors need to have. And it means that the central banks, including the Fed, need to do that very gingerly.

16:27There's a lot of volatility in the treasury market and the Fed is letting its balance sheet come down. Other investors are having to pick up the load. It's easier to expand the balance sheet than it is to bring it down. So using the Fed's balance sheet while it's necessary is not a painless solution. And I would argue it's not the best solution anyway. We can do better by improving market structure, pushing out into the extreme tails the number of events in which the Fed needs to step in and buy. Silicon Valley is selling you a future where you're obsolete or worse, identical. At Palantir, they're witnessing something different and revolutionary.

17:06From re-industrializing the nation's defense base to shipyard workers building faster and frontline workers boosting productivity, AI is transforming work across the nation. AI is not replacing American workers or flattening them into conformity. It's unleashing what makes each one irreplaceable. Their judgment, their craft, their creativity. When American workers become more powerfully themselves, they own the future. Palantir, making Americans irreplaceable.

18:04Terms apply. Guarantee applies to base monthly rate and stated discounts only. Add-on prices additional. Offers in January 5th, 2026. Why don't we talk about the market structure? And maybe before we start talking about improvements that could be made, could you give us the sort of lay of the land when it comes to how treasuries are traded today? So there's primary dealers for the new issuance, and then there's your sort of run-of-the-mill dealers for secondary market trades. But talk to us how it works right now. Terrific. Well, it's an extremely complex structure, but it can be summarized pretty simply.

18:35There's two segments of the market. There's the inter-dealer market in which the dealers trade among themselves. And then there's the customer to dealer market in which investors around the world trade with dealers. Notably, investors do not trade directly with other investors. There is no all-to-all trade in the US treasury market. No matter whether you're an insurance company, a hedge fund, a foreign exchange reserve manager, you are going to be buying and selling with a dealer. If you're a dealer, on the other hand, there is a very active interdealer market for the on-the-run securities. Those are the latest issues of the treasury.

19:11There's an order book market, which is a high-frequency trading market run by Brokertech, which is a subsidiary of the Chicago Mercantile Exchange, where you have the same kinds of high-frequency trading that you see in the stock market. The only other participants on the broker tech market are high frequency trading firms, sometimes called principal trading firms, like Jump, like DRW, like Citadel, firms that have a very special purpose of intermediating in the inter-dealer market, taking little bid offer spreads from the dealers and from each other. That's the basic structure of the market. Again, the notable feature is if you're an investor, you can trade only with a dealer.

19:52If you're a dealer, you have the ability to lay off positions in the inter-dealer market. Again, before we get into sort of like optimal structure, I'm actually just curious, you mentioned that you spent the last year at the New York Fed looking into this and sort of getting this. What did you do? How did you go about your research? How much is it sort of like a sort of statistical-based analysis versus how much was it talking to dealers and understanding how they operate? I'd just be curious about the research process. For this particular project, it was a combination of meeting and discussing what needed to be done with the economists that I mentioned earlier.

20:28And those would be weekly meetings, pretty in depth, where we would go through what we've already learned and what we need to do next. And that happened for six months or so. And at the same time in the background, we're collecting volumes of statistical data. The Fed, because it's a member of the official sector has access not only to its own data, but to exceptionally fine-grained data at the transactions level. Let me give you one example. There is a data set called Trace, which records every single trade in the treasury market with a few minor exceptions. Those data are only available to the official sector.

21:10They're not available to the public. And by the way, I disagree with that policy and we could talk about that. I think actually contributes to the problem of illiquidity. But in any case, the Fed, as a member of that official sector group, can go to its sister agencies in the federal government and say, look, we have this project. Here's its objectives. We want to use these trace data to analyze liquidity in the US treasury market. And then we get feedback saying, yeah, this looks good. The way that data are being presented will not reveal proprietary information. So go ahead. And then we can do the same thing with dealer balance sheet data.

21:49We can get exposures of the dealers not only to treasury securities, but to agency mortgage-backed securities, which turned out to be another big load on their balance sheet, particularly during March of 2020. we can go to a wide range of data sets. And we wrote a paper that explains the extent to which we access all of these data, bring them together. We developed 18 different liquidity metrics and many different metrics on how dealer balance sheets are being loaded. And then we would analyze these using reasonably intricate econometric methods like quantile regressions and a number of other statistical approaches.

22:34And then we would start to see the patterns emerge very, very clearly that I described. Two key patterns that came up over and over again in our discussion meetings were A, volatility seems to explain most of the variation in liquidity, but B, when it doesn't, it's dealer balance sheet loading that explains the remaining part of illiquidity. It's a highly nonlinear effect. when dealer balance sheets are normally loaded, they don't contribute to illiquidity. But when they're reaching their extremes, where dealers are handling more treasury trades and more agency MBS trades than they've handled in the past, then you see illiquidity go up well beyond the level predicted by volatility.

23:19So after analyzing all these data and discussing what's driving these, then we turn to writing up our results. And there's a lot of iterative work there. which you can see in the paper that we wrote. So you can see the dealer balance sheets on a daily basis, not just at quarter end? Not quite. We can only see dealer balance sheets on a weekly basis because the Fed has a data set called FR2004, which collects those data only on a weekly basis. And summaries of those data are available publicly on the New York Fed's website. So going back to this dealer balance sheet issue, I mean, this is something that has come up basically ever since the 2008 financial crisis.

24:01And there have been a lot of complaints from the dealers about all this new regulation that limits their ability to take risk on their balance sheet. And the argument for doing that has always been one of financial stability. Well, we want the banks to be safer. And if they have to cut back on their intermediation capacity in the market, maybe that's a fair trade. How do you thread the needle between those two issues, especially in a market as important as treasuries? Okay. It's very tough because those much more demanding capital requirements and other requirements that came in after the financial crisis have clearly reduced liquidity in a broad set of financial markets.

24:44It's glaringly obvious. However, we can't afford to return to the pre-Leman days in which dealers would expand And there are balance sheets for a few basis points of arbitrage creating financial instability. So while those new capital requirements are necessary for protecting the economy from collapse of the financial services sector, we do need to substitute for the liquidity that's missing in other ways. There is one capital regulation that I think is not necessary. And that's the one you mentioned, Tracy, the supplementary leverage ratio. That rule penalizes the provision of liquidity even for very safe assets.

25:31Let me give you an example. When the Fed was buying treasury securities from mid-March, it bought within three weeks nearly a trillion dollars of treasuries. And one might think, oh, thank goodness, that's making more space on dealer balance sheets for other positions. However, from the viewpoint of that capital regulation, there was really not much change at all because the Fed paid for those trillion of treasuries with a trillion of reserve balances. And reserve balances, although perfectly safe and liquid, have the same impact on dealer capital requirements as the treasury securities that they replaced.

26:10So there wasn't really, from the viewpoint of the supplementary leverage ratio, much benefit of the treasury's purchases. there were benefits in other respects because treasuries are risky and dealers were relieved of that risk by the Fed's trades. But from the viewpoint of that supplementary leverage ratio, it was very unfortunate. And I and others have argued that the SLR, supplementary leverage ratio rule, should be replaced with higher risk-based capital requirements. Sorry, can you explain that when you say replaced with higher risk-based capital requirements? The capital wouldn't be calculated on the basis of the size of your total balance sheet, but on the riskiness, the actual makeup of the balance sheet.