Below are Mr Tombini’s remarks in policy panel 2.12 (“Global financial stability and risks: What has happened and what’s next?”).
The recent environment of high interest rates, still high inflation and moderating growth in many parts of the world poses a threat to financial stability. This is especially salient if we consider the high levels of debt in many countries and sectors. Indeed, over the last year we have witnessed a number of episodes in which higher interest rates led to turmoil in particular sectors.
The first one is the crypto winter. Higher interest rates made investment in crypto assets less attractive and thus contributed to the price declines since the peak in November 2021. Of course, higher interest rates are only part of the story. Massive governance issues and design flaws also played their role.
The second was the United Kingdom’s liability-driven-investment crisis in September and October 2022. Pension funds were caught wrong-footed by an increase in gilt yields and had to sell assets to meet their hedging obligations. The Bank of England was forced to step in and purchase some of these assets in order to prevent overshooting.
The third, and arguably most serious, episode was the banking turmoil in March of this year. Silicon Valley Bank and other US regional banks faced a run because of large valuation losses on their holdings of government debt. The resulting turmoil in markets led to a run on Credit Suisse, a large global bank struggling to find a viable business model.
Let me go through some of the mechanisms by which higher interest rates affect financial stability.
The most immediate way, and the one behind the UK pension fund crisis and the demise of Silicon Valley Bank, are valuation losses on assets. How quickly these losses materialise in banks’ earnings and capital depends on the accounting treatment of interest-sensitive assets. For Latin American, Canadian and larger US banks, authorities require mark-to-market accounting for securities held both in the trading and banking book when “available for sale”. This results in a rapid recognition of the effects of valuation changes on capital. However, this does not apply to medium-sized banks in the United States – some of which are actually quite large by the standards of other countries. In fact, we have seen unrecognised losses increase again, to $558.4 billion at end-Q2 (hold-to-maturity: $309.6 billion; available-for-sale: $248.9 billion) (FDIC (2023)). Since then, Treasury yields have increased by approximately 1 percentage point, suggesting that unrealised losses could be considerably higher today.
Higher interest rates also affect financial institutions’ interest income and expenditure. In most emerging market economies, banks rely heavily on interest income. However, interest rate risk is largely managed with the provision of short-maturity loans and the use of time deposits, resulting in a low sensitivity of interest income to policy rates.
High interest rates also increase the debt service burdens of households, firms and governments. This comes at a time when high inflation has been eroding the purchasing power of salaries and moderating growth reduces actual and prospective earnings as well as tax revenues.
We have not yet seen a major rise in non-performing loans in most countries. In Latin America, non-performing loans have remained low by historical standards. They have also barely increased since central banks started hiking rates.
Banks’ provisioning policies suggest that they expect defaults to remain low in the coming quarters. Also, the disinflation and potential rate cuts suggest that some borrowers could soon see relief as well.
But this does not mean that we are out of the woods and should discard credit risks.
First of all, credit risk takes longer to materialise than valuation losses, from both an economic and an accounting point of view.
Second, the relatively low level of non-performing loans observed so far may still be somewhat supressed by the various monetary, fiscal and regulatory measures during the pandemic. For instance, many fiscal support programmes for borrowers are being phased out.
Third, banks’ benign loss projections may not fully reflect macroeconomic risks. A recent BIS working paper found that banks are good at predicting which assets are riskier than others, but they are generally not able to predict changes in aggregate losses (Birn et al (2023)). In fact, for the vast majority of banks, the correlation between the evolution of expected and actual losses over time is not different from zero! This should make us sceptical about banks’ ability to predict the impact of higher interest rates on credit losses.
Fourth, aggregate figures may hide pockets of vulnerability. While overall credit growth in the Americas was moderate in recent years, there were booms in specific categories. For instance, consumer and credit card lending in Brazil and Mexico have grown fast. In Canada, the high stock of mortgages could create large rollover risks for households. And in the United States, riskier market segments such as high-yield debt, leveraged loans and office real estate are already experiencing higher default rates and are forecast to go notably higher.
Finally, interest rates could remain high for some time. It is true that some central banks in Latin America and some other emerging market economies have started reducing interest rates. Markets are pricing in rate cuts in major advanced economies next year. But these are modest and will not take interest rates to the lows we have seen before, let alone during the pandemic. Furthermore, we cannot rule out a scenario in which inflation remains stubbornly at levels above central banks’ targets, requiring them to keep rates high for an extended period.
We could also see considerable yield volatility in the short term. A build-up of leveraged positions suggests that some sectors may have grown quite vulnerable. A case in point is the large speculative positions that leveraged investors have built up in US Treasuries, reaching around $600 billion in net short futures positions Much of this is probably driven by highly leveraged basis trades, which exploit price discrepancies between futures and the underlying cash bonds. Similar positions had accumulated in the run-up to the repo market turmoil of September 2019 and the Treasury market dislocations of March 2020. This points to heightened vulnerability of a sudden disorderly reduction in leverage if margin requirements were to spike as a result of yield volatility.
So what can and should policymakers do?
Importantly, tighter financial conditions are not an accident but an intended and essential element in containing inflation. Cutting rates would thus go against the objective of bringing inflation back to target.
But there are other tools. Policymakers can use prudential tools to create buffers at both lenders and borrowers. For example, a few months ago the Central Bank of Chile set a countercyclical capital buffer to increase banks’ ability to withstand scenarios of stress in the Chilean economy.
It is also important to enable the financial system to take losses and restructure or write off bad loans to create balance sheet and management capacity for new lending. Capital buffers help in this regard, but only if banks have built them up sufficiently, and are allowed to draw on them as needed. Measures to encourage the identification and clearing of compromised loans through efficient bankruptcy and restructuring could also help accelerate the deleveraging process and set the stage for new growth.
References
Birn, M, R Corrias, C Schmieder and N Tarashev (2023): “Banks’ credit loss forecasts: lessons from supervisory data“, BIS Working Papers, no 1125.
Federal Deposit Insurance Corporation (FDIC) (2023): “FDIC-insured institutions reported net income of $70.8 billion in second quarter 2023”, press release, 7 September.
