Time has passed and the dust has settled, at least a bit, on the Silicon Valley Bank experience. There are important lessons to be drawn from this bank run alone and the policy response.
This post reflects my joint work with Hubert Kempf and much appreciated conversations on this topic.
Enriching the Theory of Bank Runs
A earlier post talked through the logic of the basic economic model of bank runs, coming from the work of Diamond and Dybvig (hereafter DD).1 The Nobel prize in Economics made clear that this was a fundamental contribution to our understanding of banks and their inherent fragility. But models are abstractions, at least based on our recent experience, the DD framework needs to be enriched to provide both an understanding of these events and the policy responses. This is part of what makes doing economics fun.
Diamond and Dybvig Model
This economic model explains the illiquidity of banks and its consequences for banking fragility. Banks have a choice between short term liquid investment and long term illiquid investment. The returns are certain with the long term investment providing a higher return. All else the same, banks would prefer to invest deposits in these longer term assets and thus earn higher returns.
But depositors, households in the model, have random liquidity needs. The bank meets these needs by investing some of its deposits in a liquid asset, just avoiding costly liquidations of the long term investment.
The bank optimally selects its portfolio to meet the normal demand for liquidity of its depositors. The bank is solvent and sufficiently liquid. Normally.
But there is a chance that depositors will panic, turning the normal demand for liquidity into a bank run. In such a situation, the bank maybe unable to meet the demand of all depositors even after liquidating its illiquid assets. In the end, the depositors, acting in their self interest, made the right choice: run on the bank when everyone else does. But collectively, there is a loss: a solvent bank failed due to its illiquidity.
In the DD framework, the problem of a run can be solved through the provision of deposit insurance. An iron-clan guarantee that the government will provide deposit insurance in the event of a run is sufficient to avoid the run. What is important though is that the government’s commitment to deposit insurance includes a willingness to use taxation as needed to finance the deposit insurance.
Some Missing Pieces
Looking at the SVB experience through the lens of the DD model, there are clearly some elements missing in that framework. This is a work list for those involved in building these models.
Monetary policy impacts the value of liquid assets. The tightening of the FED raised interest rates and thus lowered the prices of liquid assets. All else the same, the reduction of the value of liquid assets makes banks, such as SVB, more vulnerable to runs.
Even with deposit insurance, there are depositors over the limit. In the standard version of the DD model, once there is credible deposit insurance, runs no longer happen as everyone is covered. But that is because it is assumed that all deposits are insured. The SVB episode taught us that this is not the case.
In the standard DD model, banks have no access to equity if they become unstable. In the SVB episode there was an attempt to raise more equity to meet the needs to depositors. But this source of funding evaporated. The need to understand the interaction of the decisions of equity investors along with depositors is made clear by this episode.
The standard interpretation of the DD model is that depositors are households. For SVB they were largely firms, with loans from that same bank. Evidently, perhaps for incentive reasons, there was a link between making a loan to a firm and requiring funds to be deposited at a bank. This connection is absent from the standard DD model.
Lessons for Policymakers
The policy response of the US government can be neatly summarized by two points:
the cap on deposit insurance is (apparently) gone;
taxpayers are not at risk.
If in doubt, here is a quote from the US President:
First, all customers who had deposits in these banks can rest assured — I want to — rest assured they’ll be protected and they’ll have access to their money as of today. That includes small businesses across the country that banked there and need to make payroll, pay their bills, and stay open for business.
No losses will be — and I want — this is an important point — no losses will be borne by the taxpayers. Let me repeat that: No losses will be borne by the taxpayers. Instead, the money will come from the fees that banks pay into the Deposit Insurance Fund.
Pheww … this is good news. Right?
Do these interventions stabilize banks? Seems like an important question: all too often the policies we adopt to deal with one crisis sets in motion the next one.
Answering this question and thus assessing the policies builds upon the DD model. Let’s start from the second point about protecting taxpayers and work back to the first about protecting all depositors.
No Taxpayer at Risk
Normally, one would think that in the event of a run, taxpayers have to bear some of the burden of providing deposit insurance. If the government has to provide say $10 trillion in deposit insurance in a systemic run, then a source for this must be taxation. It could be immediate taxation or debt financing of the deposit insurance today with taxes coming in the future.
So, if, under the current US plan, taxpayers are off the hook for financing deposit insurance, then where will the resources come from? Could the answer be through the resources of the Federal Deposit Insurance Corporation (FDIC)? The fund is created through contributions of member banks. Depositors of these banks receive protection through this fund.
But the unfortunate reality is that the resources of this fund are small relative to potential depositor needs. From the FDIC, the fund has a target to be able to protect 2% of the deposits in the US. This is surely enough money if one bank fails. It is far from enough if all banks fail.
The FDIC states that its insurance is backed by “ … the full faith and credit of the United States government.” This is a good thing given the relatively small size of the actual funds at its immediate disposal.
But this backing is what the recent policies of the US government have taken off the table. If no taxpayers are at risk, then there must be no tax revenue that will flow to the FDIC in the event of a run. That is fine, as long as the run is not too large.
The problem here is that a small run can become a big run due to contagion. Imagine there is a run and a bank is shutdown, requiring some FDIC insurance payments. To make the point directly, suppose that those payments alone required the 2% of the deposit insurance fund. This means that the remaining deposits in all other banks are not longer insured. This is a recipe for additional runs.2
No Depositors at Risk
The second part of the policy apparently extends deposit insurance to everyone, without limit. Given that the run at the SVB was in part driven by uninsured depositors, this appears to be an easy way to stabilize the banks.
But this extension of deposit insurance must be credible. That is, individuals must believe that in the event of a run their deposits will indeed be insured.3
But, returning to the first point, where is this insurance supposed to come from? If taxpayers are protected and deposit insurance coverage is extended, the fund of the FDIC is simply inadequate. So, with taxpayers out of the equation, it seems that this promise of full insurance is empty: no runs are prevented.
What if taxpayers were not protected and instead the FDIC had the backing of the US government? That means the government would raise taxes or issues debt (so raising future taxes) in response to a run. If this was credible, then that would be enough to avoid the run.
Is this promise credible? That is, would the government actually go ahead and raise taxes to pay off depositors?
We have analysed this question in our joint research.4 We argue that the taxation needed to finance deposit insurance may entail a redistribution from poor (Main St.) to rich (Wall St.). The magnitude of this redistribution depends on the relatively size of the deposits being insured and the progressivity of the tax system. If the tax system is not very progressive (we study the case of a lump-sum tax), then the provision of deposit insurance does imply that the rich get a lot more from their deposits being insured compared to the poor. This may be socially undesirable. If so, this means that in the event of a run the government may not have the incentive to provide the promised deposit insurance.
While the response to the SVB run did not take away deposit insurance, the extension to supposedly uninsured depositors made clear that the guidelines for the provision of deposit insurance are, just that, guidelines. Clearly the government seems willing to decide after the fact exactly what insurance to provide. Raising doubts about the credibility of deposit insurance is a potential cost to this intervention.
And this is related to the relaxation of the cap. A limit on deposit insurance also limits the funds that are transferred from the relatively poor households to rich depositors. An increase in the cap means more redistribution towards the rich. All else the same, this reduces the credibility of promised deposit insurance, either in whole or perhaps just through the adoption of a cap ex post.
There is another policy tool at play here. Tight monetary policy impacts the fragility of banks. The monetary authority needs to keep this in mind. If the FED (ECB) chooses to “fight” inflation through high interest rates, the resulting reduction in the value of government debt impacts the balance sheets of banks. This additional (perhaps new) channel of monetary policy needs to be taken into account in making assessments of the effects of higher interest rates.
Going Forward
Honestly, it is not clear where we stand with the provision of deposit insurance in the US. It is clear what we need:
clearly stated policies;
credible policies;
a stable banking system.
It is also clear how to achieve these objectives. I advocate for the sound advice of economists rather than the quick solutions provided by the political process.
Of course the FDIC can respond by replenishing its fund by demanding more contributions from the remaining banks. But this itself might hasten the instability, putting more banks in trouble.
Recent testimony by Treasury Secretary Yellen appears to take a step back, linking the provision of deposit insurance above caps to the determination of systemic risk.
For those interested, check out: Cooper, Russell, and Hubert Kempf. “Deposit Insurance and Bank Liquidation without Commitment: Can We Sleep Well?” Economic Theory 61, no. 2 (2016): 365–92. http://www.jstor.org/stable/24735338.
This post is welcome as it shows that we should worry about the policy responses to the SVB failure. There is an additional reason to worry that Russell has not mentioned: the different positions taken by the various top policymakers which can be seen as opposite and thus raise the issue of the quality of the cooperation between them in case of a bank run or a banking panic. Answering a question from Simon Rabinovitch, US economics editor for The Economist, Jerome Powell, the chairman of the Fed, said depositors “should assume” they are safe. Around the same time Janet Yellen, the treasury secretary, said that expanding insurance to all depositors is not under consideration. As the editor-in-chief of the magazine wrote, « They can’t both be right ! ».
This is disturbing for two reasons. The first is, as said above by Russell, that claiming that depositors « should feel safe » is ambiguous enough to be a clear sign of the unwillingness to commit and therefore spread defiance toward the banking sector, or at least to its weaker part.
The second reason is that the plurality of opinion from the two top policymakers in the US is weakening further the trust in the solidity of the US banking sector. The willingness of the Treasury secretary not to lift the cap on deposit insurance is understandable. The dimension of a bank panic ( a generalized bank run, not targeting one particular bank but the entire banking system) cannot be foreseen ex ante. It may be so large that it would create havroc on the financial system and in case of a pledge to insure all deposits an unsustainable burden on the federal Treasury. But it also shows that the claim by Biden that the no tax will be used to insure banks is shallow. On the opposite, the assertion by Jerome Powell that depositors should feel safe is both a tentative to instill optimism in the depositors and play on their beliefs and a way to put pressure on the Federal Treasury pointing to its responsibility (and not the Fed) in the case of a run or a panic. The combination of both claims is a further example of the impossibility to commit on the need to insure bank deposits. The contradictions between the two statements prove that the cooperation between the two major public authorities concerned by the stability of the banking system is likely not to be smooth and harmonious.
HubertKempf.