Regional Banks: The Dynamics of Net Interest Margins at Selected US Banks
SEE LAST PAGE OF THIS REPORT Howard Mason
FOR IMPORTANT DISCLOSURES 203.901.1635
March 6, 2014
Regional Banks: The Dynamics of Net Interest Margins at Selected US Banks
- The role of CMA, MTB, and KEY as favored “asset sensitive” plays on rising rates leaves them over-valued relative to our regression model, and particularly to BAC and COF. Ironically, over the last 12-months BAC and COF were the only banks in our sample (along with RF) to show margin improvement and outperformed peers on both the asset- and liability-sides of the balance sheet.
- The exhibit below illustrates this by breaking down the change in net interest margin (NIM) into contributions from interest-earning (IE) assets and from liabilities, and further unbundling the liability side into contributions from interest-bearing (IB) liabilities and NIB funding (e.g. non-interest-bearing deposits and equity).
- In our margin model, described in the Appendix, the change in NIM is the sum of the asset and liability contributions along with the change in the slope of the swap curve (measured as the difference between the 3-year swap rate or 3YS and 6-month Libor or 6ML); this was 0.45% over the period of the Exhibit.
- Given short rates fell over the last 12-months, margin-underperformance by “asset sensitive” banks is consistent with consensus. However, this appears coincidental: the data show no correlation between the standard measure of asset sensitivity (i.e., the tilt to C&I loans) and the change in net interest margins, and we do not expect a correlation to appear despite the more pronounced rate moves discounted in the forward markets (indicating a move in 6-month Libor to 1.1% by mid-2015 versus 0.35% in 2013Q4 down 0.2% from the year-ago quarter).
- Regardless, all interest-earning (IE) assets re-price eventually so that, even if asset-sensitive banks respond to rising rates by delivering strong near-run earnings, it is a double-count to value these earnings at a premium multiple.
Exhibit 1: Change in Net Interest Margin, and Margin Components, for 12-Months through 2013Q4
Among our sample of banks, only BAC, COF, and RF saw year-on-year increases in net interest margin through 2013Q4. BAC and COF outperformed peers on both the asset side of the balance-sheet, with a decline in asset spreads that was half of sample average, and on the liability side. Exhibit 1 above illustrates this by breaking down the change in net interest margin into contributions from interest-earning (IE) assets and from liabilities, and further unbundling the liability side into contributions from interest-bearing (IB) liabilities and NIB funding (e.g. non-interest-bearing deposits and equity). The change in net interest margin (NIM) is the sum of the asset and liability contributions along with the change in the slope of the swap curve (measured as the difference between the 3-year swap rate or 3YS and 6-month Libor or 6ML); this was 0.45% over the period. Our margin model is described more fully in the Appendix.
The exhibit also shows the tilt of each bank to C&I loans which investors use as a marker of asset sensitivity; the three names with the heaviest tilt (CMA, MTB, and KEY) all trade at a premium to regression-based valuations (see Exhibit 2) based in part on their role as favored bank plays on an up-rate environment. We disagree with these valuations on theoretical and practical grounds and see BAC and COF as undervalued relative to CMA, MTB, and KEY.
Exhibit 2: Regression-Based Valuations for Select US Banks
Rethinking Asset Sensitivity
On theoretical grounds, all IE assets re-price eventually so that, even if “asset sensitive” banks deliver stronger short-run earnings, it is a double-count to value these earnings at a premium multiple. On practical grounds, there was no correlation between the tilt to C&I loans and changes in the asset-spread over the last twelve months. It is possible this is because rates were relatively stable (with 6-month Libor falling 20bps to 0.35% and the 3-year swap rate increasing 25bps to 0.73%) and that the more pronounced moves called for by the forward market (which is discounting 6-month Libor at 1.1% by mid-2015) will drive the correlation higher. However, the data leave us skeptical. For example, over the last 12-months short rates have fallen and yet the asset spread of the 5 banks in our sample with the least tilt to C&I loans showed an average decline that was near-50% more than that for the 5 banks with the greatest tilt to C&I loans
Furthermore, in an up-rate environment, liabilities are likely to play as significant a role in margin performance as assets. Specifically, non-interest-bearing (NIB) funding balances, to the extent they are retained, will make an increasing contribution to net interest margin as short rates increase. Coincidentally, the three banks in our sample (CMA, MTB, and KEY) with the greatest tilt to C&I loans also have the greatest tilt to NIB funding (see Exhibit 3). The open question is whether these NIB balances will prove sticky or migrate to interest-earning products as rates rise.
Exhibit 3: NIB Funding Relative to IB Liabilities for a Selection of US Regional Banks
Appendix: SSR Model for Net Interest Margin of US Regional Banks
Our margin model unbundles bank net interest margins into three components (see Exhibit A1 which shows the aggregated results across US commercial banks reporting to the FDIC). When summed together and with the slope of the yield curve (defined as the difference between the 3-year swap rate and 6-month Libor), these three components yield the net interest margin.
- The IE asset spread defined as the difference between the asset yield and the 3-year swap rate (3YS); we use a 3-year term because this is the average duration of bank asset books.
- The benefit from NIB funding defined as the cost of funding in the bank market (assumed to be 6-month Libor or 6ML) weighted for the ratio of NIB funding to IB liability balances
- The IB liability spread defined as the difference between the cost of IB liabilities and 6-month Libor (6ML).
Another way of framing this is to consider a hypothetical bank which has access only to the bank-funding markets and so borrows at 6-month Libor and lends at the 3-year swap rate. An actual bank outperforms this hypothetical bank to the extent its asset franchise allows it to generate assets that yield more than 3-year swaps and its liability franchise allows it to borrow at rates below 6-month Libor. Of course, we should risk-adjust the asset yields for credit losses but a simplified model without this adjustment is sufficient to make two key points: that the composition of bank margins varies meaningfully through time and that bank margins are dependent on the slope of the yield curve, not merely the level of short and long rates.
For example, in 2007, there was a 60:40 ratio between the margin contribution of asset and liability franchises (across both non-interest-bearing and interest-bearing liabilities) while the slope of the curve had a negative impact. In 2010, the liability franchises made no margin contribution while the slope of the curve contributed 0.9% of total bank margins of 3.6%. In 2013, asset franchises are generated over 90% of bank margins given that in the current low-rate environment banks are not being rewarded for their liability, and particularly deposit, franchises and given that the curve has little slope to it (so that the core bank “carry-trade” of borrowing short and lending long is relatively unattractive).
Exhibit A1: The Components of Bank Net Interest Margins
- The “gap ratio” reported by banks is also sometimes used to measure asset-sensitivity but, as shown in our note of October 29, 2013 “Margin Sensitivity Driven by Deposit/Loan Ratio not Gap Ratio” it does not explain a meaningful portion of the variation in net interest margins.