This article continues the discussion started in Part I of this series on a banking debates — but does not finish it. The previous part gave a fast-paced overview of how generic financial firms that operate in fixed income view the world, along with some of the special properties of banks versus those other investors. In this, I return closer to an ongoing debate in economics: what limits bank lending? This article attempt to give a simplified view from the perspective of an individual bank — without attempting to discuss the whole financial system.
I expect that many economists will be unhappy with the discussion. Allegedly, I will have missed important factors. Those will be discussed in a follow up article, but they are not answering the question of how a bank looks at the problem. The actual perspective of a bank with regards to its balance sheet strategy will have no resemblance to Economics 101 stories.
The Missing Constraints Puzzle
If we read texts by many serious conventional economists about banking, they will insist that there are constraints on bank lending. (Admittedly, some mainstream economists have moved towards heterodoxy on this topic.) Central bank policy is the driver behind these constraints, although the exact mechanism depends which strand of neoclassical orthodoxy that the writer subscribes to.
If central bank policy steered loan growth as described, one can make an empirical prediction: we should see continuous reports of banks announcing that they are unable to extend loans, since they are continuously bumping into “constraints” set by central bank policy.
However, I think it is safe to say that outside of a financial crisis or a regime of quantitative credit controls, one will never hear bank loan officers at well-run banks turning way all borrowers. Which is to say, the empirical prediction of conventional banking analysis is incorrect. Although one might say that the heterodox thinking on constraints fits the story, I believe that there is a much cleaner way of explaining the puzzle. I will return to this puzzle after I run through how a bank sees the problem.
Finally, my italicised statement is deliberately convoluted, as there are important disclaimers that need to be understood.
No quantitative credit controls. If regulators explicitly limit the size of loan books, we expect banks to hit those limits.
Well run bank needed. An incompetent bank (think of the small banks in the United States, like Eva and Oliver’s Bank of Hooterville) might do any number of stupid things, and an individual bank teetering on insolvency might stop lending activities as a desperation move. (This points to the reason why banks do not announce that they are stopping lending: it is a signal that something has gone horribly wrong.)
No Financial Crisis. If all banks are in trouble, each bank is free to turn away borrowers, since there are no good banks for customers to run to. Growing up, my mom used to say to me “Would you jump off the Redwood Bridge if all your friends did it?” If you want a career as a senior banker, you want to answer “yes” to such questions.
All borrowers need to be turned away. Banks have diversification risk limits, and will cut off borrowing to certain classes of borrowers. For example, I believe this happened when foreign bank subsidiaries operating in Japan hit their risk limits with Japanese banks, forcing them to buy treasury bills at negative rates even though the policy rate was still 0% at the time.
I will now outline the typical structure of a large (traditional) bank — ignoring financial subsidiaries like broker-dealers. I want to emphasise that I am giving a simplified generic description, based on my experience as an analyst, as well as based on discussions with portfolio manager colleagues who mainly had bank treasury operations backgrounds. My terminology may be shaky, different banks have different structures, and I am deliberately ignoring some of the complexities involved.
A large bank has a hierarchical structure, split by legal jurisdiction. A parent company might span multiple currencies, but each traditional banking subsidiary necessarily is run a stand-alone business, although it might draw on external financing hedged into the local currency. The bank will operate multiple branches, and loan officers have hierarchical seniority, with the allowed loan size increasing as one moves up the hierarchy.
Each division of the bank runs its own book of business, and typically operate in a floating rate world: their assets and liabilities are internally swapped into floating rate equivalents. The Treasury team of the bank is a central team that collects the information and transactions of the other units, and then undertakes operations in the capital markets to manage the bank’s aggregate risk book.
The Treasury team will not directly communicate with loan officers, beyond the most senior team. There are too many loan officers spread over the country, and the loan officers might leak information if they knew the global positions of the bank. As such, a central team will set global policies on loan pricing and standards that are then pushed down to the loan officers. An individual loan officer then tries to make good loans within that global framework. (This means that any economics professor who insists that loan officers look at the bank’s position before making any loan is strictly incorrect.) In the case of a very small bank or very chunky loans, the senior team might have to keep an eye on the global position of the bank.
Bank Balance Sheet Constraints
Banks are creatures of regulation. They have special advantages versus other financial firms, but they pay the price of having regulators poring over their every move. Although almost everything they do has some regulation associated with it, I will just discuss the basics of the regulation of their balance sheets. I am delegating “bank reserves” to an appendix, since it is unclear to me which developed countries still have required reserves.
As discussed in Part I, the main risks a firm faces is insolvency (losses dropping equity to zero or negative) or illiquidity (an inability to make a payment). Each of these risks is targeted by regulation.
The next layer of constraints would be implicit ones created by rating agency methodologies. Banks need to keep a relatively good rating in order to be able to profitably run their business, since a corporation’s spread is somewhat tied to credit rating. (Ideally, credit analysts at bond trading shops have internal ratings that they use to rank credits, but those typically end up close to the rating agency ratings for higher quality credits.) Lending to customers below your cost of capital is a negative margin business, and you can’t make it up on volume.
Furthermore, banks have their own risk management frameworks that are layered on top of regulations and rating agency guidelines. Unless the bank aims at explicitly breaking the law, those internal risk management frameworks are more stringent than the regulatory ones — which represent minimums.
One of the things to keep in mind — and which is not kept in mind by many economist discussions — that banks do not have complete control over their balance sheet. As emphasised in the previous article, they are in the liquidity management business, and they need to react to often unpredictable customer actions. (Banks do try to predict behaviour, as certain things do follow patterns.) As such, for a well run bank, it will always start the business day with buffers between its balance sheet condition and the regulatory minimums (with perhaps a few exceptional limits that can be controlled, like hitting an end of day settlement balance target).
Regulations that are aimed at insolvency include the following.
The main set of regulations revolve around the ratios between risk-weighted assets and bank capital. The premise is that risky assets cannot get too large versus the “capital buffer” of the bank. Bank capital is confusing (including to me), but it includes equity as well as instruments like preferred shares and deeply subordinated debt. The idea is that the the bank can stop payments on those instruments without impairing senior debt holders and depositors. (Remember that bank deposits are a liability of the bank, even though they are are an asset for depositor.) The “risk weighting” of assets is also somewhat confusing, but the idea is that every class of bank asset has a weighting of its riskiness. If an instrument’s weighting is 50%, then it’s risk-weighted size is one half the amount of the position in the instrument. This risk-weighting is needed to avoid obvious problems. Imagine that a bank gets a large inflow of deposits (a liability). If the inflow was into an asset at accounted at 100% of its value, this means that the ratio of assets to capital would fall — which would mean that getting deposits would be “bad.” However, the bank can buy nice 0% risk weighted Treasury bills with the deposit inflow, and the amount of risk-weighted assets would not change. (Realistically, those Treasury bill investments would be drawn down later, which would lower the capital ratios — but at a pace the bank controls.)
Concentration risk. There are limits on the maximum loan size for banks. Any thought experiment involving extremely large loans are fictional, since they are barred by these risk limits. Furthermore, the hierarchical nature of loan officer organisation means that loan sizes are limited by the finite amount of loan officers.
Diversification risks. Banks will have limits on how large exposures are to certain classes of borrowers. Other than concentration risk, this is one of the few constraints that might be hit in practice.
For solvency risk, there would be requirements for minimum holdings of “liquid financial assets” (which would be jurisdiction dependent) based on the profile of their liabilities. (More short-term liabilities implies a greater need for holding liquid assets.) These liquidity regulations are specific to jurisdictions, as money market trading structure varies from currency to currency.
Banks also need to worry about other risk measures like currency and interest rate risk. Although these constraints influence pricing, they are not the topic of consideration here. All that needs to be said is that the interest rate swap and currency swap markets are deep for very good reasons.
What Happens When a Bank Extends a Loan?
We can finally get to the detail that causes great excitement in economists’ discussion of banks (but it shouldn’t): what happens when a bank extends a loan to a customer?
The simple version of the immediate effect of extending a $100 loan is as follows:
The customer gets an increase of $100 in the deposits held at the bank (an asset), but now has a new $100 debt liability. Therefore, the customer balance sheet expands by $100.
The bank gets a $100 increase in deposit liabilities, but gets a new $100 loan asset. The bank’s balance sheet also increases by $100.
What causes excitement is that this balance sheet growth “comes out of thin air!” However, this excitement is misplaced — instantaneous creation of financial assets happens all the time, it is just that people seem to have only a hazy idea how businesses in the real world operate. I will get back to this point in the follow up article.
This immediate transaction will interact with bank capital regulations: the risk-weighted assets rise, but capital does not. Therefore, the loan extension eats in the capital buffer of the bank.
We then expect the customer to spend the proceeds of the loan. Unless the proceeds goes to another customer of the same bank, there is an outflow from the bank to the payments system. This will then eat into the liquidity buffer of the bank.
At this point, the Economics 101 fan will say that the extension of the bank loan is hitting constraints! Although some variant wording of such a statement might be correct, it is not how a bank would describe the situation.
Other than a mega-transaction that requires coordination with the Treasury, no individual loan is big enough to make a noticeable dent in the capital/liquidity buffers a well run bank starts the day with. Bankers do not worry about the balance sheet impact of individual loans, the senior team looks at the entire book of loans. Furthermore, one day’s output by lending officers is not enough to move the dial on the entire bank’s balance sheet (unless something has gone horribly wrong in risk management): banks have to worry about the trend in their loan book growth. As the loan book expands, it tends to eat up capital and liquidity buffers. Part of running a bank is deciding upon a strategy to manage loan book growth. More on this point below.
The liquidity effect of outflows of loan extensions is greatly over-rated in economists’ stories. Banks are in the liquidity management business. They do not just hand out spot loans — they also hand out/sell credit lines, which are options on loans. These are non-trivial. If we look at page 90 of the 2021 Annual Report of RBC (Canada’s largest bank), the off balance sheet exposures for undrawn credit lines/financial guarantees are C$300 billion — which is 27% of the total deposits of the bank. Given the magnitudes of such contingent exposures, the bank is not spending time worrying about the fact that the proceeds of your $1 million dollar mortgage are likely to go walkies.
In summary, the effects of loan book growth are known, and are handled by bank’s growth strategy. A well run bank will not hit its regulatory minimums for capital and liquidity ratios in practice. The next section explains this further.
“We Ain’t Hitting No Stinkin’ Constraints”
A bank might allow its loan book to grow fast enough that its capital and liquidity ratios trend lower. (Assuming the bank is profitable, the profits would naturally improve capital and liquidity ratios, so growth needs to be fast enough to overcome this.) It then makes strategic transactions to periodically fix its ratios.
Securitisations. Selling off risky assets in securitisations is a two-fer: it reduces risky assets to improve capital ratios, and raises liquidity.
Issue term debt, and/or raise rates paid on term deposits to improve liquidity.
Issue “capital securities” to shore up capital ratios.
The toughest nut to crack is bank equity. Ideally, you want to do this out of retained earnings. Most banks have a relatively high target for returns on equity. For example, if we look at page 6 of the RBC Annual Report, we see that they have a return on equity (ROE) target of 16%. If a bank gets a 16% ROE and pays out half of that in dividends, that means that equity growth is 8%. Such a growth rate is more than enough for a large bank in the current relatively low nominal GDP environment, but it might take a lot of securitisations to allow very high lending growth.
The thing to keep in mind is that other than equity issuance, all are available if the bank has “market access”: is it able to issue securities at acceptable prices? From the perspective of a banker: this is always assumed to be the case. Why is this assumed? Banks are in the liquidity management business. If a bank loses the ability to tap markets, it is rapidly on the way to achieving ex-bank status. (Once again, a central bank bailout of the entire banking system is a “get out of jail free” card.)
I argue that the correct phrasing from the individual bank’s perspective is as follows: the main constraint on the bank is the need to keep access to funding markets. So long as it does that, it can issue securities to keep capital and liquidity ratios on target.
If we drop the focus on the “constraints” and use a more sensible phrasing, we can ask what factors determine bank loan growth in aggregate? The heterodox phrasing is often something like “demand from borrowers.” I would argue that the cleanest version is as follows.
An individual bank has constraints on the maximum loan size it can extend in a day, determined by regulations and internal structure (e.g., finite number of loan officers, each with their own lending capacity). In practice, actual loan growth will not even be close to this theoretical maximum. Instead, loans are extended based on the interaction between borrowers and loan officers, which will take into account the standards of lending which in turn affects the risk-based pricing for the loan (and whether the loan can be made at all). Even if a loan “would be nice to have” for a borrower, a sufficiently high interest rate will make taking it uneconomic.
But, But, Funding Isn’t Guaranteed!
In the next article, I will turn to situation of the overall financial system. Why is it that banks “with market access” blithely assume that they will be guaranteed to tap funding? Since that requires a discussion of the entire financial system, we need to go away from the internal perspective of the bank.
Appendix: Ye Olde Bank Reserves
Courtesy of out-of-date economics textbooks, “bank reserves” often pop up in discussion. These “required reserves” are not the same thing as “loan loss reserves”: which are how banks account for expected credit losses in a book of loans.
This topic is discussed in my book Understanding Government Finance. The system in the United States used to be that banks needed to hold settlement balances at the Fed equal to 10% of certain classes of deposits. This was different than the Canadian system (before things got messed up in 2020), where banks were expected to have a $0 settlement balance at the end of the day.
In practice, the two systems were dealt with in a similar fashion. The part of the American system that economists like to skip is that required reserve balances were determined with a lag, and so each day, the bank just had to hit a fixed target. All this means is that a certain fixed portion of the bank balance sheet had to be frozen as settlement balances at the central bank, after which they behaved like the Canadian banks trying to hit that target.
Of course, any bank with “market access” is able to undertake transactions to hit that fixed target.
The problem is that certain economists believe that the central bank can arbitrarily set the amount of reserves in the system, and so banks would somehow adjust lending to hit reserve targets. In practice, it did not work that way — even banks that targeted the money supply effectively changed interest rates to try to influence money growth (in the same way they set interest rates in the hope of influencing the price level).
The reason why this subject comes up is that many economic models — both neoclassical and heterodox — have an over-simplified description of “money,” and there is a simple relationship between the money supply and the policy rates. These models are apparently taken literally, and thus some people seem to believe that central banks can choose to either set a policy rate or the level of the “money supply.”Email subscription: Go to https://bondeconomics.substack.com/