Although central bank lender-of-last-resort facilities offer the ultimate backstop of banking system liquidity, the reality is that everyone involved views such operations represent a failure on the part of the banks — they will only exist further at the whims of the government. The role of regulators and risk managers at banks is to avoid that outcome. In this article, I discuss how the system is supposed to work.
Editorial note: I was hit by a number of distractions last week. This article is my first pass on bank liquidity management, which is an important topic in my banking primer. This article might overlap my earlier articles, but I will need to put them together to reduce the repetition. I started a piece on inflation, and hope to get back to it by the end of this week.
Aside: Liquidity Definition
One of the challenges to discussing “liquidity” is that people in finance use the term in two ways. I will offer two definitions.
A financial instrument is liquid if market participants can trade the instrument in what are seen as “large” sizes (“large” dependent upon instrument) without the price changing too much. (A bank deposit on a non-failed bank is always liquid by this definition, since withdrawals occur at par.)
A firm is liquid if it is able to make contractual payments (of all sorts) over the near run.
Taken together, a firm can be seen as “liquid” if it has positions in “liquid assets” that are large enough to meet its upcoming contractual payments.
The first definition (market liquidity) is very hard to measure and not everyone agrees on whether particular markets are liquid or not. As for the second, whether a firm is deemed “liquid” depends upon analysis of its financial statements that typically follow conventions that depend upon the industry of the firm. For example, analysts might look at the ratios between the holdings of short-term assets versus short-term debts.
I am not going to be able to offer an in depth guide to how banks manage their liquidity in practice, partly for the reason that different banks operate in different ways. However, my aim is not to provide a tutorial on how to set up a bank, instead to offer a guide as to how banks fit in with the macroeconomy.
My complaint with conventional “Economics 101” discussions of banking is that one could paraphrase the discussion of bank lending and liquidity as follows.
A bank is approached by a customer asking for a loan. The bank evaluates the loan.
The bank then looks at its liquidity position: does it have excess liquidity (typically phrased in the now anachronistic term — does the bank have excess reserves)?
If the bank has the reserves to meet the expected cash outflow when the customer spends the loan proceeds, it will offer the customer the loan at some interest rate.
It is entirely possible that a really small bank might operate in this fashion, but this is misleading for the large banks that dominate developed countries.
The story is better understood as follows.
The bank starts with a “liquidity buffer” that is above its regulatory minimums, as well as above the limits set by the bank’s internal risk management policies. If the bank does not start in this position, it needs to do something to correct the situation very quickly.
For a large bank, most loans are extremely small relative to the bank’s overall liquidity position. Loan officers will be posted at branches, and they will be making a steady flow of loans, with pricing under guidelines set by senior bank officers.
If lending activity ramps up for a period of time, the cumulative effect of cash outflows would cut into the bank’s liquidity buffer margin above its minimums. However, outflows due to lending are not the only thing affecting the liquidity position — customer deposits can fall for other reasons (e.g., changing bank), and bank debt instruments entail contractual payments.
The bank will raise cash in the money markets for what are seen as short-term liquidity issues. However, if the liquidity impairment is projected as being long-term, it will need to issue long-dated instruments (including selling securitisations of loans), on top of offering higher yields on long-term deposits.
If the bank cannot raise cash in either long-term or short-term debt markets, congratulations — it is an ex-bank.
Note that the above discussion is solely related to liquidity — the bank also needs to keep capital ratios above regulatory minimums. Most “Economics 101” stories about banks discuss “reserves,” but not “bank capital.” Bank capital cannot be shored up by short-term movements; the bank either needs to issue capital instruments or get risky assets off the balance sheet (securitisations) if retained earnings are not enough to keep capital ratios at acceptable levels.
The key take away is that liquidity management is a process that looks at all changes in a bank’s liquidity position, and banks cannot look worry about isolated transactions in isolation.
Aside: Are Bank Loans Always Small?
I asserted that bank loans were small relative to the bank’s overall balance sheet. That may not always be the case — but it is a situation that modern North American banks would want to avoid.
- Non-bank sources of finance for large firms might not be available. Continental European financial systems tended to be dominated more by banks, as well as earlier eras.
- Small banks may have no choice in the matter.
- The loans might be part of a deliberate corporate strategy, such as in the keiretsu system in Japan (often translated as “the convoy system”) where a group of firms had interlocking relationships, and the grouping often contained a bank that financed other group members. The popping of the bubble in Japanese asset prices (land and equities) showed the weakness of the arrangement — the entire group ended up largely crippled if bad loans within the group meant that the bank was unable to extend new loans.
The initial reaction to liquidity management is to model the contractual inflows and outflows of the bank. This is what most firms would do: look at all known contractual payments to be made and received, and see whether there are any deficiencies in the near run. The simple household example would be dealing with a large future payment, like a roof repair. If there is not enough surplus cash flow coming in ahead of the payment, one might need to sell long-dated assets (possibly the house, which kills two birds with one stone).
For some levered financial firms, this could be a straightforward exercise — compare projected cash flows from fixed income holdings/lease payments versus cash flows from borrowings. For banks, the problem is that demand deposits are payable on demand. In the worst case, they could all theoretically disappear overnight.
In practice, this does not happen (outside of old movies and fevered internet theories). Even if a run occurs, it will take some time to get momentum. Depositors tend to keep fairly stable amounts on deposit. This explains why Basel III guidelines will refer to customer deposits as “stable,” as opposed to “non-stable” wholesale financing. It should be noted that one of rare developed bank failures due to liquidity issues (and not being shuttered due to negative equity) was Northern Rock, and that run happened in wholesale lending markets.
Risk managers and regulators therefore need to model scenarios for outflows from deposit liabilities to have a realistic idea of the liquidity position of the bank. On paper, the best models would be put together by data scientists crunching the extremely large database of bank transaction history. However, one cannot be too trusting of complex models of economic behaviour, and such modelling would end up being bank-specific — which would not be useful for regulators, who need to apply a “one size fits all” approach. That is, even if the bank’s internal tools are used, those tools need to be comparable to some standards.
I cannot claim any specific expertise in the area, but my understanding of the situation is that banks and regulators have a mix of sophisticated models of differing classes of assets and liabilities as well heuristics based on financial ratios. (E.g., divide deposits into classes, look at the ratio of the amounts of those deposits versus liquid assets.) Although the sophisticated models might make the life of the bank Treasury easier, my bias is that I would still rely on the financial ratios for overall risk assessment.
Another important area to model are the bank’s wholesale funding profile (i.e., funding from the “non-bank financial sector”). Since wholesale lenders are more concentrated than the many bank depositors, runs are more likely possibility. The standard strategy is to make sure that the bank has diversified its funding sources, both across markets and also by lender. So even if one market is disrupted, there will hopefully be another source of funding. This will again result in a variety of metrics to assess liquidity risks.
The analysis will end up focussing on both base case outcomes for inflows and outflows that would drive the liquidity management strategy, as well as risk scenarios, where plausible adverse outcomes are simulated and it is seen whether liquidity buffers are adequate.
The main macro implication of this discussion is that banks have to spend a good deal of effort to model their liquidity position. Outside a crisis, the models will tend to be adequate, and they will only matter to the banks and regulators. Banks are expected to have a safety margin on top of what the model says they can get away with. To what extent there are risks with individual banks, my guess is that they will crop up with aggressively expanding smaller banks, as the larger banks will have a large experienced bureaucracy watching liquidity.
Banks can tap long-dated funding markets faster than they can grow their lending books, as lending decisions still have to processed. As such, the main constraints on lending growth are the availability of capital (which can also be raised) as well the demand for funds the bank deems to be creditworthy. This means that central banks cannot hope to influence loan growth via messing around with the monetary base.
What about crises?
If a single bank is hit by some unusual transactions that drain liquidity, if it is viewed as credible in markets, it will generally be able to raise funds in wholesale markets, possibly at a somewhat higher spread. But if there are clouds over the bank’s prospects, it might find itself cut off from those wholesale markets. In this case, it might appeal to the central bank — but the central bank and other bank regulators will need to need to put the bank management’s feet to the fire. If there is a serious problem with the bank, it should be shut down.
A different scenario that has caught the most attention after 2008 was the case where pretty much all the banks ran to the central bank. This puts the central bank (and regulators) in an impossible situation: although shuttering one bank might be contemplated, attempting to restructure every bank is going to cause a major disruption to economic activity. This means that all the bankers end up getting bailed out.
Industrial capitalism does not sit serenely in equilibrium; in the real world, commerce is generally supported by rampant credit growth, deranged speculation, and wishful thinking. As long as the music is going, everyone keeps dancing. However, the music will eventually stop, and the credit markets will seize up. When this happens, the way to minimise adverse effects on bystanders is to prop up the banking system. Sure, one might hope that regulators are not as actively terrible as they were in 2008, and one can inflict some pain on the idiots involved (e.g., use contingent convertible bonds1), but lender-of-last-resort operations are going to happen.
In modern financial systems, regulators and bank risk managers have pushed most of the concentrated credit risk within the financial system to (non-bank) fixed income markets. Since problems almost invariably arise with those concentrated risks (loans made to the most exuberant borrowers), the credit problems show up in the non-bank fixed income markets. These problems will then rebound to banks, since they need to use the wholesale funding markets to diversify their funding away from deposits.
As such, I think it is a mistake to get excited about the plumbing of the financial system in macro analysis. Other than in the case of fraudulent management of financial institutions (which is a risk), credit markets blow up because non-financial firms are defaulting. Those defaults are usually the result of over-expansion of mortgage debt by households, or overly optimistic industrial firms (or both). That debt expansion will show up in the national accounts.
This is my minimal introduction to this topic. The only way I see that I could expand it is to offer a summary of one of the frameworks involved (probably Basel III) — which the Moorad Choudhry Anthology provides. I would need to take a look at my manuscript to see whether such an expansion is helpful.
A contingent convertible (a “co co”) is a bank bond that is converted into equity in the bank when regulators deem a financial crisis has arrived. This has the effect of diluting the shareholders in the banks, and inflicting a haircut on the bond investors who bought the convertible.