Counterparty Risk CVA Eduardo Canabarro Global Head of Risk Analytics Morgan Stanley, New York
Disclaimer This presentation contains statements and views of the author only. It is not intended to represent the views of Morgan Stanley. 2 2
Introduction OTC derivatives are efficient and effective tools to transfer financial risks between market participants As a byproduct of such transfer, they create credit risk between the counterparties They also increase the connectedness of the financial system Banks have built sophisticated frameworks to manage their counterparty credit risks Typically, a large bank has many thousands of counterparties, trillions of dollars of derivatives notional and billions of dollars of credit exposures to their counterparties In this presentation we ll cover counterparty risk pricing (aka CVA), hedging, stress testing, capital, and CCPs 3 3
Counterparty exposures: bilateral and market-driven Typically, both counterparties face credit risks with respect to each other Counterparty exposures are driven by market risk factors It is necessary to measure potential future exposures (PFEs) beyond the current level of exposure 4 4
Simulation of PFEs Banks use Monte Carlo methods to simulate the future values of the portfolio of derivatives with a counterparty 5 5
Thousands of simulated market paths The paths start at the current value of the portfolio and they end at zero, when all trades in the portfolio of trades with the counterparty have terminated 6 6
EPE and ENE For each point in time on a simulated market path, we calculate the exposure as the max(value of the portfolio, 0) Expected Positive Exposure (EPE) is our average exposure to the counterparty, across all paths, at each point in time Expected Negative Exposure (ENE) is the equivalent of EPE, from the perspective of our counterparty 7 7
EPE and ENE The EPE and ENE profiles are central to the calculation of CVAs In sophisticated CVA models those profiles are calculated conditional on the credit spreads of each counterparty 8 8
Credit Valuation Adjustment (CVA) Bank A has a portfolio of OTC derivatives with Counterparty B CVA is the adjustment to the risk-free value of the portfolio of OTC derivatives between A and B to reflect the market value of the bilateral counterparty credit risks faced them Eduardo Canabarro and Darrell Duffie, Counterparty Risk: Measurement and Pricing, 2003. http://www.darrellduffie.com/uploads/surveys/duffiecanabarro2004.pdf 9 9
Economic intuition If Bank A faces more credit risk than its Counterparty B, the CVA is negative, i.e. it reduces the value of the OTC derivatives from the perspective of Bank A If Bank A faces less credit risk than Counterparty B, the CVA is positive, i.e. it increases the value of the derivatives from the perspective of Bank A 10 10
CVA is part of the valuation of derivatives CVA is an integral component of the value of derivatives Ideally, CVA should be part of each trade s valuation model The reason it is calculated separately is that there are portfolio effects that transcend the valuation of each trade (e.g. netting and margin agreements) CVA can be attributed to each trade on a marginal contribution basis 11 11
CVA volatility Banks that calculate CVA are subject to the volatility of market prices They need to hedge their CVA s risks The 2008 financial crisis showed that CVA-related losses can be much larger than default losses CVA risks include changes in the credit spreads of the counterparties as well as changes in the market prices that drive the underlying derivative exposures 12 12
CVA risk management The technology to mark to market and hedge CVA has evolved over the last 20+ years Investment banks started pricing and hedging CVA around 1990 Litzenberger, R., Swaps: Plain and Fanciful, Journal of Finance, vol.47, pages 831-850, 1992. Sorensen, E., and T. Bollier, Pricing Swap Default Risk, Financial Analysts Journal, 50, pp. 23-33, May-June 1994. Duffie, D. and M. Huang, Swap Rates and Credit Quality, Journal of Finance, v. 51, pp. 921-949, 1996 More recently, many more banks are pricing and actively hedging their CVAs 13 13
CVA calculation In concise notation: CVA E A s A E B s B E A is the present-valued expected exposure faced by counterparty B with respect to Bank A; s A is the market loss rate (i.e. the product of risk-neutral PD and risk neutral LGD) of A E B is the present-valued expected exposure faced by A with respect to B; s B is the market loss rate of B. 14 14
Example 1 E A = $200 s A = 2% E B = $100 s B = 5% CVA = 200 x 0.02 100 x 0.05 = 4 5 = -$1 The CVA is a negative adjustment to the risk-free value of the portfolio of trades as seen by Bank A because Bank A faces more credit risk than Counterparty B If the risk-free value of the portfolio were -$50, the portfolio would be worth -$51 for Bank A and +$51 for Counterparty B. Both counterparties agree with this value 15 15
Example 2 Now, suppose that Bank A exits the portfolio of trades with Counterparty B by transferring it to Bank C C has s C = 5% and from C s perspective: CVA = 200 x 0.05 100 x 0.05 = 10 5 = +$5 To effect the transfer, A pays +$51 to C C is a worse counterparty than A and it has to pay $6 to B in order to compensate B for the drop in the value of the portfolio of trades from $51 to $45 All three parties break even and they agree with the transactions 16 16
CVA risk sensitivities CVA E A s A E B s B a) Sensitivities of the CVA with respect to the credit spreads: CVA CVA EA EB sa sb b) Sensitivities of the CVA with respect to the underlying exposures: CVA EA c) Cross-convexities: s A CVA E B s B CVA E A s A 1 CVA E B s B 1 17 17
Should banks hedge their CVA? If the bank marks to market its CVA and the bank does not hedge it, it will experience P&L (and earnings) variability Importantly, in a trending and deteriorating credit market environment, the bank could suffer substantial cumulative CVA losses In the 2008 crisis, some banks lost many billions of dollars in CVAs. This was particularly the case of banks that did not actively hedge their CVAs 18 18
CVA hedging: challenges The hedges of the CVA include hedges of the market risk factors that drive the underlying exposures and hedges of the credit spreads of the counterparties There are important cross-gammas which can be of substantial size when the changes in spreads and exposures are large During the 2008 crisis, due to the large size of the CVAs and the high volatility of markets (i.e. large ΔE and Δs), the cross-gammas created difficulties for CVA desks that were dynamically hedging the CVAs Eduardo Canabarro, Pricing and Hedging Counterparty Risk: Lessons Re- Learned?, Chapter 6 in Canabarro E., editor, Counterparty Credit Risk, Risk Books, 2010 19 19
Should banks hedge their own spread? ΔCVA / ΔE A = s A ΔCVA / Δs A = E A Δ 2 CVA / (ΔE A Δs A ) = 1 Changes in the exposure E A can be hedged by taking positions on the market risk factors that drive the exposure Changes in Bank A s own loss rate s A are more challenging to hedge. The systematic risk component can be hedged. The bank-specific, idiosyncratic risk component is more difficult to hedge By hedging the systematic component of their own credit risk, banks can realize the value of the liability CVA 20 20
Bank 1: mainly systematic spread risk Bank s CDS versus CDX index spread 300 Bank 250 CDX 200 150 100 50 21 0 Jun-05 Dec-05 Jun-06 Dec-06 Jun-07 Dec-07 Jun-08 Dec-08 Jun-09 Dec-09 Jun-10 21
Bank 2: some idiosyncratic spread risk Bank s CDS versus CDX index spread 700 Bank 600 CDX 500 400 300 200 100 22 0 Jun-05 Dec-05 Jun-06 Dec-06 Jun-07 Dec-07 Jun-08 Dec-08 Jun-09 Dec-09 Jun-10 22
Bank 3: more idiosyncratic spread risk Bank s CDS versus CDX index spread 1400 Bank 1200 CDX 1000 800 600 400 200 23 0 Jun-05 Dec-05 Jun-06 Dec-06 Jun-07 Dec-07 Jun-08 Dec-08 Jun-09 Dec-09 Jun-10 23
CVA desks Some banks have opted for a central CVA desk Others have opted for various CVA desks deployed within their main derivatives units CVA desks provide counterparty credit risk protection to the derivatives trading desks They manage the risks of the CVA on an ongoing basis They are subject to market and credit risk limits and usually do not have a revenue budget 24 24
CVA risks There are important risks that often fall outside of the scope of the risk measurement frameworks: wrong way out of the money replacement costs dynamic hedging It is not what we know, but what we do not know which we must always address, to avoid major failures, catastrophes and panics. Richard Feynman, physicist 25 25
Wrong-way risks There are wrong-risks that are specific to CVA hedging. Example: crowded counterparty risks When a counterparty has entered into similar and large OTC derivatives trades with many banks, the dynamic hedging programs of the banks will create wrong-way risk Usually, those wrong way risks do not show up until credit spreads and/or exposure have grown to some large levels During the 2008 crisis this occurred with respect to monoline insurers as well as other concentrated counterparty exposures 26 26
Wrong-way risks The CVA wrong way risks are dynamic That is, they are a feature of dynamic hedging strategies They are different from the wrong way risks as usually defined in the Banking Book context They can be large, i.e. non-local, if there is illiquidity in exposure or credit spread hedges 27 27
Out-of-the-money risks Potential exposure models used for CVA calculation are not good predictors of massive market dislocations CVA traders need to be cautious in the pricing and hedging of out-the-money counterparty exposures The ability to hedge those exposures in the future, as they grow, needs to be assessed prudently considering the overall liquidity of the market The profitability of such trades needs to be evaluated considering the potential CVA risks and dynamic hedging costs 28 28
Replacement costs Potential future exposure and CVA models account for the benefits of collateral in the calculation of counterparty exposures The models measure the residual exposures after the consideration of collateral Banks should not underestimate the all-in costs of replacing trades with a defaulted counterparty Especially when that counterparty is a large market participant and its default can impair the liquidity and increase the volatility of the markets where the derivatives trade (e.g. Lehman) 29 29
Dynamic hedging costs The risk management of CVAs requires dynamic rebalancing of the hedges When the counterparty exposures and the credit spreads of the counterparties are large and volatile, the rebalancing requirements can be intense and costly The high cost is due to illiquidity, wide bid-ask spreads and overall market impact of the hedging program, especially when in crowded risk situations Dynamic hedging costs are usually not explicitly captured in the CVA pricing models but they can be the most relevant cost component of large, concentrated CVA risks 30 30
Simulation of dynamic hedging costs We can use Monte Carlo simulation models to assess the size of the costs of replication over the life of the CVA hedging program The models incorporate the market frictions and provide a realistic description of the probability distribution of potential CVA hedging costs During the 2008 crisis, the costs of CVA hedging proved to be quite material in some cases Eduardo Canabarro, Dynamic Hedging Costs of CVA, in Canabarro E. and M. Pykhtin, editors, Counterparty Credit Risk, Risk Books, forthcoming 2014. 31 31
CVA Stress tests Stress testing is a fundamental component of a sound CVA risk management program The fundamental goals of the stress test framework should be: - Identification of concentrations of market and credit risks - Identification of out-of-the-money exposures - Identification of wrong-way risks - Identification of potentially large dynamic hedging costs of CVA 32 32
Capital on CVA: advanced approach Basel 3 defines CVA using the Basel 2 IMM EE profiles. The market risk of CVA is then measured by the bank s VaR model IMM exposures for risk sensitivities VaR for credit spread risk Only spread risk; no exposure risk Single name and index hedges VaR and stressed VaR, times 3 Need IMM + VaR model approvals 33 33
Capital on CVA: standardized approach Direct formula for the capital: h = 1 year w i based on rating of counterparty M maturity factor B notional of hedges See Michael Pykhtin, Model foundations of the Basel 3 standardized CVA charge, Risk Magazine, July 2012. 34 34
Computational effort Data Sourcing Typically 2-10M trades, 2-10k netting sets and margin agreements, market data Simulation of Markets Typically 1-2k paths of 2-5k risk factors over 100 future dates per path Trade Pricing Typically 2-10M trades, over 1-2k paths at each of 100 dates Exposure and CVA calculations Typically 10k netting nodes Back of envelope numbers: 2M trades x 2k paths x 100 dates/path = 400B pricings 400B pricings x 0.00001 sec/pricing = 400k secs = 111 CPU hours This is just for one calculation we need many more calculations to obtain CVA risk sensitivities. 35 35
CVA systems CVA systems are complex and computationally demanding Banks with large OTC derivatives franchises have invested large resources to build up these systems over the last 10-15 years 36 36
CVA systems It is important to engineer the CVA system and models for computational efficiency and speed Various techniques have evolved to enable fast calculations Data storage strategies for trade and netting set data and parallel processing are key elements 37 37
CVA systems The banks that implemented the most successful CVA systems were the ones that pursued: Modularization Parallel processing capability Scalability Pragmatic analytics as simple as possible; but not simpler. - Einstein 38 38
Central Counterparties (CCPs) Clearing Members (CMs) face a CPP instead of facing each other directly Multilateral netting, margin requirements, capital buffers, and high operational standards reduce the connectedness of the financial system There will be trades left outside of the CCPs 39 39
CCPs favorable aspects CCPs are critical components of the global financial and payments systems They are vital to financial stability They enable multilateral netting and collateralization They promote transparence and standardization of trades They provide capital buffers to absorb counterparty default losses They reduce connectedness and systemic risk 40 40
CCPs becoming large exposures Since 2009, inter-dealer clearing of OTC derivatives has accelerated It is expected to continue increasing The largest counterparty risks faced banks are rapidly shifting from peer banks to CCPs A typical large bank is a clearing member of tens of CCPs and it is likely that its top 5-10 counterparty exposures are already to CCPs today 41 41
CCPs capital and margin Basel 2 did not charge regulatory capital on CCPs Basel 3 charges capital on exposures to CCPs: about 20% EAD, IMM based Initial margin for OTC is typically at 95-99% confidence level, 5-day market move Margin may also consider liquidity characteristics, risk concentration and product-specific features 42 42
CCPs loss waterfall Defaulting CM margin Defaulting CM s guarantee fund CCP s equity capital (small) Guarantee funds of non-defaulting CMs Additional calls for capital on non-defaulting CMs (unlimited liability) 43 43
CCPs - concerns Specialization Fragmentation Competition Too big to fail 44 44
Counterparty Credit Risk Measurement, Pricing and Hedging Edited by Eduardo Canabarro This book describes the methods and practices used to manage OTC derivative counterparty risk and the performance of those methods during the 2008 financial crisis. It covers topics in counterparty risk measurement, CVA, CVA hedging, credit derivatives, collateralization, stress testing, back testing and integration of counterparty credit risk into economic capital frameworks. Experiences and new ideas on models are discussed by a group of world-class experts. The content of the book is particularly relevant in light of the Basel 3 rules on the regulatory capital on counterparty risks. The book contains a wealth of insights that can be useful for practitioners, regulators, consultants, accountants, lawmakers, auditors and researchers to understand the substantive, and often technical, issues related to counterparty risk management. Chapters by: Aaron Brown Eduardo Canabarro Guanghua Cao Patrick Chen Eduardo Epperlein Jon Gregory Andrew Hollings Gregory Hopper Sean Hrabak Phillip Koop Darren Measures Shankar Mukherjee Evan Picoult Michael Pykhtin Dan Rosen David Rowe David Saunders Alan Smillie Svein Stokke Yi Tang Lauren Teigland-Hunt Dan Travers Katsuichiro Uchiyama Andrew Williams Wei Zhu Online: riskbooks.com/counterparty-credit-risk-2 ISBN: 978-1-906348-34-2