Monitoring and managing interest rate risk along the normalisation path
Keynote speech by Andrea Enria, Chair of the Supervisory Board of the ECB, at the Deutsche Bundesbank symposium “Bankenaufsicht im Dialog”
Frankfurt am Main, 8 November 2022
[The first paragraph of this speech was updated on 8 November at 11:10 to correct an editorial mistake.]
Inflation in the euro area has reached heights not witnessed since the 1970s. And here in Germany, such levels have not been seen since 1951. But while inflation rates this high are a distant memory and do not feature in the historical data of banks’ internal models, we are not in completely uncharted territory.
Historically, bank credit has been closely aligned with nominal GDP growth. In the 1970s, high inflation brought about higher nominal GDP, which in turn led to a rise in loan growth. And this is exactly what we are seeing in Europe today. At the same time, higher interest rates boost net interest margins, as the increase in interest rates that banks can charge for new loans typically exceeds the rise in their funding costs. One historical example of this is the United States between 1965 and 1980, when inflation surged and both nominal GDP and bank profit growth trended upwards.
But historical experience also shows that disinflationary policies can have severe negative consequences for banks’ balance sheets. They can cause asset quality deterioration, as seen at the beginning of the 1980s, when monetary policy tightening in the United States caused a devaluation of Latin American currencies and led to Mexico defaulting on US dollar-denominated funding obtained from US banks. Disinflationary policies can also exacerbate banks’ vulnerability to both interest rate risk in the banking book and to ill-advised asset and liability management strategies. The crisis of the US savings and loans associations in the 1980s and 1990s is an example of what can happen when, in the face of rising interest rates, financial entities mostly hold long-dated fixed rate assets and fund themselves through deposits that cannot be effectively managed and repriced.
Inflation directly affects banks’ balance sheets through multiple channels. For example, rising price levels have a direct impact on banks’ operational costs, including IT costs and staff costs. The impact on the latter could be particularly pronounced if wages increase in parallel to prices, as has been the case in periods of history when labour unionisation rates were higher and wage indexation mechanisms stronger. Higher prices can also erode households’ income and firms’ revenue, threatening their debt servicing capacity and, with it, the quality of banks’ assets.
But I will focus my remarks today on rising interest rates. In my view, and also based on historical experience, this is by far the most important implication for banks of the current inflationary environment.
Rising interest rates have been an important driver of European banks’ improved profitability in 2022. In the first three quarters, most banks posted profits that were above market expectations, and both banks and market analysts expect the positive effect of interest rates on net interest income to continue into 2023. This positive outlook is well grounded in baseline macroeconomic projections and accurately reflects the strength European banks have built in their balance sheets in recent years.
However, there is a worrying dissonance between these positive expectations and the unique combination of risks we are currently facing. Growth prospects have continued to deteriorate this year, while inflation rates and inflation projections, and with them the level of interest rates, have surged. This is happening in an environment of historically high levels of indebtedness, in the presence of several pockets of heightened credit and counterparty credit risk for banks, and with reduced room for monetary and fiscal support measures. These are all elements that call for prudence, careful risk management and enhanced supervision.
The ECB is currently finalising a review of interest rate and credit spread risk management practices among a sample of banks that are particularly exposed to those risks.
Our latest evidence suggests that, from the perspective of banks’ short to medium-term capacity to generate earnings, the euro area banking sector would remain broadly resilient to a textbook 200 basis point interest rate shock, even in a baseline scenario of an economic slowdown such as the one included in the ECB staff macroeconomic projections from September this year. Profitability would increase, driven by net interest income, while capital adequacy would only marginally worsen.
While this confirms the positive market expectations for bank profitability, there are good reasons to ask banks to strengthen their focus on monitoring and managing interest rate risk. This is a delicate moment and past mistakes must be avoided.
First, irrespective of prudential and accounting regimes, banks should not disregard the impact that rising rates typically have on the present value of their net worth – also known as their economic value of equity, or EVE for short. Lower EVE means worse long-term earnings and capital adequacy prospects, which in turn undermine the sector’s capacity to attract investment. Our evidence shows that, for some business models, a standard interest rate shock could produce material depletions in net worth.
Second, banks should adopt sound and prudent asset and liability management (ALM) modelling practices if they want to correctly capture the shifts in consumer preferences and behaviour that occur when interest rate regimes change. They should also carefully monitor the multifaceted risks that arise from using derivative transactions for hedging purposes, including the potential for this hedging to become ineffective.
Third, credit spread risk should be appropriately measured and managed, including for sovereign debt securities and other instruments accounted for at amortised cost. According to our evidence, two things in particular can result in a significant underestimation of risk and a misleading representation of stable regulatory metrics: (1) excluding material portfolios of instruments accounted for at amortised cost from the monitoring and measurement of credit spread risk; and (2) calibrating internal stress test exercises that do not mirror the severity of historical stress episodes. The exclusion of material portfolios could be especially relevant if banks needed to mark these assets to fair value later on, for example in the context of distressed sales or mergers & acquisitions.
The impact of rising interest rates on banks’ balance sheets
Rising interest rates following a very long period of low rates are a desirable outcome for banks. When interest rates rise, banks see both their earnings and their capital adequacy affected in the short term. Net interest income typically increases, as assets can usually be repriced faster than liabilities, particularly for banks that grant a higher proportion of loans at floating rates. At the early stages of the adjustment, this effect tends to outweigh the negative impact on both the valuation of instruments, such as bonds, that are accounted for at fair value, and on the creditworthiness of banks’ counterparts in the lending or capital markets business, which may lead to an increase in their cost of risk. Capital ratios can decrease, mostly driven by increasing risk-weighted assets.
From a longer-term perspective, banks’ EVE also varies, because irrespective of their accounting treatment all assets and liabilities that are sensitive to interest rates will undergo an adjustment of their present value as rates change. Where the sensitivity of a bank’s assets to interest rates is higher than that of its liabilities, and hedging is not sufficient or is ineffective, the bank operates with a so-called positive duration gap and its net worth decreases with increasing interest rates. Lower EVE can quickly become a concern because it translates into lower earning capacity and capital adequacy over the long term.
Banks tend to assess their exposure to interest rate risk by focusing mostly on the short-term earnings-based perspective. But international standards and European regulations and guidelines are clear about the need for banks to include the EVE-based perspective in their interest rate risk monitoring and management frameworks. There are good reasons for this. Banks sometimes actively manage the accounting treatment of certain instruments when interest rate or credit spread dynamics become a concern, so as to isolate their profit and loss account and regulatory capital figures from negative impacts. We have seen evidence of this recently, when increased volatility on financial markets prompted some banks to reclassify leveraged finance exposures from fair value accounting to amortised cost so as to contain the balance sheet impact.But the sense of safety and stability this brings them is misleading and can be very short-lived. In fact, in a regulatory and supervisory environment that increasingly favours transparency, any losses that threaten a bank’s medium-term viability become visible to investors and depositors, ultimately affecting the bank’s funding conditions, its attractiveness to investors and, in the most severe cases, the stability of its business. This happened, for instance, in the early stages of the sovereign debt crisis, and led the European Banking Authority to require banks to ensure adequate capital coverage of valuation losses on sovereign exposures, irrespective of their accounting treatment.
As I said earlier, in the baseline scenario of a slowdown in economic growth, a textbook 200 basis point shift in the yield curve would likely have an overall beneficial impact on banks’ profitability over a three-year horizon and a marginal negative impact on their capital adequacy. The effect on capital would reflect revaluation losses and, in particular, inflation in risk-weighted assets driven by increasing credit risk parameters. Our analysis shows that profits would increase because the boost to net interest income would more than compensate for the revaluation losses and the lagged increase in cost of risk. Net trading income would also make a positive contribution to profits, owing to increased trading volumes.
On average, capital depletion would be higher for certain business models, such as consumer lenders and promotional and development banks. For consumer lenders, cost of risk and loan demand tend to be very sensitive to increases in interest rates and funding mostly relies on wholesale markets. Promotional and development banks typically rely less on deposits than other banks and tend to have longer-dated exposures to governments and other public entities. The tail of the distribution of the effects on capital would be fatter for retail lenders and diversified lenders, as 25% of banks with this business model would experience a negative impact on their Common Equity Tier 1 (CET1) capital exceeding 200 basis points. So while this analysis confirms the overall positive message, it also highlights that there could be significant differences in how a significant increase in interest rates would affect the balance sheets of individual banks, which is what matters for a prudential supervisor.
In their regulatory reporting based on June 2022 data, banks confirmed that their net interest income would react positively to a 200 basis point shift in the yield curve. The reported data show that the repricing gap, i.e. banks’ capacity to reprice assets faster than liabilities, has recently improved, particularly for short maturities. But the same standard interest rate shock would, on average, have a negative impact on the EVE of most banks, with the 20 most strongly affected banks seeing equivalent CET1 depletions ranging between 100 and 400 basis points. The impact tends to be higher for banks with business models characterised by a higher average duration gap, such as retail, diversified and small market lenders.
Banks are well aware of the main second-round effects and structural changes that could accompany the interest rate normalisation process. In particular, they know that along the normalisation path asset quality will likely deteriorate in step with increasing cost of risk, but they also consider that credit risk would remain manageable. Banks also recognise that pressure to reprice deposits may increase as extraordinary monetary policy support is withdrawn, households’ and firms’ savings decrease and market competition for deposits increases. Finally, they understand that consumers’ preferences may change with the shift away from the “low for long” interest rate environment, leading for instance to higher loan repayments and faster withdrawals of (non-maturing) sight deposits.
However, our targeted review of banks’ practices in the area of interest rate and credit spread risk management shows that a considerable number of banks in the sample (approximately one-third) do not have a properly defined framework for assessing the most important second-round effects and structural shifts that could accompany an increase in interest rates.
In particular, when banks model customers’ preferences and behaviours, we found that some of their ALM models are calibrated using historical data which are biased towards past periods of low interest rates and they do not take sufficient account of the lack of information on phases when interest rates increased. Furthermore, a few of these models are not validated, back-tested and re-calibrated sufficiently often. Some are based on aggressive assumptions, such as quite high repricing maturities for various types of non-maturing deposits. In such cases, the ALM framework of the bank may fail to capture very relevant dynamics – for instance deposit withdrawals that typically characterise periods of increasing interest rates – and may severely underestimate the sensitivity of the bank’s balance sheet to increasing interest rates, both from an earnings-based perspective and an EVE-based perspective.
But improved ALM modelling is only a prerequisite to managing the sensitivity to interest rates that stems from the structure of a bank’s assets and liabilities. On the whole, we observe that banks have repositioned their hedging books towards derivatives with a positive sensitivity to interest rates, thus increasing their overall preparedness for interest rates hikes. Some business models rely more on derivatives to limit the depletion of banks’ EVE, in comparison with the lower natural hedge provided by behavioural items. For instance, both consumer lenders and promotional banks are characterised by relatively high sensitivity of contractual items to interest rates. While promotional banks almost exclusively use (large volumes of) derivatives, consumer lenders see their sensitivity to interest rates mostly offset by modelled adjustments to behavioural items.
While hedging is overall an effective tool for managing interest rate risk, the increasing notional values of traded derivatives and the related risk management frameworks merit close monitoring. Given the leverage embedded in those positions, it is essential to understand not only what the implications of abrupt price corrections could be for banks’ exposures, but also the extent to which hedging could become ineffective. Our review of hedging books has highlighted that risk management practices are deficient in certain areas, such as stress testing frameworks and the governance of what appear to be widespread strategies for transferring risk from the banking book to the trading book for hedging purposes.
Together with interest rate risk, our targeted review also covered credit spread risk, which has recently become an integral part of the broader regulatory focus on market-like risks in the banking book. In times of macroeconomic and financial turmoil, such as we are currently observing, financial spreads widen, including those on sovereign exposures, with the potential to affect both banks’ earnings and their EVE. We find that the average sensitivity to credit spread shocks of balance sheet items measured at fair value is rather limited, with differing outcomes across business models. The level of sensitivity is higher for custodian banks, corporate/wholesale lenders and promotional banks, while it is particularly low for investment banks, asset managers and consumer lenders. However, in the case of credit spread risk, too, we are concerned about some deficiencies observed in risk management and accounting practices which might undermine banks’ ability to properly monitor and measure the risk they are exposed to.
More than 90% of the banking book is accounted for at amortised cost. Almost 75% of the sovereign exposures in the banking book, which can exhibit particularly volatile spreads when the macroeconomic outlook worsens, are accounted for at amortised cost. This accounting configuration gives a false sense of security in the face of shocks and volatility, in that actual changes in fair value are not reflected in the banks’ earnings and regulatory capital figures. Irrespective of the accounting practices used, the EVE remains fully exposed to credit spread volatility. Our own calculations show that, for some banks, their sensitivity to credit spread shocks can be many times greater when instruments accounted for at amortised cost are included in the measurement, thus capturing the adverse effect on their economic value.
Like interest rate risk, credit spread risk should be measured and monitored from both an earnings perspective and an EVE perspective. It should be mandatory to include assets measured at fair value, while assets measured at amortised cost should only be excluded where this can be duly justified. Our findings reveal that some banks measure and monitor credit spread risk from only one of these two perspectives. In addition, several others tend to exclude entire asset portfolios from either the calculation of net interest income or the calculation of EVE – most often assets accounted for at amortised cost, but sometimes also assets measured at fair value that affect profit and loss.
Another cause for concern is that the vast majority of banks falling within the scope of our review apply credit spread stress test shocks to banking book positions that are considerably milder than the fluctuations actually observed during recent episodes of turmoil.
In conclusion, while there is no denying that the exit from the negative interest rate policy and the normalisation of the interest rate environment is good news for bank profitability, it is important for banks to pay due attention to measuring, monitoring and actively managing interest rate risk.
Even if the fairly optimistic expectations currently prevailing among banks and market analysts were to be realised, there would be distributional effects, with winners and losers. If we also consider that the shift in the interest rate environment has been faster than expected and is happening in a context of heightened financial market volatility and a deterioration in the macroeconomic outlook, we cannot rule out the possibility that even the central expectation of a positive effect of higher interest rates on the average European bank might have to be revised.
The findings of our targeted review and the supervisory follow-up actions will hopefully help banks to focus their efforts when managing interest rate risk at this difficult juncture.
Thank you for your attention.