Marco Bardoscia, Gerardo Ferrara and Nicholas Vause
Participants in derivative markets collect collateral from their counterparties to help secure claims against them should they default. This practice has become more widespread since the 2007-08 financial crisis, making derivative markets safer. However, it increases potential ‘margin calls’ for counterparties to top up their collateral. If future calls exceed available liquid assets, counterparties would have to borrow. Could money markets meet this extra demand? In a recent paper, we simulate stress-scenario margin calls for many of the largest derivative-market participants and see if they could meet them – including because of payments from upstream counterparties – without borrowing. We compare the sum of any shortfalls with daily cash borrowing in international money markets.
Recipe for disaster simulation
Figure 1 summarises our simulation. We start with a scenario of sharp asset price movements taken from the severely adverse scenario of the 2018 Comprehensive Capital Analysis and Review stress test of US banks. Next, we revalue different types of derivative in light of these price moves, and hence revalue the portfolios of derivatives held between counterparties. We then set calls for ‘variation margin’ (VM) – which is collected to cover the current value of counterparty exposures – from gaining counterparties to losing counterparties equal to these valuation changes.
Figure 1: Simulation methodology
In our simulation, we cover interest rate and foreign exchange (FX) derivatives, specifically interest rate swaps, forward rate agreements, FX forwards and FX swaps. We also limit our focus to contracts held by LCH.Clearnet, which is the largest central counterparty (CCP) in these markets, and its’ clearing members, which are mostly banks and include all the major derivative dealers. Our data on the derivative holdings of these firms comes from trade repositories, where our access is restricted to trades involving at least one UK counterparty or denominated in sterling. Nevertheless, we still cover 50% of the global derivatives market in terms of outstanding volume. This is spread across more than 100 cleared portfolios (held between the CCP and a clearing member) and nearly 8,000 non-cleared portfolios (held between pairs of clearing members). Interest rate derivatives dominate the cleared portfolios, while FX derivatives form the majority of the non-cleared portfolios.
The second key ingredient of our simulation, besides VM calls between counterparties, is the cash they have available to meet these calls. Variation margins must be paid in cash, so the more of this a counterparty holds, the larger the VM calls it can pay without having to borrow. We derive cash available to meet VM calls from total cash holdings, defined as demand deposits plus central bank reserves. Although regulated institutions are allowed to let their cash holdings fall below regulatory standards at times of stress, we only use their cash ‘buffers’ on top of these requirements. This is because we want to be conservative and focus on ‘worst-case’ liquidity shortfalls. For the same reason, recognising that market participants may have other business activities that could require cash in a stress scenario, we scale these buffers according to estimates – based on Liquidity Coverage Ratio reports – of derivatives-related cash needs relative to total stress-period cash requirements.
To work out which VM obligations institutions can meet without borrowing, we need to model how payments between them are made. For this, we have designed a payment algorithm that reflects market practices. Figure 2 shows how it works.
Figure 2: Payment algorithm
Following asset price moves on a previous day, the top-left panel shows VM obligations at the start of the following day (grey arrows) and the initial cash buffers available to help meet them (blue boxes). Early in business hours, clearing members must pay their obligations to the CCP. If their cash buffers are too small to make these payments, they must borrow. In the top-right panel, CM1 and CM2 pay the CCP from their cash buffers without borrowing, but CM3 need to borrow 1 to supplement its cash buffer and meet its obligation of 3. Shortly after the CCP has received its VM calls, it pays out. This could include payments to institutions that had just paid the CCP if the inward and outward payments were in different currencies or related to different clearing services. Thus, the CCP pays CM1 and CM3 in the bottom-left panel. As the CCP takes equal but opposite positions with different clearing members, it always pays out the same amount of variation margin in aggregate as it receives. Hence, it never needs to borrow. This assumes clearing members are always able to borrow when they are short of cash. This may not be the case if their solvency was in question, but we assume it is not, as our interest is in the potential demand on money markets when derivative users face liquidity – not solvency – shortfalls.
Later in the day, VM obligations between the clearing members are settled. At this time, we assume institutions immediately pay their obligations in full if they can afford to do so using only their cash buffers. These payments supplement the cash buffers of their recipients, who in turn meet their obligations in full without borrowing if they are able to do so. We repeat this step until all institutions meet their obligations or no more recipients can make their payments in full without borrowing. We assume institutions then wait until the last moment in the day at which payment is due – to minimise interest costs – before borrowing just enough cash to help meet their obligations. Throughout this process, we do not permit partial payment of any obligations. This is consistent with our assumption that all market participants remain solvent. Thus, in the bottom-right panel of Figure 2, CM1 pays CM2, which allows CM2 to pay CM4. CM3, CM4 and CM5 then need to borrow 4, 6 and 1 respectively at the end of the day to meet their remaining obligations of 6, 11 and 3 (triangle of grey arrows). Further details on our ‘full payment’ algorithm is available in a separate paper.
Our scenario generates some very large margin calls, which sum to nearly $300 billion. This reflects the severity of our scenario, which includes some changes in swap rates that exceed historical experience. Roughly one-third of these margin calls are obligations to the CCP. Almost all of these relate to interest rate derivatives, as the CCP clears relatively few FX derivatives. The remaining two-thirds of margin calls are between clearing members, where obligations relating to interest rate and FX derivatives are netted. Before that netting, the majority of non-cleared obligations relates to FX derivatives.
Total borrowing requirements in our simulation are $54 billion. Figure 3 shows how these are spread over the corporate groups to which clearing members belong. Recall that these are worst-case estimates, as it is possible that institutions would use some of the cash buffers we assumed they held back for non-derivative cash needs to help meet margin calls. They may also dip into their regulatory liquidity requirements – as permitted – to help make payments at times of stress. Other estimates, based on broader liquidity buffers, are available in the paper. However, even in this worst case, the aggregate borrowing requirement is less than 10% of average daily cash borrowing on international repo markets. Moreover, actual daily borrowing often rises above this average with little discernible effect on the interest rate.
Figure 3: Borrowing amounts
As well as the total borrowing requirements for individual corporate groups, Figure 3 shows how these can be decomposed. First, we decompose their cash needs relating to non-cleared derivatives into ‘fundamental’ and ‘domino’ components. Fundamental borrowing is required to meet outward payments even when all inward payments are received, while domino borrowing is only necessary when some inward payments are not received (until the last moment). There is no such decomposition for cleared payments, as inward and outward payments are never simultaneously due. Any borrowing to help make cleared payments is fundamental.
Second, we decompose bilateral domino borrowing into ‘avoidable’ and ‘unavoidable’ components. For example, domino-borrowing requirements could be reduced if institutions made one or more rounds of payments they could afford from their current cash buffers, even if this meant some obligations were not met in full. Thus, in the bottom-right panel of Figure 2, CM3 could eventually pay 5 (of the 6 owed) to CM4, CM4 could pay 10 (of the 11 owed) to CM5 and CM5 could pay 3 (of the 3 owed) to CM3. This would leave unavoidable-domino borrowing of just 2. In our simulation, however, most of the borrowing requirements are fundamental and avoidable-domino requirements are relatively small at about $3 billion in total.
Our simulation suggests that large margin calls would not generate demands for additional cash that international money markets would struggle to meet. However, many financial institutions are presently holding large cash balances by historical standards, which may fall as monetary policies normalise in the major advanced economies. Hence, simulations like ours could usefully be run on a regular basis to monitor potential spillovers from margin calls to money markets. At the same time, our methodology could be expanded to include more derivatives, more counterparties and more scenarios. As one policymaker stresses, ‘don’t leave it too late, simulate!‘.
If you want to get in touch, please email us at firstname.lastname@example.org or leave a comment below.
Comments will only appear once approved by a moderator, and are only published where a full name is supplied.Bank Underground is a blog for Bank of England staff to share views that challenge – or support – prevailing policy orthodoxies. The views expressed here are those of the authors, and are not necessarily those of the Bank of England, or its policy committees.