The stress test
The FSR presents results from the Bank’s 2021 solvency stress test exercise, which examines the impact on major UK banks of a severe global economic shock.
The exercise finds that while the stress reduces banks’ capital positions substantially, “the system and all eight participating banks remain above their reference rates”. And this leads the FPC to conclude that the UK banking system “remains resilient to outcomes for the economy that are much more severe than the MPC’s central forecast”.
Given the scale of the economic shock assumed in the scenario, we feel these published results do not pass the “sniff test”. The test assumes an almost 10% fall in global GDP in the first quarter of the scenario – a 7 standard deviation shock. To put this in context, during the global financial crisis, global GDP fell cumulatively by a little over 3%.
It is of course true that global GDP fell by a similar amount in the second quarter of 2020 when the extent of the Covid outbreak became apparent. While that shock did not trigger a banking crisis, it was met by a truly unprecedented set of fiscal, monetary and supervisory policy responses – actions that cushioned the impact of the shock on households, firms, and financial institutions and markets, preventing losses from materialising and protecting the financial system to a large extent from its full impact.
While these actions were entirely appropriate once the Covid shock occurred, it would be absurd for the Bank’s stress test to bake such responses ex-ante – the point of these tests, after all, is to check whether banks have enough resilience to withstand large shocks without relying on extraordinary support by central banks and governments. And without such aggressive policy responses, it is implausible we think that a shock of this magnitude would not lead to another banking crisis.
Taking a step back, the exercise tells us one of two things about the Bank’s stress test and capital frameworks:
— Either the exercise is flawed and the banks should have failed the test, in which case the stress testing toolkit is miscalibrated and in need of urgent examination and repair;
— Or the analysis is indeed valid, in which case the Bank’s capital framework is delivering an exceptionally high level of resilience.
To be clear, the second possibility is not necessarily preferable to the first. At some point, you can have too much of a good thing when it comes to resilience. If the Bank’s analysis is valid and if there is any meaningful trade-off between growth and resilience – and we assume that the FPC believe both to be true – then the results of the stress test would appear to indicate that we have probably achieved “the stability of the graveyard”. It is therefore hard to understand why the FPC chose to tighten capital requirements in December by re-applying the countercyclical capital buffer in light of the results of the stress test.
However, rather than diluting the capital framework, we think that the focus should be on the first possibility – the idea that the analysis may be flawed. Perhaps the best way to illustrate our claim is to ask the FPC to conduct and publish a reverse global macro stress test. If the FPC believes that the analysis is valid and the banking system would be resilient to a 7-standard deviation, 10% fall in global GDP, then it should be asked to estimate the scale of a global recession that would be required to trigger a UK banking crisis within the stress testing toolkit.
There are several reasons why stress tests are likely to underestimate the impact of severe adverse shocks of this sort.
First, they ignore features of the real world that would amplify the losses banks would face in such circumstances. For instance, as far as we can tell the FSR does not consider the possibility of foreign banks collapsing in the scenario, triggering counterparty losses and systemic risk. Nor does it consider the implications for UK banks of a meltdown in the shadow banking system, as occurred when the system last faced a shock of this magnitude in March 2020. Another concern is these tests rely on banks’ own models accurately gauging the performance of loan portfolios in circumstances far more extreme than those for which the models were developed. More generally, these tests present banks with a pre-defined scenario with a known trough, underestimating the fear and risk aversion that would come with such a shock in reality.
Second, the tests are predicated on improbable and imprudent assumptions about how banks and investors would react in such circumstances. For instance, the Bank’s stress test assumes that the UK banking system would continue to operate as normal at a capital ratio of just over 7.5%. There is considerable uncertainty about what the right hurdle rate should be – but if the purpose is to identify the point at which banks would begin taking harmful deleveraging actions, then a conservative estimate would be significantly above this level. In addition, the stress test permits banks to assume they would voluntarily stop paying dividends and bonus payments in full were the stress to materialise. If this laissez-faire approach is realistic, why then did the Bank feel compelled to order UK banks to halt dividend payments in March 2020?
As a simple cross-check on the plausibility of the stress test, we can use estimates provided by Stern-NYU of the capital shortfall UK banks would face in the event of a large global financial market shock. These estimates are based on the current market value of each bank’s equity, an estimate of how the bank’s equity is likely to evolve in the event of a global market shock, and an assumption of the capital threshold a bank must maintain to be viable. If we map these assumptions to those used in the Bank’s stress scenario – in this case, a 20% fall in global equity markets and a 3.5% hurdle rate for banks’ leverage ratios – we find that UK banks would face a capital shortfall of £52bn. This shortfall is wholly accounted for by news since the start of the pandemic – it was effectively zero in January 2020. In short, the view from financial markets is that UK banks’ resilience has deteriorated significantly since the start of the pandemic.
Overall, we see limited value in the FPC continuing to conduct pass/fail stress test exercises based on arbitrary hurdle rates of the sort published in the December FSR. These exercises seem to do little more than confirm pre-ordained positions that the system is resilient to seemingly any shock. It is possible that the exercise could even prove counter-productive. If the central bank repeatedly communicates that the probability of another banking crisis is vanishingly small (i.e., would require a double-digit standard deviation macro shock) then it seems likely that investors and institutions would change their behaviour in response, either by taking more risk or expending fewer resources of their own on monitoring and managing tail risks.
Instead, as the Committee reflects on its approach to stress testing in 2022, we suggest it reorientates its efforts in a macroprudential direction – by this we mean it should use stress tests as a tool for identifying vulnerabilities; for assessing which shocks the system as a whole is most vulnerable to, and which markets or institutions its stability is most reliant on. The exploratory nature of these questions is closer to the approach taken by the FPC in its ongoing climate stress test exercise. This approach would yield information the FPC could act on to address any vulnerabilities it finds and would provide a natural complement to the bank-by-bank micro-prudential focus of the current approach.
Elevated interest rate risk
We are concerned that the risk of a severe interest rate stress – a significant and sustained increase in the level of official interest rates – is perhaps at its highest level in several decades. To avoid any confusion, we do not believe that interest rates necessarily need to rise significantly or even that it is likely that they will do so. However, financial stability policy is about preparing for what might happen and there is clearly a risk that interest rates might rise sharply in the coming years.
We are surprised about how little commentary there is within the FSR about the risk of an interest rate stress. One comment stands out to us in particular – about the share of mortgagors with debt-servicing ratios above a level typically associated with repayment difficulties: “With all other factors, such as income, held constant, mortgage interest rates would need to increase by around 150 basis points for the share to reach its 1996–2006 average of 1.8%.” We are unclear what was intended by this statement but we are not particularly reassured by it, not least given that the implied increase in repayment difficulties is non-trivial and that we do not view 150 basis points as anything like an upper bound on an interest rate stress.
Rising interest rates are not always a huge problem. Central banks will often raise rates in response to good news on the state of economy. Growth in income will tend to be healthy in these circumstances – at least at the aggregate level – and that can help households and companies to manage rising debt-servicing costs. But there may be problems even in this good news scenario: a rising tide does not lift all boats. Not everyone’s income will rise in lockstep with nominal GDP so rising interest rates can still cause stress within the household and corporate populations. However, we are worried about a different problem today: that central banks will be forced to raise rates in response to bad news.
The prime cause for concern today is inflation or too much inflation to be precise. Inflation has risen sharply around the world, not just in the United Kingdom, and that is putting the world’s central banks under pressure. Opinions differ on why inflation has surged and hence whether inflation is likely to stay too strong. Two possibilities, in particular, will alarm prudent policymakers: first, the risk that the recovery in demand has been so robust that it has already far over-shot supply causing the economy to over-heat (i.e., there is a large positive output gap); and second, that high rates of inflation today will become embedded in expectations of what inflation will be tomorrow. In the worst-case scenario, the output gap is significantly positive and inflation expectations have de-anchored. Inflation will likely stay persistently above the target in these scenarios, unless or until central banks take corrective action by tightening the stance of monetary policy which inevitably means bearing down on demand.
The transmission mechanism of monetary policy essentially works through real interest rates. Central banks rein in inflationary pressures by raising real interest rates to the point where the resulting correction in asset prices and then spending is ultimately sufficient to drag down on costs and prices. In a perfect world, the adjustment in interest rates is calibrated impeccably, causing spending to cool but not crater and inflation glides gracefully back to the target. In the real world, these disinflations can often involve a recession as spending slumps and asset prices crash as the central bank slams on the brakes. Moreover, the longer that central bankers wait in this scenario to hit the brake, the larger the correction that might be required.
The variation in official interest rates over the business cycle is determined by the state of the economy. Interest rates will be higher than usual when the outlook for growth and inflation is robust, and lower than usual in the opposite case. However, what qualifies as the usual level of interest rates (what economists refer to as the equilibrium level) can also vary over time. In recent years, that equilibrium level has drifted lower. As a result, actual interest rates have been on a downward trend from one cycle to the next and it has been reasonable to assume that interest rates may not rise that far in absolute terms in the near future, even at the next cyclical high in interest rates.
There has always been a possibility that this downward trend in the equilibrium rate could reverse at any moment in time but the risk of a reversal is also potentially elevated at the current juncture. Right around the world households have accumulated savings across multiple lockdowns when income was protected and consumption was constrained. Once the virus finally retreats in the rear-view mirror there may be a period of significant dis-savings around the globe, which in turn may encourage a surge in investment spending via a classic accelerator effect. The equilibrium real interest rate required to prevent spending from rising faster than growth in the supply capacity of the economy would then likely increase.
A reversal in the equilibrium rate would amplify the cyclical stress in interest rates discussed above: the central bank needs to raise real interest rates by a sufficient amount above the equilibrium rate to slow inflation down. If the equilibrium rate is rising then actual interest rates need to rise further to account for that.
What is particularly striking about the current situation is that financial markets are relaxed about this risk. Real interest rates are very low. As the highly-esteemed former Chief Economist of the IMF, Olivier Blanchard, observes in the context of the United States: “How much will the Fed have to increase interest rates, and for how long? Markets seem completely relaxed, 5-year real rates are close to minus 2 percent. I do not think they should be.”
Given the elevated risk of an interest rate spike, we, therefore, believe that this is a curious moment for the FPC to consult on whether the Committee’s affordability test should be retired. The reason that the FPC introduced that affordability test in the first place was to allow the Committee to provide quantitative guidance on the appropriate interest rate stress test of affordability. Indeed, if ever there was a moment for the FPC to start warning households and companies about how high interest rates could climb (from very low levels) then this it. In particular, if ever there was a moment for the FPC to insist that the calibration of the stress in the mortgage origination process reflected a considered assessment of the risks to rates then this is it.
To repeat: this spike in interest rates and the recession it potentially induces is far from inevitable. However, the probability that it proves necessary increases with the level and persistence of above-target inflation. Before the pandemic, the typical macroeconomist would have probably attached a vanishingly small probability to a scenario in which Bank Rate reached 5% over a three-year horizon. Now, one cannot be so sure.
Fortunately, the FPC does not have to rely on the typical macroeconomist to calibrate the risk of an interest rate spike. The FPC can ask the people who set interest rates for their opinion. There has long been a compelling case for the Monetary Policy Committee to publish the path that Bank Rate is expected to take over the MPC’s forecast horizon, placed within a fan chart to illustrate the uncertainty around that plan, in order to anchor expectations of the evolution of monetary policy. But there is also a powerful financial stability argument here: a probabilistic statement about the future path of interest rates that illustrates both the most likely path for rates and how far rates could plausibly rise above that mode might help households and companies make more informed decisions about their financial affairs.
Of course, the FPC would already have access to a specific stress scenario for official interest rates if the MPC placed that fan chart for Bank Rate in the public domain: the FPC could simply read off the level of Bank Rate at the top of the fan chart. The FPC should recommend that the MPC publish this information. If the MPC refuses to do so then the FPC could recommend that the MPC make it available to the FPC on a private basis. If the MPC does not currently have access to this information either – that is, if the people who set interest rates are not currently discussing the range of possible outcomes for Bank Rate in internal policy meetings, even though they already discuss and publish views on the range of possible outcomes for growth, unemployment and inflation – then a helpful public nudge from the FPC might encourage them to start doing so.
Quantifying the risk to interest rates is the starting point. We believe that the FPC would be wise to investigate the financial stability consequences of a severe interest rate stress for the UK financial system and, given the global nature of the inflation problem, it would probably pay to consider the financial stability consequences of a global rate stress too.
There are numbers of channels through which that stress will impact the economy. The obvious starting point is with the increased stress on highly indebted households and companies and ultimately rising arrears and defaults. However, even the sovereign is not immune: an increase in official interest rates will push up the cost of servicing the national debt and that might prompt a contractionary change in fiscal policy.
Transmission through the debt servicing channels in the “real economy” will tend to be slow. The FPC might be more concerned about the fast transmission through financial markets. If real rates surge higher that should trigger a severe correction in the price of a broad constellation of assets that could then be exacerbated by a decline in risk appetite and liquidity within financial markets: essentially the discount rate on uncertain returns in the future could increase suddenly and sharply. That correction could lead to complications within financial markets, particularly in the case of a global interest rate stress.
Those complications could all too easily arise in the shadow banking sector. Shadow banks perform many of the same functions as banks and therefore run the same fundamental risks as banks. However, shadow banks cannot rely on the same level of state support as the banks enjoy to help them withstand a period of stress, both before a crisis (deposit insurance) or during a crisis (access to central bank liquidity facilities). Nor has there been quite the same improvement in the resilience of balance sheets and business models within the shadow banking network to match what has been achieved within the banking sector since the Global Financial Crisis. The events of March 2020 illustrate how a severe correction in asset prices can trigger a run on parts of the shadow banking network that might otherwise have led to a much bigger financial stability problem, had it not been for the forceful intervention by central banks.
A global interest rate stress need not necessarily lead to a full-blown crisis but it is surely worthy of further investigation, particularly at the current juncture. Central bankers might also reflect on what actions market participants are expecting them to take and when and at what price, to quell panic in a future stress scenario.
 See https://vlab.stern.nyu.edu/srisk. This website allows users to obtain estimates of the capital shortfall (“SRISK”) at individual banks or at the country level under assumptions about the size of the global market shock and minimum capital requirement. The USD estimates reported on the website have been converted to GBP using an exchange rate of £1:$1.35.
 The limited coverage of this issue is perhaps particularly surprising given that several members of the FPC are also members of the MPC and you would have thought that the uncertainty around the future path of Bank Rate and the possibility that a significant rise in Bank Rate might prove necessary would be uppermost in the minds of those setting interest rates.
 Blanchard, O (2021), ‘US inflation is running higher. What should we worry about now?’, PIIE Realtime Economic Issues Watch.
 Barwell, R and Chadha, J (2013), Complete forward guidance in den Haan (ed.) Forward guidance, VoxEU.
 Sengupta, R and Fei, X (2020), ‘The global pandemic and run on shadow banks’, Federal Reserve Bank of Kansas City, Economic Bulletin, May 11.