The flaw in the 'bank run panic' narrative is not that panic is irrelevant. It is that panic is a symptom, not a cause. A recent working paper from the New York Fed, 'Depositor Behavior and the Role of Fundamentals', asserts that financial institution health—not irrational depositor fear—determines the likelihood of a run. The researchers analyzed granular deposit-level data from a major U.S. bank and found that only institutions with underlying balance sheet weaknesses experienced significant outflows during stress. Panic amplified the run; it did not originate it.
For anyone who has audited a DeFi protocol or a stablecoin in the last bull run, this finding reads less like a revelation and more like a belated admission of something we in crypto have known since the collapse of TerraUSD. The Fed is essentially saying: 'We have data showing that depositors are rational. They run when the bank is sick, not when the market is loud.' But the crypto industry has been operating under the opposite assumption for years—that runs are caused by FUD, by Twitter mobs, by coordinated attacks. The new research dismantles that excuse. It forces us to ask: if the Fed can prove this for opaque, fractional-reserve banks, why do so many crypto projects still treat health as an afterthought?
Context: The study, authored by researchers at the Federal Reserve Bank of New York, examines the 2023 regional banking crisis in the United States, specifically the runs on Silicon Valley Bank (SVB) and Signature Bank. The key finding: depositors did not flee uniformly. Instead, they concentrated outflows at banks with high unrealized losses on their securities portfolios, heavy reliance on uninsured deposits, or poor capital ratios. The paper explicitly argues that 'pathology'—institutional fragility—preceded 'panic'. It challenges the traditional view that self-fulfilling prophecies drive runs, a view that has dominated banking theory since Diamond and Dybvig (1983).
But the paper's real power lies in its implications for crypto, a domain where the 'panic-first' excuse has been weaponized to avoid accountability. When a stablecoin depegs, the project often blames a 'coordinated attack' or 'irrational fear'. When a DeFi protocol suffers a bank run on its liquidity, the team cites 'external FUD'. The Fed study now provides a data-driven rebuttal: if depositors are rational, then the underlying health of the protocol was already compromised. The 'run' was merely the inevitable consequence of that rot.
Core: Let me dissect this using the language of an audit partner, not a macro analyst. The Fed's methodology is code. The institution's balance sheet is the smart contract. Deposits are liquidity that can be withdrawn at any time—much like a lending pool on Aave. The study found that the 'withdrawal function' was called excessively only when certain 'state variables' were corrupted. These variables include: (1) unrealized losses (illiquid assets marked at distressed prices), (2) a high proportion of 'uninsured deposits' (i.e., large, uncapped whales), and (3) a low CET1 capital ratio (essentially, thin equity backing).
In crypto terms, these translate directly: - Unrealized losses -> IL in liquidity pools, bad debt in lending protocols. - Uninsured deposits -> Large, dominant depositors (whales) without 'deposit insurance'—which does not exist in DeFi. - CET1 ratio -> Protocol's own capital buffer, often measured by the 'reserve factor' or the treasury balance of the DAO.
The Fed found that for every 1% increase in the share of uninsured deposits, the probability of a run increased by 0.12% per day during stress. For crypto, this is even more extreme: because there is no deposit insurance, the entire deposit base is 'uninsured'. The logical conclusion? Every crypto protocol that relies on a single stablecoin pool or a liquidity provider with large, concentrated positions is a bank run waiting to happen. This is not a risk; it is a structural feature.
I have seen this pattern repeatedly in my audits. Consider the case of Mango Markets in 2022. The protocol allowed large depositors to borrow against their own positions, creating a circular health issue. When a whale manipulated the price, the 'depositors' (the other traders) did not panic randomly. They correctly identified that the protocol was unhealthy—because it was. The run on Mango's liquidity was a rational response to a fragile balance sheet. The Fed study would classify this as a 'fundamentals-driven run', not a panic.
The paper also notes that banks with higher levels of institutional transparency (more frequent disclosures, clearer asset quality data) experienced less severe runs. This aligns perfectly with the core thesis of on-chain auditing: transparency reduces the information asymmetry that causes runs. But the crypto industry is bifurcated. On one side, protocols like MakerDAO publish real-time risk metrics (collateralization ratios, liquidation prices) and thus enjoy higher stability. On the other side, projects like the Terra ecosystem had hidden leverage—the Luna Foundation Guard's reserves were opaque, and the reported backing was probabilistic at best. The Fed study would predict that Luna's collapse was not a panic cascade; it was the market rationally pricing in a lack of health. And it was right.
Contrarian: The bulls might argue that the Fed study actually supports the narrative of decentralization. If a protocol is truly transparent and health metrics are embedded in the code (smart contract), then under the new framework, runs should be less likely because depositors can verify health in real time. In traditional banking, even the best disclosures are quarterly. In DeFi, it's per block. So perhaps DeFi is inherently more resilient to runs than banks—if health is good.
But this is where the contrarian argument becomes a trap. The bull case ignores that 'health' in DeFi is often an illusion. Most protocols do not report 'unrealized losses' in a way that is comparable to bank accounting. A lending pool that uses a volatile asset as collateral has unrealized losses embedded in its price feed—but these are not flagged as 'risk' until the price drops. The Fed study would treat that as a hidden pathology. Moreover, 'transparency' is different from 'auditability'. The data may be on-chain, but if the protocol's accounting method is flawed (e.g., using a manipulated oracle), the 'health' reading is false. This is a blind spot that the crypto industry has not addressed.
Another blind spot: The Fed study assumes that depositors can run simultaneously. In a bank, a run depletes reserves quickly because the bank is the counterparty. In DeFi, a run on a liquidity pool is limited by the pool's size—the run can be exhausted. But this is a double-edged sword. A large run on a small pool can completely empty it, causing the protocol to halt. The 'health' factor is not just about the balance sheet; it is about the ratio of stickiness of deposits to the pool size. A pool that is 90% owned by one whale is healthy only until that whale decides to leave.
The Takeaway: The Fed study is not a paper; it is a mirror held up to the crypto industry. It shows that the myth of 'irrational panic' is a convenient scapegoat for poor engineering and worse risk management. The code speaks louder than the whitepaper, but even code can hide pathology if the assumptions are flawed. Every crypto project that blames FUD for a depeg should instead ask: 'What is our unrealized loss position? What is our capital ratio? How concentrated are our depositors?' These are not questions for a macroeconomic report; they are questions for an audit. And until the industry answers them with the cold, forensic rigor of a New York Fed analyst, the runs will continue—not as bugs, but as features of a system that refuses to admit its own fragility.
Complexity is the enemy of security. The Fed paper proves that even in the world of opaque, centuries-old banking, the simplest metric—institutional health—predicts runs better than any panic index. The crypto industry is ten times more complex, with a thousand times more opacity. The inevitable conclusion? We are not ready for the next run. And that is not panic; it is fact.
Logic does not bleed, but it does break. And when it breaks, it does so at the weakest variable—which, for both banks and DeFi, is the health of the institution when the stress test begins.