March 21, 2015

SA-CCR: risk-sensitive FCM capital but still no IM offset

Regulators have declared their intent to implement the Standardized Approach for measuring Counterparty Credit Risk (SA-CCR) in critical parts of Basel III capital rules.  This starts to go live at the beginning of 2017 for default fund capital and the leverage ratio implementation will follow.  
Here I take a high-level view of the effects on capitalization of clearing members house and client portfolios and whether capital rule changes are likely to make client clearing a commercially sustainable business.

Regulators have declared their intent to implement the Standardized Approach for measuring Counterparty Credit Risk (SA-CCR) in critical parts of Basel III capital rules.  This starts to go live at the beginning of 2017 for default fund capital and the leverage ratio implementation will follow.  
Here I take a high-level view of the effects on capitalization of clearing members house and client portfolios and whether capital rule changes are likely to make client clearing a commercially sustainable business.
Regulatory Background

Banks' internal model methods (IMM) have been developed with regulatory approval to fully model portfolio risk netting and margin offsets.  The models are complex and hard to compare between banks on a like-for-like basis. The current alternative is the current exposure method (CEM) used in cases where non-internal model methods are required or direct comparison is mandatory.   CEM has been heavily criticized because it poorly models poorly models portfolio risk netting and margin offsets.   

The Standardized Approach for measuring Counterparty Credit Risk exposures (SA-CCR) is regulators attempt to find a happy medium between the two extremes where IMM is inappropriate but CEM is too crude.  So far regulators inked the planned SA-CCR implementation in default fund capital calculations as January 1st 2017 (as part of the April 2014 final rules on bank capitalization of cleared trades). 

Next up seems to be replacing CEM in the leverage ratio calculation in response to concerns that the leverage ratio PFE calculation does not adequately reflect portfolio netting (see my post here on the client clearing specifics).  BCBS and the Fed have declared this intent and we await the completion of the BCBS work on the default fund capital implementation so that regulators can finalize plans.
Clearing model as context
As context here's a simplified model of client and dealer clearing.  Dealers either self-clear (bank member above) or as the affiliate of a bank member (Affiliated Dealer Bank above) both via the bank member house account (blue arrow).  Clients clear via the client account of their chosen clearing member (red arrow).
SA-CCR – Rates Clearing Summary
Whilst the mathematics is not as complex as IM models or EEPE calculations, SA-CCR is by no means vacation reading.  More sophisticated netting makes it significantly more complex than CEM.  Complexities include varying approaches by asset class to portfolio netting and aggregation, cross-correlations and adjustments to notional to reflect trade duration and the close out holding period.  However, the final version still meets the goal of independence of bank-specific internal risk modeling.  
SA-CCR is an exposure at default (EAD) model which can be summarized as:
EAD = 1.4 x (RC + PFE)
where RC = replacement cost (a.k.a. positive current exposure)
  and PFE = potential future exposure   
Here's my interpretation for Rates clearing:
RC:  The change from CEM to SA-CCR doesn't make much difference to Rates clearing RC.  As for CEM daily in currency net settlement of trades and VM with zero thresholds and minimums.  This materially "nets out" trades and VM so only material element is the RC on initial margin and default fund contributions.  IM is the more material of the two.  The key point being lobbied is that the client IM is not actually an exposure of the clearing member (see IM LRE lobbying below).  
PFE:  Under CEM, PFE is fairly crude being the aggregate of trade gross notional x portfolio net to gross ratio x addon % per time to maturity bucket with no initial margin offset.    The key factors are:
  • Notional: gross notional
  • Netting approach: multiple gross notional by portfolio net to gross ratio (based on portfolio NPV as a fraction of gross positive PV) i.e. based on current exposure offsets not those in the future
  • Addon %: varies by time to maturity 0% up to 1 year, 0.5% for 1-5 years and 1.5% for over 5 years
Under SA-CCR, PFE changes significantly:
  • Notional: effective notional – this factors in trade duration, a maturity factor reflecting the 5 day MPOR and other notional adjustments – e.g. for amortizing swaps
  • Netting approach: effective notional incorporates netting within trade currency – fully within maturity bucket and partially across maturity buckets.  No cross-currency netting / correlation is accounted for.
  • Addon %: a flat 0.5% reflecting interest rate volatility as a whole (maturity effects are instead modeled in the notional and netting calculations)
  • PFE Multiplier: There is also a PFE multiplier which factors in the risk reduction effect of any over-collateralization
The message is that SA-CCR EAD imperfectly but largely models portfolio netting but still does not incorporate initial margin offsets.  For most portfolios there's much more PFE netting benefit from portfolio netting than initial margin offset.  But that doesn't mean that IM offsets are not worth having.
IM LRE lobbying
Beyond SA-CCR implementation the other two currently open strands of lobbying are the effort to exempt segregated client IM from LRE RC and the effort to allow held IM to offset LRE PFE.  The first of these has very broad consensus – even including CFTC Chairman Massad.  It is specific to client clearing and could be done simply by adjusting clearing specific capital rules here which already similarly exempt the client portfolio legs facing the CCP.  The second has a narrower consensus and would take more work to implement including re-work to the guts of SA-CCR.  The first looks more likely and faster to implement than the second.
Per my other post – here – I expect segregated client IM is exempted from RC but IM offsets will remain.  Let's assume this is the case for the purpose of discussion.
Post-SA-CCR how will member capitalization look?
First let's summarize the counterparty exposures between the players involved (see diagram).  Notice that inter-company cleared transactions drop out in consolidation at the firm level (every exposure being equal and opposite to the other entity's).   So for simplicity I've effectively ignored them.  Here our bank member has a house account portfolio exposure and is also exposed on the associated IM put up.  The bank member also has default fund contribution exposure to the CCP and exposure to the client mitigated by IM.  Meanwhile the CCP is exposed to the bank member on both the member's house account and the members client portfolios (mitigated by IM, default fund contributions and even the members capital in the case of client default).  The client is exposed to the default of the CCP though the assumption is of no exposure to the bank member given porting should take care of the member's default.
Next let's summarize the bank counterparty risk driven risk weighted assets (cRWA).   Notice we are only focused now on the bank's RWA so the client and CCP exposures are not included.   Not that I have highlighted in bold outline the element which seems to be prominent.  The other two numbers are either mitigated by IM or given a preferential 2% risk weight and can be considered relatively immaterial.  Default fund capital however is sized on SA-CCR EAD without mitigation of IM and distributed by member share of default fund contribution.   This is why I've highlighted it as a material capital cost.
Now let's summarize the bank leverage ratio exposure (LRE).  Again we're only looking at bank LRE.  Given the lack of IM offset in any of these numbers and given that IM is much bigger than DFC in raw number terms, we can downplay the LRE on default fund contribution relative to the other numbers.  The members own portfolio is likely to get more netting benefit from SA-CCR but won't likely become immaterial.  
Finally for completeness let's cover off the CFTC minimum capital rule.  Right now 8% of total IM is the minimum level (across client and house account portfolios).  The diagram is not of much use (except perhaps to show how little this approach is related to the actual exposures of the member).

Summarizing member capital charges
For the purpose of summary I ignore in the diagram below the capital charges where there are IM offsets (RWA facing clients) or a special 2% fixed risk weight (IMM risk on house portfolio and IM facing the CCP) or the raw amounts are simply small compared with others (DFC LRE RC).  If we do this we can summarize the material bank member capital charges as follows: 
1. LRE on house and client portfolios (SA-CCR RC (member only IM) and PFE (risk)) – a combination of the two numbered arrows below
2. cRWA on default fund capital on house and client portfolios (SA-CCR PFE (risk))
3. CFTC minimum capital on house and client portfolios (8% of IM)
It's important to realize that these numbers are not additive to one another – rather capital can be somewhat reused to meet each.  This means that the important one depends case-by-case on each bank's situation.  One bank may be leverage ratio constrained while another maybe RWA constrained and a third may be CFTC minimum capital constrained.  A firm may be constrained by more than one or all three.  
It is unclear to me which of the three charges will be more prominent but we can clearly say that all three charges are an increasing function of IM or portfolio risk unmitigated by IM.   
Capital optimization refocuses on risk reduction
A shift away from riskless compression?  In a CEM based LRE world riskless compression is heavily incentivized as reducing notional reduces LRE PFE.  In an SA-CCR based LRE world not so much.  SA-CCR based PFE will mainly be optimized by optimizing risk on the portfolio.  I expect this will lead to a decline in benefit from and volumes in riskless compression offerings such as TriReduce and CCP coupon blending.   This could be the platform for existing and emerging risk reduction tools coming to the forefront of portfolio optimization.  "Risk reduction tools" in the background include:
  • Elective backloading of legacy or non-mandated trades (a largely spreadsheet driven process to broker portfolio backloads)
  • CCP spread trading to reallocate risk between members portfolios in different CCPs (e.g. as provided by Tradition SEF)
  • Emerging risk compression tools e.g. LMRKTS and TriBalance (powerful tools which will have their day)

These tools will increase in use but the incentive for buy side firms to use them is much less than that for banks.

Over-collateralization?  Under SA-CCR over-collateralization by the client would give PFE capital relief.   Anecdotally some of this is already happening for RWA optimization purposes.  This is no way to fully eliminate PFE because the multiplier is floored at 5% and the function is exponential so you get less and less capital reduction for each $ of over-collateralization.  For example, I calculate that if the client is overcollateralized by half the add on then you get a 22% reduction in PFE.  But how much is the add on?  And anyway how many clients have plenty of IM funding in unleveraged securities?   They may do now but by the time SA-CCR is implemented client clearing will be much more IM intensive and bilateral IM will be in play also.

This is one tool but no silver bullet.
Incentivized client portfolio risk reduction?  It's not obvious why many clients would on their own care much about the cleared portfolio risk.  
  • CCP IM is some incentive to reduce CCP portfolio risk but clients so far have shown little discomfort in meeting IM requirements.  The margin financing and collateral transformation mooted bonanza has in fact been a muted bonanza.  This may change for some clients as client clearing reaches a steady state and the quantum of client cleared risk becomes much larger than today.  
  • As bilateral mandatory IM arrives this will lead clients to focus on clearing more trades given netting benefits on total IM.
  • Many clients are hedging non-OTC risk or trading directionally and are therefore inherently limited in optimization of portfolio risk
So it seems logical that members selectively incentivize clients to reduce CCP risk – either through raising risk-based clearing fees or by paying one-off fees to clients incentivize backloading of legacy portfolios.  However, the largest clients with most of the risk tend also to have the most negotiating leverage on clearing terms.
Overall, whilst a return of focus to risk reduction is generally right, it seems clients will be unlikely go all the way to optimizing members ROE problem.
Will the client clearer capital problem go away?
Whether the capital costs here force more dramatic decisions is partly about standalone clearing business ROE but also about the business division and firm level impact on total capital and ROE.   If clearing is a tiny business in the context of the firm a poor ROE and a small incremental capital usage amount can be tolerated.  If clearing capital requirements or ROE start to shift the firm level capital and ROE needle significantly this is when senior management will start to get involved.  I doubt whether this is the case for any players today with client clearing still a small fraction of it's eventual size.  
What I do know though is that if you put this together with a future where client risk dominates CCP risk in a mature state, you can see that clearing member capital charges for clearing will be dominated by client risk even outweighing their own portfolios in influence.
The linked article also indicates that client clearing will grow by 5x or more bigger than today.  Putting this together with the recent Citi article (subs required) the impact on clearing fees for clients even after SA-CCR has been implemented and client IM exempted from RC might increase annual clearing fees for clients for a given portfolio by 5x or more in many cases.  Combining these two factors simplistically indicates a 25x increase in member clearing fees for clients in a mature state.  That sounds like a bad sign at least for the large client clearing players.
Will some client clearers head for the door?
So far BNY, State Street and RBS have pulled the plug.  Of these only RBS was significant and we can put this into a special case category given the UK governments direct stake in and influence on the business.  A big sign will be whether one or two of the other G15 banks pulls out.   If they do this could be the start of a very bad time for client clearing (see outline at the end of my recent post).   The first question will be which other bank(s) will port in their client portfolios.  Inward porting will involve inherited client risk driven capital costs as well as the portfolio itself and thus be unappealing for most banks. 
At minimum this would not be pretty.
Fundamental change?
If client clearing is heading for some kind of crisis, I wonder whether there are other ways than risk optimization and yet more capital rule adjustment which are more palatable than the crisis.   I suggest CCPs and members could explore the possible restructuring of the client clearing model itself.  The purpose would be to reduce or remove counterparty risk from clearing members resulting from client portfolios.   ICE has made a start here with its "sponsored principal" model in which CCPs more directly face the clients and may manage client defaults directly.  
My thought is that we could go further than this and remove in a restructured model any member counterparty risk driven by client portfolios.  And with it the associated capital charges.
SA-CCR implemented as intended by regulators will substantially reduce the burden of leverage ratio and default fund capital.
The consequence once the capital relief is banked will be that all three remaining material capital charges on clearing members will be substantially driven by client portfolio net risk and will have no relief from initial margin offsets.
A positive effect is to shift focus towards portfolio risk reduction (away from the current distraction of notional reduction), client cleared risk and IM inherently resist optimization.
When combined with the future dominance of cleared risk by client portfolios it is clear that client clearing businesses will drive the lion's share of cleared product capital costs.
It looks at least possible that post SA-CCR capital charges mean client fees have to go up to much to generate sustainable ROE.
This suggests a different approach than adjustment of capital rules is required: restructuring the clearing model to remove member counterparty risk driven by client portoflios – and the associated capital charges.

Most Viewed


Related Articles

June 30, 2022

SIMM Falls Short says PRA Letter to Banks


Risk Management

June 28, 2022

FMSB Statement of Good Practice on Trading Platform Disclosures



June 20, 2022

Regulatory change and data fragmentation are key challenges for 85% of firms