FVA - Putting funding into the equation 3rd Marcus Evans Derivatives Funding Valuation conference September 2013

Contents I Background II FVA Calculation III Operating Model

FVA – Funding Valuation Adjustment Background What is FVA? FVA – Funding Valuation Adjustment Funding-Spreads widened Consideration of funding cost in pricing and valuation is highly relevant since the crisis in 2007/2008 The introduction of OIS Discounting was a first step to consider funding cost in valuation for the collateralized derivatives business FVA is the next step on our way to fully consider funding cost in pricing and valuation. Basically it deals with the consideration of funding cost of uncollateralized derivatives. In the market practitioners currently develop conceptual FVA calculation approaches and first start implementation New concepts for valuation

Background FVA in the press Hull and White debate Goldman story Regulation Banks at risk of FVA arbitrage Unfair value: FVA's hidden charms Goldman and the OIS gold rush: How fortunes were made from a discounting change

Background Discounting Collateralized Derivatives Fixed at 5% v market level of 2% Bank A Bank B PV: €+10m IRS PV €-10m 3M LIBOR PV: €-10m Cash Collateral PV: €+10m EONIA on Collateral Assume that for counterparty A the PV(A) is positive (e.g. €+10m). The collateral account holds the total PV of the swap (bank A: €+10m, bank B: €-10m ) The collateral amount is available to bank A. On return, bank B receives the overnight rate on the collateral amount from bank A. Consequently, the collateral mechanism can be interpreted as a funding mechanism at overnight rate, transferring liquidity from bank B to the bank A. Notice that as PV(A) is positive, (under assumptions of the example) bank A has paid a net cash amount to bank B since the inception of the swap. This suggests that bank A has in fact a funding demand. By no-arbitrage arguments, the CSA collateral rate and the discounting rate of future cash flows must match (hard mathematics!) Discounting of collateralized derivatives with the collateral rate (OIS)

Background Discounting Uncollateralized Derivatives An uncollateralized swap has no implicit collateral funding mechanism The funding cost of an uncollateralized swap are therefore the bank’s individual funding cost (e.g. LIBOR + 100 bp) The price of an uncollateralized deal is related to the cost of funding There is currently no clear market view on valuation methodology and/or how the cost of funding is actually determined. Most practitioners agree that own funding cost need to be considered. Risk of double counting effects between CVA, DVA and own funding cost / FVA Fixed at 5% v market level of 2% Bank A Client IRS / client derivative PV: €+10m PV: € -10m 3M LIBOR 3M LIBOR + 50 bp Notional (€ 10m) Funding Capital Markets / Investors

Mirroring Client derivative? Background Funding effect from derivative transactions cannot be estimated by collateral position Typical situation: Uncollateralized client trade, hedged back-to-back with a CSA counterparty – one-time change in market interest rates PV of uncollateralized client derivate assumed to have become large and positive (from bank’s perspective) No collateral received, no influence on cash position Future positive net product cash flows will be received from client PV of hedge derivative is negative Collateral amount equal to hedge PV has been transferred to hedge counterparty. Negative cash position Collateral earns OIS (EONIA) interest All future product cash flows will be neutralised by collateral paid back (and vice-versa) Net effect: positive PV needs to be funded until realization by uncollateralized product cash flows Client Uncollateralized Derivative Client Business Bank BU/Desk Treasury Funding Collateralized Derivative Hedge trade Mirroring Client derivative? Hedge Counterparty Important: The hedge in general does not involve same fixed rate coupons (re-couponing macro hedge) This means the collateral position in general is not useful as an estimator of FVA - only uncollateralized trades must be considered

Background Advancements in derivates valuation: CVA, DVA, Funding Since 2000: Consideration of counterparty credit risk in derivatives valuation Consideration via CVA (Counterparty Credit Valuation Adjustment) CVA is market price of counterparty credit risk / cost of hedging Derivates valuation with Counterparty VA Derivates valuation without Counterparty VA CVA = + Introduction of bilateral CVA Consideration of market price of own default risk via DVA (Debit Value Adjustment) Derivates valuation with Counterparty VA Derivates valuation without Counterparty VA CVA DVA = + + Since 2007/2008: Rise in bank funding cost and introduction of funding in valuation Consideration of funding cost/benefit in derivatives valuation applying a FVA (Funding Valuation Adjustment) Derivates valuation with Counterparty VA and Funding VA Derivates valuation without Counterparty VA and Funding VA CVA DVA FVA = + + + Double Counting Issue

Partially collateralized Background Landscape of collateralized and uncollateralized derivatives business Fully collateralized Partially collateralized Uncollateralized Assumption of “perfect” CSA, i.e. bilateral CSA, no thresholds Future product cash flows are fully funded / invested by the collateral received / posted External interest on collateral defined by CSA-rate, e.g. EONIA Unilateral CSAs, thresholds and/or minimum transfer amounts Product and collateral cash flows do not match No CSA Future product cash flows need to be funded / invested Market practice for valuation Closed formula: discounting on CSA/OIS- curve already implemented by most institutions No generally agreed market practice yet Heated discussion between practitioners and in academia Focus of this talk Open issues How to handle initial margins / independent amounts and/or a liquidity reserve held to cover future market value or resp. collateral volatility? Open issues Consistent internal pricing framework: Link to accounting, Link to funds transfer pricing FVA, CVA and DVA - how to avoid double counting? FVA is relevant for all kinds of derivatives

Contents I Background II FVA Calculation III Operating Model

Putting funding into the equation: KPMG’s view on FVA General principles for FVA calculation Proposition 1 Proposition 2 Proposition 3 Proposition 4 Proposition 5 Valuation models need to take into account all cash flows associated with a derivative - including collateral flows and funding cost and benefits Assumptions concerning effective life times of uncollateralised derivatives positions will influence FVA From a bank management perspective a close alignment between the rules for FVA calculation and the bank’s funds transfer pricing rules would be desirable Despite the bank- specific nature of FVA calculated using a bank’s own funding curve this can often still be justified as a reasonable estimate of funding cost and benefits in IFRS fair value calculations In general FVA needs to be determined on a portfolio level with subsequent allocation to individual trades Differences between FVA and DVA and their consolidation (Active) management and governance Proposition 6 Proposition 7 Proposition 8 Proposition 9 FVA has potential overlaps with both CVA and DVA that need to be taken care of in a consistent pricing/valuation framework One way to avoid double counting between FVA on the one hand side and CVA/DVA on the other is to take one as the increment to the other. Whilst DVA is an accounting requirement it should not be considered in managing the business There are advantages to managing FVA centrally, and close alignment to the CVA desk can be efficient

Transparency in funding cost will lead to re-couponing Proposed FVA framework (proposition 2) Close alignment between funds transfer pricing and FVA means application of a real funding curve in a two way market Central counterparty Standard CSA Restructuring Transparency in funding cost will lead to re-couponing Decrease in uncollateralized trading volume Re-negotiation of non-standard CSAs already ongoing

Proposed FVA framework (proposition 3) Close alignment between funds transfer pricing and FVA means application of a real funding curve in a two way market Consider liquidity cost and benefit It is not only fair but necessary to charge for liquidity need and to compensate liquidity benefits. As valuation and pricing should be consistent with internal incentives cost and benefit must both be part of FVA Price cost and benefit on bank’s own (internal) funding curve For risk management reasons and for internal transfer reasons the bank‘s own funding curve (which is in general not the bank‘s senior unscured curve) must be applied Only under application of the bank‘s own funding curve an internal transfer of liquidity term risk towards treasury is fairly priced internally Divide between term funding and liquidity buffer component As the term funding component should be transferred to treasury to allow netting with other positions in the bank and to enable central management the buffer cost are a different „beast“

Funding/ Investment of derivative and collateral CFs Proposed FVA framework (proposition 3) Bank management profits from consistency between FVA and internal liquidity transfer pricing (FTP) Core elements of FTP Implications for FVA Liquidity is managed by a restricted set of capital markets transactions (term funding, asset investment, ...) Internally liquidity is transferred between businesses at one price (FTP) Any liquidity (spread) risk is transferred to a central Treasury function To avoid arbitrage, liquidity from trading activities need to be priced similar to other business areas (e.g. credit business) Therefore, the FVA calculation needs to be aligned with concepts of FTP This leads to a number of benefits, e.g.: Avoiding the large bid/ask spread currently observed in the capital market by netting on bank or even group level No arbitrage between different businesses Clear rules on handling liquidity in foreign currencies Clear rules on allocation of the cost of liquidity buffers Basis for FVA Business Unit Treasury Term funding Markets Funding/ Investment of derivative and collateral CFs Investment Liquidity transfer Other BUs FTP FTP

Proposed FVA framework (proposition 5) Generally FVA is calculated on portfolio level and not on single transaction level Portfolio-view Calculation of the expected or relatively sure liquidity position to determine the term funding component Calculation of unsure liquidity streams for liquidity risk control, if neccessary buffer Consideration of portfolio effects as, for example, non-standard-CSAs, e.g. Onesided CSAs, high minimum transfer amounts and threshold amounts Partial collateralization of the portfolio with assets of different quality Modelling of expected product life time taking into account future restructurings, break clauses and trends as for example central counterparty trading FVA is a valuation component with significant risks of higher order (cross gamma), analytics and risk control similar to CVA Distinction between deterministic and volatile funding effects Consideration of CSAs Assumptions on the life time of the portfolio or product Cash Flow / Liquidity profile + Risks of higher order (e.g. cross gamma) Contractual CF   Modelled CF time -

Proposed FVA framework (Proposition 6) Double counting of FVA and DVA is somewhat obvious but also CVA needs to be considered Assume that a bank will receive “N“ at time “T“. The bank prices this under consideration of counterparty credit risk / CVA and FVA Position causes funding need and bears credit risk Interest rate FVA* CVA Look at a degenerated balance sheet Balance Sheet Assets Liabilities Funding cost depends on the banks asset quality. This means that CVA and DVA and FVA have joint risk drivers. Simply summing up the effects leads to double counting. Loan position Credit Risk Funding Position Spread driven by asset’s credit risk *Simple term funding FVA

Not considered in this schematic diagram Proposed FVA framework (proposition 7) Double counting can be avoided by defining and calculating FVA, CVA and DVA as mutual increments FVA Positive and negative funding effects must be netted => FVA is based on net expected flows/exposure EE x fs Not considered in this schematic diagram CVA DVA EPE x ccs ENE x ocs Funding = FVA + liquidity buffer ENE and EPE add up to EE Replication does not result in the sum of liquidity and credit premium* We find it most practical to define CVA and DVA as increments of the (symmetric) FVA * E.g. a funding ticket is not traded by the price of funding cost + DVA FVA FVA on the basis of EE and funding spread “fs“ EE x fs CVA on the basis of EPE and residual credit spread “ccs -fs“ CVA Funding Spread DVA DVA on the basis Basis of ENE and residual credit spread “ocs -fs“ EPE x [ccs-fs] ENE x [ocs-fs] Counterparty Credit Spread Own Credit Spread

Transfer scheme is based on established core competences Proposed solution (proposition 9) A central FVA control is advantageous; the use of the CVA infrastructure may be an efficient way of implementation Liquidity spread risk FVA Trading unit FVA Management A/P Mgt. Treasury Management of market risks Gain/cost from the transfer of the FVA Complex „hybrid“ risk Risk drivers are liquidity curve and market risk drivers as e.g. interest rates and cross gammas Quality of the risk position comparable to CVA Management of the banks liquidity position Control of the liquidity spread risk Control of the maturity transformation Transfer scheme is based on established core competences Remark The scheme allows, besides the use of available core competences, also the use of available infrastructure. This may be of help at employing portfolio simulations or scenario engines and the related aggregation tools as well as analytics for the FVA management.

Contents I Background II FVA Calculation III Operating Model

Operation Model Is it necessary to establish a FVA desk? Tasks of the FVA desk Analytics FVA calculation Sensitivities and scenario calculations Risk Management Hedging of term funding (spread risk) with treasury Differentiation between term funded liquidity and liquidity buffer Consulting Understanding Cross Gamma risk Guidance for trade pricing and (spread) risk mitigation in trading Benefit of FVA desk Special core competence Specialized in complex hybrid risk position Owner of dedicated analytics Enables active risk management Takes hedge decisions with deep portfolio insight Helps trading comply with liquidity (spread) risk limits in an optimal way

Operation Model What are the key tasks and competencies of a FVA desk? FVA „cost/benefit“ Liquidity spread risk Trading unit FVA Management A/P Mgt. Treasury Liquidity buffer cost Derivatives trading in the market No market access Hedging via internal funding tickets Decision on hedge strategies term hedge / buffer Consulting role for trading and treasury Limit management for buffer funding of FVA desk to make sure treasury leads the term transformation Market access to secured and unsecured funding FVA desk is not a liquidity trader but transforms the liquidity position into term funding an liquidity buffer cost and transfers it to treasury

Market Risk Management Mobile +49 172 3006809 matthiaspeter@kpmg.com Partner Market Risk Management Mobile +49 172 3006809 matthiaspeter@kpmg.com © 2013 KPMG AG Wirtschaftsprüfungsgesellschaft, is a subsidiary of KPMG Europe LLP and a member firm of the KPMG network of independent member firms affiliated with KPMG International, a Swiss cooperative. All rights reserved. This document is confidential and its circulation and use are restricted. KPMG and the KPMG logo are registered trademarks of KPMG International. The KPMG name, logo and “cutting through complexity” are registered trademarks or trademarks of KPMG International.