Collateralisation: CVA & FVA - Murex - Alexandre Bon


Published on

OTC Collateralisation : implementations issues in the context of CVA & FVA
- The ideal CSA hypothesis : Imperfect collateralisation for credit mitigation and/or funding
- FVA vs. CSA-discounting
- Implications in terms of curves calibrations and management
- FVA for Cleared positions
- FVA or CSA-discounting : which funding management model?

Published in: Economy & Finance
1 Comment
No Downloads
Total views
On SlideShare
From Embeds
Number of Embeds
Embeds 0
No embeds

No notes for slide

Collateralisation: CVA & FVA - Murex - Alexandre Bon

  1. 1. OTC COLLATERALIZATION Implementation Issues in CVA & FVA Frameworks Alexandre Bon The CVA Conference: CVA & Liquidity Implementation Stream March the 23Rd, 2012 London
  2. 2. Preliminary remarks  Credit and Funding: – Two sides of the same coin – Inherently asymmetric and a portfolio level issue  Where is the risk-free asset?  What about the « Law of One Price? » – Market equilibrium price – close-out value – profitability measure  Speed of innovation vs. financial engineering entropy Copyright ® 2012 Murex S.A.S. All rights reserved 2
  3. 3. Agenda  CSA clauses & variations  Modeling collateralized exposures: CVA vs. FVA  The Standard CSA: challenges and systems implications  A word on clearing  CVA/FVA organizational frameworks Copyright ® 2012 Murex S.A.S. All rights reserved 3
  4. 4. Introducing the Collateral Agreement  A bit of terminology : Collateral, ISDA & CSA – – Non-mandatory appendix to the ISDA master agreement (which enforces closeout netting for eligible contracts) – Defines the scope and terms of bilateral remargining agreement – CSA is the most common (but not the sole) collateral agreement for OTC derivatives –  CSA : credit support annex Standard templates but varied implementations (differences in clauses and jurisdictions) Main clauses – Eligibility of positions to close-out netting and collateralization – Eligibility of pledge-able assets and applicable haircuts – Remargining process : valuation, frequency, settlement, reconciliation, dispute resolution – Determination of the collateral balance: symmetry, thresholds, independent amounts, minimum transfer amounts, rounding rules… – Legal framework: pledge or title-transfer, rights of re-use & rehypothecation – Remuneration of the collateral account (most often based on an OIS rate, but not always!) Copyright ® 2012 Murex S.A.S. All rights reserved 4
  5. 5. Collateralization & typical portfolio mix  An institution’s OTC portfolio will commonly contain a mix of: – Bilateral CSAs with 0 threshold and daily margining (cash) – Positions cleared on CCPs : daily or intraday exchange of Variation and Initial Margin – CSAs with asymmetric terms • One-way with SSAs • Over-collateralized agreements (IAs, Thresholds, IM) and security collateral (e.g. PB agreements) : small funds and corporates – No CSA – Multiple “collateral sets” with a single credit entity (by products : CSA, GMRA, OSLA, GMSLA … or entities)  Some local variations, but the interbank market is mostly on bilateral CSAs and daily cash margining  Imperfect collateralization bears additional risks & and warrants further valuation adjustments (credit and funding) Copyright ® 2012 Murex S.A.S. All rights reserved 5
  6. 6. Modeling Credit Exposures  Monte Carlo Simulation: – Generate thousands of risk factors paths across hundreds of time-points – Revalue each transaction on each node – Aggregate at each node the transactions MtMs taking into account credit risk mitigants to get the counterparty’s portfolio value – For each time step, take the histogram of Credit Exposures across scenarios to derive Expected Exposure and Liability profiles (as well as PFE or other statistics) MtM1 (1) MtM2 (1) . . . . . . Expected Exposure MtMN (1) Copyright ® 2012 Murex S.A.S. All rights reserved 6
  7. 7. MC system implementation questions  Models & calibration choices are important But do not underestimate the other big issues:  Data Quality & Flows  System(s) interoperability issues  Evolution and maintenance Copyright ® 2012 Murex S.A.S. All rights reserved 7
  8. 8. CVA - Collateral Modeling Counterparties  Netting Nodes Margining Nodes Eligibility rules CSA   Exposures are mapped to Netting and Margining sets Upon close-out the collateral balance is offset against the netting node positions ISDA - ABC No collateral Bank ABC GMRA GMRA No netting No collateral Copyright ® 2012 Murex S.A.S. All rights reserved 8
  9. 9. CVA - Collateral Modeling  Exposure at t is the difference of the Close-Out value of the portfolio and Outstanding collateral balance  Considering the simulation date correspond to the close-out date following default one identify the previous effective re-margining date  Common modeling options Margin Period of Risk previous remargining dispute fail grace period close-out Unsecured exposure (collateralised set) Exposure Collateral Balance Simulation date Ti-1 Threshold Simulation date Ti Ti - MPR Copyright ® 2012 Murex S.A.S. All rights reserved 9
  10. 10. Stylized example with Currency Swaps  CCS vs. FX reset swap  FX reset swap has an inbuilt collateralization feature  Market risk : low FX Delta for the FX reset swap Copyright ® 2012 Murex S.A.S. All rights reserved 11
  11. 11. Stylized example with Currency Swaps  CCS vs. FX reset swap  Impact of collateralization on the CCS risk (10 days MPR) Copyright ® 2012 Murex S.A.S. All rights reserved 12
  12. 12. Stylized example with Currency Swaps  CCS vs. FX reset swap  FX reset swap exhibits slightly larger collateralized exposures and liabilities, due to larger CF settlements.  Margining-square not a desirable feature  Side-note : collateralization is a great hedge – –  Copyright ® 2012 Murex S.A.S. All rights reserved CCS Bilateral CVA drop of 60%, FX delta and DV01 near 80%, CR01 near 80% The case for simple upfront reserving for collateralized exposure, vs. dynamic hedging of non collateralized exposures. 13
  13. 13. Stylized example with Currency Swaps  CCS vs. FX reset swap  Impact of reduced re-margining frequency (1 week to daily) Copyright ® 2012 Murex S.A.S. All rights reserved 14
  14. 14. CVA – Collateralization Efficiency Main collateralization risks and issues:  Lack of standardization across CSAs  Costs and risks of operation (bilateral & 0-threshold)  Concentration risk  Credit dependent clauses  Eligibility of collateral assets & haircuts  Execution : rounding, split differences, disputes – In practice collateral amount will never exactly match the exposure levels – The former are typically ignored in the model, the latter managed by adjusting the MPR of problem counterparties (dispute history).  Rehypothecation and re-characterization risks  Gap Risk – Model risk & close-out value – JTD & Wrong way risk Copyright ® 2012 Murex S.A.S. All rights reserved 16
  15. 15. One-Way Collateral Agreements Part I  Sovereigns, Supranationals and Agencies (SSAs)  Small non-bank counterparties without a collateral management function  Potentially large exposures for the un-collateralized party  Bilateral CVA ~ Unilateral CVA Exposures vs. Liabilities distributions Copyright ® 2012 Murex S.A.S. All rights reserved 17
  16. 16. Collateral gap risk  Instantaneous jump in exposure and counterparty default leaving a portion of the portfolio un-collateralized • More prevalent with imperfect CSAs (large thresholds & MTAs, longer remargining frequencies) • Credit protection bought from related entity • Simply settlement effects (warrant special treatment?) • Liquidity effects upon counterparty default Copyright ® 2012 Murex S.A.S. All rights reserved 18
  17. 17. Rehypothecation  Rehypothecation: –  Right of re-use: –  The collateral taker uses pledged assets as security for his own obligations to a third party Covers rehypothecation as well as any use of the collateral asset in line with ownership of the property (e.g. sale, lending to a third party) Depending on the jurisdictions and legal phrasing, collateral exchange can be performed under: – A Title Transfer Arrangement (implicit re-use rights) – A Pledged Collateral Agreement, where the rehypothecation right may be explicitely granted (often the case with non-bank counterparties)  Similar question with cash collateral and margin segregation  Some remarks: – Rehypothecation and “re-pledging chains” have played an essential part in providing liquidity (and leverage) to the financial markets – The GFC showed how damaging the combined effects of reduced collateral velocity (cf. Lehman close-out) and collateral squeeze (haircuts) can be in a systemic shock. – Not a desirable feature from a CCR mitigation point of view, but forfeiting this right represents a funding cost. Copyright ® 2012 Murex S.A.S. All rights reserved 19
  18. 18. Risk-free or risky close-out  ISDA documentation not 100% clear on how we should price the liquidation value of derivatives.  Open issue for default close-out as well as valuations for Unwinds and ATEs  Introduces a recursive pricing issue  Theoretical justifications for both approaches: the need for another Valuation Adjustment (RVA/ATEVA) ?  Practical questions: – Pricing of DVA or funding cost in distressed markets – Joint-default – Going-concern collateral balance is determined based on risk-free valuation Copyright ® 2012 Murex S.A.S. All rights reserved 20
  19. 19. From Credit to Funding Credit risk Discounting Calibration Valuation Copyright ® 2012 Murex S.A.S. All rights reserved 21
  20. 20. Funding benefit & funding cost  Classical pricing models assume that we can borrow/lend at a risk-free rate.  Post crisis, financial institution fund with significant spreads. – This credit/liquidity component appears in LIBOR basis spreads (OIS/LIBOR and LIBOR of different tenors) OIS vs. LIBOR 3M spread Copyright ® 2012 Murex S.A.S. All rights reserved
  21. 21. Funding un-collateralized trades  In any derivatives contract future cash-flow exchanges need to be “funded”.  A bilateral position with an open negative MtM can be seen as an overnight loan granted by the counterparty : logically this funding benefit is financed at our cost of funds.  A positive MtM represents a funding cost : by unwinding the trade and investing this amount with my treasury (or buying back my own bond issue) I could get the same rate.  Hence an uncollateralized transaction’s Cash Flows should be discounted at my senior unsecured cost of debt.  Neglecting the CDS-Bond basis, my senior unsecured cost of funds is in line with the assumptions PDs and Recovery of the CVA calculation. Hence at a single contract level (i.e. single deal or netting set): DVA = Funding Benefit. Copyright ® 2012 Murex S.A.S. All rights reserved 23
  22. 22. Funding cost: the (non-)effect of netting  2 parties A & B have two exactly offsetting trades but no netting agreements between them: – Both parties will have non-zero CVA & DVA terms (and bilateral CVA) – They both have 0 funding cost as CFs will offset.  In practice whether a set of transactions is covered by a close-out netting provision (ISDA) or not, has no implication on their funding cost (and thus the discounting curve to be used)  For non-fully netted portfolios the Funding Benefit is not equal to DVA! – No close-out netting agreement – Multiple netting sets Copyright ® 2012 Murex S.A.S. All rights reserved 24
  23. 23. Funding collateralized trades  If a CSA is in place, the “lender” typically receives a collateral for a value ~ equal to the MtM of the position, either as: – a Cash amount, which can be re-invested (overnight) and on which a prespecified interest is paid back to the poster (typically compounded OIS index). – a Security. If the CSA agreement allows for re-hypothecation, that collateral can be repo-ed to another party to fund at a much lower rate than an unsecured funding rate. – Simplifying assumptions: 0 Thresholds & MTAs, daily remargining, one currency, no haircuts on securities, no dispute…  Hence CSA-covered positions can be funded by using an OIS discount curve  Non CSA-covered positions are funded using the internal cost of funds (senior unsecured debt) Copyright ® 2012 Murex S.A.S. All rights reserved 25
  24. 24. The Ideal CSA Hypotheses  Bilateral Agreement  Continuous Margining  Instantaneous settlement of margin calls  0 Threshold and Minimum Transfer Amount  No Independent Amounts  No haircuts  Cash (or equivalent instrument) collateral, independent from exposure  No valuation differences  No disputes  Netting set = Margining set  No Initial Margin  CCR : – – No posting of Initial Margin with risky entities –  No rehypothecation / segregation of collateral accounts No settlement risk on margin flows Funding : – Rehypothecation / no segregation of collateral accounts – No posting of Initial Margin – Single risk-free collateral asset (e.g. no currency basis arbitrage) Copyright ® 2012 Murex S.A.S. All rights reserved 26
  25. 25. New FO and Risk systems needs  Front-Office systems require flexible curve allocation mechanisms: – –  Collateral documentation is pricing data! Rule-based dynamic allocation of curves based on both the leg currency and underlying collateral currency Proper allocation of risk and sensitivities –  E.g. Uncollateralised CMS swap (CMS rate derived from collateralised instruments) Need a multiple curve calibration engine: – – Wider selection of curve building instruments – Copyright ® 2012 Murex S.A.S. All rights reserved Able to detect the dependencies Simultaneous bootstrapping of all involved curves with accuracy and speed. 27
  26. 26. Pricing example Uncollateralized USD CSA  In-the-money XCCY swap EUR/USD with 5Y outstanding maturity  P&L impact of 36bp  Forward MtM, vs. Expected Exposure & Expected Liability evolution. Copyright ® 2012 Murex S.A.S. All rights reserved 28
  27. 27. Pricing in a Multiple Curve Environment  Forwarding curves are derived from collateralized quotes – – E.g. EONIA and EURIBOR 3M, then EURIBOR 6M vs. 3M… –  Joint bootstrapping of discounting and forwarding curves Triangular calibration with XCCY basis curves or markets with varying liquid swap tenors depending on the horizon. Different discounting curves depending on the CSA clauses. – EURIBOR swap collateralized in EUR is discounted on an EONIA curve – EURIBOR swap collateralized in USD is discounted on a EUR/USD XCCY basis curve built upon a USD Feds Funds curve. Copyright ® 2012 Murex S.A.S. All rights reserved 29
  28. 28. Pricing in a Multiple Curve Environment  Pricing of Exotics requires multi-curve evolutions for pricing of exotics and CVA estimation. – Current standard market practice: deterministic basis spreads curves on top of a risk-free OIS curve – Currently testing a HJM 2F stochastic basis spread model, calibrated to historical data (results to be presented soon).  Another difficult question pertains to correlations (OIS-LIBOR spread vs. rates, bond-CDS basis vs. LIBOR basis and credit…) Copyright ® 2012 Murex S.A.S. All rights reserved 30
  29. 29. Limitations of the deterministic spreads view?  Even plain vanilla swaps require several curves to be priced  Using deterministic spreads obviously implies a perfect correlation between various curves used in the model.  Is such a constraint acceptable, in light of recent market changes ? Copyright ® 2012 Murex S.A.S. All rights reserved 31
  30. 30. Limitations of the deterministic spreads view?  Empirical tests on recent data showed that: – Deterministic spreads are a decent approximation when pricing collateralized callable fix-float swaps (reassuring for CVA) – However for basis products both the spread volatility correlation between rate curves have a significant impact on the valuation Copyright ® 2012 Murex S.A.S. All rights reserved 32
  31. 31. Main issue: the CSA is not perfect  When exposure is in-between thresholds, we fund at LIBOR + spread and not at OIS flat  Non-cash asset: haircuts and rehypothecation rights?  Choice of collateral currency: – E.g. steep XCCY basis spreads with the 2011 EOY USD squeeze – Apparently comparable to a contingent Bermudan XCCY swaption on the portfolio (hint at American Monte Carlo pricing) – In practice varying implementation approaches – However, uncertain execution / enforceability • Different legal interpretations (US vs. UK law – do we require the consent of the receiving party? Is full substitution always possible when there is no margin call?...) • Will the collateral management team deliver the adequate collateral? – Will the issue disappear with the Standard CSA?  Does the local market even have a liquid OIS instrument?  One-way CSA case is another tricky case of funding asymmetry (one threshold pushed to infinity) Copyright ® 2012 Murex S.A.S. All rights reserved 33
  32. 32. One-Way Collateral Agreements Part II  Funding cost at OIS flat  Funding benefit at unsecured debt level  Double hit: CVA & FVA Usually SSAs will have much lower credit spreads than the institution so the Funding risk effect would dominate the Credit risk one.  Difficult to value in the simple discount switch setting, however actual quoted price is unlikely to be the “fair-one”. Borrow at LIBOR + spread Receive funding at OIS flat Copyright ® 2012 Murex S.A.S. All rights reserved 34
  33. 33. Introducing the Standard CSA  New collateral support annex protocol promoted by ISDA  Aim to standardize valuation practices – Specify OIS discounting – Remove the collateral switch optionality – Align CSA to the margining mechanics of CCPs  0 Threshold, no MTA, daily margining  Cash collateral only for variation margin  Phased implementation in 2012 : transactions can be moved from legacy CSA to S-CSA  Transactions pooled in 5 Designated Collateral Currency buckets Copyright ® 2012 Murex S.A.S. All rights reserved 35
  34. 34. Introducing the Standard CSA Phase I Discounting on USD Feds Funds & corresponding FX basis curves Local currency OIS discounting (EONIA, SONIA…) CHF trades EUR trades GBP trades CHF collateral balance EUR collateral balance GBP collateral balance JPY trades JPY collateral balance CCS and other currencies USD collateral balance Margin Calls / Deliveries Counterparty Herstatt Risk! Copyright ® 2012 Murex S.A.S. All rights reserved 36
  35. 35. Introducing the Standard CSA Phase II CHF trades EUR trades GBP trades CHF collateral balance EUR collateral balance GBP collateral balance JPY trades JPY collateral balance CCS and other currencies USD collateral balance Margin Calls / Deliveries Safe settlement: PvP platform operated by ISDA Swap margins to USD (ISA method) Counterparty Copyright ® 2012 Murex S.A.S. All rights reserved 37
  36. 36. Moving to S-CSA: system implication   Straight-forward adaptation of the CVA Monte Carlo Engine thanks to dynamic construction of the netting and margining sets (rule-based) In practice, need to follow closely migration of trade blocks (by products, entities) from legacy CSA to SCSA margining. Counterparties Netting Nodes Margining Nodes CSA – EUR ISDA - ABC … CSA – USD* Bank ABC Legacy CSA … No collateral Copyright ® 2012 Murex S.A.S. All rights reserved 38
  37. 37. S-CSA implementation challenges ISDA : “Regardless of approach, firms will need to undertake considerable internal technology and process re-engineering work to implement the SCSA.”  Collateral systems impact: – Electronic messaging – Exposure pooling and collateral accounts by currency buckets & flexible mechanism to migrate positions off legacy CSA – Mandatory OIS discounting – Implementation of ISA & PvP processes  Front-office: – Availability of collateral eligibility criteria at point of pricing – Discount curve allocation mechanism based on CSA / SCSA mappings – For a period of time maintain local OIS curves and Basis OIS curves  CVA / FVA units – Consistent mapping of the positions to currency buckets – Multiple margining sets per netting set – Value margin conversion via ISA-type method and capture FX risk over MPR Copyright ® 2012 Murex S.A.S. All rights reserved 39
  38. 38. FVA - Collateral Modeling  An alternative approach to the Discount Method consists in looking at the question from a portfolio level by representing the funding cost as another valuation adjustment (the OIS curve providing a proxy for the risk-free rate).  Evolve market rates and explicitly model the collateral balances and a funding strategy. E.g. – –  Collateral balance : funded at OIS flat Portfolio value – balance : shortfall funded at own cost of funds Extend existing CVA simulation framework since this will provide: – A consistent pricing framework for CVA and FVA (calibration, deal aging and termination events) – The CVA engine already has all required business logic (margining set mapping, curve and spreads evolution) – A validated & controlled infrastructure : inter-system data flows, interfaces, reconciliation processes – A low & managed TCO, as one can leverage existing infrastructure (e.g. grid, GPU farm) : running FVA calculations on top of a CVA simulation is computationally efficient (provided i. consistent modeling assumptions and ii. that collateralised positions are already included) Copyright ® 2012 Murex S.A.S. All rights reserved 40
  39. 39. FVA - Collateral Modeling  Rates curves are evolved jointly Counterparties    Netting Nodes Collateral Balances are obtained at the margining node level Collateral assets are funded at the Agreement’s specified rate source Margining Nodes CSA – EUR ISDA - ABC … CSA – USD* Bank ABC Legacy CSA … Collateral shortfalls funded on funding curve No collateral Copyright ® 2012 Murex S.A.S. All rights reserved 41
  40. 40. FVA - Collateral Modeling  Practical simulation implementation : DVA is not the FVA benefit (MPR vs. Settlement lag). Margin Period of Risk Settlement Lag Collateral Funding Collateral Balance (FVA) Collateral Balance (CVA) Simulation date Ti-1 Ti - MPR Ti - SL Copyright ® 2012 Murex S.A.S. All rights reserved Simulation date Ti 42
  41. 41. FVA - Collateral Modeling  Reducing the MPR (10 days) to the Settlement lag (3 days) halves the DVA estimate.  Final FVA impact would be stronger on portfolios with imbalanced EPE/ENE profiles or asymmetric collateral terms (thresholds, IAs, oneway CSA). Copyright ® 2012 Murex S.A.S. All rights reserved 43
  42. 42. Managing consistently Funding & Credit Credit Funding Collateral  Stating the obvious: need consistency between valuation practices & business models  Collateral optimisation, transformation  Funding vs. liquidity  Hedgeable costs vs. friction and cost of doing business Copyright ® 2012 Murex S.A.S. All rights reserved 45
  43. 43. Managing consistently Funding & Credit  Discount curves approach vs. Global revaluation model  Asymmetric or symmetric FVA expression: – As a funding cost Risky Value = Risk-free Value – CVA + DVA – FCA* – As a funding adjustment Risky Value = Risk-free Value – CVA + FVA  System implementation – The same simulation can provide DVA and FVA broken down in FCA & FBA – DVA can be computed as mandated by FAS157 (topic 820) / IFRS 13 – FVA can be outsourced to and hedged by a dedicated CFU – CVA desk can hedge unilateral CVA or unilateral CVA + DVA-FBA basis – FO incentivization through marginal pricing  Basic modeling and implementation questions: – Credit Liability does not have to coincide with the amount to fund (nonnetted trades, netting set fragmentation) – Margin Period of Risk adjustment vs. Settlement Lag – CDS – Bond basis spread Copyright ® 2012 Murex S.A.S. All rights reserved 46
  44. 44. Modeling framework vs. Mandates  Model choices, system parameterization should be made with their users mandates in mind – Risk management vs. CVA desk vs. Funding – Minimizing accounting P&L volatility, JTD risk, Liquidity risk, Funding costs, and/or capital requirements – Incentivizing risk-takers – Active hedging (of which components?)  Some obvious impacts – Choices of metrics, data inputs, adjustments – WWR, basis and cross-gammas – P&L management tools – Fees management, data workflows, systems interoperability Copyright ® 2012 Murex S.A.S. All rights reserved 47
  45. 45. What about FVA for CCP-cleared products?  Margin requirement broadly split in IM and VM  IM typically large and aimed at covering gap risk over the auctioning period (so as to preserve default funds contributions)  IM models are typically VaR-based (adjusted with credit and liquidity factors)  The IM funding requirement will then depend on the “directionality” of the cleared portfolio!  Should this additional cost be modeled on an incremental basis (consistent with CVA and OTC FVA), or handled as a post trade operational cost? Incentives may differ depending on the institution.  Extending the CVA/FVA model to provide estimation of forward Initial Margin requirements would require a forward approximation of the margining sets VaR. Computationally, the issue is similar to the estimation of the incremental RWA cost of capital. Copyright ® 2012 Murex S.A.S. All rights reserved 48
  46. 46. Variation Margin and Initial Margin  FVA for cleared products should, in theory, account for the incremental cost of funding of the Initial Margin 5 days to close the auctioning process High C.L. VaR Initial Margin Variation Margin Position at Ti-1 Position at T Copyright ® 2012 Murex S.A.S. All rights reserved 49
  47. 47. Questions and practical issues  Vanillas are de-facto level-2 derivatives and market prices are not transparent (difficulty to unwind off-market positions)  Broker quotes need to be reinterpreted (e.g. B&S vols)  New premium quotation modes (unfortunately not applicable for all types of options)  Sensitivities and hedge ratios differ between collateralized and uncollateralized cases  Perfect hedge can only be achieved under identical collateralization terms  Pricing effects are complex to quantify for imperfect collateralization cases and embedded optionalities  Difficult /costly hedging of basis risks  Convexity and wrong-way effects deemed small (valuation impact smaller than bid-ask) but traders need to be aware of them  Which CSA clauses should be modeled / can be hedged ? Copyright ® 2012 Murex S.A.S. All rights reserved 50
  48. 48. Questions and practical issues Internal organization challenges:  Need to implement consistent pricing of new transactions, unwinds and legacy books  Fair pricing of internal positions  Ownership of the funding issue and hedging  Ensure that the Collateral Management & Treasury functions provide optimal funding (as supposed in the pricing)  Integration of data flows and inter-operability : both a processes and systems challenge !  Establish clear-cut transfer pricing and cost management policies Copyright ® 2012 Murex S.A.S. All rights reserved 51
  49. 49. Two modeling and organizational models     Discount curves method Simpler to implement in a crude way, additional complexity with curves management and FO assignments Trade level pricing compatible with local models.   Requires significant investments (starting with a simulation framework)  Global hybrid pricing consistent across desks and with CVA.  Flexible handling of CSA agreements and explicit modeling of the funding strategy Fail to account for corner cases – asymmetric funding terms – convexity effects (e.g. spread / rates correlation) – Global FVA/CVA exposure method – liquidation value at risky value Reproduces the previous method results under specific case – Can include funding impact of credit mitigants  Non-explicit link with DVA  Isolates clearly funding cost from valuation and CVA  Deal-level and easily understood by traders   Funding and convexity risk owned by the traders Portfolio-level, cost reallocated to the trades (like CVA)  Works best with smaller decentralized operations well collateralized Funding and convexity risk transferred to a centralized Funding / Treasury desk  Works best when bringing together Treasury, CVA and Collateral trading operations  Open question : what should be the regulatory treatment of the FVA market risk in the second setting? FVA VaR integrated in the IMA model? Copyright ® 2012 Murex S.A.S. All rights reserved 52
  50. 50. Some useful references Collateralization & Counterparty Risk:  D. Brigo & A. Pallavicini (2011) – Arbitrage-Free Counterparty Risk Valuation under Collateral Margining  D. Brigo (2011) – Counterparty Risk FAQ: Credit VaR, PFE, CVA, DVA, Closeout, Netting, Collateral, Re-Hypothecation, WWR, Basel, Funding, CCDS and Margin Lending.  J. Gregory (2009) – Being two-faced over counterparty risk  J. Hull & A. White (2011) – CVA and Wrong Way Risk.  ISDA (2011) – Overview of ISDA Standard Credit Support Annex (SCSA).  M. Pykhtin (2010) – Collateralised credit exposure, in Counterparty Credit Risk, edited by E. Canabarro, Risk Books.  M. Pykhtin & D. Rosen (2010) – Pricing Counterparty Risk at the Trade Level and CVA Allocations. Books:  G. Cesari & al. – Modelling, Pricing, and Hedging Counterparty Credit Exposure.  J. Gregory – Counterparty credit risk – The new challenge for global financial markets. Wiley Finance. Copyright ® 2012 Murex S.A.S. All rights reserved 53
  51. 51. Some useful references Collateralization & Funding:  C. Burgard, M. Kjaer (2011) – In the Balance.  C. Fries (2010) – Discounting Revisited: Valuation Under, Funding, Counterparty Risk and Collateralisation.  M. Fuji, Y. Shimada & A. Takahashi (2010) – Collateral Posting and Choice of Collateral Currency.  A. Green (2011) – Engineering a CVA and FVA solution, talk given at the WBS Discounting and Funding conference, November.  D. Loiseau (2012) – Introducing Stochastic Spreads in a Multi-Curves Framework, Murex talk given at the MathFinance Conference, March.  M. Morini & A. Prampolini (2010) – Risky funding: a unified framework for counterparty and liquidity charges.  V. Piterbarg (2010) – Funding beyond discounting: collateral agreements and derivatives pricing, Risk Magazine, February issue.  Risk Magazine (2011) – The evolution of swap pricing. Nick Sawyer, March issue.  M. Singh & J. Aitken (2010) – The (sizable) Role of Rehypothecation in the Shadow Banking System. Copyright ® 2012 Murex S.A.S. All rights reserved 54
  52. 52. THANK YOU 55