Learning from the dash for cash – findings and next steps for margining practices - speech by Sir Jon Cunliffe

Speech

It is now nearly two years since the ‘dash for cash’ that swept through financial markets at the onset of the Covid pandemic. A great deal of work is underway by central banks and the international regulatory community to learn the lessons of that episode. And I want to talk today mainly about one element of that work: the examination of the role played by margin payments, both cleared and uncleared.

But first, it is important to remind ourselves of just why we are engaged on this work and why it matters that we learn the lesson of that short lived but potentially extremely destructive episode. Unlike the Great Financial Crisis (GFC), more than 10 years ago, one senses that the 2020 ‘dash for cash’ is fading from memory.

There are obvious reasons for this. The disruption did not last long. By early May 2020, order had returned to financial markets. There were no spectacular bank failures. Indeed, strengthened greatly by the post-GFC reforms, banks were able to meet the huge drawdown of credit facilities at the beginning of the crisis. Unlike 2008, the solvency and liquidity of the banking system was not at the very forefront of market and regulator concerns.

And, of course, the shock itself did not originate in the financial system but rather in a once-in-100 year pandemic that has left us with many other, more powerful, and disturbing memories.

But it is worth refreshing our memories about just how extraordinary were the events of late February and March 2020.

The realisation that we were facing a global pandemic – and that, in order to contain it, large parts of the global economy would need to be shut down – sparked a reassessment of financial asset values and, as one would expect, a ‘flight to safety’ by investors.

In February 2020, as you would expect, risky asset prices fell sharply. At the same time, the price of traditionally safe assets rose sharply and yields fell – the yield on US treasuries for example fell by around 100 bps. This flight to safety was also reflected in the gold price, which rose by around 4% over this initial period.

In the early weeks of March however – from the 9th to the 18th – a different dynamic took hold; an accelerating ‘dash for cash’ in which, given the illiquidity of other markets, investors sold their most liquid assets, driving safe asset prices down. A dash for cash which turned into a stampede.

The numbers are impressive. Bond yields reversed their movements from the early part of the pandemic, with 10 year US treasury yields rising by 65 bps in this period and 10 year gilt yields moving 64 bps higher. Bid-offer spreads on US treasuries widened by a factor of 10. The gold price fell by 12%.

Investors suddenly looked to redeem from money market funds (MMFs). Sterling MMFs saw outflows of 11% in just over a week. Similar outflows were seen in the US prime MMFs.

This self-reinforcing dynamic was broken by massive central bank intervention. On 19 March, the Monetary Policy Committee (MPC) announced the stock of asset purchases would increase by £200bn of gilts. We also launched special repo operations (the Contingent Term Repo Facility), a credit scheme (the Covid Corporate Financing Facility), extended term credit to banks (the Term Funding Scheme with additional incentives for SMEs) and joined other major central banks in activating the swaplines. In the US, amongst other actions, the Federal Reserve announced a $500bn asset purchase program, quintupled its repo operations (from $100bn to $500bn) and established a temporary repo facility for international authorities with a view to support treasury market functioning. In Europe, the Pandemic Emergency Purchase Programme was put in place to purchases eligible assets up to €750bn. As a result, order returned to financial markets and subsequently the growing realisation that governments would step in to limit economic damage further restored confidence.

What lesson should we learn from that episode? I have, since those events, heard two views that, as a central banker responsible for financial stability, give me considerable concern.

The first is that there was ‘nothing really to see here’. After a brief period of disruption, understandable given the nature of the shock, markets returned to normal and indeed were able to support the economy through the pandemic. Implicit in that view, of course, is that absent central bank intervention markets would have returned to order and stability.

I simply do not believe that is the case. One cannot of course know for certain what would have happened if central banks had stood aside. But by the time central banks were forced to step in, market flows had become almost entirely one way. In a period of huge uncertainty, once the flight to safety had become a dash for cash, there was no-one other than central banks able and willing to step in to catch the falling knife. Absent the massive intervention I described above, core markets would have continued to seize up and the liquidity crunch would have become worse.

Indeed, though concerned with bank rather than market liquidity, the experience of the Great Financial Crisis was certainly that once powerful and adverse liquidity dynamics set in, they do not go away by themselves.

The second view that gives me some concern is that in a severe shock, in a tail event, it is the role of central banks to underpin market liquidity. To put that view in its extreme form, there is little to learn from the ‘dash for cash’ and no policy action is required. The market was hit by an exogenous shock, central banks intervened promptly, as they were required to do, calm was restored and markets could function again. Job done.

Up to a point, I have more sympathy for this view than for the first. But it is only up to a point. It is certainly unrealistic to expect the private sector, be it banks or markets, to be able to self-insure against all liquidity shocks, no matter how severe. That, indeed, is the underlying rationale for central banks backstopping the liquidity of the banking system.

But the private sector, be it banks or non-banks, need to insure adequately against severe but plausible shocks. That is the basis on which we stress test and regulate banks. And, again as we have done on the banking side, we need to tackle features of the system that amplify and reinforce liquidity shocks.

This is not simply about avoiding the moral hazard that can result if market participants believe they can simply leave the job of insuring against liquidity shocks in severe, market wide, stress events to the public sector. It is also because we should not presume that central banks will always be able to step in with massive bond purchases as they did in March 2020.

At the onset on the pandemic, in the second week of March 2020, the Bank of England cut interest rates to close to the effective lower bound. Only a week later we announced further action to purchase £200bn of gilts.

The first decision was motivated by the need to support the economy through the pandemic and through the hit to demand that was on the horizon. However, the second, as the MPC made clear at the time, was motivated by the urgent need to reverse the increasing and unwarranted tightening of financial conditions driven by the disruption of the dash for cash.

Though distinct, these two motivations – to keep demand and supply in line on the one hand and to restore market order on the other – clearly went in the same direction in the circumstance we faced at the time.

But that may not always be the case. At a time of rising, externally generated price pressures or when demand is stronger than supply, inflation-targeting central banks may not find it so easy to provide massive injections of liquidity to restore market functioning.

So it is important that the private sector, banks and non-banks, insures itself prudently against severe but plausible liquidity shocks, and that we do learn the lessons of March 2020.

In that respect, we need to look at what drove the demand for liquidity in the 2020 episode, whether there were unnecessary constraints in the supply of liquidity, and crucially whether there were amplification mechanisms that turned the flight for safety into the dash for cash. And to take action, where necessary, to avoid similar dynamics in future.

This is the context in which to see the Financial Stability Board’s (FSB’s) programme of work on the March 2020 market turmoil. The first stage of this work, the ‘holistic review’ of the March market turmoil, was published in November 2020. The review analysed the dynamics at play in the non-bank financial intermediation (NBFI) sector and set in train further international work, on various vulnerabilities that were exposed. This work covers: money market funds, on which policy recommendations have been published; the liquidity demand on -ended investment funds; the structure and drivers of liquidity in core bond markets during stress; USD cross-border funding and its interaction with emerging market economies; and the impact of margining practices, on which I will now focus.

In order to assess the role played by margin, a group of senior representatives from central banks and market regulators was established, which I co-chair with Russ Behnam of the Commodity Futures Trading Commission (CFTC). The group was tasked with examining the margin calls during the March and April 2020 stress period and their impact on market participants in derivatives and securities markets, and, in light of that assessment to make recommendations for future work on policy.

In making the assessment, we were concerned essentially with four questions:

• First, what actually happened to margin during the period across a wide range of clearing services and asset classes, as well as in non-cleared derivatives?
• Second, did the dynamics of margin models lead to unnecessary pro-cyclicality and amplification of the liquidity stress?
• Third, were market participants adequately prepared for the increased margin requirements seen during the period of stress?
• Last, if market participants were not adequately prepared, was this due to a lack of availability of sufficient information on how margin might increase in market stress – or, was it was due to a lack of willingness to prepare or hold sufficient cash in reserve for such an eventuality?

Before setting out the findings of this work and the next steps, it is worth making a few general points on margin.

The aim of the reforms to margin standards after the Great Financial Crisis was to ensure that derivatives trades were prudently, transparently, and efficiently collateralised. One of the key drivers of the financial crisis was the opaque network of insufficiently margined bilateral derivatives contracts and the impact of large margin calls – both actual and anticipated – on counterparty credit risk.

Prompt and transparent margining and, where possible, central counterparty (CCP) clearing helps avoid the situation in which the perceived stress on one major participant can cause a panic which amplifies stress across the market due to uncertainty about who else, is exposed – and whether those exposures are adequately protected by sufficient margin.^{footnote [1]}

As a result of those reforms, the financial system entered the Covid crisis in a much stronger position with regard to collateral and counterparty credit risk, with, for example, at least $1trn in additional collateral against over the counter (OTC) derivative exposures.^{footnote [2]} Very significant and sudden changes in asset prices did not lead to counterparty credit risk – actual or anticipated.

Margin call is of course pro-cyclical, insofar as it reflects movements in markets. This is not a bug in the system, it is a feature. As markets move, losers have to pay variation margin to winners. And as risks grow, initial margin rises to protect the members of a clearing house from the consequences of a future default by other members.

This characteristic of margin is unavoidable if we are to ensure that collateralisation keeps pace with risks and counterparty credit risk is minimised. But it is right to try to dampen this dynamic to the extent practicable and to avoid unwarranted procyclicality.

Against that background, let me set out the findings of the FSB-mandated group on margin and the proposed next steps.

I should say that the group’s assessment, the ‘Review of Margining practices’ was based on a very extensive and granular quantitative data set drawn from a wide range of CCPs and clearing services and on detailed qualitative surveys of CCPs, clearing members and clients. I will only set out the headlines; the full report was published, for consultation, in October last year and is available on the BIS website.^{footnote [3]}

As you would expect, given the scale of the market shock, margin calls in aggregate were large over the stress period. Initial and variation margin increased significantly, with the increase in IM totalling around $300bn in March 2020 and the increase in VM flows peaking at $140bn during the height of the stress in mid-March.

In cleared markets, initial margin increased by $300 billion over that period. Excess collateral also increased by $115 billion at the peak of the crisis, meaning the overall increase in collateral pre-positioned at CCPs of $415 billion. To put this into perspective US open-ended funds (OEFs), which represent around half of the global OEF market, saw a similar magnitude of outflows in March, at just under $350 billion.

The increase in initial margin called was driven by volatility and CCPs margin models rather than by changes in positions. Volatility was highest in exchange traded derivative and in equity products. The increase in margin requirements did not exceed the overall volatility observed in the relevant markets.

Whilst this increase in margin requirements did not exceed the volatility observed, our analysis found a wide dispersion in the level of increase in CCP initial margin requirements across different firms and clearing services, as well as both within and across different asset classes.

This dispersion may be indicative of several things. Clearing is a practice applied to a wide range of products, services and margin models in a way that reflects the nature of market and underlying asset. For example, volatility in energy markets reflects the volatile nature of the underlying physical products, which can be affected by political events or natural phenomenon, and is more likely to experience dramatic peaks and troughs.

However, it also reflects the variance across CCPs in their selection of key model parameters, such as the choice of lookback periods, the incorporation of stressed market data, the base level of initial margin, liquidation period and confidence level. While there is clearly no one right model for all products, the choice of model parameters makes a material difference to the way in which margin models respond to volatility.

In comparison to cleared markets, initial margin requirements from uncleared derivatives remained relatively static throughout the period, increasing by just $3.2 billion despite the large moves in asset prices. This was likely as the non-cleared margin models employed are less reactive and were slower to incorporate the March volatility into their calculations.

Variation margin calls were large during the period, peaking at $140 billion in mid-March. Variation margin, however, is not determined by margin models but rather is calculated directly from the changes in market prices. There appears however to have been a number of operational issues around the extensive use of intraday margin calls, particularly ad hoc and late in the day calls which market participants found hard to predict and which complicated liquidity management.

We found that the level of preparedness of participants to meet their margin requirements in the March-April 2020 period varied. On the whole, market participants, with a small number of exceptions, were able to meet the elevated margin requirements seen during the stress period. While margin calls increased during the first half of March 2020, total margin outflows as a percentage of liquidity resources for even the top third quartile of respondents most affected did not exceed 2.5%. However, this outcome obscures the reality that many participants found doing so was challenging. Whilst most bank and non-bank intermediaries surveyed said they did not experience material issues converting high quality liquid assets into cash to meet margin requirements there were challenges for some and the intervention of central banks was a significant factor in enabling some participants to convert non-cash assets.

We also noted that a lack of transparency for market participants on important aspects of margin models and processes was reported by many survey respondents, making their liquidity management more complicated and potentially amplifying market stress. This lack of transparency has a number of aspects.

CCP initial margin was considered by many banking and non-banking participants (covering clearing members and clients) to be insufficiently predictable, with considerable uncertainty over the scale and the speed of increases they may be likely to face over a given time period, or for a given level of volatility. We received similar feedback on the predictability of the timing and frequency of intraday variation margin processes during stress. In turn, this would inform and improve their liquidity management processes.

CCPs do already provide information on their margin models via their public disclosures, and in many cases share margin simulators with market participants to enable them to anticipate, and prepare for, changes in initial margin. However, clients’ responses to the survey we conducted suggest that the transparency of initial margin models and variation margin practices varies significantly across firms and jurisdictions.

The picture on clients is more varied and the data less reliable. When investigating clients’ preparedness, our surveys revealed that data is only sparsely available, and doesn’t allow for a comprehensive picture of preparedness to meet margin calls in the NBFI sector. But there was evidence that some clients faced liquidity needs materially greater than anticipated. And turning high quality assets into cash to meet margin requirements was a problem for some participants.

Drawing on the findings in the report, CPMI, IOSCO and the BCBS have consulted on six potential areas for further work. I will group these under three heading: procyclicality; transparency; and preparedness.

First, we need to ensure that the necessary pro-cyclicality of margin does not add unnecessarily to a systemic stress.

On variation margin, this means tackling the issue on predictability and preparedness identified but also considering ways toward more efficient collection and distribution of variation margin around the system.

On initial margin, it means looking at the trade-off between levels of margin in business as usual. This is why we have proposed an in-depth evaluation of the responsiveness of centrally cleared initial margin models to market stresses, with a focus on impacts and implications for CCP resources and the wider financial system. This would look into the degree and nature of CCP margin models’ responsiveness to volatility and other market stresses – including impact, costs and benefits of this responsiveness for CCP resources and the wider financial system.

Second, transparency. Participants need to have sufficient information, data, and tools to be able to plan ahead – they need to be able to anticipate what margin calls might look like under stress, and how pro-cyclical they may be. We need to examine carefully whether more could be done, in addition to the information already made available by CCPs, to help market participants to plan ahead.

Ideas that could be looked at in this area include: developing good practices for margin tools and simulators and disclosure of key modelling choices; as well as promoting the use of consistent metrics and disclosures concerning procyclicality, responsiveness to volatility and model performance.

Finally, though reducing the impact of margin call in stress and increasing transparency will help, much will depend on the liquidity management of clearing members and clients. It is worth noting that in our examination we found that there was little reliable data available on client preparedness for stressed margin call.

Future work on in this area could: seek to identify enhancements to liquidity preparedness, including liquidity measures for NBFIs; and look into the effectiveness of intermediaries’ provision of liquidity to clients during stress, also considering how clearing members can encourage and facilitate greater liquidity preparedness of clients.

This is part of the broader issue that I noted at the outset about the ability of non-bank finance to insure against severe but plausible liquidity stress. It is an issue that the FSB should take forward as part of its work programme on enhancing liquidity resilience in the NBFI sector.

Prompt and transparent margining and, where possible, central clearing are essential to the resilience of the financial system. In times of stress, margin requirements will increase – as I said earlier that is a feature of the system, not a bug. But at the same time we need to do everything possible to ensure that the impact is not unnecessarily pro-cyclical, that market participants have the tools to prepare and manage that impact and that they do so.

The March 2020 ‘dash for cash’ has given us a valuable lesson, we should not waste it.

The views expressed here are not necessarily those of the Bank of England or the Financial Policy Committee. I would like to thank Sarah Crowley, David Macdonald, Shriya Mrug, Jack Hillier, Barry King, Sam Riley, Dan Wright and Cormac Sullivan for their help in preparing the text. I would also like to thank Andrew Bailey, Andrew Hauser and Christina Segal-Knowles for their helpful comments on the text.