This article continues the previous discussion of bank net interest margins. In it, I discussed how changes in the yield curve changed the net interest margin (NIM) for banks. This showed up historically — when bank balance sheets were shattered by the combination of holding long-dated mortgages with low fixed coupons versus having a sky-high short-term rate imposed by deranged Monetarists. In this article, I address a common macroeconomic story: yield curve inversions cause recessions by the alleged effect on NIM. As a spoiler, I do not think that story holds water in “modern” banking systems.
Animal Spirits?
The first thing to do is to distinguish the mechanism described here from another similar-sounding story — one that I see as plausible. The alternative mechanism is as follows. Yield curve slopes are widely followed indicators, and if key slopes invert, many people will take that as a recession signal. Bank loan officers could easily be in that population — and recessions tend to lead to increased credit losses. They would predictably tighten up lending standards. Given the importance of credit to finance fixed investment, this then leads to a recession.
This is a story about “animal spirits,” and fits into post-Keynesian views about the business cycle — which I describe in Recessions: Volume I. (The projected Volume II is still missing key content, but it is projected to be heavily weighted towards interest rate effects. As such, a version of this discussion might make it into that text.) As I see it, “animal spirits” are the key not-directly-measurable variables behind post-Keynesian thinking about business cycles — as opposed to r* (“neutral interest rate”) in neoclassical thinking.
To the extent that this story can be quantified, one could look at the yield curve, the results of loan officers surveys (e.g., percentage that are tightening standards), and the growth of bank credit.
Stock Versus Flow
In macro analysis, I often see references to the aggregate net interest margin of banks, as in the series above. (I showed this figure in the previous article. It was produced by the FDIC, but discontinued in mid-2020.) The problem with this series in this context is that it is based on the total stock of assets and liabilities on banks’ balance sheets, whereas the effect that we are looking for is supposed to be based on the flow of new deals.
As I previously discussed, “modern” banks now manage their interest rate risk (often undertaken by a specific asset-liability matching (ALM) desk). To the extent that interest rate is immunised — assets matched against liabilities with the same interest rate profile — changes to the yield curve (both changes to the level or slope) do nothing to the interest margin on an economic basis. (Accounting principles might create phantom gains/losses that are ultimately reversed.)
In order for the short rate to 10-year slope to matter (for example), the bank needs to be running a book of 10-year loans that is funded by short rates. This of course creates a duration mismatch. But even if this mismatch is allowed, changes to the slope is going to be a small factor in the global interest rate margin when the level of rates is changing. The entire loan book would have large change to a move in the short-term rate, whereas only a small portion of the book would be rolled over each year at the new slope. For example, if the bank had a book of loans originally issued with a 10-year maturity and the book size was constant, 10% of the loans would roll over each year. In practice, the rollover would tend to be higher — loans might be paid off early, and the book size will likely grow over time.
As such, I would expect the aggregate NIM to basically move with spread trend of mainly short-term loans, along with some jumps up/down in response to rapid short-term rate changes if there is a global duration mismatch. If I squint at the chart, the data seems roughly consistent with that story, and I do not see a reason to waste the reader’s time pursuing it.
What about the behavioural effect on new loans? I will now go back to bank structures to explain why I think that story does not really work.
Transfer Pricing
I will return to a topic mentioned in my introduction to bank treasury operations: transfer pricing. The first thing to note is that different banks will have different institutional structures, I am giving a simplified overview of how most large banks are structured. I accept that I may be mangling some of the jargon, and over-simplifying the particular tasks of different teams.
The bank treasury team sits in the middle of all the divisions within the banks, with some desks raising funds in different markets, and others consuming the funds (e.g., commercial and household lending groups). They effectively act as market makers within the bank. The pricing signals provided by transfer pricing will influence the behaviour of the other teams within the bank without having to have long discussions between multiple teams.
The pricing of internal funds is a zero-sum transaction for a bank within a legal jurisdiction, so the transfer pricing is only of external interest when it touches divisions in different countries — as taxable income is transferred across borders. (The taxation effects of internal transactions of multinationals is the main source of interest in the term “transfer pricing.”)
The lack of visibility of transfer pricing does not mean that it is unimportant — poorly thought out transfer pricing can lead a bank to drifting into awkward positions that can cause difficulties, as in the Financial Crisis. One typical issue seen across firms in that era was that liquidity risk was not being properly priced internally, leaving many firms scrambling for liquidity once it was no longer possible to fund some classes of risky assets.
Simple Example For A Lending Division
How a bank communicates its internal pricing is going to be institution-specific. However, we can think of it being like a market maker in fixed income markets — the treasury team sends out a pricing sheet giving levels for buying/selling funds, with the pricing depending upon the risk characteristics of what the funding is used from (or where it is being raised).
A lending team might be confronted with a “pricing run” for high quality industrial loans which includes the following entries (where “FR” = whatever floating rate the bank uses as an internal entry).
- 6-month loan: FR + x%.
- 5-year loan, floating quarterly payments: FR + y%.
- 5-year fixed rate: {market 5-year swap rate} + z%.
The lending group then will offer customers who meet their lending requirements a menu of lending options, which would be the above pricing plus whatever credit premium they think they can get away with. E.g., borrow 6-months at FR + x% + s%.
The profit/loss for the lending group will be determined by the excess returns of their loan book over the internal costs of funds they used to fund the loan book. From a simplified accounting perspective, this will equal excess spread they charged clients, less credit losses and operating expenses. From the perspective of the lending team, they do not need to care where the transfer pricing costs come from.
Some added observations.
- Although the 6-month and 5-year loan are both floating rate, the liquidity profile is different. Unless the bank had just accessed cheap 5-year funding, we would expect the spread on the 5-year term (y%) to be higher than the six month (x%). The difference between y% and x% might be called the “liquidity premium” or “term premium.” (I prefer term premium, given the multiplicity of meanings to “liquidity premium.”)
- The pricing should be dependent upon the risk characteristics of the loans — although that would depend upon the bank’s institutional structure. (For example, could imagine a simpler structure where the lending group has risk limits on the entire loan book, and it has a flat cost of funds that reflect the average risk profile of the loan book.) Riskier loans imply more capital requirements — which raises the weight cost of capital of the bank. But even if the capital requirements for two classes of loans might be the same, transfer pricing might be even used to discourage the growth of certain classes of credit risks.
- The lending team’s funding of the loans matches the loan payments. This means that loan officers are duration and liquidity matched at all times — so they have to make their profits by evaluating credit risk, and not making bets on interest rate changes. Allowing different divisions to make interest rate bets against each other is a great way for a bank to pay lots of bonuses based on activity that generates no aggregate profit for the bank.
- Banks may have funding groups operate in the opposite manner of this loan group example — they go out and gobble up needed funds in various funding markets, and their profits are determined by how much they can get funding below the pricing of the treasury group. The objective for pricing here is to provide an incentive to diversify funding sources to reduce the risk to the bank.
Hedging with Swaps
Let us imagine that a bank pushed out $50 million in loans with a 5-year fixed rate. The loan group would get 5-year funding from treasury, which means that the treasury is receiving payments of the 5-year swap rate +z%. This would initially be “funded” by the $50 million increase in deposits of the borrowers at the bank. Once the loan proceeds are spent, some of those deposits would be transferred to other banks, and so the treasury team would need to get the funding team to get their hands on short-term funding. This raises funding for the treasury team, balancing out its position, as well as the position of the bank.
The treasury team funding profile mismatch (long-dated asset against the lending team, short-term liability with the funding team) corresponds to the bank’s net position (since the other divisions have matched profiles versus treasury). The asset-liability matching (ALM) team would probably need to do something about the mismatch.
One option to eliminate the interest rate risk is to enter into a $50 million 5-year swap contract — where the bank pays fixed and receives the floating rate. These swap payments will cancel out the fixed payments received on the loan for the bank, and for the treasury team’s internal positions.
So long as the swap market is functioning — which admittedly is not always the case — executing the hedge is essentially trivial. However, unless the financial crisis is extreme (like 2008), the freeze ups in the swap market tend to be short-lived. The bank is therefore not deeply concerned whether the bank customer picks the fixed coupon option.
Inversion and Bank Behaviour
It is now possible to see why I am unconvinced that yield curve inversion has a causal effect on modern developed banking systems.
- From the perspective of loan officers, they just offer loans based on a markup over the transfer price of funds. They are going to offer customers multiple options, so they are going to worry about creditworthiness, and not the yield curve.
- If the bank has access to the interest rate swap market, it can easily dump off the interest rate risk associated with fixed rate loans. So, there would be no reason to be stressed about customers taking that option.
- Residential mortgages can feature longer fixed coupon terms. However, developed countries have deep markets for securitisations (the residential mortgage-backed security — RMBS — market). The bank will typically warehouse a small amount of interest rate risk until enough mortgages are gathered to be securitised.
- The bank can raise longer-term fixed financing — term deposits, or bonds. This creates a hedge that gives capacity for more fixed interest lending.
Large banks in developed countries can now easily access all of the above options, and all banks could at least attempt the last option (secure long-dated fixed funding).
What happens if the bank is not able to hedge the interest rate risk? The solution is for the treasury team to move the transfer pricing — raising the spread z% in my example. Customers are not guaranteed fixed spreads to government securities; the banks just need to nudge them towards floating loans. Although one might object that the bank cannot raise that spread without being uncompetitive, that objection is not consistent with the theory that nobody is willing to lend at a fixed interest rate for long terms.
In summary, the bank is either going to be indifferent to customers opting for fixed rates, or at worst, would adjust the spread accordingly.
Concluding Remarks
As a result of the behavioural analysis here, I see no reason to believe that yield curve inversion will have a mechanical effect on bank lending in the modern era in the developed countries. (Things were different before the 1990s.) The quenching of animal spirits by recession fears is going to be far more important for determining loan growth (both on the demand and supply of loans side) — although quantifying “animal spirits” is always going to be a slippery task.