Initial Margin on Uncleared OTC Derivative Trades: The ISDA Standard Initial Margin Model (“SIMM”)

Following the Financial Crisis the Financial Stability Board (“FSB”) of the G20 nations at its Pittsburgh Summit in 2009 determined that one key component of its response was to force
April 25, 2016 - Editor

Following the Financial Crisis the Financial Stability Board (“FSB”) of the G20 nations at its Pittsburgh Summit in 2009 determined that one key component of its response was to force certain standardised Over the Counter (“OTC”) derivatives into Central Counter Parties (“CCP’s”) or “clearing”.

Following the Financial Crisis the Financial Stability Board (“FSB”) of the G20 nations at its Pittsburgh Summit in 2009 determined that one key component of its response was to force certain standardised Over the Counter (“OTC”) derivatives into Central Counter Parties (“CCP’s”) or “clearing”.

One part of this strategy is to impose stringent margin requirements on non-standard OTC trades which remain outside clearing. These proposals cover both Variation Margin (“VM”) and Initial Margin (“IM”).  VM settles profits and losses incurred between counterparties to date, with IM seeking to provide a cushion for potential future losses.

The Working Group on Margin Requirements1 (“WGMR”) consulted widely and received a great deal of industry feedback culminating in a final policy framework published in March 20152.  A Quantitative Impact Study (“QIS”) was also undertaken; highlighting that the effect on the industry will be very significant. Although starting on 1 September 2016, the proposals are to be phased in, and will not be implemented in full until 1 September 2020. Nevertheless, according to ISDA3 roughly 95% of “clearable” Interest Rate Derivatives are currently cleared4.

The proposals raise many questions, key amongst these are:

  • Which IM requirements will apply to uncleared OTC derivatives?
  • How can the IM calculations be agreed by parties to the contract? Large dealers have proprietary models whose details they do not wish to disclose.
  • How do these work across different jurisdictions given the fact that the rules are different across jurisdictions?
  • What collateral will be acceptable to post as IM?

Initial Margin Model

To help provide answers to these issues ISDA initiated a WGMR Implementation Program to facilitate the implementation of the margin rules across jurisdictions5. These are currently in draft form and may be subject to change. They do, however, set out current industry thinking and a framework under which IM can be calculated and agreed. A further objective of this initiative is to reduce disputes though the transparency of the model approach. The model is called the ISDA SIMM (Standard Initial Margin Model).

The concept of an IM model is not new. The London International Financial Futures Exchange introduced SPAN6 margining (niftily known as Liffe-Span) in 1990 following its use in Chicago. The idea is simply to look at an aggregate portfolio, apply shocks to the underlying risk factors and find out the maximum loss through some aggregation process. Key to this is the notion of a “margin period of risk”, which is the period which needs to be covered by IM to allow the portfolio to be liquidated or the risks immunised. Clearly a model for OTC derivatives will be much more complex than one for exchange traded instruments. By definition, OTC derivatives are non-standardised and illiquid and therefore harder to hedge or immunise.


This is certainly not a “one size fits all” approach. There are a number of choices and so counterparties will need to check carefully, and agree, a number of details. The OTC “world” is divided into five risks7, six risk classes8, four product classes9 and two different ways of handling equity risk factors. The key to the risk aggregation methodology is to estimate the correlations between various inputs. There are literally hundreds of these. The correlation matrix of risk sensitivities across different time buckets contains 120 factors, 18 of which are zero and one is negative. So the correlation between 11 year and 15 year buckets is 18% whereas between 11 year and 16 year buckets is zero.

No doubt a great deal of hard work has gone into calculating these correlations, and will continue to be invested. Those of us who have spent several decades in the markets know two things. Firstly historic correlation is easy to calculate but of limited relevance, whereas future correlations are highly relevant but not measurable. Secondly, correlations are highly stable until they become unstable. Shocks are sudden, violent and unpredictable. The fact that these inputs are hard to predict does most certainly not mean that it should not be attempted, simply that the limitations of the modelling approach should be fully understood.


In October 2015 ISDA announced that it had appointed ICE Benchmark Administration Ltd to build and operate the crowdsourced utility for the SIMM. It is interesting that the methodology chosen is not one of those widely used. Value at Risk (“VaR”) usually comes in three flavours: Parametric (also known as variance-covariance), Historic Simulation and Monte-Carlo Simulation. The method most widely used is Historic Simulation10. Regulators are moving from VaR to Expected Shortfall. VaR gives a probability of losses not exceeding a certain amount to within a given confidence limit. However, it gives no information whatever about the likely losses when the specified amount is exceeded. Expected Shortfall attempts to remedy this by predicting the loss distribution in the tails (outside the specified amounts). It is also interesting that the draft ISDA SIMM document annotates the term ISDA SIMM with a note stating “patent pending”. Patenting a new methodology at this time may well be commendable. However, given the timescale and widespread use of an existing methodology, is it a bridge too far?


One of the criteria for determining the eligibility of a derivative to meet the clearing mandate is the liquidity of instruments and the ability of a CCP to hedge or immunise in case of default. A large portfolio of liquid 20 year “on-the run” interest rate derivatives may well meet this test. However, in a year the portfolio may become 19 year residual maturity, off-the-run and illiquid – what happens then? Either a CCP may be able to “put” the portfolio back to market participants causing tremendous disruption, a risk impossible to measure and capitalise, or there will be a “Hotel California” model (remember the Eagles?) meaning that once a derivative is in clearing it can never get out. In the latter case CCP’s must be capitalised and have the risk management capabilities to manage large illiquid portfolios, with a potentially large requirement for collateral.

Risk Weights

The approach of determining risk weights prescriptively is bold, but one followed by the Securities and Futures Authority many years ago. Volatilities, like correlations, are easy to measure historically but unmeasurable (other than from market pricing) for the future. The British Pound is currently categorised as “regular volatility”. Imagine how a Brexit could change that. The point is that, if parameters are to be centrally measured and maintained, recalculation needs to be frequent and accurate. The fact alone that this is a Herculean task again does not mean that it should not be attempted. There is also the question of ownership of the data and who will underwrite losses arising from errors or omissions.

Model Approval

Obtaining, and maintaining model approval for the calculation of IM is non-trivial. There are a number of issues including:

The current WGMR proposals require firms to apply to each jurisdiction in which they operate for approval. There is no concept of “Home/Host” whereby models approved by the Home regulator would automatically be accepted by Host regulators. This would make the process potentially much longer, complex and expensive as different jurisdictions apply differing standards, data requirements and languages. A common ISDA SIMM approach should certainly make this process much easier. However the challenges of obtaining model approval should not be underestimated. Regulators advise that, when considering a model, only 30% of the process addresses the model. The remaining 70% is based on systems and control, management, supervision and other firm-specific issues. The ISDA SIMM approach should certainly reduce the 70% part, but will certainly not eliminate it.

Model approval may be restricted to certain instruments, risks and data sets. Thus two firms facing each other would need to check the full details of each other’s approvals before IM could be agreed. The types of collateral acceptable with models approved by different regulators may differ (see below). Hence two firms facing each other have a great deal of details to check.  The benefits of an industry model are derived from the efficiencies of being applicable to all and not needing to be tweaked in respect of each counterparty relationship.  Clearly the specific nuances of a party’s model approval are problematic in this regard.

Dispute Resolution

The rules for the calculation of Counterparty Credit Risk (“CRR”) involve significant add-ons in respect of unresolved disputes including those relating to collateral. It is therefore extremely important that, in the face of all of the issues outlined above, collateral disputes can be quickly resolved. One aim of the ISDA SIMM, which is to be highly praised, is a significant reduction in disputes through the use of (more) transparent models. Institutions who do not use the ISDA SIMM my find it hard, if not impossible to agree collateral calls on a daily basis.


It has long been recognised that the imposition of margin requirements on uncleared OTC derivatives will create a very significant requirement for “acceptable” collateral. In order to mitigate the ensuing collateral shock the WGMR has, quite rightly, attempted to widen the classes of collateral which are acceptable to cover IM. One consequence of this is a proposal that individual jurisdictions set their own lists of acceptable collateral. After all, if covered mortgage bonds are liquid in certain Nordic countries, why should the Nordic regulators not accept these? No reason at all other than, a firm running a centralised collateral management unit (by far the most efficient way to manage collateral) may face real challenges. Collateral considered acceptable in one jurisdiction may well not be considered as acceptable in another.

The consequence is that we may move from a world of Black and White, where each piece of collateral is either acceptable or not acceptable, to a world which might be characterised as “Fifty Shades of Collateral”. Clearly a legal documentation nightmare.  There is already such a major dependency on the legal documentation aspect of the margin reform regulation implementation and it will be certainly interesting to observe how firms manage this.


This is an area fraught with difficulty. Although the intention of the FSB is hard to argue against, its implementation shows that the devil really is in the detail. Firms, not limited to banks, face an array of hurdles and some will inevitably require outside expert assistance. These include model approval, data sources, risk categorisation and legal documentation management together with building the remainder of the 70% of the model approval not covered by the ISDA SIMM itself. The problems of adapting to a new methodology should also not be underestimated.

The ISDA SIMM is a welcome initiative and one which clearly has a long way to run. In view of the WGMR implementation timetable, this should be at the forefront of the industry’s mind. If it is not, it may already be too late.


  1. A group formed in 2011 by the BCBS, IOSCO and other interested parties
  2. BCBS publication d317
  3. The International Swaps and Derivatives Association Inc.
  4. ISDA Derivatives Market Analysis: Interest Rate Derivatives, 21 January 2016
  5. Draft ISDA SIMM Methodology V 3.8 documents published 12 January 2016
  6. Standard Portfolio Analysis of Risk
  7. Delta risk, Vega risk, Curvature risk, Inter-curve basis risk and Concentration risk
  8. Interest Rate, Credit (Qualifying, Credit (Non-Qualifying), Equity, Commodity and Foreign Exchange (“FX”).
  9. Interest Rates and FX, Credit, Equity and Commodity
  10. See Amir Khwaja of Clarus Financial Technology November 2015


This article was first published in edition 6 of Rocket, our magazine. Download available Rocket editions here, and save your up to date address in your profile to to indicate your interest in receiving a printed copy of the magazine. Copies are also available to purchase and subscribe to via the shop.

To save your address into your profile:

  1. Visit the home page
  2. Click Account (in the middle of the row of black buttons)
  3. Click Edit Profile (in the row of buttons at the top)
  4. Click Reader (top right)
  5. There you can see your profile, with a box for your address – complete it accurately, and click Save


Most Viewed