Pure Agency: Reducing Client Clearing Bank Capital Burdens
Post SA-CCR capital drivers
For what looks like the last time, bank regulators are working through material changes to the Basel III capital rules that affect banks' OTC client clearing businesses. The changes center on the implementation of the Standardized Approach for measuring Counterparty Credit Risk (SA-CCR) in the revised default fund capital calculation and the leverage ratio PFE calculations. The former aspect is planned to go live on January 1st 2017 whilst the latter is a declared intent of bank regulators but doesn't have a firm live date yet.
In my recent post on the subject I interpreted that the leverage ratio adjustments may include exempting segregated client initial margin (IM) from replacement cost but that IM is unlikely to ever be allowed to offset the PFE either in the leverage ratio or default fund hypothetical calculation. The benefit will be to largely curtail the inappropriate influence of raw portfolio notional as a significant driver of banks reg. capital. However, with notional tamed, all of the clearing-related regulatory capital (leverage ratio, default fund capital and CFTC minimum capital) will be driven by cleared portfolio exposures.
An interesting related point is that, once client clearing reaches a mature state, CCP portfolio risk will be dominated by client portfolios. Taking the two points together, we can infer that all banks' client clearing related reg. capital will be mainly client-driven.
In other words by fixing the CEM problems with leverage ratio and default fund calculations, we are not surprisingly returning to the concern that bank members subsidize client risk which was a top concern before the onset of the hard leverage ratio trumped it.
The key rider is that if client IM won't offset these capital charges so the only way to reduce capital further will be to directly reduce client-driven counterparty exposures.
Rising clearing fees
Recent Risk.net articles (subs. required) confirm that banks are raising clearing fees to help them reach target ROE somewhere above 10% despite leverage ratio capital costs:
- Goldman Sachs is adding a 75bps charge on cash IM for non-top tier clients
- Citi published figures indicate 3x or more higher clearing fees might be required to generate adequate ROE even after the SA-CCR changes
When default fund capital and CFTC minimum capital become significantly driven by client risk in a mature state, this can only lead to further ROE degradation or higher fees. So we certainly don't know yet whether this approach will work. Even if it does it could lead to very skewed pricing between top-tier clients who likely bring the bulk of the client-driven CCP risk and non-top tier clients who likely bear the brunt of the fee increases.
Even if fee increases ameliorate client clearing businesses' ROE, they may not go far enough to fix the "porting disincentive". When a member defaults the client clearing model assumes that another member accepts the defaulting members client portfolios. The new member takes on the capital burden of the portfolio but doesn't get to charge ticket fees. Since inward porting is not mandatory to the other members, porting could fail and the clients of a defaulting member would themselves be closed out (even though they have not defaulted). In this way clients bear "porting risk" in the current client clearing model. The risk is material because current pricing models are dominated by upfront ticket fees so there's likely no way the accepting member can generate reasonable returns on the capital burden taken on. Only by substantially shifting fees from ticket fees to maintenance fees can porting be adequately incentivized.
Non-bank provided client clearing
To recap: presuming that IM offset is not allowed of SA-CCR PFE, the only way to materially eliminate client-driven capital charges for clearing is to eliminate the associated client-driven counterparty exposures from banks. One way is to get the banks out of client clearing.
One way is for buy side firms to self-clear. Recent clearing regulations have enabled buy side firms to become CCP members but take up has been negligible. Buy side firms likely don't want the burdens which go with being a member which include:
- The demanding settlement and treasury management requirements to fund a direct-debited set of settlement accounts daily without fail on pain of default
- The obligation to bid in a default auction, book won portfolios and participate in periodic dress rehearsals
So beyond a few aspiring non-bank market makers, e.g. Citadel, as a blanket solution this seems a remote prospect.
Another way is for select non-banks to provide client clearing instead. From a regulatory point of view this would just push the exposures into the shadows where they would not be subject to bank capital rules. CCPs would only be comfortable with non-banks which could fulfill the above member obligations - hence perhaps the lack of willing non-banks. Banks would also lose a means to maintain a client relationship in an increasingly anonymous trade execution world where conventional sales forces have less interaction with clients.
In short, clients don't want to self clear and no one wants non-banks to take over client clearing provision.
CCP innovation of the client clearing model
So far the only example of CCP innovation in progress that I'm aware of is ICE's sponsored principal model (Risk magazine subs. required). In summary this seems to differ as follows from the regular model:
- The client and CCP settle the portfolio directly (without the clearing member being involved)
- Default fund contributions (DFCs) are still paid by the member but are calculated discretely for the client portfolio (as opposed to as part of the member default fund contribution calculation)
- The member can optionally turn over client default management to the CCP (in which case the defaulting client discrete DFC is used to cure losses directly after the defaulting client's IM)
Part of the purpose is to alleviate the capital burden on the member but it is not clear exactly how. What is clear is that the member still has exposure to the client. Even if the member turns over defaulting client close out to the CCP, they are exposed to losses on the discrete client DFC. So client-driven exposures and therefore likely capital burdens have not been entirely removed.
However, I believe the initiative to address this should partly come from one or more CCPs as rulebooks will have to change.
Pure Agency model
The current client clearing model is a "hybrid" of principal and agency counterparty exposures. The clearing member has some client-driven "principal" exposures facing the client (portfolio default exposure) and facing the CCP (exposure to mutualized losses on DFCs arising from client portfolios). Other client-driven exposures are "agency" e.g. the clearing member is not exposed on the client portfolio in a CCP default.
The above prompts the idea of making the model "pure agency" by removing the two principal exposures in parentheses above by legal restructuring?
An approach would be a combination of CCP rulebook changes and member client clearin agreement changes so that:
- CCP and client become direct principals of one another. Client default is called and managed directly by the CCP. The bank clearing member is not exposed to client portfolio default losses.
- Client pays a discrete DFC with associated mutualized loss exposure. Bank clearing member's DFCs / mutualized losses are limited to their own portfolio and are no longer client-driven. This model may not suit all clients. Legal restrictions may include restrictions disallowing e.g. 40 Act funds from taking mutualized risk associated with DFCs. Even if not legally prevented some clients may elect to stay with the current model even if this involves paying higher member fees.
Aside from these changes lets suppose the rest of the model stays the same. In particular, trade flow and normal times settlement via the clearing member continue with the clearing member acting as trade capture and settlement agent but with no principal risk. At the point of trade acceptance the client and the CCP become principals of one another per the rulebook - hopefully not much operational change is required. Normal times margin call settlement would also stay as is:
- The CCP direct-debits the clearing member bank's settlement accounts for client margin calls as today.
- The member guarantees the client's settlement performance outside client default. This may includes daily funding of the direct debit account and tackling the cost of regular 1-day funding lag between CCP margin settlement and the margin call from the client the following day. It may also include managing liquidity risks through funding buffers, liquidity backstops and credit lines.
Counterparty exposure impacts
The pure agency model impact on the exposures of the bank member, CCP and client are summarized in the bullets below:
- Bank member - eliminates default exposure to client; eliminates client-driven elements of DFC and thus the client-driven mutualization exposure to the CCP
- CCP - eliminates exposure to members being brought down by clients; takes on exposure to the client from the member (with additional mitigation from the client's DFC)
- Client - eliminates porting risk exposure to the clearing member; takes on mutualization exposure on its DFC to the CCP
So far we have removed the client risk driven exposures from the bank member (see diagram),
Eliminating capital charges
The pure agency model is not worthwhile if it doesn't enable the capital burden to be removed by changes and approvals from banks' internal policy groups and possibly regulators. Once pro forma changes to the rulebook and clearing agreements have been produced the following hoops may have to be jumped through:
- Bank internal legal and risk approval for agency treatment. A risk and legal opinion would be needed that there was no principal risk on pure agency model client portfolios and DFCs.
- Bank balance sheet accounting policy. A firm agency opinion above should enable the client portfolio, IM and DFCs to be removed from the US GAAP and IFRS balance sheet based and also enable removal from Basel III capital (leverage ratio, client-facing RWA and default fund capital) as well.
- Basel III capital rules. It is hoped that removal of the relevant pieces of the capital burden is largely a consequence of bank policy approval. However, given the new model is structurally different there may be a need for some level of representation and approval in the Basel III and bank regulators capital rules and some dialogue with regulators as a result.
- CFTC minimum FCM capital rules. These may need change to excluding client IM for client portfolios under the pure agency model from the 8% of total IM calculation.
These policy changes may take considerable cross-discussion and analysis among banks, clients, CCPs and regulators. Having looked at the changes in a bit more detail I don't see fundamental stumbling blocks. It is hoped this can be done in a reasonable timeframe.
Assuming approval is obtained the diagram below illustrates the removal of the red shaded client-driven capital burdens from bank members under the Pure Agency model. (A fuller elaboration of the bank member capital charges in the current model is provided in my recent post.)
Impact on commercial incentives
It's worth contemplating an incomplete transition to a Pure Agency model. As noted before some funds are prevented by legislation or investment management mandate from contributing to mutualized default funds. The end state reality would likely consist of a mixture of clients adopting the pure agency model and clients who stay with the hybrid model. The hope is that take up is significant enough to make a major difference to client clearing business economics. In particular I assume that take up cannot be mandated and therefore the move to the new model needs to incentivized by commercial considerations.
Some obvious immediate commercial impacts to note:
- Sustainable bank member ROE: Client clearing ROE is rendered acceptable by reducing the capital usage for a given client portfolio rather than increasing the fees per the current approach. Members would offer both models but higher fees for the hybrid model than the pure agency model. This difference could easily be 3x or more (based on the above Citi article) so this would be a strong adoption incentive.
- Plentiful supply of client clearing and porting: With its current substantial capital requirements only the largest banks are capable of offering the service and even they are struggling. Removing the counterparty risk driven capital requirements may enable more banks to provide the service or at least enable those already in the business to remain in it.
- Lower client IM MPOR: Given the porting disincentive is eliminated - the risk of porting failure is largely eliminated. Hopefully this can justify to UK regulators that client IM MPOR can be reduced from 7 days back to 5 days as the 2 extra days are no longer needed to accommodate trying but failing to port.
This looks promising, but how do the pros and cons stack up overall for each actor? Here's a quick summary for each.
1. Clients lower fees and 15% lower IM (for UK CCPs) may well compensate for DFC funding of approximately ~2% of IM and the associated mutualization exposure
- A key related point is that as total DFCs will increase much more slowly than total portfolio risk. You can imagine that not many client portfolios will be larger in risk terms than the existing clearing member portfolios. Yet total risk will maybe increase by a factor of 5x from today as client clearing matures. The DFCs are sized by the largest two portfolio worst case losses which perhaps won't go up so much given that dealer portfolios are already quite large in risk terms and likely to go down over time in response to incentives. However total IM will increase substantially with client clearing maturity. Where DFCs are maybe ~6% of IM today they might be ~2% in future. (Figures illustrative only).
2. CCPs: Removal of member capital constraints from expanding their businesses may be enough. The reduced protection of their skin in the game from client defaults may weigh on their mind.
- Today CCP skin in the game is only impacted if the member also defaults and burns through its IM and DFC.
- Time-efficient default management processes can limit the number of days close out takes and therefore limit the number of days of market move the defaulted parties IM is exposed to.
- Tightly controlled normal times risk management can also help. If they operate net settlement of date-synchronized cash flows with trade currency variation margin and pre-fund initial margin, exceedences of IM by fails prior to default can be limited to market move VM fails on exceptionally large market moves.
3. Members might transform the return economics of the client clearing business. Their main concern may be that CCP risk and default management of clients is well organized (including some of the items above) to mitigate mutualization risk given other members are no longer protecting their DFCs from mutualized losses if their clients default.
4. Regulators might be pleased to avoid an FCM concentration crisis (or EU equivalent) and therefore enable relatively unconstrained expansion of client clearing under the G20 mandates. They may also relish the prospect that clearing fees may be lower in general and less skewed between the largest clients and other clients.
I summarize in the diagram below the pros and cons for the four actors in fuller form.
Caveat: figures are indicative SWAGs from experience only. Further analysis would be required to substantiate the commercial case.
Further analysis and implementation approach
Of course a lot more analysis is required to define the risk, legal, operational and policy aspects of the approach and to plan and estimate implementation and to land the full business case. I have not included my further thoughts here at this stage. However, I have captured 20 further implementation points and haven't yet found a gotcha.
As such I believe at this stage the Pure Agency model merits wider discussion with a view to exploring the possibility of taking it on.
- It is possible or even likely that capital rule changes and fee increases alone will not deliver sustainable economic returns from client clearing.
- If not the elimination of client-driven portfolio exposures from bank members by client self clearing or client clearing by non-banks does not look promising.
- The pure agency client clearing model outlined here may be a viable way to enable the substantial elimination of client-driven counterparty exposures and the associated capital burdens from bank clearing members whilst still enabling them to provide client clearing.
- The model may also make sense for clients, CCPs and regulators too.
Perhaps one CCP and a small number of clearing members are interested in taking this idea forward?