US Banks: The Coming Tailwind to Bank Revenues and Earnings from Rising Net Interest Margins

Print Friendly
Share on LinkedIn0Tweet about this on Twitter0Share on Facebook0



July 22, 2013

US Banks: The Coming Tailwind to Bank Revenues and Earnings from Rising Net Interest Margins

  • Over the next two years, we expect bank net interest margins to increase to 4% from the current 3.2% (in aggregate for banks filing regulatory “call” reports with the FDIC). This forecast is based on the rate-market’s discounting of an increase in 6-month Libor to 1.2% from the current 0.4% and an increase in the 3-year swap rate to 2.2% from the current 0.8%.
    • A 25% increase in bank net interest margins over the next 2 years, together with an increase in earning assets, implies a 15% increase in revenues (since net interest income is approximately one-half of revenue). There is a highly leveraged effect on earnings.
  • The components of the forecast margin increase are: 0.8% due to steepening of the yield curve; 0.7% due to increasing profitability of deposits as short rates increase; less 0.7% due to tightening spreads on both loan book and investment securities book.
    • The Exhibit below presents a decomposition of bank net interest margins into contributions from the loan franchise, the deposit franchise, and asset-liability management or “ALM”; the margin and spreads are presented on a core basis to exclude the impact of trading assets
    • The ALM spread consists of a passive component arising from the slope of the yield curve and an active component arising from, for example, investment securities and term-funding.
  • The Appendix describes our use in the margin model of spread-components (normalized to have core earning assets as the denominator) which, unlike raw spreads, are additive and capture mix as well as pricing effects; and our use of 3-year swaps and 6-month Libor as reference rates for the loan and deposit books respectively.

Exhibit: Spread Contributions to Core Net Interest Margin for FDIC-Reporting US Banks

Source: SNL, SSR Analysis: Reported Net Interest Margin as a % of Average Earnings Assets including the trading book, not Core Average Earning Assets which exclude the trading book


Exhibit 1 shows the interest-rate market is discounting that 6-month Libor will be 1.2% by July 2015 (versus 0.4% today) with a slope between the 3-year swap rate and 6-month Libor of just over 1% (versus 0.4% today). As discussed below, the effect of these changes in interbank rates would be dramatic for bank net interest margins.

Exhibit 1: Past and Forward USD Interbank Rates

Source: Capital IQ, Bloomberg

Impact of Interbank Rates on Bank Net Interest Margins

At 3.2%, bank net interest margins today are nearly as low as the levels during the 2008 financial crisis which, in turn, marked multi-decade lows (see Exhibit 2). We truly are in an extraordinarily difficult environment for net interest income which is approximately one-half of bank revenues and, as discussed above, the rate markets are discounting a significant, positive change in the operating environment. An increase of 0.8% in 6-month Libor and 0.6% in the spread between the 3-year swap rate and 6-month Libor would be meaningful for bank net interest margins.

Exhibit 2: Reported Net Interest Margins (as % Average Earning Assets) for US Banks

Source: FFIEC, Fed of St. Louis; Shaded Areas Represent Recessions

A Model for Bank Net Interest Margins

To understand the sensitivity of bank net interest margins to changes in interbank rates, we need to unbundle the margin into its components: the loan spread (representing the value of a bank’s loan franchise); the deposit spread (representing the value of a bank’s deposit franchise); and the asset-liability management or “ALM” spread (including the value generated, at least with a positively-shaped yield curve, by typical bank practice of borrowing short and lending long). We work with the core net interest margin and average earning assets or which respectively exclude the impact of interest income and assets from the trading book.

Our margin model, described in the Appendix, unbundles the above margin components with the results for banks filing regulatory “call” reports with the FDIC (and so excluding bank holding companies which file FRY9-C reports with the Fed instead) presented in Exhibit 3. We note that the results are expressed in terms of spread contributions with a common denominator of core average earning assets, and so are additive. (A common example of this contribution approach is the cost of funds which is reported as a ratio of average earning assets, not interest-bearing liabilities, so that it may be deducted from the yield on earning assets to generate the net interest margin).

The key device for unbundling margin components is to compare results to those of a hypothetical “reference” bank lending at the 3-year swap rate (3YS) and borrowing at 6-month Libor (6ML). Hence, for a specific bank or sample of banks, we define the loan spread as the loan yield less 3YS and the deposit spread is 6ML less the rate paid on deposits; for non-interest bearing or “NIB” deposits, the spread is simply 6ML. As discussed above, the spread contribution is the raw spread after normalization so that the denominator is the average of core earning assets not of the relevant liabilities.

In addition to lending at a spread over interbank rates through their loan franchise and funding at a spread under interbank rates through their deposit franchises, banks generate profits through asset-liability management. Specifically, they typically borrow short (at a reference rate of 6ML in our model) and lend long (at a reference rate of 3YS in our model) and so capture the spread, at least in a positively-sloped yield curve environment, between 3YS and 6ML. In practice, banks take active duration and credit bets relative to our model reference rates (including, for example, through their investment securities portfolios and through long-term funding) so that the “ALM” spread from asset-liability management activity will not match the slope of interbank curve between 3-years and 6-months: as shown in Exhibit 3, our margin model captures this variance through an “Active ALM” spread contribution.

Exhibit 3: Spread Contributions to Core Net Interest Margin for FDIC-Reporting US Banks

Source: SNL, SSR Analysis: Reported Net Interest Margin as a % of Average Earnings Assets including the trading book, not Core Average Earning Assets which exclude the trading book

The Margin Tailwind to Bank Revenues and Earnings

Since 2010, net interest margins for banks in our sample have fallen by 50 basis points almost entirely driven by a flattening of the yield curve so that the ALM spread contribution declined by 60 basis points through Q1 2013 (despite an offset from active duration and credit bets); the difference, along with a 10 basis point decline in loans spreads, is made up by increasing spreads on interest-bearing or “IB” deposits as banks have reduced the rates paid by more than the decline in 6ML. The spread-contribution from NIB deposits is essentially flat as a favorable mix-shift has offset the decline in 6ML. We note the impact of deposit-pricing and mix-shifts is not explicit in Exhibit 3 but is in the Appendix.

The key conclusion is that the flattening of the yield curve since 2010 has created significant pressure on bank net interest margins. The rate-markets are now discounting a steepening of the curve so with the spread between 3YS and 6ML increasing ~80 basis points by July 2015. There could be an additional lift to net interest margins of 70+ basis points if 6ML increases, as the rate-market is discounting, to 1.2% in 2015 from 0.4% today; note that the margin does not rise by the full amount of this increase because the spread contribution needs to be adjusted to reflect the fact that represent ~90% of the funding for earning assets and because, as rates rise, there is likely to be an unfavorable mix-shift from low-rate to high-rate deposits.

We can expect some offset to this 1.5% possible margin improvement from a declining spread contribution on the investment securities portfolio since this will likely not re-price at the same rate as 6ML. If we assume a zero (but not negative) from the “active” component of the ALM spread contribution, we are still left with an increase in net interest margins of over ~1%; allowing for some compression, albeit uncertain, in loan spreads reduces this to, say 80 basis points. This may prove conservative because banks lag re-price deposits which is why deposit spreads were depressed as short rates fell sharply in 2008; of course, lag re-pricing will tend to push deposit spreads higher as short rates back-up over the next couple of years.

If our estimate for an 80 basis point increase in net interest margins proves accurate, it will lift net interest margins by 25% (from the current 3.2% to 4%). Allowing for some growth in earning assets, this will lift revenues by 15% (since net interest income is approximately one-half of total bank revenue) and earnings before tax and extraordinary items by near 50% (since earnings are less than one-third of bank revenues and there is no variable cost is associated with margin expansion).


Modeling Bank Net Interest Margins

Our model for bank net interest margins works on the core margin (i.e. excluding the impact of the trading book[1]) and has two defining features.

First, we use a contribution approach to ensure that the denominator for all our spread-contributions is the same and equal to core average earning assets; this means the contributions are additive where raw rates and spreads will not be if they have different denominators. A common example is the reported cost of funds which uses average earning assets or “AEA” as a denominator (and so can be subtracted from the yield to obtain the net interest margin) not the average balance for interest-bearing liabilities. An additional consequence of normalizing the denominator is that the cost of funds captures the impact of mix-shifts in funding balances as well as changes in rates paid; this is generally true of spread-contributions.

Second, we use the 3-year swap (“3YS”) and 6-month Libor (“6ML”) rates as reference rates for earning asset and funding liability balances respectively. We then define the “loan spread” as the difference between the loan yield and the 3-year swap rate and the “deposit spread’ as the difference between the rate paid on deposits and 6-month Libor; for non-interest bearing liabilities, such as non-interest bearing deposits and equity, the spread is simply 6-month Libor.

These raw spreads are balance-weighted in accordance with our contribution approach to generate the loan spread contributions of Row 6 in Exhibit A1 and the deposit-spread contributions for interest-bearing or IB deposits and non-interest bearing or NIB deposits in Rows 8 and 9; furthermore, because spread-contributions are additive, we can compute the overall deposit spread-contribution shown in Row 7. Below, we discuss the asset-liability management or “ALM” spread.

Exhibit A1: Contribution Analysis of Bank Net Interest Margins: Spreads are as % of Core Average Earning Assets

Source: SNL, SSR Analysis

Asset-Liability Management or ALM Spread

If a bank actually operated according to our reference rates, that is to say targeted a duration of 3 year for earning assets and a duration of 6-months for liabilities, then the margin would be equal to the sum of: the asset spread, the liability spread, and the “curve spread” being the difference between the 3-year swap rate and 6-month Libor. In practice, of course, banks actively manage the gap between the duration of assets and deposits directly and through ancillary portfolios such as the investment securities portfolio and term funding (comprising non-deposit IB liabilities).

To capture these “active” asset-liability management effects, we define the ALM spread-contribution as the sum of the curve slope (see Row 11) and an “Active ALM” spread-contribution from these additional books (see Row 12). Finally, there is a typically small “other” contribution (see Row 15) arising, for example, from items such as loan and lease fees that are included in interest income but not in a loan yield and from funding, such as equity funding in 2012 for example, that is not captured in other rows.

  1. There is a lack of uniformity for disclosure of interest expense on the trading book. Therefore, we assume it is funded at the lower of 6-Month Libor and the rate paid on FF Purchase/Repo
Print Friendly