ISDA 2016: Implementing Margin Rules for Non-Cleared Swaps

On September 20, 2016, I attended the ISDA 2016 Annual Europe Conference, titled The Future is Now: Implementing the Margin Rules for Non-Cleared Swaps. The Conference Agenda is here complete with the list of panelists for each of the sessions:

• Fireside Chat: The Future is Now
• The ISDA SIMM: Setting a New Standard
• The ISDA SIMM: Available at a Terminal Near You
• Collateral Management Solutions: Is Help on the Way?

My notes of the key points that I found interesting are as follows.

Fireside Chat: The Future is Now

Scott O’Malia, ISDA CEO, set the scene noting that from September 1, nineteen Phase 1 firms in the US, Japan and Canada were live with ISDA SIMM for Non-Cleared Swaps Margin and invited the two panelists to give their thoughts.

Eric Litvack, SocGen and ISDA Chairman, spoke on how this day had been a long time coming and the implementation going down to the wire with regulatory approval of the ISDA SIMM model in August by the appropriate regulator for each firm and jurisdiction, The first day was dicey, in that even though all (most?) agreements were signed off, in some cases the account set-up at a custodian was not complete or the information had not made it back to the Front Office. An estimated 90,000 trades were conducted by Phase 1 firms in the first 10 days of trading.

Kieran Higgins, RBS and ISDA Board Member, spoke on some issues with disabling or enabling legal entities on electronic platforms, not many margin disputes encountered, consistent delivery of VM, but in-consistent delivery of IM due to the afore-mentioned custodian issues, some ambiguity on which life-cycle event on an existing trade meant it could no longer be grand fathered and would have to be submit to margin, not yet seen a change in pricing from one mid to 19 different mids, but expect this to happen as portfolios build up and dealers look at the margin impact of new trades.

Then the following points were made by one or other of the above three:

• European deadline for Phase 1 firms is next, exact date not fixed, thinking is Jan/Feb 2017
• Unlike the US where pre-approval of the use of ISDA SIMM as an internal model was required, in the EU firms do not need pre-approval but need to be able to defend their choice of model
• March 2017 big-bang for VM when all firms needs to exchange VM is the big looming issue as each dealer may have several thousand clients to deal with and make sure agreements are in place
• Amending current Collateral Support Agreements (CSAs) requires the negotiation of terms which have economic impact (e.g. optionality in currency of delivery) and this is time consuming and much harder than signing new CSAs for new trades, which has no economic impact, so a new impetus and imperative to move to standard terms in documentation
• Most clients already exchange VM, so the major change for them is the process to pay and receive IM
• Phase 1 firms that licensed ISDA SIMM got access to model backtesting results that they could share with regulators to greatly help with the internal model approval process
• Relatively few model based disputes have been seen in practice, disputes down to the population of trades differing due to cut-off times or jurisdiction
• Cross Currency Swaps not covered adequately, pending an update to ISDA SIMM wrt basis risk
• Transparent and simple model to implement, someone has even done it in Excel! (a reference to our blog)
• The change in pricing from one mid to each dealers own mid will first be seen in portfolio trades (e.g. novations/assignments),  as the “axe from a margin perspective” is considered, an additional cost
• Manifestly higher Non-Cleared margin cost will encourage a further move to Clearing

Scott ended with a line that neatly captures ISDA’s efforts, “Documents that are Standard, Calculations that are Standard and now the Process needs to be Standard”.

The ISDA SIMM: Setting a New Standard

Paulo Peres, JP Morgan and Tomo Kodoma, BAML gave an overview of ISDA SIMM:

• ISDA SIMM is a Variance-Covariance VaR model
• Similar to and an offshot of the FRTB model (see my article)
• 99% confidence level and 10-day MPOR
• Backtesting results over a year show it to be a conservative model
• “Not a pretty car, but will take you to where you want to go”, a reference to the simplification of the real world that underlies a Var-Cov VaR Model, e.g. Normal distribution, Correlation assumptions,
• Effort to implement is mostly about how to pull all the risk sensitivity inputs together from a firms systems
• In some cases the calculation of these risk sensitivities needs to be changed to be SIMM consistent
• Common Risk Interchange Format (CRIF) is the ISDA SIMM standard format for these risk sensitivities
• The need for this data at a trade level as opposed to commonly used book or desk levels
• ISDA SIMM being a portfolio risk based model leads to between 5 times to 30 times lower margin than the Standardized Schedule Margin model which is percent of notional based

Next Paul Shah, Nomura and Stewart Quinn, UBS also joined the conversation for a discussion on Model Approval and Testing:

• As ISDA SIMM is a simple model, the implementation is the key to getting internal model approval
• Soundness of data feeds, soundness of risk sensitivity calculations,
• Are a firms approved internal models being used,
• Ability to monitor the actual coverage of margin and if it falls short, a process for additional margin
• ISDA Unit tests and backtesting results are important
• Regulators are satisfied in general that ISDA SIMM is conceptually sound and fit for purpose
• Fifteen Phase 1 firms have been running ISDA SIMM for a few months prior to September 1st
• And checking their actual portfolios in the real world and comparing results, making sure that risk sensitivities and margins match and that margin cover is adequate
• This gave a lot of confidence in the use of the model from September 1st.

The either Katherine Darras, ISDA or Tara Kruse, ISDA, (my notes are illegible on this point) spoke on the Crowd Sourcing Utility :

• For the correct identification of risk buckets for Credit and Equity
• One firm might assign a name to High Yield, another the same name to Investment Grade
• This has a significant impact on the margin and results in margin disputes
• ICE Benchmark Administration runs this Utility
• Firms vote for each name on the appropriate the risk bucket to use
• A Consensus view is determined and sent to each firm
• Over 11,000 results being published daily
• Not every firm using the data votes, but firms are encouraged to do so

One of the delegates confirmed to me over lunch that prior to the use of the Crowd Sourcing Utility, there were lots of differences in risk buckets and margins, which have now been eliminated.

A final free-fall:

• SIMM Governance by the Executive Committee
• Model is calibrated to a 1 year stress period and 3 year normal period
• Annual re-calibration exercise
• Custodial set-up, in Phase 1 for 19 firms, something like 150 trading pairs but for VM big-bang in March 2017, thousands of firms and tens of thousands of pairs
• VM Custody is 1 sided while IM Custody is 2-sided (what I need to collect, what I need to post)
• FX haircuts not common for VM but are for IM
• Some Life-cycle events, e.g. fixings do not result in grandfathering being lost
• Others, re-couponing, increasing notional, assignment result in the trade being subject to new margin rules
• VM practice prior to September 1, was for calls to be met T+2, so overnight calc, next day to check and resolve disputes and then settle following day, with 85% of margin calls settled on T+2
• In the US 30% of calls are met same day and outside US this is 10%
• So the requirement for VM to settle T+1 from March 2017, is a big change in practice

A final explanation of Why ISDA SIMM is Parametric and not Historical Simulation (see my article on Why the ISDA SIMM Methodology is Not What I Expected and the more recent Uncleared Margin, ISDA SIMM and FRTB SA), which was explained as a quick model to use pre-trade and huge savings in Data and Infrastructure costs over Historical simulation as ISDA calibrates model parameters and makes these available to firms.

A memorable comment from Paulo, said with so much passion, that I want to believe him.

I paraphrase his words, “The largest industry effort in the 20 years that I have been working”.

Phew! Two panels and so much new detail, onto the next.

The ISDA SIMM: Available at a Terminal Near You

This was a vendor panel, the first of its kind at an ISDA Annual Conference, according to the chair, George Handjinicolaou of ISDA, with five vendors talking about their experience and purpose in using ISDA SIMM.

Full disclosure, I am conflicted on this panel, so will only summarise my notes on Project Blazer from Mark Demo of Acadiasoft as this is by far the most important vendor initiative to date:

• Acadiasoft and TriOptima collaboration
• In June 2015, raised funds and more partners
• Working Groups, with over 325 sessions to date!
• 800 industry professionals have been involved
• 6,000 pages of supporting docs – gulp, I won’t be reading those, still ploughing through Tolstoy this week 🙂
• Collateral Hub in use by 35 major banks and 22 Counterparty Groups
• Phase 1 firms send trade risk sensitivities each day
• Margin using ISDA SIMM is calculated
• Return results and reconciliations to each firm
• Am sure more detail on the process is on the Acadiasoft website

Thats it, now lunch for me, but for you please keep reading.

Collateral Management Solutions: Is Help on the Way?

Unfortunately  the lunch was too good, the coffee not strong enough, so my notes are very brief here.

Coupled with the fact that I kept meandering off into regretting why I had sold USD and bought GBP this morning at 8:30am when the mid was 1.3043, only to see it now at 1.2964. Not that it was worth that much to me in monetary terms, but having waited more than a week for GBP to break 1.30 and not having the patience to hold out, it is the emotion that counts. Good job I am not a trader. Onward.

• Amy Caruso, DTTCC, said that these Margin Rules would cause more change and more impact than the hitherto US regulations on Mandatory Clearing and Mandatory Execution.
• David White, TriOptima, that 75% of all bi-lateral trades are reconciled on triResolve and improving the process of dispute resolution.
• Nathan Ondyak, LCH, on the need to create a common technology stack for Cleared and Non-Cleared and the pitfalls in trying to automate a process as is, rather than replace entirely e.g. dispute resolution does not exist in the cleared worlds as the CCP establishes the acceptable mark.
• Jonathan Cooper, Broadridge, collateral optimisation starts with good inventory management covering Repo, Securities Lending and Derivatives, more real-time the better, including what has actually settled.

Much more was said by all, but apologies to Hugh Daly of Message Automation, Helen Nicol of Lombard Risk and Darryl Twiggs of Smartstream, I did hear your words but have nothing down on paper to recall specifics.

Lots of good sentiment on vendors collaborating and integrating to provide for the electronification (?) of collateral management.

Finally the moderator, Steven Kennedy of ISDA, posed the obligatory block-chain question, which I think one or each of Darryl, Jonathan and Nathan game-fully answered or punted, I forget which.

Roundtable Discussion on Non-Cleared Margin Implementation

And last but not least, a roundtable moderated by Scott O’Malia on implementation with more detail on Legal and Operational issues and Buy-side input.

Mary Johannes of ISDA and Head of the ISDA WGMR Initiative spoke on the SIMM Oversight Committee, the need for a firms Model Validation team to involved early in the process, ISDA documents and backtesting results get you 80% to what you need for Regulators but Model Validation team is crucial. Regulators appreciate the size of the challenge and the work effort undertaken by ISDA and its member firms.

Paul Shah, the need to enrich trade date with regulator specific details (CFTC, Prudential, Japan) and that there have been fewer disputes since September 1, than in the test period, either due to the small product set and number of trades executed in the period and perhaps a greater scrutiny now real production and not testing.

Mark New of ISDA, spoke on the documentation efforts and how the new ISDA VM protocol allows many people to amend their legal documentation all at the same time and these are applicable to all types of firms. However the IM Docs were specialised for Phase 1 firms and see the need for new revised IM documents as further phases of firms have to comply.

Tracey Jordal of PIMCO, on the need for Legal, Operational and Business to all have input on proposed changes and the un-expected impact on PIMCO even of the September 1, 2016 deadline that was not meant to be a relevant date for them. It became more difficult to do Novations/Assignments of deals, a common practice at PIMCO as Dealers were either not willing to step into a deal or were quoting a cost to do so (as the trade would now be subject to IM). Standardised CSAs are the way to go for the industry. PIMCO as a large asset manager, does not actually hold client assets but directs custodians to move assets and practices vary a lot, a move to margin on T+1 is challenging as clients may have assets that do not settle T+1 and are not really interested in the arcane details of how the Margin rules impact their assets.

Paul Shah, the dealer market uses cash for VM, so T+1 is practical, while for IM can run a buffer to cover the expected days business and not face intra-day calls. The charging of IM covers risk and provided integrated into Capital rules should result in the freeing up of Capital for the business to trade more. Non-Cleared Margin pricing is rather like CCP Basis or Physical Commodities where the point of delivery is important to the price. Higher margin, will help with a push to Clearing for more products, even those without a mandate but a CCP needs to ensure sufficient liquidity in that product first as should liquidity dry up in future, than managing a member default would be very difficult.

The End

Thats it, thank you for reading to the end.

I certainly learned a lot from the event.

Errors, omissions, in-accuracies in the above are all mine.

I will end by quoting Scott O’Malia.

Documents that are Standard, Calculations that are Standard and now the Process needs to be Standard.

There is more work to be done.