Challenges in Counterparty Credit Risk Modelling

Transcription

1 Challenges in Counterparty Credit Risk Modelling Alexander SUBBOTIN Head of Counterparty Credit Risk Models & Measures, Nordea November 23 th, 2015

2 Disclaimer This document has been prepared for the purposes of providing information on Nordea Group (Company) and its activities. The following disclaimer applies to this presentation and any information provided regarding the information contained in this presentation (Information). The Information does not constitute an offer, invitation, solicitation or recommendation in relation to the purchase or sale of securities in any jurisdiction and neither this presentation nor anything in it forms the basis of any contract or commitment. It does not contain assessment of risk or profitability on any investment product. This document may not be reproduced, distributed or published for any purpose without the prior written permission of the Company. Copyright Nordea,

3 Today s Agenda Briefly about CCR what are we talking about? New institutional & regulatory environment New economic environment Collateral modelling Summary 3

4 Counterparty Credit Risk 14

5 The OTC market End users 5

6 Introduction to counterparty credit risk Nordea s perspective: Wants to avoid the credit risk that occurs out of market risk / market movements Customer trade introduces market risk Hedge trade off-sets Nordea s market risk PL = 0 No market risk but What if customer or hedging counterparty defaults? We loose MV when MV>0, Only nothing if MV=0 Nordea needs to control, manage and mitigate the counterparty credit risk 6

7 What is counterparty credit risk? Credit risk on counterparties trading OTC derivatives is called Counterparty Credit Risk (CCR) Credit losses may occur when a counterparty defaults Size of potential loss is unknown The loss at default depends on the market value MV of the OTC derivative contracts with the counterparty and the recovery rate R The loss at default is Loss = max MV, 0 1 R We need to estimate how the MV may develop! 7

8 Overview of model framework Generate future states of the market Distributions and measures Revalue trades in futures states Aggregate according to legal agreements 8

9 Model framework - what s needed? Today s market data and simulation models Measure specifications Generate future states of the market Distributions and measures Monte Carlo simulations: correlations and time grid Revalue trades in futures states Aggregate according to legal agreements Trade data, pricing models, scenario consistency Agreement details, opinion on legal enforceability, model for collateralization 9

10 CCR measures require complex business and IT infrastructure, with a lot of stakeholders Regulators Netting, Collateral, Margin calls Trades Back Office/ Collateral Management / Legal Integration to CCR engine Computation engine CCR indicators Regulatory capital computation Risk Control & Reporting Extractions Pricing Front Office data Data transformations Scenario Simulation and pre-deal check Market data Internal market data storage Simulation - Traders in markets External sources of market data Calibration Model parameters Documentation 10

11 Capitalization of Counterparty Credit Risk Losses from defaults => Counterparty credit default risk charge Losses from deterioration of credit quality of counterparties => CVA capital charge Regulators Advanced vs standardized approaches 11

12 The world is changing and models also have to! New market environment New regulation & regulators New management expectations CCR Models & Measures 12

13 New regulation 14

14 General context CRDIV/CRR International Convergence of Capital Measurement and Capital Standards (Basel I) Market risk amendment to the Capital Accord A revised framework is published (Basel II) Basel II enters into force International framework for liquidity risk measurement, standards and monitoring (Basel III) Extended accord for Basel III Process to monitor members implementation of Basel III Fundemental review of trading book, Leverage ratio framework & disclosure equirements, Review of Risk models under way Basel Timeline 14

15 Impacts on CCR Basel III More requirements to collateral modelling More attention to backtesting CVA risk capitalized Strong incitation to central clearing + clearing obligation in EMIR Post Basel III New and more risk-sensitive standardized method CVA risk is going to be part of the new market risk framework under Fundamental Review of the Trading Book Margin reform on OTC trades Initial margin requirements and Standardized Initial Margin Model (SIMM) by ISDA New way of computing variation margin 15

16 Capital computation for an IMM bank is extremely complex Normal Calibration Stressed Calibration Internal Model Default Risk Charge Regulatory Capital CVA Risk Charge Internal Limit Management Normal EAD Normal RWA Stressed EAD Normal Risk Weights Stressed RWA Normal EE Normal Credit Spreads Stressed EE Stressed Credit Spread Potential Future Exposures Max of the aggregated sum RWA Sum RWA Utilizations 16 18/08/2014

17 Central counterparts far from being simple Textbook picture More realistic picture 17

18 CVA risk charge Background: CVA volatility in general not captured in market Risk VaR The risk charge for Credit Value Adjustment was introduced in CRD IV Exemption for non-financials has been granted in EU (but not elsewhere) The current framework has received much criticism Only risk coming from credit spreads Regulatory vs Accounting CVA exposures A complete revision of the framework was put forward in June 2015, in conjunction with a Quantitative Impact Study Timeline: 2014, 1 Jan: CRD IV 2018 alt 2019: revised framework for CVA RC 2016, Q3: excessive CVA 18 10/11/2015

19 CVA Risk Charge Challenges Modelling challenges CVA is very exotic product, computing all sensitivities is not a trivial task Which CVA model to use? If internal model, how to do PnL attribution for backtesting And more general considerations It makes OTC derivatives business more costly Gives stronger incitation for hedging CVA PnL Will more hedging result in more demand for CDS, but still limited supply? What about liquidity of the credit market (already very poor in Europe) Is it going to increase Wrong Way Risk and systemic risk in the whole banking system 19

20 New economic environment 14

21 Low interest rates and inflation in EU Interest rate (IR) curves represent by far the most important type of risk factors, driving counterparty risk exposure Developments since 2007 have made the IR modelling world much more complex. The value of the most vanilla flow would now depend on Clearing Collateral Currency of collateral Henrard, M. (2014). Interest Rate Modelling in the Multi-curve Framework: Foundations, Evolution and Implementation. Palgrave Macmillan. Pallavicini, A., & Tarenghi, M. (2010). Interest-rate modeling with multiple yield curves. Available at SSRN Nominal rates levels have fallen to extremely low to zero in many developped world currencies, which is not only a challenge for banks profitability, but also for risk and other modelling teams. Christensen, J. H., & Rudebusch, G. D. (2014). Estimating shadow-rate term structure models with near-zero yields. Journal of Financial Econometrics, nbu

22 Swaps rates in free fall 22

23 Counterparty risk perspective When we forecast rates, it is the probability distribution of the rates and not the value at some particular future date that matters To evaluate the quality of the forecast, we check how likely it is that actually observed values come from the distribution, predicted from the model. If we have many observations, we can do backtesting. Many forecasting horizons (from several days to several years) are important and must be treated with the same framework. For very long term horizons, common sense and not backtesting matters. =>Difficult to model something we have very little experience of and hardly can backtest 23

24 Challenges in Internal Models 14

25 Diverging practices across banks The Basel Committee published a report on the regulatory consistency of riskweighted assets (RWAs) for counterparty credit risk on 2 Oct 2015, further called Report [1]. The Report is a part of its wider Regulatory Consistency Assessment Programme (RCAP), which is intended to ensure consistent implementation of the Basel III framework. A hypothetical test portfolio was used to examine the variability in banks practices for derivatives exposure assessment. The study shows considerable variability in the outcomes of CCR models 25

26 An example of considerable variability 26 14/04/2011

27 Future collateral modelling as the main reason Settlement period liquidation period Based on: Alexander Sokol Leif Andersen Michael Pykhtin. Modeling Credit Exposure for Margined Counterparties,

28 Getting more complicated with Initial Margin Initial Margin is based a 10-day VaR calculation on specific buckets of trades (not the entire netting sets) Live September 2016 for the biggest institutions Challenges How to project at future dates on each scenario? How this would affect the default schedule picture? Can counterparty stop paying variation margin but pay IM, or vice versa? Can it be used a pool to cover losses on all trades? (Irrespective of the buckets, on which it has been computed) Will it really be standardized? More generally, are tighter collateral requirements transforming some CCR into liquidity risk, and in a stressed situation, can this liquidity risk turn back into counterparty risk? 28

29 Summary 14

30 Some hot topics on our agenda Central counterparts Margin reform and collateral models New CVA risk framework Risk factor modelling and backtesting in new economic environment Concentration risk, systematic risk, wrong way risk Increased focus on stress-testing 30

31 Expressing the challenge in one sentence Being compliant while enabling business and providing opinion on risk 31