Recent Posts

Tuesday, June 25, 2024

Primer: Bank Liquidity Risk

This article is an unedited draft from my banking primer manuscript. It probably needs more work, but I will not be able to look at again for awhile.

One of the main economic functions of banks is providing liquidity to other actors – i.e., ensuring that clients can get funding on short notice. Banks are only able to do this by themselves carefully managing liquidity risk. Although the central bank can bail out the banking system if something goes horribly wrong, the expectation is that private banks should manage liquidity risk on their own.

Two key terms (that are sometimes confuse) that come up in discussions of liquidity are insolvency and illiquidity.

  • A firm is illiquid if it does not have cash on hand to meet immediate payment obligations. The firm may still have positive equity. Either the firm manages to cut a deal with creditors to keep going, or they may force the firm into bankruptcy (or alternative legal way to restructure payments). The firm still exists if it enters bankruptcy – bankruptcy is a legal status that prevents creditors from seizing assets in a disruptive fashion. Bankruptcy procedures vary by jurisdiction, but the usual objective is to try to preserve as much value in the enterprise as possible. However, if the outlook is dire, the firm may be liquidated and creditors paid off in order of priority.

  • A firm is insolvent if it has negative equity. Although the firm is barred from paying dividends, it is theoretically free to operate normally. However, the only financial institution that can hope to operate with negative equity is the central bank – lenders generally refuse to lend money to insolvent financial firms. A non-financial firm might be able to survive as the balance sheet value of assets may be far lower than their true economic value, and so the negative value of equity is misleading.

The meaning of “liquidity” is somewhat uncertain within most financial and economic commentary, but liquidity risk is a meaningful term. It is the risk of becoming illiquid (by the above definition).

Liquidity Risk in Non-Bank Finance

Understanding liquidity risk of financial firms that do not take deposits is the best starting point. We assume that we have a firm that owns debts instruments that have fixed payment terms, and it in turn issues debt instruments in wholesale markets that have fixed payment terms. (For now, we put aside any uncertainty about future floating interest rates.) Since both assets and liabilities have fixed payment schedules, we can project incoming and outgoing cash flows under the assumption of no credit losses on assets. (Credit losses are credit risk, which is a separate risk management concern.)

Using the cash flow projection, we can see if there are net outflows that cause the firm’s cash to run out over some horizon.

For example, imagine a firm has $100 in cash on Monday. On Tuesday, it expects to receive $50 from a borrower paying back a loan it has. On Friday, it must make a payment of $200.

The firm’s liquidity position is fine from Monday to Thursday – it will have $100 cash on Monday, and $150 after the borrower repays the loan. The problem shows up on Friday – it must make a payment of $200, and it will only have $150 in cash (unless it undertakes some transaction in the meantime).

The firm has a few basic strategies to deal with this projected shortfall before Friday.

  • Sell another asset for at least $50.

  • Borrow against another asset (that is not already being used as collateral) for an amount greater than $50, with the time of repayment beyond Friday.

  • Issue new debt for an amount greater than $50 with a maturity beyond Friday. In particular, the firm may be able to roll over the borrowing from the original source that lent the $200.

  • Draw on a credit line, which is the last line of defence. (Alternatively, run to a central bank liquidity facility is possible, which is an option that is always open to banks, but sometimes central bankers have made their facilities more widely available.)

One key point is that most capital market instruments have fixed payment terms, and so we can project future cash flows on such assets with the only uncertainty from a liquidity perspective being credit risk. Some instruments may allow flexibility in payments (termed embedded options) that can be dealt with by using a bit of fixed income mathematics. The most economically significant form of embedded option is the ability of homeowners to prepay residential mortgages (which keeps a lot of American fixed income quantitative analysts employed). However, prepayments matter for interest rate risk – from a liquidity risk perspective, the payments just occur faster than expected, and so they reduce the risk of a cash shortfall. What matters for liquidity risk is the option to extend payments. It is therefore no surprise that this is an expensive option to add to a debt contract.

Maturity Mismatch

One key empirical regularity of finance is that we see that there is a great interest in holding short-dated assets that are used as a cash buffer, while borrowers tend to prefer longer-dated maturities on their borrowing. This creates a maturity mismatch between the supply and demand in the “lending market.” The financial sector sits in the middle of this mismatch and attempts to make a profit by bridging it – hopefully without blowing itself up. Traditional banks famously do this by taking on deposit liabilities and holding long-maturity loans, but the non-bank financial sector also is important as it issues money market instruments.

Imagine that we control a non-bank financial entity and want to finance an attractive loan. The borrower is an industrial firm that wants to borrow $10 million and will pay the one-month reference interest rate plus 100 basis points (1%) as interest. We are incredibly sure that this issuer will not default on this loan. We can reliably borrow using 1-month commercial paper at a spread of 20-50 basis points over the one-month reference rate. What the reference rate is depends upon the jurisdiction, it probably would have been LIBOR (London Interbank Offered Rate) in the old days. For our purposes, it is safe to assume that the reference rate will move up and down with the central bank’s policy rate, although things are slightly complicated by the fact that central banks policy rates tend to be overnight, while we are looking at a one-month reference rate. (Assuming that there are no other premiums, the one-month rate equals the expected (geometric) average of the overnight rate over the following month.)

What we can do is make the loan, and issue $10 million of commercial paper that is rolled over every month to finance the position. So long as we can issue the commercial paper at a spread of less than 100 basis points, the transaction generates a steady interest rate profit. At the loan maturity, we use the proceeds to pay off the commercial paper borrowing.

Opportunities like this exist because investors park considerable amounts of cash in money market funds, and those funds need to buy high quality short-term instruments like our 1-month commercial paper. This large pool of buyers that are forced into a small subset of total debt instruments creates a mismatch that results in the spreads of high-quality money market instruments being lower than other parts of the credit market.

There are three risks to this transaction.

1.      The industrial firm defaults. We are exposed to the credit losses on the loan. This is termed credit risk.

2.      The spreads for our firm’s commercial paper rises above 100 basis points. We would then be facing an interest rate loss. Although this running loss would likely only be a small percentage of the $10 million position size, the entire business model of the firm may be at risk unless this situation reverses. This risk might be called “cost of finance risk.”

3.      We are unable to roll over the 1-month commercial paper. This could either be the result of our firm’s credit quality weakening, or a widespread seizure of the money markets. We would need to find an alternative way of funding the $10 million loan (or somehow sell it to repay the commercial paper). This is called “rollover risk” (we are unable to roll over the commercial paper).

However, the transaction is not exposed to interest rate risk – which refers to movements in the risk-free yield curve (or the reference rate). Both the asset and liabilities have interest rates that are effectively a spread over the reference rate, so we are immune to movements in the reference rate. This is even though the asset we own is a 1-year debt, while we are issuing 1-month commercial paper to finance it. One common error I see in discussing this topic is that people compare the maturities of assets owned by banks versus the maturities of their liabilities (deposits having an immediate effective “maturity”) and state that this exposes the banks to interest rate risk. To the extent that the assets have interest rates that float, they are matched to their liabilities that also float. (Although demand deposits – chequing accounts – often do not pay interest, saving accounts do.)


The analysis of traditional banking liquidity risk is complicated by the properties of deposits. As noted above, a demand deposit is a liability of the bank that the holder can redeem at any time. (Some deposits have limitations on when they can be redeemed.) Theoretically, a bank might need to cover all deposits simultaneously being lost in the same day. In practice, such a total loss is unlikely, but banks can experience rapid deposit losses (called bank runs).

Traditional banking would not be a viable business if banks prepared for a total loss of all deposits. (That said, crank economists propose full reserve banking to be able to weather such a scenario, as discussed in Section TK.) Instead, the banks need to have prepare for plausible deposit outflow scenarios. They will need a liquidity buffer based on this analysis. I would divide the type of analysis into three broad categories.

  1. Simple rules based on high level balance sheet items. The analysis will usually look at the ratios between deposits (and other short-term liabilities) and assets that are deemed to be “liquid.” Since they are expressed as ratios, they can be called “liquidity ratios.” These liquidity ratios are what people were stuck with in the era before widespread use of digital computers in risk management and financial analysis. These ratios are probably adequate for high-level analysis by outsiders, like bank equity analysts.

  2. More complex liquidity ratios that are mainly used by regulators that are supposed to better match banking risks based on experience. (These measures tend to be great at identifying problem banks in past financial crises.) Since the regulators look at these ratios, bank management has to keep track of them as well. These ratios are generally “one size fits all”: they are the same for all banks (of a certain size) within a banking system. (Banking regulations can distinguish between different classes of banks, with larger banks generally kept on a tighter leash.)

  3. Liquidity risk analysis done internally by the bank, which should be done using a variety of techniques. For example, historical data might be used to define probability distributions of potential deposit losses over a horizon. This analysis would ideally include seasonal effects that a particular bank’s customers create. It can also include scenario analysis, which looks at what would happen if a particular historical crisis was replayed (or plausible unusual behaviour by depositors). Regulators might want to see the results of such analysis, but they cannot rely on it for the simple reason that the bank has an incentive to understate risks in presented data.

An example of a simple liquidity ratio is the Liquidity Coverage Ratio (LCR). The ratio is the stock of “high quality liquid assets” divided by the net cash outflows over the next 30 days. As described in the book Bank Asset and Liability Management, the Hong Kong Monetary Authority (HKMA) took a two-tiered approach to applying this ratio in 2015. Twelve large local banks (liabilities above HK$250 billion) had to have keep a LCR of 100%, while other banks kept the earlier target of 25%. Since larger banks cause much more grief if they fail, they are held to a higher standard.

Attempting to describe the details of other liquidity ratios that are in use are beyond the scope of this book. The regulatory environment constantly changes, and so the ratios I might cover might no longer be in use by the time it is read. Additionally, describing all the variables in the formulae would require delving further into bank accounting, and require more definitions. However, these details would not be needed in the rest of the text.

Bank Runs

Large deposit outflows – bank runs – feature strongly in popular financial/economic commentary. People have seen too many old movies of small American banks getting wiped out by a line of people standing in line to withdraw their money. Every so often, one sees people who tend to go on about “fractional reserve banking” cooking up schemes to collapse the banking system by getting a handful of people to withdraw their savings from their banks.

The problem with the American experience is that not many other banking systems feature banks that just serve some small town. For such a bank, having a good portion of its client base standing in line at its one branch to withdraw funds is a death sentence. For a large bank, this scene would need to be repeated at a lot of branches across the country.

For non-microscopic banks, runs are a threat – but these runs happen in the wholesale funding markets. Institutional lenders spend their day obsessing about credit quality of their lending targets, and they are a small, interconnected group. (Fixed income investing outside of high-risk lending is a low margin, economies of scale business. This means that the investor base is far more concentrated than in equities, which also means that information spreads very quickly.) We do see retail bank clients waiting to withdraw funds from a failing bank – but that usually happens long after the bank was effectively crippled by losing access to the wholesale funding markets. The retail deposit run looks cool on news reports, but it is not adding much information.

(Outside of the floating currency sovereigns, retail depositors lining up to withdraw funds due to issues with currency conversion is common. Banking with foreign currencies is an inherently bad idea, and so such events are not surprising. Such events are outside the scope of this text.)

In 2023, there were bank runs by depositors in the United States which attracted a lot of attention. These were regional banks that were hit hard by interest rate losses. However, one of the key banks in the drama was unusual in that it had clients in the technology sector that kept deposits at the bank that were well above the deposit insurance limit. This is a form of wholesale funding (from firms and individuals with rather dubious cash management strategies), even though they were depositors. Not many banks manage to rope in that many dysfunctional clients, and the more usual outcome is that the wholesale funding run happens in other markets. There is a New York Fed staff report by Cipriani, Eisenbach, and Kovner that discusses the run (see references), and it notes that the run took the form of large depositors fleeing.

Concluding Remarks

Bank liquidity risk management is a core part of banking, along side credit risk analysis. Although it is possible to develop sophisticated models of the usual behaviour of bank counterparties, there is still a large amount of guesswork about their behaviour during times of financial stress.

Since the objective of this book is not to provide a manual providing the details for running a bank, we can just focus on the following. Banks need to keep on hand liquid assets to meet potential outflows. The minimum amount of liquid assets required will be set by regulatory ratios, but a sensible bank is going to have to keep a buffer above those regulatory minimums. How much of a buffer is required will depend upon the analysis and strategy of bank management.

References and Further Reading

  • Cipriani, Marco, Thomas M. Eisenbach, and Anna Kovner. 2024. “Tracing Bank Runs in Real Time.” Federal Reserve Bank of New York Staff Reports, no. 1104, May.

  • More details are in the books The Moorad Choudhry Anthology and Bank Asset and Liability Management, full references in Section 2.2.

Email subscription: Go to 

(c) Brian Romanchuk 2024

No comments:

Post a Comment

Note: Posts are manually moderated, with a varying delay. Some disappear.

The comment section here is largely dead. My Substack or Twitter are better places to have a conversation.

Given that this is largely a backup way to reach me, I am going to reject posts that annoy me. Please post lengthy essays elsewhere.