Many people seem to think that – as a new BIS working paper concludes – banks benefit when monetary policy tightens and interest rates rise (especially from a low level). Do they? In some instances, perhaps, but as a general principle, surely not.
A casual glance at recent U.S. stock market behavior seems to support the idea that higher interest rates would be good for banks now. When the Federal Open Market Committee decided not to hike interest rates on September 17, the S&P500 dropped by 1.85% over two days, while the KBW index of bank stocks fell by 4.85%. A week later, when Fed Chair Yellen speaking about inflation dynamics expressed her continued expectations for a rate hike this year, the S&P500 edged lower, but the bank index rose by nearly 2%.
The simple interpretation of these events is that investors believe looser policy hurts banks and tighter policy helps them. But one alternative is that investors are responding to news about the economic outlook more generally. We know that banks and the economy do well and suffer together. When it chose not to hike rates on September 17, the FOMC emphasized that “recent global economic and financial developments may restrain economic activity” and promised to monitor “developments abroad.” Chair Yellen’s September 24 speech – her first since early July – instead emphasized the economy’s “solid prospects.”
So, without a careful analysis that distinguishes the range of factors affecting bank stocks, recent performance can’t resolve the question of whether policy tightening per se helps banks.
There are two ways to pose and address this question. The first is to ask how it is that policy changes influence earnings and profitability that banks themselves report in their published accounting statements. Our reading is that this is what the BIS paper studies. The second approach is to examine the impact of interest rates on the ability of banks to withstand negative shocks: that is, how monetary policy changes influence banks’ true net worth and, hence, financial stability more generally. Because of some important accounting conventions that we will describe, the answers to these two questions can be quite different.
Our objective, then, is to analyze how it is that interest rate changes influence bank performance by examining the impact of three possible changes in the yield curve: (1) a rise in short-term rates that leaves long-term rates unchanged (a yield curve flattening); (2) a rise in long-term rates that leaves short-term rates unchanged (a yield curve steepening); and (3) a parallel upward shift in short- and long-term interest rates (a yield curve slope that is unchanged).
To understand the impact of each of these, we can look at the three principal ways in which interest rate changes affect banks.
First, banking is a spread business. Banks profit from the gap between the interest rate they receive on their assets (mostly loans and securities) and the rate that they pay on liabilities (mostly deposits). Because banks engage in maturity transformation – with assets having a longer maturity than deposits – this lending spread will generally fall when the yield curve flattens.
Second, since banks hold fixed-income instruments (bonds and loans), higher interest rates across the yield curve reduce the present discounted value of their assets. And, the longer the duration of their assets (relative to their liabilities), the bigger those losses will be. Similarly, banks face greater risk when the volatility of their portfolios (and the correlations among their assets) rise.
Third, rising interest rates are often a signal of stronger economic growth. An improved growth outlook benefits banks in several ways. It reduces the risk that borrowers will default, increasing the value of bank assets. It increases credit demand at any given interest rate, potentially raising the volume of their lending business. And, it is associated with greater stability, which makes bank assets more liquid. (See here for evidence on the procyclicality of bank profits.)
Importantly, since higher policy interest rates are normally associated with a flattening of the yield curve (and, when introduced after a long period of low rates, a rise in financial volatility), the first two of these mechanisms work to harm banks, reducing both profitability and net worth. Only the third mechanism, which is not about interest rates at all, can help them.
So, why do people think that a Fed rate hike will help banks? One possible argument – supported by the BIS paper – is that at near-zero interest rates banks cannot mark down their deposit rates sufficiently to maintain their interest rate spread as lending rates fall. From this perspective, as policy rates start to rise, banks would not need to raise their deposit rates, so their interest margins and profitability would go up. This would be possible if deposits are sufficiently sticky, meaning that investors are not very active in looking for higher yields as market interest rates rise.
At first glance, this focus on the zero-lower bound for bank deposit rates seems consistent with the decline of U.S. banks’ net interest margin (the gap between the weighted-average yields on their fixed-income assets and liabilities) to record lows in recent years (see upper chart). However, aside from the financial crisis, the trend of net interest margin compression is now two decades old, having begun around the time that Congress eliminated constraints on interstate banking in 1994. It is far from clear that this trend will reverse as interest rates rise. Indeed, the yield curve today is already a bit steeper than the average over the past three decades (see the lower chart), so banks may be compelled to compete more aggressively than usual for depositors.
U.S. Banks: Net Interest Margin, 1984-2Q 2015
U.S. yield curve gap between 10-year and 3-month Treasury constant maturity rates
Another possibility is that the belief in a positive link between the level of interest rates and bank performance arises from certain aspects of bank accounting. Here, the fact is that banks are not required to mark the value of most of their assets to market. Without going into the mind-numbing details, if a bank expects to hold a security to maturity in what is known as the banking book, it can account for it at “historical cost” – that’s the value it paid. This means that when interest rates rise, so that the market value of a long-term bond falls, banks are not required to report a loss. Consequently, without detailed information on bank balance sheets (including their derivatives exposures), outsiders using accounting data only are not in a position to evaluate the true impact of the interest rate increase on banks’ net worth.
That said, banks do provide information in their annual reports on the sensitivity of their net interest income to interest rate changes, including yield curve flattening, steepening and parallel shifts. For 2014, all of the Big Four U.S. bank holding companies (Bank of America, Citigroup, JPMorgan, and Wells Fargo) reported that positive interest changes would boost their net interest income. Consistent with the BIS paper, it would seem that they are assuming that they can raise their lending rates (or the yield on their securities portfolios) by more than their deposit rates will increase.
Only one of the Big Four reports the impact of a rise in interest rates on overall equity – taking into account the change in the market value of its assets (in line with second principal mechanism). Citi estimates that a positive 100-basis-point yield curve flattening, steepening, and parallel curve shift would lower its Common Equity Tier 1 capital ratio by 28, 18, and 44 basis points respectively. Even after adjusting for taxes, the estimated portfolio loss is more than twice as large as the projected net interest income gain.
We suspect that this pattern – in which, once the asset values are marked to market, higher interest rates reduce bank capital – is quite common. For example, for 2014, Barclays reported the impact of a 100-basis-point upward shift in the yield curve on its equity capital. In their simulation, pre-tax portfolio losses (£3,756 million) swamp gains on pre-tax net interest income (£170 million), resulting in a decline of equity of £2,726 million (after taxes). This would result in a decline of 4.13% in Barclays’ regulatory capital. More broadly, the Bank of Japan in its April 2015 Financial System Report estimated that the aggregate losses on Japanese banks’ bondholdings from a 100-basis-point yield curve steepening and parallel shift would total ¥4.8 trillion and ¥7.5 trillion, respectively.
The BIS paper finds otherwise: when interest rates rise (particularly from low levels), the “permanent” increase in net interest income exceeds the one-off decline in asset value, resulting in a positive return on assets. However, since the study is based on bank accounting data, which we know does not require that banks recognize valuation losses on assets in their banking books where most securities are held, this is a statement more about the impact on publicly reported earnings than about the true impact on a bank’s net worth from a rise in the central bank policy rate.
Regulators appear to share the expectation that higher rates are a threat to bank capital. When they develop scenarios for bank stress tests, the unfavorable settings are typically ones in which interest rates rise even when the economy and other asset prices fall. For example, in the Federal Reserve’s 2015 Comprehensive Capital Analysis and Review (CCAR), the adverse scenario had the end-2017 3-month and 10-year U.S. Treasury rates rising to 5.3% and 5.8%, respectively, at the same time that the unemployment rate rose to 8% and inflation spiked to 4%.
Of course, financial crises are frequently deflationary, so they tend to be accompanied by a plunge in policy rates that steepens the yield curve. (This is in fact the pattern in the “severely adverse scenario” of the 2015 CCAR.) But the purpose of cutting rates in these circumstances is to provide support for a collapsing economy and financial system, not to impose even greater strain on banks. If higher interest rates really have a positive net impact on banks’ profits and equity, central banks would have far less reason to ease monetary policy in a crisis.
We don’t rule out that a Fed rate hike from such a low level – independent of the message that it sends about future growth – could help banks. If they (as well as many bank analysts and the BIS) are correct that deposits are sticky, net interest income probably will rise. But there remains good reason to doubt that the overall impact on properly measured bank equity (again, excluding the signal about economic growth) would be positive.
To provide more clarity about this important issue, we hope that regulators will require all large banks to report the sensitivity of their equity capital to interest rate shifts, taking account of the impact on the market value of their assets. And, it would be helpful if published stress test results were sufficiently detailed to allow us to separate out the impact of interest rate changes alone from everything else in the very detailed and complex test scenarios. It would be nice to quantify the distinct impact of each of the three mechanisms on overall bank well-being. Doing so could be instructive both for bank managers and their shareholders.