Markets

Bank runs aren't madness: This model explained why

By Anil K Kashyap     
June 17, 2014

From: Magazine

Run on East Side Bank, New York, February 16, 1912. Photo by Bain News Service via Library of Congress.

Douglas W. Diamond's research agenda for the past 30 years has been to help us understand what banks do, why they do it in the way that they do, and what the consequences of these arrangements are. A few of the questions that Diamond, Merton H. Miller Distinguished Service Professor of Finance, has addressed include: Why does it make sense for banks to offer deposits that make them vulnerable to runs? Why do they specialize in extending certain types of loans? Why do deposit-taking and loan-making go together? How do banks exacerbate financial crises? And how might banks best be regulated? His work has greatly influenced the direction of academic research on intermediation, and has had considerable practical influence on policymakers wrestling with these issues.  

The first in a series of papers laying out his research agenda, “Bank Runs, Deposit Insurance, and Liquidity,” written with Philip H. Dybvig, appeared in the Journal of Political Economy in 1983. Banking research prior to this point was pretty primitive. Banks were perceived as organizations that facilitated certain transactions for their customers, not as intermediaries that performed a unique service.     

Diamond and Dybvig were pathbreaking when they proposed that banks specialized in creating liquid claims against illiquid assets. Banks, in the authors’ view, allowed their customers to hedge the need to access funds on short notice. In this interpretation, banks didn’t provide access to proprietary investment opportunities—rather they created value through the structure of the bank liabilities that they offered to their customers. Banks improved upon the outcomes that individual depositors could achieve by investing elsewhere, by recognizing that while each depositor may have uncertain liquidity demands, few would need money on short notice. By pooling many of the deposits, the banks essentially offered insurance—they gave each depositor the right to withdraw those deposits upon demand, while relying on the fact that few depositors would need to do so.   

The key idea is that customers are not sure how soon they might require their savings, and are willing to pay a bank to manage this risk. The banks provide this service in part by investing in illiquid assets that earn a high return—real estate, for example. 

The banks then offer depositors an interesting contract. Say you’re in a six-month certificate of deposit. If you’re patient and leave your money in for the full six months, you accept a less-attractive return than you could have gotten by aggressively investing on your own. If you’re impatient and withdraw early, the bank will penalize you by offering a lower interest rate that nevertheless beats what you would have gotten if you had conservatively invested your money yourself.

This liquidity insurance is attractive, but it comes with a dark side. If all depositors were to ask for their money back at the first opportunity, a bank could not possibly repay them all. An inevitable consequence of what banks do to help their customers, by offering a short-dated claim against a longer-dated asset, makes them vulnerable to the possibility of a bank run.  

In this model, pure panic can cause a run. If enough depositors become concerned that a bank may experience more withdrawals than it’s prepared for, it becomes rational for depositors to try to get their money back rather than wait and risk getting nothing after the bank’s other depositors have withdrawn their money. Illiquid assets such as real estate need not be risky, and patient depositors may not need their money right away. Yet those patient depositors might ask to withdraw their money, simply to avoid losing it altogether.    

In the event of a run, the economy suffers in various ways. Savers are harmed because some of them may not get repaid. The banks scrambling to repay depositors will liquidate investments at fire-sale prices.

But Diamond and Dybvig’s framework also offers some insights about regulatory options for controlling runs. Traditionally, banks dealt with doubts about their solvency by “suspending convertibility,” meaning that they simply prevented customers from withdrawing their savings. While this stops a run by brute force, it means that some customers who need to access their savings are unable to do so. The model captures the perceived pros and cons of suspending convertibility. 

The model also can be used to evaluate deposit insurance. In the model, the mismatch between assets (loans) and liabilities (deposits) is simply a liquidity issue, which deposit insurance can help address.

However, if the high-return asset comes with some fundamental risk, then there is a possibility that a loan the bank makes will default. Although deposit insurance makes depositors patient and less prone to runs, it can also incentivize a bank to take excessive risk, because the bank knows that a risky investment won’t spark a run.  

The basic structure of the model has become a platform on which hundreds of other banking-related models have been built. That is a testament to the fact that Diamond and Dybvig’s model appears to be simple, but can be extended to deliver insights about much more complicated environments.  

Their perspective on banking informs economists both about historical episodes and the recent global financial crisis. Since Milton Friedman and Anna Schwartz published their pioneering book A Monetary History of the United States, economists have understood that runs can lead to devastating consequences. Prior to Diamond and Dybvig, most discussions of runs sought to explain why they were so destructive, but there was no model explaining why banking institutions should exist, despite the risk of runs. 

Many accounts of the recent financial crisis point to the Diamond-Dybvig model to help make sense of it. For example, then–Federal Reserve Chairman Ben Bernanke cites it at the Federal Reserve Bank of Kansas City’s Annual Economic Symposium in August 2009, explaining, 

. . . the events of September and October [2008] also exhibited some features of a classic panic, of the type described by Bagehot and many others. A panic is a generalized run by providers of short-term funding to a set of financial institutions, possibly resulting in the failure of one or more of those institutions. The historically most familiar type of panic, which involves runs on banks by retail depositors, has been made largely obsolete by deposit insurance or guarantees and the associated government supervision of banks. But a panic is possible in any situation in which longer-term, illiquid assets are financed by short-term, liquid liabilities, and in which suppliers of short-term funding either lose confidence in the borrower or become worried that other short-term lenders may lose confidence.  Although, in a certain sense, a panic may be collectively irrational, it may be entirely rational at the individual level, as each market participant has a strong incentive to be among the first to the exit.

One remarkable feature of the crisis is how much liquidity transformation was taking place outside of the traditional commercial-banking system. Policy observers perhaps did not immediately appreciate this as it was happening. But as Bernanke points out, people soon understood the elements of the crisis, as outlined by the Diamond-Dybvig model. 

In the second half of 2007, the asset-backed commercial-paper market collapsed. In this financing structure, which had emerged to avoid regulation, banks created financing vehicles that bought assets such as mortgage-backed securities, financed by issuing commercial paper with a much shorter maturity (often less than one month). The banks collected fees for arranging the payments and avoided having to hold capital, as they would have had to do if they had kept the securities directly on their balance sheets. 

When investors began to fear that the underlying assets might be riskier than anticipated, they refused to renew the funding. Within six months, more than $400 billion that had been flowing through this market disappeared. It was exactly the kind of loss of confidence that translated into an effective run, as proposed by Diamond and Dybvig.  

The crisis also exposed the vulnerability of firms that relied on repurchase agreements to fund themselves. Bear Stearns, Lehman Brothers, and other investment banks lost access to funding when this market seized up. Diamond and Dybvig’s basic observation, that funding longer-term assets with shorter-term debt is always risky, was also exposed in this case.  

In the wake of the crisis, regulators have turned to mandating how much banks can finance illiquid assets with short-term liabilities. The Diamond-Dybvig model helps us understand how challenging this will be. Some of this financing activity may happen due to regulatory arbitrage, but some might also occur because depositors may prefer to have access to their money on short notice. Thus, limiting the amount of this type of funding comes with costs, and quantifying those costs will not be easy.  

The Diamond-Dybvig model is remarkable in how broadly it informs our thinking. Besides explaining what banks do (provide liquidity), the model improves our understanding of the recent global financial crisis, and provides guidance about potential regulatory alternatives. It is appropriately hailed as a seminal contribution.

Works cited

Ben Bernanke, “Reflections on a Year of Crisis,” Speech at the Federal Reserve Bank of Kansas City’s Annual Economic Symposium, August 2009.

Douglas W. Diamond and Philip H. Dybvig, “Bank Runs, Deposit Insurance, and Liquidity,” Journal of Political Economy, June 1983.

Milton Friedman and Anna Jacobson Schwartz, A Monetary History of the United States, 1867–1960, Princeton: Princeton University Press, 1971.