An evolving financial system: Don’t leave it too late, simulate | VOX, CEPR Policy Portal
Editor’s note: This column is based on remarks delivered at the “Conference on Non-Bank Financial Institutions and Financial Stability” held at the Bank of England on 28 September 2018 (accompanying slides here).
Ten years ago, the financial system was in crisis. A complex network of connections cascaded chaos through the system, turning $500 billion of losses on sub-prime mortgages into $4 trillion of write-downs on assets globally (IMF 2009, Mishkin 2011). Those connections had been created by derivative transactions and shadow banking activities: securitisations funded by short-term debt instruments; constant net asset value money market funds; and proliferation of credit insurance. Bailouts and a credit crunch ensued. And the economic legacy of the crisis continues to be felt to this day.
Much has changed in the last ten years. The banking system is safer. Ten years ago, the UK banking system faced losses (in the period 2008-10) of £200 billion but had a capital base of just £100 billion (Figure 1). Today that position has been turned on its head. Banks have shrunk in size. If they lost the same share of their risk-weighted assets, they’d now write off just over £100 billion. And that compares to an increased equity capital base of over £200 billion.
Figure 1 Losses versus capital for major UK banks(1)
Notes: (1) Barclays, HSBC, Lloyds, Nationwide, RBS and Santander UK. (2) For 'crisis', 2008-10 impairment and market losses. For 'if crisis repeated now', crisis losses adjusted for changes in (estimated) risk-weighted assets. (3) Basel III common equity tier-1 capital. Estimated values for 'crisis' data.
Sources: Bank of England, banks' financial reports and Bank calculations.
At the same time, post-crisis reforms have made the financial system simpler, making it less likely to amplify underlying losses as it did a decade ago. As we’ll see below, derivative markets have been transformed. And shadow banking is a shadow of its former self, having shrunk and been reformed (Figure 2 left panel).
Figure 2 Global shadow bank and non-bank financial assets
Notes: (1) These data are based on the Financial Stability Board’s split of shadow banking by economic function (EF). ‘Broker-dealers’, ‘securitisation vehicles’ and ‘finance companies’ respectively correspond to EFs 3, 5 and 2, while ‘money market funds’ are in EF1. EF4 (credit insurance) is omitted, as it is relatively small. (2) Investment funds other than hedge funds and money market funds.
Sources: Financial Stability Board and Bank of England calculations.
In part as a response to these reforms, the structure of the system has evolved. Non-banks such as pension funds, insurance companies, and investment funds have grown in importance (Figure 2, right panel). That shift has reduced reliance on leverage and short-term funding in the system. As Adrian (2017) argues, there has been a “generalised ‘flight to simplicity and transparency’ in the intermediation of non-bank credit”.
As non-bank intermediaries have grown, so has market-based finance of the wider economy. This market-based finance – issuance of corporate bonds and other securities – accounts for all of the net finance raised by the UK corporate sector since the end of the crisis. The outstanding stock of bonds and commercial paper of £490 billion in the UK is now bigger than the outstanding stock of bank loans to businesses.
The system today is more diverse and, as a result, better able to deliver the macroprudential objective: serving, rather than sideswiping, households and businesses in bad times as well as good.
But that does not mean macroprudential policymakers can rest on their laurels.
Their growing importance to the economy means that markets can, like banks, be systemic (Tucker 2017). Their behaviour, just like that of banks, will determine whether the system can deliver the macroprudential objective.
The potential threat to their ability to support the delivery of this objective stems from firesales (Shleifer and Vishny 2011, Brazier 2018): sudden sales of assets on such a scale that prices are driven down and market functioning disrupted. These can sideswipe the real economy by raising funding costs and restricting access to market finance for significant periods (Bond et al. 2012).
The assessment of such risks is complicated. The past is unlikely to be a good guide to the possibility of this happening in the future. The changed (and constantly evolving) structure of the financial system means these markets may behave differently now. Vigilance is needed. But monitoring whether markets do serve households and businesses during a stress does not amount to vigilance. By then it is too late.
Vigilance means asking “what could go wrong?” before the bad times come along. That’s precisely the approach taken to systemic banks. Stress testing is the simulation of bad times to see whether banks could continue to serve households and businesses throughout. We must now extend the principle of simulation to markets.
The Bank of England is beginning to develop such simulations and has currently two works in progress. One looks at a specific issue in detail: simulating derivative variation margin calls to assess the risk that they could prompt asset firesales. The second takes a higher-level view of the corporate bond market to assess the potential for feedback loops between the major participants that create firesale dynamics.
Simulation 1: Variation margin calls on derivatives
The transformation of derivative markets in the past decade means that the past cannot be a good guide to the future.
Requirements to clear standardised derivatives with central counterparties have stripped back the network of connections in the system: complex webs have become more like hubs with spokes (Figure 3). As importantly, this change in structure has supported proper collateralisation of derivative positions, with variation margin called each day and initial margin to supplement that. Globally, over $1 trillion more of collateral is posted against derivative exposures than it was a decade ago.
Figure 3 Counterparty network for UK credit default swaps before and after post-crisis reforms(1)
Notes: (1) For ‘before’, positions as of early 2009. For ‘after’, positions as of early 2009 but cleared at current rates of clearing for new trades in order to proxy the future steady-state network. The figures show only the CCP (in green), the largest 16 dealers (in blue) and 24 randomly chosen other market participants (in purple).
Sources: Depository Trust & Clearing Corporation and Bank of England calculations.
To assess the risk, we run a simple simulation of what margin calls various non-bank entities could face in a market stress. And we ask whether and what they might need to sell to meet those margin calls. We look at the largest UK insurers, and the biggest derivative users among UK pension funds and investment funds (amounting to 73 institutions in total). We focus here because banks must have a buffer of liquid assets to meet demands, including margin calls.
Drawing on a variety of data sources, including regulatory returns and transaction-level data reported to EU trade repositories, we simulate the variation margin calls on single-currency over-the-counter (OTC) interest rate swaps and forward rate agreements. We find a 25 basis point increase in interest rates in all currencies and at all maturities would result in a total margin call of £3.5 billion. For the 10-year sterling swap rate, such a shift is a 1/1000 day event.
The margin call is dwarfed by the £299 billion of cash and government bonds that these non-bank institutions hold (Figure 4, left panel).
Figure 4 Simulated interest rate derivative margin calls for UK non-banks compared with their available liquid assets(1)
Notes: (1) Derivatives covered are OTC interest rate swaps and forward rate agreements. ‘Available’ liquid assets are total liquid assets scaled by the ratio of the notional amount of derivatives covered in the simulation to total derivatives held (where notional amounts are adjusted using coefficients from the Basel 3 Current Exposure Method).
Sources: Trade repository data (DTCC & Unavista), Bloomberg, FCA AIFMD data, FCA UCITS data, Morningstar, Pension Funds Online, ONS, PRA Solvency II data and Bank of England calculations
Of course, there may be other demands on those liquid assets from other parts of the derivative book. And the liquid assets must sit with the same entity as that facing the margin call. So we dig down to the individual entities and assume that only a fraction of liquid assets can be used for interest rate derivative margin calls. But even so, the total shortfall between margin call and available liquid assets is a mere £50 million.
Even for a 100 basis point move in all yield curves, the shortfall in liquid assets is just £0.9 billion (Figure 4, right panel). Any sales of less liquid assets, such as corporate bonds, would be unlikely to challenge those markets.
Although there are refinements we can make to these simulations (which cover, after all, only some institutions on some products at a particular point in time), they have yet to find evidence of firesale risk arising from variation margin calls. But that conclusion can’t be guaranteed to hold for ever as derivative positions and asset holdings shift. Repeated simulation will allow us to monitor how the risk is evolving. And that would certainly be aided by the development of consistent, comprehensive data covering the derivative positions and liquid assets of non-banks.
Simulation 2: Firesales in corporate bond markets
The first simulation tried to simulate one thing in isolation but, of course, in reality nothing is isolated. Simulations of markets under stress will ultimately need to capture interactions both within and between markets.
So in parallel we are developing less granular but broader simulations of systemic markets. The starting point here is to consider how participants in corporate bond markets might interact in a stress; to explore whether behaviour that is individually sensible can collectively generate firesale dynamics. We are beginning to simulate how open-ended bond funds, dealers, hedge funds, and long-term investors – like insurance companies and pension funds – could interact.
These suggest that firesale dynamics could develop as an initial fall in corporate bond prices prompts the market to enter a feedback loop (as in Baranova et al. 2017). To generate this, a number of things would need to happen.
The initial price fall would need to prompt investors in open-ended investment funds to redeem some of their holdings. The incentives to do this, which can be created by an inconsistency between redemption terms and fund pricing, are documented in Goldstein et al. (2017). These sales force the funds to sell bonds at short notice, driving prices down further.
The increase in market volatility would prompt dealers to minimise their bond inventories. This effect could arise if they manage their inventory risk using value-at-risk metrics based on market volatility. It would be reinforced if dealers treated their exposures to hedge funds through repo financing arrangements in the same way and raised repo haircuts as a result.
The result would be a feedback loop between greater selling pressure by funds and withdrawal of liquidity provided by others (Figure 5, left panel). Initially plentiful market liquidity could disappear when it mattered most.
Figure 5 Simulation of fire sales in the UK investment-grade corporate bond market(1)
Note: (1) Investment-grade bonds in any currency issued by UK corporations.
Sources: Bloomberg, Morningstar, SNL, Bank regulatory returns, Purple Book, FCA Hedge Fund Survey, SEC Private Fund Statistics and Bank of England calculations
The market outcomes would lean heavily on the behaviour of long-term investors. But unless these investors are quick to exploit value opportunities, we are able to simulate initial price shocks being amplified by a factor of three (Figure 5, right panel).
These simulations remain in development. As they develop, they will capture a broader set of institutions and assumptions about their behaviour. And they will start to come together as we capture the real-world linkages between derivative markets, cash markets, and financing markets (Figure 6).
Figure 6 Schematic for simulation of systemic markets
Our ultimate ambition is simulation of the linkages that drive behaviour in systemic markets. We know that no simulation can ever hope to capture all the intricacies and imperfections of the world. They will not tell us what willhappen to markets in a stress. But they will help us – and those who trade in markets – to monitor what couldhappen in a stress. That can inform both firms’ own risk management and help policymakers to ensure the financial system of today can serve households and businesses in bad times as well as good. Simulation allows corrective action to be taken before it’s too late.
So as the system evolves, let’s not just wait. Let’s simulate.
References
Adrian, T (2017), “Shadow Banking and Market-Based Finance”, speech prepared for the 33rd SUERF colloquium, Bank of Finland, 14-15.
Baranova, Y, J Coen, P Lowe, J Noss and L Silvestri (2017), “Simulating stress across the financial system: the resilience of corporate bond markets and the role of investment funds”, Bank of England Financial Stability Paper No. 42.
Bond, P, A Edmans and I Goldstein (2012), “The real effects of financial markets”, Annual Review of Financial Economics 4: 339-60.
Brazier, A (2018), “Market finance and financial stability: will the stretch cause a strain?”, speech at Imperial College London, 1 February.
Goldstein, I, H Jiang and D T Ng (2017), “Investor flows and fragility in corporate bond funds”, Journal of Financial Economics 126(3): 592-613.
IMF (2009), “Crisis and Recovery”, World Economic Outlook, April.
Mishkin, F S (2011), “Over the cliff: From the subprime to the global financial crisis”, Journal of Economic Perspectives 25(1): 49-70.
Shleifer, A and R Vishny (2011), “Fire sales in finance and macroeconomics”, Journal of Economic Perspectives 25(1): 29-48.
Tucker, P (2017), Systemic Risk Council Letter to Treasury Department.