Why Financial Crises Recur

Why do financial crises recur with the proliferation of every new type of runnable short-term debt? In the 18th and 19th centuries, the short-term debt at issue were “banknotes” that circulated as a medium of exchange. Then came bank runs on account-based “demand deposits” during the 20th century. In this century, we have witnessed runs on “money market funds” and “repos” (and demand deposits still). And, during the past half decade, we have seen repeated runs on circulating digital short-term debt known as “stablecoins.”

We explore this question in our new article, “Why Financial Crises Recur.” To be sure, we are not the first to inquire. In the late 19th century, William Graham Sumner lamented that the literature on financial crises “for fifty years had repeated the same inferences, lessons, and warning; but all the doctrines of currency have to be learned over again apparently every ten or fifteen years, if indeed they are ever learned at all.” Over the past two decades, prominent legal scholars have offered explanations rooted in a flawed legislative process and a fragmented regulatory process. We do not disagree with these explanations but given what has occurred over the past few years—in the aftermath of runs on stablecoin issuers, crypto banks, and Silicon Valley Bank—we believe there is still a lack of substantive understanding with respect to the fundamentals of money creation and bank runs.

At a high level, we argue that generation after generation of lawmakers tend to make two (admittedly intuitive) errors when crafting financial regulation. The first error is overemphasizing the importance of transparency and underappreciating the role of opacity. Here, opacity refers to the difficulty of valuing assets, which then implies that liabilities cannot be easily valued.

To make short-term debt effective as money, its price should not fluctuate—that is, the price of short-term debt should not change. A one-dollar banknote should always hold its price at one dollar; withdrawals from a bank account should always be one-for-one. But how is that possible? Based on a simple understanding of supply and demand, we know that the price of a good or service moves in response to new information. For example, the price of mangoes rises if the market learns of a mango shortage caused by drought. Or the price of gold falls when market participants learn of the discovery of new gold mines. So, how is it possible for runnable short-term debt—a product whose value should, in theory, fluctuate along with the value of the assets backing the debt—to trade at a constant price?

The answer is that no one should have an incentive to produce information about the backing for the money (i.e., the banks’ assets); and everyone should know that no one has an incentive to produce that information. If that is the case, then we say that the money is “information insensitive.” And achieving information insensitivity requires a level of opacity that is not commonly accepted in a market economy. We are not, to be crystal clear, arguing that opacity by itself is sufficient to make short-term debt stable. Opacity must be paired, and has been paired, with a credible supervisory regime. (We provide a detailed discussion of this in our article, with analysis on the origins of bank examinations, the role of clearinghouses, and the bank examination privilege.) But opacity has been repeatedly underappreciated and underutilized.

The second error is believing that strengthening individual financial entities is sufficient to guard against system-wide risk. Here, a brief discussion of impending stablecoin legislation is instructive. (In our article, we also discuss this second error in the context of double liability regimes for banks prior to 1933 as well as the pivot to bank capital regulation in the late 1980s.)

Successive attempts at regulating stablecoins have correctly identified the need to improve the quality of the stablecoins’ backing assets. Doing so would improve the resiliency of stablecoin issuers. But underlying these attempts is a belief that system-wide runs would be eliminated altogether. That is not the case. Improving the resiliency of an individual stablecoin issuer is not sufficient to ameliorate systemic risk.

Suppose each stablecoin issuer holds only Treasuries in its portfolio. Viewed in isolation, holding Treasuries is an excellent choice for the issuer, but play out the macroeconomic shock that hit Silicon Valley Bank in 2023.  If interest rates rise, then the value of the Treasuries—the assets meant to back the stablecoin’s peg—will fall.

If this happens to all stablecoin issuers, then they will seek to buy new Treasuries that are of higher value. As bond prices fall, issuers will need to buy more Treasuries. When interest rates are sufficiently high and the value of Treasuries are sufficiently low, holders of stablecoins will have to decide if their issuer has the cash to buy more Treasuries. At that point, holders of stablecoins might ask questions about what is in the rest of their issuer’s portfolio and if the issuer is able to hedge any risk. Once the “No Questions Asked” condition is violated, holders of stablecoins will begin to run.

Rising interest rates is only one scenario where a common factor could affect all stablecoin issuers. There are others. Perhaps money market funds are misvalued and stablecoin holders do not know which issuers are most affected. Or perhaps there is a shortage of safe assets, and other firms have a demand for the exact same securities. In short, attempts to regulate stablecoins, like the GENIUS Act, may be able to guard against idiosyncratic risk, but it cannot prevent systemic risk. As we have previously written, the way to truly combat the systemic risk created by stablecoins is to create a central bank digital currency—that is, digital money issued by the central bank or backed fully by central bank reserves.

In sum, there have been bank runs and financial crises for at least two hundred years. To many, it may seem that crises are inevitable, and nothing can be done about them since their causes are idiosyncratic. We do not believe that is the case. Crises can be prevented with the right regulatory interventions. But doing so would require a course correction along two fronts—a course correction that is likely to strike many as counterintuitive.