Not Your Nan’s Bank Run…
Balance Sheet Liquidity and Bank Runs
NotCentralised is seeking to shine light on some of the more technical aspects of banks and money markets, for the benefit of those in web3 or those yet wondering what the heck went on with Silicon Valley Bank, Silvergate and Credit Suisse etc.
We are considering how money works in a “post Triple S” regime. What do I mean? I mean, we are wrestling with threats to the foundation of fractional reserve banking, using fiat currency, in the wake of the failures at Silvergate, Silicon Valley Bank and Signature bank.
Banking is exceptionally complex in detail and yet simple in principle. The principle is:
I am a bank and you will trust me to look after your money, while I use that money to generate returns for my shareholders.
A bank is thus a money transformation function, at its core, built on promises. A confidence trick.
In this note, we explore challenges to the established regulatory model around liquidity, which is the core issue contributing to the undoing of the Triple S.
Liquidity Coverage Ratio
The LCR, as it is called, is a global regulatory requirement under the Basel III regime, which seeks to ensure that any bank has a sufficient amount of liquid assets on hand to honour depositors’ requests for money back. This LCR framework seeks to avoid or minimise the problems of bank runs. A bank run is a commonly used phrase, but what is it?
A “run” occurs when confidence declines in a given bank’s solvency, causing depositors to seek their money en masse, yet the bank has lent or invested that on-demand money into less liquid assets, thus cannot honour all those demands. The stereotypical imagery here is a line of people around the block waiting to get physical cash out of a branch. In practice, this is now primarily a digital phenomenon, certainly in the “developed” economies.
These runs occurred during the 2008 financial crisis, for example at Northern Rock in the UK, a building society turned bank that failed. In response, global regulators enhanced the requirements for liquidity, through the LCR regime.
The LCR is a stress test for each regulated bank. A hypothetical exercise, which simulates a bank run over the course of 30 days. It seeks to guesstimate how much customer demand there would be for cash withdrawals in that stressful period, and ensure banks have sufficient assets on hand to honour those claims. This is known as the NCO amount — Net Cash Outflows. The assets required to cover this modelled scenario are called HQLA — High Quality Liquid Assets. In more technical terms, the LCR = HQLA / NCO. In Australia this number on aggregate for all the ADIs (authorised depository institutions) is about 1.3x.
HQLA
Now, of course “high quality” is a relative term and therefore different national regulators have different sets of assets which qualify. The three categories of liquid assets, in descending order of quality, are called level 1, level 2A and level 2B. In Australia for example, assets that are acceptable include State and Federal government bonds, raw cash and balances with the RBA. These are level 1 assets. Interestingly in Australia we lack enough level 1 HQLA assets to satisfy the LCR requirements as a whole for the domestic banking system.
So far, so good. In theory then, if the modelling is representative of a 30 day run, and if the assets held in the HQLA bucket do their job, there would not be a liquidity crunch. Obviously the IF statements here are where things break down. The NCO assumption is critical in this equation, and is what proved so wildly wrong in the case of SVB, with an infamous one day drain on deposits of US$42bn.
NCO
The BIS states that NCO is calculated by “multiplying the outstanding balances of various categories or types of liabilities and off-balance sheet commitments by the rates at which they are expected to run off or be drawn down” (my emphasis; LCR40 Cash Inflows and Outflows, BIS 2019). Janet Yellen herself stated she was surprised at how quickly the SVB deposits fled, exposing a critical weakness in LCR regimes.
We must acknowledge how prolific internet and mobile banking usage is in 2023. I suspect all of you reading this have a mobile banking app, or use web-based banking tools. Branch closures are a global phenomenon in light of superior remote banking technology — in Australia, 30% of the nation’s bank branches have closed in the 5 years to February 2023 (ABC News, 2 March 2023). Apple Pay is prolifically used in Australia, further showing the truly digital nature of banking. On top of this, we must add the rapid — viral — dissemination of information in the TikTok and Twitter age. Banks have always been confidence tricks, therefore once that confidence starts to crack, social media is likely to catalyse a more rapid rush for the exits.
In light of this, and the realised experience of SVB, it is a reasonable assumption that bank runs in a stress scenario will be quicker and larger than the assumptions in the LCR regime. It would be fascinating to interrogate the departments in the banks tasked with estimating this pace of net outflows — what if it’s twice as severe, three times etc? The shape of a bank run is likely to be front-end loaded, not linear, over this 30 day window.
Bank payments often require several days to finally settle and are not processed outside of banking hours in many cases. And yet, technology is perfectly capable of running 24/7 payment regimes, such as those we see in crypto. There seems to be a growing gap between regulatory regimes, perhaps even the knowledge and attitudes of regulators themselves, compared to the underlying technology available.
Wrapping Up
Banks fail fast. They fail due to liquidity not loan asset quality, which evolves relatively slowly. Newer banks or newer payment technology are being held back by outdated regulatory assumptions. This retards overall growth potential and customer utility. Without embracing new technology primitives, users will vote with their wallets and move their payment requirements elsewhere, or will suffer a weaker UX. It is likely that the assumed NCO numbers for many banks are understated, given the experience with SVB and the now-prolific use of mobile and online banking. This in turn suggests systemic stress, due to insufficient HQLA on hand to honour outflows in a bank run, is much more likely today.
Thanks for reading this far!
Bish, Numbers, Capt Defi, Web3PO.
