Silicon Valley Bank (SVB) was a bank with $212 billion in assets and $175 billion in deposits that was admittedly ‘special’: very specialised in tech, with mostly uninsured start-up deposits rather than retail deposits. It did not aggressively search for deposits by offering particularly high interest rates, so it did not ‘gamble for resurrection’. Instead, its problem stemmed from its asset side, which did not primarily consist of loans but was instead mostly standard and liquid securities that were poorly hedged as far as interest rate risk was concerned. The recent increase in interest rates combined with poor hedging lowered the value of its assets, and eventually led to insolvency when some depositors sought to withdraw their deposits.
SVB clearly looks like an example of US failure in both regulation and supervision. Together with other banks, SVB had successfully lobbied Congress for weaker regulation, which allowed it (and others) to rely on held-to-maturity accounting, and to be exempted from the Basel liquidity coverage ratio (LCR) requirement.
In terms of crisis management, after unclear communication by the authorities (“FDIC will not do a bailout but will try and help uninsured depositors, through a dividend”), the weekend of 11/12 March 2023 ended with the decision to fully guarantee uninsured deposits too, so there was a bailout (which the authorities may or may not recoup in the future through fees levied on the banking sector as a whole). This decision echoes the earlier ones for hedge fund LTCM or investment banks Bear Stearns and Merrill Lynch (and differs from the Lehman decision), with a rationale that is both similar to (avoiding runs) and different from (the goal this time is to protect tech startups rather than financial creditors) these three previous cases.
General principles
Banks engage in maturity transformation, which can lead to runs. As Diamond-Dybvig (1983) showed in their Nobel Prize-winning contribution, multiple Nash equilibria exist under incomplete deposit insurance. This is even truer in a phone/internet banking era: SVB witnessed $42 billion deposits being withdrawn on 9 March 2023 alone!
Risks of contagion do exist, but it is still worth asking why the failure of a medium-sized special US bank like SVB can lead to big banks in Europe losing more than 10% of their stock market value in a couple of days. One reason of course is that some of the factors having affected SVB – in particular, the current cycle of interest rate increases and the risk of recession – can potentially impact every bank. This situation echoes to some extent the savings and loans episode of the 1980s in the US, where these specialised savings institutions were collateral damage of the anti-inflation policy of the Fed. Markets are naturally volatile, even more so when investors start asking themselves whether banks have engaged in imprudent maturity transformation, which supervisors may have unduly allowed. Moreover, for every institution that has hedged itself against an increase in interest rates, there is a counterparty that has accepted this risk and may, or may not, have hedged it, and so on. This can lead to nervousness (“are we in 2008 again?”) and liquidity problems for some banks that inevitably turn into solvency problems. At this point, panics can become self-fulfilling when depositors, i.e. those who can run, start doing so when they are less than fully protected.
When considering bailout versus bail-in/bankruptcy, the financial authorities are facing the familiar trade-off between financial instability and moral hazard. One should not forget the lesson of the 15 September 2008 Lehman episode, when the decision to impose losses on short-term depositors was made in order to avoid a bailout, but which ended up being the costliest bank failure for taxpayers by creating panics and deepening the Great Recession.
Of course, moral hazard is an important problem which should not be underestimated, and ex-post deposit guarantees increase it and are therefore problematic. However, in order to address this issue effectively, one needs to concentrate losses on those investors who cannot run, that is, equity, bondholders and term depositors (and possibly sue top management and board members for misbehavior). Spreading the pain among those who can run is definitely a counterproductive idea. Anyway, one can certainly not hope to impose ‘serious discipline’ on banks through bailing in short-term deposits held by individuals, or by (large or small) firms for their ‘ordinary business’ (i.e. to make payments to the owners of their premises, their staff and their suppliers, and to receive customer payments). Typically, such depositors do not have the expertise, nor should it be their ‘job’, to monitor banks, hence they tend to react ‘randomly’ by running, which will typically be destructive for the economy. This is the idea underlying the ‘representation hypothesis’ put forward by Dewatripont and Tirole (1994): one should delegate the monitoring function of short-term depositors to other parties, namely, a combination of long-term private investors and supervisory and resolution authorities.
This should be done with a combination of instruments: (1) appropriately calibrated solvency ratios (risk-weighted ‘capital’, non-risk-weighted ‘leverage’, and overall ‘loss absorbency’, which adds to capital subordinated debt instruments TLAC and MREL);
(2) liquidity ratios (the one-month LCR and the longer-term net stable funding ratio); (3) sufficient marking-to-market of assets; (4) intrusive-enough supervision (with good stress tests); (5) good resolution plans and thresholds; and (6) appropriate macroprudential buffers. The Basel Committee and the Financial Stability Board have come up with an important package of rules in this respect. Pushback from industry, however, has weakened this package (the EU is still Basel III non-compliant, and the US has many banks that ‘escape’ a number of Basel III rules, which technically apply only to ‘internationally active institutions’).
Two lessons for prudential regulation, one for Europe and one for the world
In the EU, we do not have a credit institution like SVB, and supervision appears better when compared to the SVB debacle (even though Basel III compliance would be desirable).
This being said, the first lesson from this episode is that it should increase the resolve of public authorities to further improve the resolution framework in the EU. As discussed for example in Dewatripont et al. (2021), the Banking Recovery and Resolution Directive (BRRD) has worked ‘backwards’ by legally preventing from its beginning in 2016, “even under extraordinary circumstances”, any bailout before 8% of the unweighted balance sheet of a troubled bank has been bailed in. This ‘8% bail-in rule’ would make sense but only if all EU banks had 8% of long-term subordinated securities that could be bailed-in (i.e. would belong to MREL). However, today, some banks cannot satisfy the 8% bail-in rule without hitting short-term depositors. Under BRRD, for such banks the US approach to SVB is legally unavailable.
Some progress has been made over the years in raising loss absorbency. It would therefore be good if EU authorities were to: (1) communicate about the percentage of banks/banking assets that already do satisfy this 8% long-term subordinated claims condition (in principle, at least all GSIBs and all banks above €100 billion of assets); and (2) in order to avoid deposits moving to safer/too-big-to-fail institutions, announce a plan that would temporarily suspend this 8% condition until it is reached thanks to a precise publicly announced timetable, potentially forbidding to distribute dividends until it is achieved.
The second lesson from this episode, relevant not just for Europe but worldwide, is that the current regulatory treatment of short-term deposits that (large and small) companies rely upon for their ordinary business is flawed and needs to be adapted. Protecting such company deposits only up to $250.000 or €100.000 means forcing them to face unnecessary risks, and for some big companies, this is almost like not insuring them at all, and this can be economically very costly.
There are several possibilities to address this. A first one is to significantly increase the protection of these company deposits, and to price this risk by charging banks deposit insurance fees, just like for currently insured deposits (making sure of course that banks in trouble get resolved promptly, otherwise these fees become irrelevant as soon as the bank is becoming insolvent). Calibration of insurance thresholds for these deposits is obviously a challenge which should be addressed carefully. A natural approach would be to link the insurance threshold to company size, corrected for business model. In any case, it should be actuarially fairly priced.
A second, less ‘radical’ approach would be to reduce the intrinsic risk faced by short-term company deposits by: (1) raising the volume of claims junior to them (as stressed in our first lesson); and/or (2) strengthening the LCR requirement by raising the expected monthly outflow rates (which currently ranges from 5% for SMEs up to 40), given the speed of withdrawals observed at SVB, which has been boosted by new technologies.
Banking regulation and supervision cannot eliminate all risks. However, the above two lessons offer clear directions to make the banking system safer and better equipped to support the real economy.
References
Dewatripont, M, L Reichlin and A Sapir (2021), “Urgent reform of the EU resolution framework is needed”, VoxEU.org, 16 April.
Dewatripont, M and J Tirole (1994), The Prudential Regulation of Banks, MIT Press.
Diamond, D and P Dybvig (1983), “Bank Runs, Deposit Insurance, and Liquidity,” Journal of Political Economy 91: 401-41.
