Table of Contents
Darrell Duffie is the Adams Distinguished Professor of Management, Emeritus, and Professor of Finance, Emeritus, at the Stanford Graduate School of Business. Trained as a civil engineer, Duffie has established himself as one of the world’s foremost authorities on the inner workings of financial markets. He chaired the Financial Stability Board group that guided the global transition away from LIBOR, served as president of the American Finance Association, and wrote the seminal textbook Dynamic Asset Pricing Theory. He earned his Ph.D. from Stanford in 1984 and has been on the faculty ever since.
This conversation follows Professor Duffie as he examines the cyber threat to America’s largest banks, the limits of the country’s fiscal space, and the financial system he would design if he started from scratch. Finally, Duffie considers what AI means for the next generation of economists and reflects on the quiet reward of hearing from former students years later.
The following interview was lightly edited for length and clarity.
Jack Murawczyk: In 2012, you acknowledged that the financial crisis caught almost all economists by surprise, including yourself. What do you think economists might be failing to see today?
Darrell Duffie: The big economic issues of the day are plain to see for everybody. So I don’t think we’re missing the big ones – like sovereign debt risk and geopolitics and artificial intelligence or climate change.
What might not be quite as obvious is the threat of cyber risk to the financial system. I worked a few years ago with Josh Younger on the exposure of the banking system to cyber risk and what might happen if there were a successful cyber attack on one or more big banks.
Subsequent work by others, like Thomas Eisenbach, Anna Kovner, and Michael Lee at the New York Fed, shows that it would only take one or two very large banks to suffer a cyber attack to cause a major crater on the whole U.S. financial system. Of course, policymakers and academics are kind of thinking about that in the background, but I don’t think it’s as present in their minds as a very significant risk to the economy as it might be.
Now, if you talk to central bankers, they’re worried witless about cyber risk, and they think about it all the time. Same thing with CEOs and boards of directors of large operating companies. I know that they devote an enormous amount of time – that’s one of their primary concerns from the risk management perspective. Academics are aware of this, but I don’t think they’re as threatened by it as I am.
JM: You’ve warned that without fiscal correction, the United States would eventually end up in a debt crisis. The U.S. hasn’t experienced a classic sovereign financing crisis in the modern era, and unlike countries that have defaulted on foreign-currency debt, it borrows in its own currency. What would an American debt crisis actually look like?
DD: Well, some would say we’re already in a debt crisis, with over 100% debt-to-GDP and a very large fraction of our government’s expenditures merely paying interest on the debt. But in the sense of a classic debt crisis – where the government’s having trouble rolling over its debt, auctions are failing, yields are skyrocketing – the United States has a lot of room to go. It’s called fiscal space. The U.S. could in principle get much, much deeper into debt.
So that’s good news and bad news. It’s good news in the sense that there’s probably quite a lot more give in markets to accept Treasury debt, and the U.S. could manage to dig a much deeper hole than it already has dug. That’s also the bad news, of course, because it means that someone – future generations, perhaps our grandchildren – will have to deal with that.
JM: How much fiscal space do we have before the bond vigilantes start knocking? What would the indications be?
DD: Well, nobody knows for sure, because Japan has double the debt-to-GDP of the U.S., and it’s been pronounced to be on death’s door for decades, and it hasn’t failed.
It’s partly a confidence game, related to when international bond investors lose faith in U.S. Treasury securities. And it can be a self-fulfilling prophecy. If bond investors think that other bond investors won’t be willing to roll over the debt, then they won’t either. In principle, it could happen at almost any time.
As for the evidence, you’ll see yields climb sharply. What we’ve seen in the last few months is nothing compared to what we could see. We could see auction fails: the government is unable to get the entire issue absorbed by investors. And you would see a lot more anxiety in Congress about how to reduce deficits. The trade-offs could cause tremendous stress on society, just as you’re seeing today in France: demonstrations, strikes, protests, bankruptcies, city and state governments getting into deep trouble, and maybe some of them not getting bailed out by the federal government.
But those disasters are far away. There are plenty of opportunities for the government to get its fiscal act together. The problem is that politicians react to what voters do, and voters tend to take a long time to recognize these risks.
JM: Still, when you look at the convenience yield, the premium investors pay for the safety and liquidity of Treasuries over corporate bonds or German Bunds, it seems to have shrunk.
DD: It’s disappeared.
JM: Is that not an indication of some form of debt crisis?
DD: The so-called exorbitant privilege has disappeared, which means the U.S. is no longer getting the special deals it used to from scarcity in the supply of Treasuries relative to the demand for safety and liquidity. But that doesn’t mean Treasuries are trashed or that we’re in a crisis. Yields could go much higher. And a lot of angst will be expressed before either the bond market breaks or the government does something about it.
JM: What is driving the yield spike?
DD: Imagine yourself as an investor who gets a call from the Treasury Secretary saying, “I’m the government’s chief bond salesman. I see that your hedge fund has $10 billion of the 10-year Treasuries. I think you should take another $5 billion.”
Let’s suppose that the Treasury Secretary has convinced you that inflation risk is well under control and there’s no chance of a sovereign default. And you say to the Treasury Secretary, “I need more, Mr. Secretary. I picked $10 billion as how much I want at a 5.2% yield. If I were to go to $15 billion, it would only be because you’re offering a higher yield.”
So from that conversation, you can see that supply really is a very dominant factor in all of this. I’m not saying there’s no inflation or there’s no sovereign debt risk, but the supply of bonds on the market is a huge driver of yields.
JM: The Treasury still carries its gold at $42.22 an ounce, a price set by statute in 1973. Marking it to market would hand the Treasury roughly $1 trillion overnight. Could you walk us through the financial plumbing if that happens? Is this clever fiscal policy or monetization in disguise?
DD: Well, I’m not sure exactly why the government put a statutory price on gold, but they could change that tomorrow if they wanted to. Maybe it’s the safety valve: the debt cap gets hit, and then they say, “Well, let’s pass a new law saying we’ll mark that gold to market.” But honestly, I don’t know why Congress wouldn’t do that, and why in the first place it wants that gold marked at a fixed price.
It would reduce the debt, but not materially. It’s the primary surpluses that count: the amount of money that the government takes in minus what it actually spends other than on interest expense. That’s what really matters.
JM: Apollo’s Torsten Slok has raised the prospect of an “agentic bank run,” in which AI agents constantly move household cash to wherever it earns the most. Much of the banking industry assumes deposit inertia. How seriously should we take the idea of financial instability from AI agents, in banking and beyond?
DD: About a year ago, I told a banking industry group that AI could help depositors sweep their funds out of low-interest bank accounts and into other banks, money market funds, or other investments earning higher interest. Eyebrows went up around the room.
I think this is a significant threat over the long term, not of bank runs, but to banks’ bottom line. Net interest margins will go down.
Will it be big? Not for a long time. American depositors have already proven that they’re pretty sleepy at making these kinds of decisions. And agentic commerce isn’t close to being sufficiently resilient to cyber and impersonation risk, so the safeguards around it are going to be extremely carefully done. But it adds to the pressure on banks. Some people think stablecoins are a threat. I don’t think they’re much of a threat at all. But I do think neobanks will give conventional banks a run for their money.
JM: Why not stablecoins?
DD: There are no killer applications in the U.S. domestic economy. People aren’t going to buy them as an investment, and if they’re only buying them so they can move them easily, banks aren’t going to be threatened by that. Merchants need to be in a position to accept them for payment. And there’s a lot of concern that they don’t pay any interest.
JM: Do you think that’s permanent, or is it purely a function of the GENIUS Act?
DD: It’ll be temporary. It’s not really as much an issue of the GENIUS Act – which doesn’t help, by the way, because it rules out the payment of interest. But it’s more of an issue of technology. It’s not offering much in domestic payment systems that can’t be offered by banks. It starts to look a little bit more convenient for certain applications, so banks can meet that need by improving their technology. So that’s really the main value of stablecoins: poking the fork into the banks to get them to move faster on innovation and convenience.
JM: And internationally?
DD: That’s probably a different story. There’s lots of research showing that cross-border payments of stablecoins, especially in emerging markets, are a useful innovation. And in economies with volatile currencies or untrustworthy banking systems, they’re a good substitute for local currency deposits.
JM: If you could wave a magic wand to design the global financial system from scratch, what would you eliminate? Would we still have banks, central clearinghouses, stock exchanges, and the same monetary arrangements?
DD: Are you familiar with Pix? It’s a payment system used in Brazil: an app everyone has on their phone and desktop that allows them to pay each other without necessarily engaging with the bank, even though the money they’re moving is their bank deposits.
I can see that happening in the United States in the future. If I could, as you say, wave a magic wand, that would be one of the first things I would do: give every American an app that would allow them to pay or receive money from anyone else in the economy instantly. We don’t have that today. The back end for it is called FedNow, so that’s a service banks could provide to their customers.
That would be a major improvement on the retail side. And it would be much less regressive for lower-income Americans who don’t have credit cards and aren’t getting the 4x miles or the 2% cash back. It would democratize access to the payment system in a way that would help all Americans.
On the wholesale side, we’d be able to do all kinds of programmable financial trading. Take portfolio trading: I want to get certain elements of a portfolio from one market participant and the rest from others. But I don’t want to buy one batch from you and then be over a barrel with the next seller. I’d rather conduct all of these trades together, all or none, in one go. That’s called smart contracting. In my recent work with Chaojun Wang, an economist at the Wharton School, we worked out the theory supporting that approach, which could be available today if securities markets were on programmable ledgers.
The Depository Trust Company, which holds securities in custody for most Americans, records investors’ claims on a conventional ledger. So if you say you own so many shares of Apple, what you really have is a claim to shares held in custody, maybe multiple levels down, at DTC. Starting this month, the DTC can make that ledger programmable for you. You can say, “I want my shares of Apple held in programmable form.” Apple doesn’t need to do anything different. But now you can write a contract with another investor, a broker-dealer or an investment fund to buy or sell shares 24/7, much more efficiently than today.
Although I have to say, American securities markets are already quite effective. It’s not as if this is a desperate need for the economy.
JM: What are your thoughts on building a regulated market for equity perpetual futures?
DD: I think perps are terrific. They don’t have to be provided on Kalshi or Polymarket; they can be provided through normal derivatives exchanges. Although the Chicago Mercantile Exchange seems to think that they don’t belong on one. I disagree.
One argument is that they’re not a proper futures contract. But there’s nothing in the Commodity Exchange Act that says they’re not. For certain types of hedging or speculation, they’re a very natural product. And I think you should allow innovation, properly regulated, in financial markets. U.S. securities markets are very effective at allowing innovation, so I expect perps will come into play. I don’t think they should be on prediction markets, though. That’s not a natural role.
So I’m all in favor. Now, whether there will be a huge demand to trade those, I don’t know, but I don’t see any reason why they wouldn’t be a natural element of our modern securities and derivatives markets.
JM: Kalshi and Polymarket trade contracts on Jerome Powell’s delivery of “good afternoon,” Federal Reserve interest rate decisions, and CPI prints. You’ve spent a career on market design. Are these event contracts a real financial market, a form of legalized gambling, or something in between?
DD: Well, they combine a lot of motives. Sports gambling is still the majority of position-taking in those markets. As you know, there are several district court cases over whether state gambling authorities should regulate Kalshi, or whether it should be regulated by the CFTC, which is traditionally a derivatives and commodities regulator. So far, the cases seem to be going against Kalshi. They prefer to be regulated by the CFTC, the courts have not gone along with that, and I suspect that’s the way it’s going to end up. They’ll be able to provide sports event contracts regulated by the state, and if they provide commodities contracts, they might have two regulators.
The CFTC could regulate contracts for delivery of commodities or financial instruments, like those traded on the Chicago Mercantile Exchange today: foreign exchange, stock market indices, interest rates. Single-stock futures or single-stock perps might take more work. They should be under SEC regulation, and I think the courts would also push in that direction.
So, to summarize: if prediction markets want to innovate, that’s terrific. Political or economic events are a potentially important source of information. And if they want to push further into financial markets, they’re going to have multiple regulators.
JM: I noticed that in your recent paper on a liquidity-saving payment mechanism, you and your coauthors acknowledge using ChatGPT to help demonstrate mathematical results. Do you see AI fundamentally transforming how economics and mathematical research are conducted?
DD: Yeah. I was just having lunch with a young economist today, describing my concern about early-career economists making their way in research and teaching with the advent of AI, because it is transformative. It’s going to change the way economic research is done, and the way we teach.
I’m toward the end of my career. I’m still active in research, and I’m happy to use AI as a tool. But if you’re a young economist and most of what you do is problem solving, quantitative and empirical analysis, gathering data sets and hypothesis testing, you’re going to get run over.
The path to career success in economics is now going to favor economists who are imagining new questions to ask, and new worlds in which we may all be living in the future. It’s going to be very difficult for a lot of economists to adjust to that.
In many ways, this is going to increase the quality of education, but changing our teaching methods is going to be a lot more work and a big hurdle to get over.
I can’t overemphasize how transformative this is for university education.
JM: To conclude, what’s the kindest thing someone has done for you?
DD: It’s happened many times that a student, years after leaving Stanford, sends me a note right out of the blue saying, “I just want you to know that I’m using what you taught me every day in my work, and it has been valuable.” That is just very special.
