Article: Decision Support and Analytics for Clearing08 June 2014 | Peter Walsh
New regulations are transforming how banks and the buy-side process Over the Counter (OTC) derivatives business, and in particular making the relationship between process choices and cost of the highest importance. For instance, U.S. regulations insist that the journey from execution to clearing take mere seconds, which makes limit checking pre-trade vital to smooth the path for each trade.
This environment generates a new raft of strategic and tactical decisions to be made and monitored by Clearing Brokers, and their clients, around the choice of venue for clearing and managing the economics of their portfolios. For a user of both exchange traded and OTC products decisions such as:
- Which Central Counterparty (CCP) should I use?
- Do I have sufficient limits with a SWAP Execution Facility (SEF), Clearing Broker or CCP to keep trading?
- Is the margin call from the Clearing Broker correct and valid? (Either overnight or intraday)
- How will my margin requirements change over the next 3/6/12/24 months?
- How can I reduce my margin requirements?
- How much capital should I be holding for my cleared / un-cleared business?
- How do I monitor all these parameters in real-time? What controls can I put in place?
For a Clearing Broker, issues such as:
- How do I explain to a potential client what their margin will be?
- How can I help a client manage their limits and Initial Margin (IM)?
- Managing intraday limits for a client
- Being able to perform as a backup Clearing Broker when offered a client portfolio
Before performing any high level analytics, you need a platform into which you can acquire, store and manage market data, credit data, reference data and trade data required for pricing, analysis, counterparty and internal organisational management. Data needs to come from multiple sources, be put through a quality process to handle missing or inaccurate data, and stored in such a way as to make building curves and performing pricing, simple. It is easy to under-estimate how difficult this task can be, as the quantity of data to be received and checked can be large, and the effects of small errors can skew pricing across a whole firm. An example would be rounding errors on a single curve point, causing the pricing of OTC Rates and Fixed Income trades to be incorrect.
The next component is pricing models, which cover both OTC and Exchange Traded Derivatives (ETD) products. Many firms trade across all assets classes including Interest rates, Credit, Equities, Bonds both OTC and ETD and need to value everything as inputs to their analytics tools. Whilst trading desks tend to be aligned around the vertical markets, risk decisions cut horizontally across all of these which mean your platform needs comprehensive capabilities for vanilla OTC products, but also complex options and strategies.
Once the foundations are in place, adding your portfolio into your platform is necessary. From a trade processing point of view you need to align portfolios with the actual book structure of your firm, and record that within your Risk platform. For analytics purposes, you need to cut across the book structure to look for wider opportunities to combine business at CCPs or by offsetting risk. Your risk platform needs tools to import or take feeds from your internal systems, in real-time, across all assets classes, and should have flexibility of file format including FpML, FIXML or other flat file formats as well as flexible real-time application programming interface (API’s) which use technologies such as C, C++, C# , Java and increasingly web-based API’s.
Next you need to configure your risk models, for cleared and un-cleared business. At the moment un-cleared business tends to be margined relatively simply using net mark-to-market or a percentage of notional. For cleared business you need access to models which replicate those used at CCPs, in particular Standard Portfolio Analysis of Risk (SPAN) and Value at Risk (VaR) for both ETD and OTC products.
The table below gives some examples of the configuration of VaR for various CCPs, which on the face of it is a complex list, but in reality are built on the same foundations with specific design decisions made by each CCP to express their view on how to measure IM.
Figure 1 – Public Data on Initial Margin Models
The intermediary role
In the U.S. the clearing model treats the futures commission merchant (FCM) as an agent, so has the advantage of not being a party to the trade in-between the client and the CCP. In Europe a Clearing Broker sits between the client and the CCP so becomes a party to two trades which then impacts costs of margin, capital and credit lines. A client will look to their intermediary to provide access to clearing and therefore must help the client manage their margin and credit lines, to avoid failed trades.
Flavours of VaR, Historical Simulation, Monte Carlo and SPAN
Value at Risk (VaR) is a statistical technique used to measure and quantify the level of financial risk within a firm or investment portfolio over a specific time frame. VaR is used by risk managers in order to measure and control the level of risk which the firm undertakes. The risk manager's job is to ensure that risks are not taken beyond the level at which the firm can absorb the losses of a probable worst outcome. Given the focus on capital and margin calculations it is worth understanding the basics of how Historic Simulation (HS) and Monte Carlo (MC) VaR are carried out, as management decisions, corporate appetite to risk and relationships with trading and clearing partners are all impacted by these fundamental tools. An appreciation of SPAN is also important as this is another tool typically used by CCP’s to establish margin amounts.
With HSVaR the model takes the past few years of market history to re-value a portfolio, and derive a predicted loss with an assumption on a holding period and confidence level. The choice of market history, either a continuous period or selected stressed periods gives flexibility, as does the length of the periods. Whilst history is not a guide to future market scenarios, the way markets have behaved in the past gives the model a basis to illustrate possible outcomes.
MCVaR uses a synthetic approach to generating the underlying scenarios, by stepping through time using random market scenarios, and recording the terminal profit or loss of a trade. Repeating these random paths many times gives a wider range of possible outcomes, and can be used for portfolios containing options where historical data doesn’t hit the strike price. MCVaR is the most computationally expensive as there are few computational shortcuts using this method.
SPAN, traditionally used by Exchanges for Futures and Options, is an approach is to simulating a small set of scenarios where the price of the contract, or the volatility of the market are varied in permutations, to explore how the portfolio behaves. The permutations are usually price moves up or down of 0, 1/3, 2/3 or 3/3 of a ‘scanning range’, plus each of those with an upward or downward move in volatility giving 13 unique combinations of market scenarios. SPAN can also take into account opposite positions in different months of the same contract, and the closeness of a contract to physical delivery, plus other adjustments measuring specific risks for options.
It seems beholden upon many of us in the capital markets to learn and understand how these models works, the opportunities and threats they bring to the capital markets – as these are at the core of our lines of defence to risk.
For bilateral OTC business now, the margin requirements are not imposed by any regulator, so rely upon the policies of the two parties to negotiate and implement a Credit Support Annex. From 2015 firms must adopt new rules from the BIS to apply Initial Margin to their un-cleared bilateral portfolio. This will create considerable new work with all OTC users as the rules are not simple. The BIS proposal requires IM to be calculated:
- Separately for each asset class: Interest Rates & FX, Equity, Credit and Commodities
- Historic scenarios must include a ‘stressed period’ and have not more than 5 years of history
- Scenarios cannot be weighted or filtered – meaning periods of high volatility will drive the IM amount for long periods (See top of page 12 where the text says “Additionally, the data within the identified period should be equally weighted for calibration purposes.” In the BIS PDF bcbs261.pdf on their site)
- The model must be approved by your regulator
- Amounts of IM must be exchanged gross between parties, and held such that they are available quickly in a default – implying the use of a third party custodian
Observers have pointed out that given the data inputs to operate a typical historic VaR model, and the range of products in a typical OTC portfolio, the unique approach firms will take to valuing trades, and calibrating their VaR model, it is unlikely two parties will arrive at a similar figure for the IM on a bilateral portfolio – something which suggests the use of third parties to serve pairs of bilateral firms.
VM & IM prediction
U.S. regulations have moved the OTC market towards that of an Exchange and CCP, by making the time gap from execution to clearing short, and raising the indirect penalty for failing to clear a trade, that of paying a termination fee. Firms have moved to place pre-execution limit monitoring in place to avoid failed trades, but this needs to take into account the two main limit criteria, that of the intermediary (Clearing Broker or FCM) and the intended CCP. More than one firm offers market infrastructure to enable execution platforms to manage limits, but all of these rely again on calculating the effect of adding a trade to a portfolio and the resulting economic effects.
Risk platforms need to provide the ability to react in ‘real time’ to trade events, such as answering the question “which FCM should I use for this trade” or “what will the effect of this trade be on my IM at any eligible CCP?”. At the moment the pools of liquidity at each CCP tend to be discrete, but over time if more banks and more sell-side firms sign up for multiple CCPs for the same product set e.g. Interest Rate Swaps, there will be an increasing benefit to modelling the effect of each trade prior to execution.
The other area for modelling and prediction is on intraday funding for margin calls. Some CCPs update their curves intraday to capture market movements to monitor margin requirements and make intraday margin calls. Firms may leave intraday calls to chance, but having the ability to replicate the CCP approach, and anticipate intraday funding needs may avoid late funding charges, and enable pre-funding at lower cost.
Razor provides the capability to compute a the IM and VM on your portfolio in real-time to underpin these decisions. Feeding the same trades as those cleared into Razor, after each novation event, will keep both Razor and your cleared portfolio in-sync, you then have a real-time dashboard providing risk managers with the ability to predict intraday margin calls due to market moves or additional trades.
Another factor in meeting CCP margin liabilities is selection of assets, being cash or securities. Each CCP has a different schedule of eligible assets, each with their own haircut to cover price volatility. Factors to be considered to find the ‘cheapest’ asset include the funding cost, the interest paid by the CCP on cash, the CCP haircut, the match of assets versus liabilities (not all assets can cover all liabilities), and the repo rate of securities – in which case all available assets can be ranked to indicate the optimum asset to deliver. This approach can be re-iterated each day for assets already held at the CCP to suggest candidate assets to withdraw and replace, to continue to ‘optimise’ the assets you hold at each CCP.
Prior to entering into a CCP relationship, it is wise to analyse their margin models to understand the relative cost of your portfolio in IM. At the same time, you will also be considering your choice of intermediary (an FCM or Clearing Broker) who may have obtained Regulatory approval to also offer cross-margining between the cleared and un-cleared portfolios. Once you are have an active CCP portfolio, managing IM becomes a material cost reduction, through two effects; 1) understanding which trades in your portfolio are key to your IM amount and 2) knowing how your IM will change over time as trades mature.
A risk platform should offer these tools, in the case of understanding the output of a VaR calculation, you will want to see data on the specific trades included in the final IM figure, captured by the 99% confidence level (or whatever the CCP uses) such that you can choose to put on risk reducing trades (at a cost) to bring down your IM.
In case 2, your risk platform should have the capability to explore the future of your portfolio, using Monte Carlo VaR and by dynamically rolling forward todays date, make a prediction on your IM level over time. Knowing how your IM behaves over the next months or years may enable portfolio adjustments to anticipate maturing trades, but also to predict funding costs.
Relying upon external parties to tell you the risk in your portfolio isn’t the way forward – firms need to equip themselves with the right tools to manage their portfolio in all the circumstances described above, in order to make decisions which can lead to cost reductions. Decisions you will need support for include:
- Calculating the impact of each trade on capital and IM funding costs, using Historical, Monte Carlo or SPAN methodology.
- Optimising the CCP venue or bilateral trade choice to minimise capital, collateral and IM funding costs
- Predicting future IM requirements as portfolios mature
- Reducing margin/collateral overheads. Intra-day analysis improves liquidity and collateral funding requirements
- Pre-deal limit checks to avoid failed trades
The Razor Risk platform can deliver this, and much more.
Razor Risk is a trade-mark of Razor Risk Technologies Pty. Limited.