It’s practically an article of faith in some quarters that higher interest rates benefit banks. But is this really the case? Recent experience has led some to challenge this assumption. Rates have started to trend up again with 2024 rate cuts looking less frequent and less likely. A closer look at the evidence suggests that the impact of higher rates on banks is more ambiguous than many assume.
The Wall Street Journal wrote in December 2021 that “Banks yearning for interest rates to rise appear on the verge of getting their wish.” The logic is that higher market rates allow banks to charge more on loans and receive higher yields on investments, while the rates they pay on customer deposits lag. Significantly. Even after the sharp and protracted rise in market interest rates, Chase and Bank of America pay only 0.01% on Savings. (CDs are another matter.) A recent article in Investopedia confidently stated that “A rise in interest rates automatically boosts a bank’s earnings.” (Emphasis added.)
This rosy picture contrasts with the three large bank failures in 2023. Higher rates played a role in each failure. A more recent WSJ article (4/12/2024) suggests higher rates might not be quite the boon they once were, as net interest margins at the three largest U.S. banks shrank during Q1 2024.
There are two primary types of interest rate risk measures: earnings at risk (EAR) and economic value of equity (EVE). EAR focuses on potential changes to net interest income, usually over the next year. EVE focuses on changes to the present values of assets and liabilities, which ultimately affect the economic value of equity. The two measures don’t necessarily show similar risk exposures, in terms of either direction or severity. Publicly available IRR measures can be hard to come by. As I’ve noted in an earlier post, banks’ IRR disclosures leave a lot to be desired. Disclosures are usually limited to EAR and moderate rate shocks of 100 basis points.
The OCC publishes a quarterly Interest Rate Risk Statistics Report. The report shows EAR and EVE sensitivity for national banks and federal savings associations across a range of rate shocks, as indicated below:
The EAR data suggest rising rates would, on average, lead to a modest improvement in NII. A significant minority of banks (25th percentile), would see shrinking margins, especially for more severe rate changes. The EVE data show that rising rates would reduce economic value for the typical bank. These exposures get quite large for about a quarter of banks.
There are some significant limitations with the data. State-chartered institutions, which accounted for all the big 2023 failures, are excluded. OCC primarily collects the IRR data during examinations rather than quarterly, so some of the data may be up to 24 months old. Most significantly, the data only cover community and midsize banks.[1] OCC also collects IRR information for the largest banks, but they are not included in this report.[2] Based on what I did see while I was at OCC (albeit more than a year ago), EVE exposure to higher rates is also common at larger banks.
There are typically two approaches to estimating EAR: static and dynamic. Static assumes a fixed balance sheet. Runoff, in terms of amortization, prepayments, and deposit attrition, occurs but is replaced on a like-for-like basis. Dynamic allows for growth and changes to balance sheet composition. Some dynamic approaches incorporate assumed management responses to changing rates, which are somehow always optimal. This has always struck me as akin to peeking at the answer sheet rather than a true risk assessment. Better managed banks report EAR under both static and dynamic approaches.
Historical net interest margins can indicate the earnings impact of higher rates, if not economic value or more structural interest rate risk. Peer group data from the Bank Holding Company Performance Report (BHCPR) can show some broad industry trends. A single peer group can cover a wide range of banks. One includes all banks with over $10 billion in assets. The most recently available report for that peer group shows an uptick in NII during 2022 for the median large bank. However, NII was flat during 2023 and remained well below 2019 levels, when rates were considerably lower. In addition, the proportion of loans increased from 58.67% to 63.56%. Short-term investments fell by nearly 42% in 2022. These changes may explain the jump in net interest income more than an asset-sensitive balance sheet.
The 2021 Wall Street Journal article also reported that “some bankers have said they plan to redeploy that money [cash] into securities as soon as interest rates rise.” It depends on how soon. Bankers loading up on MBS when rates had risen to, say, 4%, would live to regret that decision. Both the assumption that higher rates are good for earnings and not disastrous for investment portfolios relies on the assumption that rate increases would be both moderate and short-lived. Decisions to reclassify investments as held-to-maturity rest on the same assumption. An underwater investment portfolio? A bank can muddle through until rates come back down.
But what if the rate increase is secular and not cyclical? The focus is usually on the target Fed Funds rate or some other short-term rate, but longer-term rates have a bigger impact on asset prices. While there have been occasional spikes in long-term market rates, they tended not to last long. Consider the trend in 30-year mortgage rates since 1985 (shown below). Rates rose by 205 basis points between December 1993 and December 1994, but fell back down 207 basis points by December 1995. Rates rose 187 basis points between December 1998 and May 2000 but then fell by 175 basis points by March 2001. The most recent rate increase was different. Rates rose an incredible 514 basis points from trough to peak. While off their peak levels, current rates remain 445 basis points above the 2021 low point.
There are a couple of key lessons that come from the banking industry’s experience with higher for longer rates. First, be wary of forecasts outside of the range of historical experience. What works for a moderate and short-term rate shock might not for a severe and long-lasting one. The relationship is more complex and ambiguous than it might have appeared. Second, be careful what you wish for.
[1] Midsize banks typically fall into the $10-100 billion asset range.
[2] OCC indicates that the Large Bank data is difficult to anonymize. Large Bank Supervision covers about 20 banks.