Cover photo

What is a Bank Run

User avatar


This year's Nobel Prize in Economics goes to Ben Bernanke, Douglas Diamond, and Philip Dybvig for their research on financial crises– what causes them, how they were developed, and what might prevent them. This caught my interest as an observer of the crypto market. How can we incorporate findings from traditional finance to improve the resiliency of crypto & DeFi?

The prize was announced at an apt time: We've seen central banks print trillions of dollars during the pandemic, rising worldwide inflation, ensuing tighten monetary policy, and a spectacular crash of UST, Celcius & co. And in the past few days, we also observed a huge bank run on FTX's liquidity.

Why banks exist

Banks solve a key problem in modern society: the need for a large liquid market between depositors and borrowers. Households want a place to park their money, in which they can withdraw their money whenever at a short notice. Meanwhile, investment projects have financing needs to become profitable in the long term.

However, there is a fundamental conflict between the interests of the depositors and borrowers. Depositors want to have their savings instantly available at any unexpected time, but the borrowers who take out a loan to finance an investment do not want to be forced to repay their loans at unexpected times, as this drastically reduces the profit they could make in the long term.

So banks arose as the intermediaries. It offers depositors instant access to their money when they want and lends out money to long-term projects. This process is called maturity transformation: Banks transform loans with long maturity into liabilities (bank deposits) with short maturity.

What is a bank run?

Bank run is depositors rushing to withdraw their deposits because they expect the bank to fail. Essentially, there are five stages to a run.

  1. Rumors about a bank's survival began circulating

  2. Depositors rush to withdraw their savings

  3. If enough people do this simultaneously, the bank's short-term reserves cannot cover all the withdrawals

  4. The bank conducts a fire sale of its assets at a potentially huge loss

  5. The bank runs out of assets, it declares bankruptcy

It need not be anything fundamental about the bank's condition. The problem is that once they have deposited, anything that causes thern to anticipate a run will lead to a run (Diamond and Dybvig, 1983).

Bank runs could happen to healthy banks as well. It could be a rumor about insufficient funds, a bad earnings report, a run at another bank, or even negative macro forecasts. Once depositors become concerned they may not get their money back, they will rush to withdraw their deposits.

When there are more withdrawals than the bank can cope with, the bank is forced to perform another "maturity transformation"– by terminating its illiquid loans early and conducting fire sales of its assets to generate liquidity.

The system that is designed to help banks generate money (the yields from the future expected cash flow of loans given out) is also what makes it vulnerable.

The catastrophic consequences of a bank run

Prior to Bernanke's research, the conventional view was that depression was caused by a shortage of money, and could be solved by printing more of it.

Through analysis of The Great Depression (1929-1941), the deepest and longest downturn in the history of the modern industrial economy, Bernanke showed that depression is mainly caused by bank collapses and the reduced supply of credit.

Because banks are crucial for connecting savers to borrowers, when banks are gone, so did their nontrivial relationships with borrowers. It takes time to rebuild screening, monitoring, and accounting knowledge of good & bad borrowers. Consequently, borrowers found credit to be expensive or unavailable, which depressed economic activity and contributed to a longer downturn.

Fear of runs led to large withdrawals of deposits, precautionary increases in reserve—deposit ratios, and an increased desire by banks for very liquid or re-discountable assets. These factors, plus the actual failures, forced a contraction of the banking system's role in the intermediation of credit (Bernanke 1983).

Risk management for crypto

Diamond and Dybvig propose one solution to bank runs: having government-guaranteed deposit insurance. In theory, this stops a run before it starts as depositors no longer need to rush to withdraw funds. In practice, once people anticipate a bank run, a run will happen. Hence, we need an insurance/backstop functionality.

Protocols must set aside a pool of money in the form of bank-run insurance, where vaults could be insured against a loss, and can be used to repay users in catastrophic events.


In traditional finance, banks take on the role of monitoring borrowers. Individual depositors aren't expected to check if their money has been well managed because it would be too difficult to do so. But who monitors the bank?

The invention of blockchain challenges this opacity. We need proof of reserves/solvency.

Publicize proof of solvency in regards to the multi-sig or treasury every quarter, showing assets > liabilities, as audited by certified accountants.

To prevent bribery, the process of monitoring borrowers should be publicly viewable, transparent, and linked to on-chain data. This is to verify that the protocols are honestly representing their liabilities.

Long-Term Capital Management, ran by the brightest minds in finance, didn't have long-term capital. They had overnight capital. When the market went against their quant models, they suffered a big loss. They had to then liquidate their assets. Market knew this and went against them, so they completely blew up.

Hypothetically, Martin Green commented, "if they had matched their liability with their assets in terms of liquidity so they have some time to liquidate their assets, they probably could have survived and made the money they thought they could've made."


The Nobel winners added not putting all your eggs in one basket should be strictly adhered:

  • Grant loans to a large number of small borrowers. Even if a few defaults on their loans, the losses will be small and manageable

  • LP responsibly across various farms

  • Limit the amount of borrowing in the protocol. Reduce overleverage by making risky speculation sufficiently costly

Some last comments

Contrary to popular opinion, I'm actually quite glad crypto isn't as big as the traditional financial markets. There are no billions of users who would've been exposed to LUNA, 3AC, Celcius, and FTX. No mass liquidation. No widespread housing crash. That would have triggered another period of Global Financial Crisis.


Press release: The Prize in Economic Sciences 2022


Diamond, Douglas and Philip Dybvig, "Bank Runs, Deposit Insurance, and Liquidity", mimeo, 1981.


  • Loading comments...