The xVA prism:
Current challenges and best
practices
6 February 2014
Agenda
1. Introduction to Numerix
2. Overview of the regulatory landscape & current market trends
3. The new margining regime: Accounting for collateral under an
unified decision measure
4. Valuation Adjustments and replication
Section I
Numerix Overview
Market leader in Advanced analytics for structuring/managing all derivatives and structured products and risk management and compliance
200+ employees globally
Integrated in 3rd party solutions (trading, risk and operations platforms)
Some of our clients Strong resource pool
Global coverage, with support across all major languages
Phds 40%
MBA/Msc 30% CFA 20%
Section II
The new regulatory framework for counterparty risk
Basel 3
Capital requirement for OTC bilateral exposures and exposures to CCPs (from January 2014)
IFRS 13
Accounting rules for CVA/DVA – since January 2013All standardized OTC derivatives should be cleared through CCPs OTC derivative contracts should be reported to trade repository
Exemptions: sovereign and corporates (thresholds, hedging) – III/IVQ 2013
Mandatory Clearing for eligible OTC Derivatives – June 2014
Margin requirement for non centrally cleared OTC – from Dec 2015
EMIR
Solvency
2
Solvency incentivizes high quality collateral selection – From January 2016
BCBS/IOSCO
Mandatory initial margin and variation margin for bilateral OTC transactions Segregation and limits to rehypotecation of the amount of collateral that• The new regulatory requirements have accelerated the development and re-definement of valuation and risk methodologies at institutions across the globe with a focus on valuation transparency, margin assessment and risk fortification.
Regulatory-Compliance
Margining & Valuation
Model Validation
Total Value Adjusment
Data & systems
Current market trends
• « New » regime based on Initial/Variation Margin and Collateral Posting • Funding inclusion into pricing aka funding monetization
• Collateral Choice in pricing and optionality (CTD) • Multi-Curve Pricing
• Confidence in the appropriateness of the model (local adaptation) • Transparency for model and parameters
• Confidence that the entity has a good understanding of assumptions, risk drivers and results
• Stress Testing
• Understanding Pricing/Risk decomposition : CVA, DVA, FVA, RVA,… • Risk Integration : market+credit+liquidity in a consistent unified
framework
• Term Structure of eg Multi year TVA
• Continued investment in data and systems, increasing
computation speed and embedding within front office tools and P&L
Longdated “Unclearables”
Source: ISDA Margin Survey 2013
Source: Tabb Group
Growth in value of collateral agreements for non-cleared OTC derivatives
Notional amount outstanding has grown over the last decade. Aggregate counterparty exposure in terms of net MtM value considering close out netting before collateral has decreased from 3,9 to 3,7 billions Usd from 2011 to 2012 (BIS statistics); collateral value reported has increased, suggesting more exposures being
Collateral Transformation & Trends
• Mostly handled in the back office
• Low necessity of margin calls (high thresholds and min transfer • amounts
•Few derivatives were collateralized
Before 2008
• Lack of transparency in Structured Credit Notes • Double Defaults
tendency
•One rate regime
• Regulatory Haircuts
2008- 2012
• Increased usage of collateral as a way to mitigate risk of counterparty default •Regulation requirements – CCPs –
Margin calls
• Divergence of rates after the crisis and complex CSAs
• Collateral shortage (high demand, rehypothecation)
• Equally important for sell-side and buy-side
• Choice of collateral significantly affects derivative pricing
•Push from traders and front-office – profit is the drive
Section III
The new margining regime: Accounting for collateral under an unified
decision measure
Initial & Variation Margins
• The initial margin is supposed to cover quality of collateral, gap risk, wrong way risk. By design the IM needs to cover the closing out of positions (without the loss) to a CCP in a worst case
scenario.
• Initial margin is an “haircut” (overcollaterisation) which is calculated at inception of the trade. The margin can fluctuate during the liftetime of the trade, given market conditions and remaining risk.
• As its colour varies, several methodologies exist, two of them are proposed for cleared and uncleared positions.
– Cleared position :It is calculated as an historical VaR
capturing a cycle (5-10 years) observation, on shifts of underlying market factors.
– Uncleared position : It is calculated as a Potential Future
Exposure (99% quantile) at 10 day horizon over a stressed market period.
• The Variation margin covers the change in the valuation of the relevant positions as with the collateral in a standard CSA.
Pricing & Ageing
Hybrid Model
Historical VaR (cleared) or PFE (uncleared)
Haircut Optionality
MTA Thresholds Collateral Data Initial Margin Variation Margin FVA
CVA , DVA , RVA, CollVA
Collateral Data
Counterparty Risk Platform
CSA-Netting agreements/ Path dependent Collateralized
Historical
VaR 10 day PFE
Risk Mitigation Strategies
IR (S)BK2F EQ Heston FX Heston IR HW1F IR (S)BK1F IR HW2F IR SV-LMM INF JY
(HW) INF JY (BK) EQ BS EQ Dupire EQ Bates EQ LSV FX BS CMDTY Black CDMTY S1F CMDTY GS2F CMDTY Heston HYBRID MODEL -100.00 -50.00 50.00 100.00 150.00 200.00 250.00 300.00
4/1/2012 8/14/201312/27/20145/10/2016 9/22/2017 2/4/2019 6/18/2020 PFE w Collateral PFE w/o Collateral
Overall Framework
EVA / Limits
PnL Variations
13
Credit Value
Adjustment
Debit Value
Adjustment
Replacement
Value Adjustment
Margining Value
Adjustment
Funding Value
Adjustment
Risk Cocktail :Valuation Adjustments
The cocktails ingredients
• The final commercial margin include the following ingredients: • CVA (Credit Value Adjustment) :Adjustment to the risk
free value made by one agent to take into account the potential default of a counterparty defaulting first.
• DVA (Debit Value Adjustment): Adjustment to the risk free value made by the agent to take into account the his potential default first than the counterparty. It also incorporates a funding angle, as bank’s default risk increases, the value of liabilities decreases.
• FVA (Funding Value Adjustment)/LVA: Reflects the cost an institution incurs when hedging an uncollateralized trade with an offsetting position on which collateral is required.
• RVA (Replacement Value Adjustment): reflects the “trigger” cost that replaces a credit quality downgrade of the counterparty with another trade from another counterparty.
• MVA (Margin Value Adjustment): reflects the cost associated with Initial and Variation Margins
• Cost of Capital : Valuation Adjustment for Regulatory Capital
Section IV
Origins of xVA
• In the modern world:
– existence of multiple rates corresponding to di
ff
erent possibilities to borrow/lend
money
• rate r
xcollateral rate (almost risk-free rate)
• rate r
rfor asset secured borrowing (“repo”)
• rate r
ffor unsecured funding
• Possibility of default and migration
the classical arbitrage-free theory should be modified
• We should modify the replication/hedge arguments to include multiple rates and
defaults
•
Definition
.
The TVA (Total Valuation Adjustment)
is the di
ff
erence between the true
(modern) price and the base one. The base price is often related with fully
Financial instrument replication (hedging) is a unique way to calculate its fair
price or adjustments xVA (CVA, DVA, FVA etc.)
Different replications lead to di
ff
erent adjustments. There is a big variety of the
adjustments (CVA, DVA, FVA, FCA etc.). Thus, it is important to identify them for
given replication techniques to avoid double-counting.
A portfolio replication (bonds) strategy is designed and used in the context of
hedging default risks for both self and counterparty.
Hedge (replication) against di
ff
erent market movements and defaults:
Credit spread and default of the counterparty and the bank Funding costs of the counterparty and the bank
Assets movements
Collateral rate movements
Piterbarg
(2010)
Borrow/lend money via the bank treasury desk with a unique “funding” rate
The bank default is already taken into account by the treasury "integrated"
rate
Burgard-Kjaer
(2013)
Borrow/lend money with external (w.r.t. the bank) parties: different rates to
borrow and lend.
The rates are related with the bank credit spread and expected recovery.
Multiple replication strategies
Double-counting counterbalancing effects
Banks are required to hold a DVA adjustment for their own credit risk on
uncollateralized trades
Funding spreads are a function of bank credit quality: by implementing FVA you
are effectively subsuming DVA on payables and accounting for funding on
receivables
Knowledge of the replication/hedging strategy:
gives the unique pricing equation and specifies the TVA
permits to identify parts of the adjustments with xVA’s
prevents from the double-counting
• if we calculate the FVA coming from Piterbarg replication, it will be double-counting to take the DVA into account;
Implementation
1. Simulate the model rates and all payment indexes
2. Build the single-rate pricing model equipped with Least Square Monte Carlo
3. Calculate future values for all instruments in portfolio on this model (can be
done independently ”instrument-by-instrument” using the Algorithmic Exposure
methods (AIM (2011))– important for parallel computation)
4. Aggregate the instrument future prices into the portfolio ones
5. Calculate the xVA from the obtained future values using universal approximation
formula (ABM (2013)) and its scripted version available in Numerix
Script example
1. Alexandre Antonov; Numerix FVA for General Instruments: Theory and Practice
PRODUCTS
NONDISCOUNTING SpreadIntegral, FVA NONDISCOUNTING CollateralUnits
TEMPORARY dt, RHS, EffectiveRate, EffectiveSpread, Delta TEMPORARY Collateral, Collateral0, V0, RHS0
END PRODUCTS
PAYOFFSCRIPT
IFISACTIVE(MarginCallDates) THEN
Collateral = CollateralUnits * CollateralAssetValue Delta = MAX(V - HighT, 0) - MAX(- V - LowT, 0) - Collateral Delta = WHEN((Delta > 0) AND(Delta < CTPYMTA), 0, Delta) Delta = WHEN((Delta < 0) AND(Delta > - SELFMTA), 0, Delta) Collateral += Delta
CollateralUnits = Collateral / CollateralAssetValue ENDIF
IFISACTIVE(ObservationDates) THEN
// Calculate true collateral amount
Collateral = CollateralUnits * CollateralAssetValue
// Calculate the true RHS (based on true collateral amount)
RHS = Collateral * r_C + (V - Collateral) * r_F
// Regularization procedure for effective rate calculation:
// Calculate regularized price floored/capped to +-Tolerance
// when V close to zero
V0=WHEN(V>=0, MAX(V, Tolerance), MIN(V, - Tolerance))
// Calculate regularized collateral (ignoring margin call dates and MTA)
// and corresponding regularized RHS
Collateral0 = MAX(V0 - HighT, 0) - MAX(- V0 - LowT, 0) RHS0 = Collateral0 * r_C + (V0 - Collateral0) * r_F
// Calculate regularized effective rate protected from division by zero
// The approximation error is small (higher order than our accuracy)
EffectiveRate = RHS0 / V0
EffectiveSpread = EffectiveRate - r_M
dt = ObservationDatesDCF
SpreadIntegral += EffectiveSpread * dt
// Aggregate the FVA
FVA -=(RHS - r_M * V) * exp(- SpreadIntegral) * DF * dt ENDIF
Section IV
Conclusions and potential extensions
Funding and Credit are untimely, implicitly linked and their interplay causes
(some) counterbalancing effects:
The Initial Margin “wave” for both cleared and non cleared instruments imply higher capital
requirements, with higher funding necessity.
FVA & DVA monetization dynamics, where DVA, generates funding benefits in asymmetric
situations.