Skip to main content Skip to secondary navigation

Fragile: Why more US banks are at risk of a run

Key Takeaways

  • Recent declines in bank asset values have significantly increased the vulnerability of the U.S. banking system to uninsured depositor runs.
  • The actual market value of assets in the U.S. banking system is $2.2 trillion lower than the stated value of these assets.
  • A substantial number of institutions are at risk of failing should there be a run on these banks by uninsured depositors.
  • Regulatory responses could involve stress tests, focusing more on the value of assets, and stricter capital requirements.

The recent rise in interest rates by the Federal Reserve has increased the fragility of the U.S. banking system to the point that a substantial number of institutions are at risk of failing should there be a run on these banks by uninsured depositors.

This vulnerability is not only due to a decline in the value of bank assets but also a related factor — the powerful urge uninsured depositors have to withdraw their funds in a perceived crisis, leading to bank insolvencies.

We present a model of solvency runs, which illustrates that interest rate increases can lead to bank runs even when bank assets are fully liquid. These solvency runs can become more widespread “self-fulfilling” solvency runs — that is, when uninsured depositors pull their funds in the belief that others will do the same, making compounding negative expectations among banks a reality.

Our model identifies the most fragile banks, which are those with low capital levels; a high ratio of uninsured deposits to assets, called uninsured leverage; and those for which the true market value of their assets is far lower than the stated value of those assets on their books.

Our analysis shows the market value of U.S. banking system’s assets is $2.2 trillion lower than suggested by their book value, accounting for loan portfolios held to maturity1. Moreover, most of these asset declines have not been hedged through the use of interest rate derivatives.

This discrepancy between the book and market value of assets means that in case of a run on these banks — as happened in the spring of 2023 at Silicon Valley Bank, Signature Bank, and First Republic — the liquidation value of assets would not be sufficient to cover both insured and uninsured deposits.

That, in turn, would lead to insolvency and could threaten the stability of the entire U.S. banking system.

This vulnerability of the nation’s banks has serious implications not only for financial stability but also for the Federal Reserve’s ability to conduct monetary policy and for government agencies to oversee the safety and soundness of the banks.

We show that the key to understanding whether these asset losses would lead to some banks in the U.S. becoming insolvent — and the likelihood of that happening — is the concept of uninsured leverage (uninsured liabilities/assets). That is, the degree to which the mark-to-market or true value of their assets is or is not sufficient to cover uninsured liabilities in case of liquidation.

Uninsured depositors are key players because those with deposits over the $250,000 FDIC covered limit stand to lose a part of their deposits if a bank fails. That potential loss gives them a strong incentive to run in times of perceived trouble — such as a decline in asset values — which could push a bank into insolvency.

We devise a simple model showing that uninsured leverage is the key to understanding whether these losses would lead to some banks becoming insolvent. And we show that a bank’s survival depends on the market’s belief about the share of uninsured deposits that will be withdrawn following a decline in the market value of bank assets.

Systemic risk

We compute the incentives for uninsured depositors to run for the sample of all U.S. banks and find that even if only half of uninsured depositors decide to withdraw, almost 190 banks with assets of $300 billion are at a potential risk of impairment, meaning that the value of their assets cannot cover even their insured depositors. 

We also find that regions with lower household incomes and large shares of minorities are more exposed to bank risk.

Overall, these calculations suggest that recent declines in bank asset values — especially in commercial real estate — have significantly increased the fragility of the U.S. banking system to uninsured depositor runs.

We briefly discuss events and subsequent research following our paper’s release on March 13, 2023, which we see as providing validity to our approach and findings.

SVB: the canary in the coal mine

Silicon Valley Bank (SVB) failed in a “run” on March 10 of 2023 following a sharp tightening of monetary policy that began in March 2022. Tighter monetary policy has a significant negative impact on the value of long-term assets on bank balance sheets. And this decline in asset values relative to liabilities can lead to bank instability through two channels. First, a bank can become fundamentally insolvent if asset values are less than the value of its liabilities. Second, uninsured depositors may run, causing the bank to fail. This is especially the case, because systemwide uninsured depositors make up about half of bank deposits (Egan et al. 2017) [1].

We develop an empirical methodology and conceptual framework to analyze the effect of rising interest rates on the value of U.S. bank assets and bank stability when bank assets are liquid and we measure the losses due to interest rate increases from Q1 2022 to Q1 2023.       

Because bank call reports do not mark significant parts of their assets to their market values, we provide a mark-to-market calculation of these losses using tradable and liquid market indexes.

Using the case of SVB, we show that the asset side alone cannot explain its failure. SVB was not an extreme outlier from the perspective of asset losses; however, it was an outlier from the perspective of its liabilities. Some 92.5 percent  of its deposits were uninsured, leading to significant withdrawals that ultimately resulted in the bank's collapse within two days. In other words, despite its liquid balance sheet, the SVB failure was the result of a run by uninsured depositors.

Computing bank fragility

We have created a model that illustrates that banks can become exposed to solvency runs when monetary policy tightens. We model the existence of solvency runs, which arise even if banks’ assets are fully liquid. The model builds on the idea of solvency runs of Egan et al. (2017) but extends it to study the role of maturity transformation, i.e., banks invest in long and short maturity assets, exposing banks to asset declines due to monetary policy. At the same time, they fund with demandable deposits. We then use the insights from the model to compute new empirical measures of bank fragility for the sample of all U.S. banks.

Long-dated assets experienced significant value declines following the monetary tightening from the first quarter of 2022 onward. In response to high inflation, the Federal Reserve Bank severely tightened monetary policy. From March 7, 2022, to March 6, 2023, the federal funds rate rose sharply from 0.08 percent to 4.57 percent. As a result, long-dated assets experienced significant value declines. For instance, the exchange-traded fund that tracks the market value of residential mortgages declined by more than 10 percent (Figure 1).

Figure 1. Potentially Insolvent Banks under Different Uninsured Depositor Withdrawal Scenario

This figure presents the number of insolvent banks (top) and their aggregate assets (bottom) associated with a given uninsured deposits withdrawal case. We consider ten cases ranging from 10% to 100% of uninsured deposits being withdrawn at each bank. The bank is considered insolvent if its mark-to-market value of assets – after paying a given share of the uninsured depositors -- is insufficient to repay all insured deposits.

Similarly, the market value of commercial mortgages indicated by the fund that tracks market value of commercial mortgages — iShares CMBS ETF —declined by more than 10 percent during this time. Long maturity Treasury bonds were particularly affected by monetary policy tightening, with 10-20 year and 20-plus year Treasury bonds losing about 25 percent and 30 percent of their market value, respectively, as suggested by iShares Treasury ETF. Overall, long duration assets similar to those held on bank balance sheets experienced very significant declines during the Fed’s monetary policy tightening.

We mark-to-market losses on banks’ assets due to interest rate increases from Q1 2022 to Q1 2023 using market-level prices of long-duration assets.

Uninsured deposits and panic runs

 But a case study of SVB confirms that banks’ asset losses and bank capitalization alone are insufficient to understand how monetary policy tightening affects bank stability.

The composition of liabilities plays a central role. About 500 banks (10 percent) had larger unrecognized losses than SVB’s. Similarly, 10 percent of banks had lower capital prior to monetary tightening. On the other hand, SVB had a disproportionally large share of uninsured funding: Only 1 percent of banks had higher uninsured leverage, defined as uninsured debt over assets in Jiang et al. (2020) [2]. For SVB, 78 percent of its assets were funded by uninsured deposits. This fact suggests that uninsured deposits played a critical role in the failure of SVB.

To analyze how monetary policy can trigger panic-induced runs in banks with liquid assets, we next developed a model that builds on the idea of solvency runs of Egan et al. (2017) [3] but extends the model to study the role of maturity transformation: that is, the fact that banks invest in long maturity assets, exposing banks to asset declines due to monetary policy and at the same time they fund with demandable deposits.

When interest rates are relatively low, bank asset values are high enough that they can survive the withdrawal of all uninsured deposits. Then, it is not rational for any individual depositor to withdraw and the bank is immune to solvency runs. When interest rates rise sufficiently, and thus asset values decline, self-fulfilling runs are possible. Banks with smaller initial capitalization, higher uninsured leverage, and a higher share of sophisticated depositors are more susceptible to such runs and insolvency.

We compute new empirical measures of bank fragility for the sample of all U.S. banks across a broad range of scenarios. We illustrate that banks were not likely to have hedged the vast majority of the decline of their assets due to the rise in interest rates. We identify which banks were at risk of uninsured depositor solvency runs under different scenarios in March 2023. We found that without regulatory intervention, even if only half of uninsured depositors had decided to withdraw, almost 190 banks with assets of $300 billion were at a potential risk of insolvency, meaning that the mark-to-market value of their remaining assets after these withdrawals would be insufficient to repay all insured deposits (see Figure 1).

Regional disparities

The risk in the banking sector due to monetary tightening is not spread uniformly; there is higher exposure in regions with more minorities and lower income households.

The most exposed counties in the U.S. have up to 13 percent of their deposits at the risk of impairment. These counties are clustered in several states, including New Hampshire, Massachusetts, Wyoming, and New York. Some counties do not have much exposure to the risk, namely Delaware, Nebraska, Arkansas, and Maryland.

Counties with a higher percentage of minorities, especially those with more than 80 percent Black and Hispanic population, tend to be more exposed to bank risk. In part that’s because in some areas of lower household incomes, bank asset values are less than the face value of their liabilities. For instance, on average, counties with more than 90 percent Black and Hispanic population have about 4 percent of total deposits at the risk of impairment. Counties with low median income are more likely to be exposed to bank risk. Regions with median annual income below $35,000 are mostly exposed to the risk, with about 4 percent of deposit at the risk of impairment.

Beyond SVB

One key challenge an empirical work typically faces is whether the findings apply in the real world. We released our study on March 13, 2023, and since then events have validated our approach and findings.

On May 1, 2023, the FDIC announced that First Republic had been closed and sold to JPMorgan Chase, becoming the third bank that failed during 2023, following Silicon Valley Bank’s collapse and the closure of Signature Bank on March 12, 2023. All three banks have similar characteristics to the banks at risk we identify: a significant decline in the value of their assets and a high share of funding coming from the uninsured depositors.

The collapse of all these banks was also preceded by significant withdrawals of funds by uninsured depositors (e.g., First Republic Bank saw almost half of the uninsured depositors withdraw). In addition, several other banks, for instance Pacific Western Bank, suffered large declines in their share prices putting them at the brink of bankruptcy with the regional banking tracking fund declining by more than 40 percent between March 2023 and May 2023. In line with our analysis, these events indicate that the financial stability risk we focus on is not an isolated phenomenon, to the Silicon Valley Bank, but affects a significant set of other banks.

Policy implications

Our findings have important implications for financial stability, regulation, and the pass-through effects of monetary policy.

First, our analysis suggests that U.S. banks have significant asset exposure to higher interest rates that can lead to insolvency bank runs by their uninsured depositors.

Second, this fragility of the U.S. banking system to higher rates can significantly constrain the conduct of monetary policy, adversely affecting its price stability objectives.

Third, our findings have implications for several short-run and longer-term regulatory responses one could consider when addressing the financial fragility risk we focus on.

In the near term, the creation of the Bank Term Funding Program in March 2023 together with other policy responses to the recent banking vulnerabilities may have put a pause on the crisis and reduced the risk of acute deposit runs across the banking system. However, these polices do not address the fundamental insolvency risk, which our analysis indicates could involve hundreds of banks. Hence, a near-term response to the crisis could involve a recapitalization of the U.S. banking system (see DeMarzo et al. 2023) [4].

In the long run, one regulatory response to the crisis could involve an increased oversight of the U.S. banking system. Regulators could adopt our methodology to stress test the banking system for a scenario of higher interest rates. This would take into account both the composition of bank assets as well as their liabilities and assess the insolvency risk due to runs by the uninsured depositors.

Regulators could also consider expanding even more the complex banking

regulation on how banks account for mark-to-market losses. However, such rules and regulations, implemented by myriad of regulators with overlapping jurisdictions, might not address the core issue at hand consistently (Agarwal et al. 2014) [5].

Alternatively, banks could face stricter capital requirements, which would bring their capital ratios closer to less-regulated lenders that retain more than twice as much in capital buffers, as documented in Jiang et al. (2020) [6]. But discussions of this nature remind us of the heated debate that occurred after the 2007 financial crisis, which many might argue did not result in sufficient progress on bank capital requirements (see Admati et al. 2013, 2014, and 2018) [7]. They also resonate well with historical studies on the impact of deposit insurance on banks’ risk-taking behavior (see Calomiris and Jaremski 2019) [8].

About the Author

Amit Seru is a SIEPR Senior Fellow, a Professor of Finance at the Stanford Graduate School of Business, and a Senior Fellow at the Hoover Institution. His primary research interest is in corporate finance. He is interested in issues related to financial intermediation and regulation, interaction of internal organization of firms with financing and investment, and incentive provision in firms.


[1] As of end of 2023:Q3 end, analysis shows the market value of assets is $2.5 trillion lower than book value.


[1] Egan, M., G. Matvos, and A. Hortacsu, 2017. “Deposit Competition and Financial Fragility: Evidence from the U.S. Banking Sector,” American Economic Review 107, 169-216.

[2] Jiang, E., G. Matvos, T. Piskorski, and A. Seru, 2020. “Banking without Deposits: Evidence from Shadow Bank Call Reports,” NBER working paper.

[3] Egan et al. (2017).

[4] DeMarzo, P., E. Jiang, A. Krishnamurthy, G. Matvos, T. Piskorski, and A. Seru, 2023. “Resolving the Banking Crisis,” policy proposal.

[5] Agarwal, S., D. Lucca, A. Seru, and F. Trebbi,  2014. “Inconsistent Regulators: Evidence from Banking,” Quarterly Journal of Economics 129, 889-938.

[6] Jiang et al. (2020).

[7] Admati, A., P. DeMarzo, M. Hellwig, and P. Pfleiderer, 2013. “Fallacies, Irrelevant Facts, and Myths in the Discussion of Capital Regulation: Why Bank Equity Is Not Socially Expensive,” Working Paper.

Admati, A., and M. Hellwig, 2014. “Bankers’ New Clothes: What’s Wrong With Banking and What To Do About It?” Princeton University Press.

Admati, A., P. DeMarzo, M. Hellwig, and P. Pfleiderer, 2018. “The Leverage Ratchet Effect,” Journal of Finance 73, 145-198.

[8] Calomiris, C.W., and M. Jaremski, 2019. “Stealing Deposits: Deposit Insurance, Risk-Taking, and the Removal of Market Discipline in Early 20th-Century Banks,” Journal of Finance 74, 711-754.

Amit Seru
Publication Date
November, 2023