When Risk Data Is Sparse: Modeling Emerging Market Credit Risk

Transcription

1 GARP Webcast When Risk Data Is Sparse: Modeling Emerging Market Credit Risk Presented by: Rob Stamicar, PhD Senior Director, Research Multi-Asset Class Risk Management Axioma Inc. Tuesday December 2, 2014 On24 Tech Tips 1 Make sure your speakers are on Hit F5 any time your console freezes For a LIVE event you should be hearing music now Use the Ask a Question feature to report issues Webcast starts at the top of the hour

2 Rob Stamicar, PhD Senior Director, Research, MAC Risk Management Before joining Axioma Rob worked as chief risk officer at Lighthouse Partners, a fund of hedge funds that specializes in managed account investments. Prior to Lighthouse, Robert worked at RiskMetrics as head of research in New York and as a risk manager at the Royal Bank of Canada. At Axioma, Rob conducts applied research projects utilizing Axioma Risk, Axioma s multi-asset class risk management platform. Rob holds a Ph.D. in Applied Mathematics from McMaster University in Canada. 2

3 References Aktug and Seyhan (2011). A critique of the contingent claims approach to sovereign analysis. Duyvesteyn and Martens (2012). Forecasting sovereign default risk with Merton s model. Gray, Merton and Zvi (2007). Contingent claims approach to measuring and managing sovereign credit risk. Journal of Investment Management, Vol. 5, No. 4. Lardy, Finkelstein, Khuong-Huu and Yang (2000). Method and system for determining a company s probability of no default. JP Morgan internal document Paras (2003). Advanced credit derivatives valuation: Bridging credit default swaps and corporate bonds. Presentation, JP Morgan Credit Conference Stamicar and Finger (2006). Incorporating equity derivatives into the CreditGrades model. Journal of Credit Risk, Vol. 2, No. 1. 3

4 Agenda Examine relationships within the fixed income, equity and credit default swap (CDS) markets Framework for translating credit information from one market to another - Relationships between markets Equity driven dynamics via a Merton Model - Sovereign Risk - Emerging Market Debt 4

5 Modeling Multi-Asset Class Portfolios Aggregation is challenging 5

6 Modeling Multi-Asset Class Portfolios Aggregation is challenging EM/Sovereign Risk 6

7 7 Big Picture: How Can Relationships Within/Between Asset Classes Help Measure EM/Sovereign Risk?

8 8 Scarce Data: How Do We Overcome Infrequent Data?

9 Agenda Examine relationships within the fixed income, equity and credit default swap (CDS) markets Framework for translating credit information from one market to another Equity driven dynamics via a Merton Model - Sovereign Risk - Emerging Markets 9

10 Example: EM/DM Correlations Yield Correlations 10

11 Example: EM/DM Correlations Yield Correlations Eurozone Debt Crisis Tapering 11

12 Some Dislocation Even Within DM Yield Correlations 12

13 And in the CDS Market More Responsive Than Bond Yields Sovereign CDS Correlations 13

14 Agenda Examine relationships within the fixed income, equity and credit default swap (CDS) markets Framework for translating credit information from one market to another - Default probabilities are links between asset classes Equity driven dynamics via a Merton Model - Sovereign Risk - Emerging Markets 14

15 How Can Relationships Between Asset classes Help Measure EM/Sovereign Risk????? 15

16 16 Translation Between Markets: Linking Markets Via Hazard Rates

17 Pricing Risky Securities With Hazard Rates Let s assume that we know the default probabilities Bond: CDS: Survival Probability Default Intensity 17

18 Translation Between Markets Example: Pricing Bonds From CDS Data Step 1: Extract risk-neutral default probabilities from CDS prices - Work backwards to bootstrap hazard rates CDS 5y compute h1 for [0, 7] 0.03 CDS 7Y compute h2 for [7, 10] CDS 10Y compute h3 for [10, Inf] h 2 h 3 Step 2: Price bond using hazard rates from CDS h Similar algorithm for pricing CDS with bond data Time to Maturity 18

19 19 Example: Spain s Sovereign Bond and CDS Spreads

20 What About Pricing Risky Securities via Equity Data? Risk-Neutral Model Merton/Structural Model Equity Options 20

21 Structural Models: Attempt to Model Asset Value of a Firm Many flavors of Merton Models exist - Stochastic interest rates/recovery rates - Complicated capital structure that encompass broad set of assets - Stochastic barriers - First passage knock-out problems Risk-neutral vs. actuarial probabilities of default - Goal is to track spreads and generate risk-neutral probabilities Calibration based on market observables - Goal is to estimate real-world probabilities of default Useful for economic capital calculations We will focus on risk-neutral probabilities 21

22 Let s Examine the Following Structural Model E2C/CreditGrades Model (Lardy, Finkelstein, Khuong-Huu and Yang) Key inputs are asset value and asset volatility which are unobservable These inputs are unobservable, estimated from - Balance sheet information - Equity prices/volatility 22

23 23 Structural Models

24 24 Structural Models

25 Agenda Examine relationships within the fixed income, equity and credit default swap (CDS) markets Framework for translating credit information from one market to another Equity driven dynamics via a Merton Model - Sovereign Risk Explicitly Implicitly - Emerging Markets Issuer specific risk Risk decomposition 25

26 Structural Models for Sovereign Risk Difficult to model sovereign risk via structural models - Indirectly: Financial sector - Directly/Explicitly: Balance sheet information difficult to obtain for a sovereign. What is the equity for sovereign? - Calibration is always an issue Inputs Asset Value Asset Volatility Outputs Default Probability Credit Spread 26

27 Linkage with Banking/Financial Sector Distressed/weak financial sector adds to sovereign risk Risk linkage (see Gray, Merton, Zvi (2007)) - Banks Sovereign Systematic banking crisis/deposit runs Non-performing loans - Banks Sovereign (implicit guarantee) Banks hold significant amounts of government debt securities During a sovereign shock, implicit guarantee increases, but government may fail to honor guarantee 27

28 28 Brazil: Sovereign and Financial CDS Spreads

29 29 Russia: Sovereign and Financial CDS Spreads

30 How Can Linkage Help? Financial/Banking sector is a suitable proxy for sovereign risk - Wealth of data Bond spreads, CDS spreads, equity prices In addition, use models to translate information from one market to another - Use information from one model in a market to estimate changes in another - Extract implied CDS spread Equity to credit via structural model Bond spreads to CDS spreads via risk-neutral pricing 30

31 Explicit Structural Models for Sovereign Risk Gray, Merton and Zvi: Local currency debt is subordinate to foreign currency debt, - Local currency debt plus monetary base is the equity Intuitively, FX volatilities are key drivers in this model Firm Balance Sheet Sovereign Balance Sheet Assets Loans Debt Equity Assets Foreign reserves Net fiscal assets Other public assets Foreign Currency Debt Local Currency Liabilities Local currency debt Monetary base Equity is a junior contingent claim 31

32 Explicit vs. Direct Approach Implied CDS: Russia Implied Sovereign CDS inline with sovereign CDS Implied CDS (from equities) more sensitive to recent events 32

33 33 Explicit Approach Implied CDS: Ukraine Before Crisis in Crimea

34 Sovereign Structural Model Explicit structural model for sovereigns - Difficult to estimate balance sheet information for a sovereign - Still can provide useful warning signals - Foreign exchange is a key driver More slides using indirect method - Should we use the Merton model for EM corporates? 34

35 Implied Sovereign CDS from Equity Data: Brazil 35 Calibration without balance sheet information

36 36 Implied Sovereign CDS from Equity Data: India

37 37 Implied Sovereign CDS from Equity Data: Greece

38 38 Implied Sovereign CDS from Equity Data: China

39 39 Implied Sovereign CDS from Equity Data: Malaysia

40 Agenda Examine relationships within the fixed income, equity and credit default swap (CDS) markets Framework for translating credit information from one market to another - Default probabilities are links between asset classes 40 Equity driven dynamics via a Merton Model - Sovereign Risk Explicitly Implicitly - Emerging Markets Issuer specific risk Risk decomposition

41 Issuer Specific Risk How should we decompose spread risk? EM equity factor models - Typically, EM equity models span countries - Over 16,000 securities Combine EM equity model with Merton Model - Translate equity factor returns into spreads Examples to follow 41

42 42 Application Value At Risk

43 JD Group 7.5% June 20, 2017 Issuer Spread Drilldown VaR95 (%) mvar95 (%) Value Leverage Growth Size Liquidity Short Term Momentum Medium Term Momentum Volatility Exchange Rate Sensitivity European Market Banks Specific Risk

44 Banco Espirito Santo 3.875% January, 21, 2015 Issuer Spread Drilldown VaR95 (%) mvar95 (%) Value Leverage Growth Size Liquidity Short Term Momentum Medium Term Momentum Volatility Exchange Rate Sensitivity European Market Banks Specific Risk

45 45 Banco Espirito Santos: Spread Decomposition

46 Stress Testing We can also impose structural relationships between different markets Correlations approaching one - Under financial distress, markets tend to move together - In less liquid markets, use information from other markets Use one market to stress another market Translate CDS spreads into bond/sovereign spreads Use equity information to infer risk of bond spreads 46

47 Concluding Remarks: EM/Sovereign Risk via Market Data Broader and deeper view of EM/Sovereign risk by analyzing multiple markets - Detect changing relationships within and between markets Translation/transmission framework between markets is linked by riskneutral default probabilities 47 Benefits of implied spread measures: - Sparse/illiquid data Incorporate equity/cds market information to predict changes to potentially less liquid domestic currency debt - Enhanced stress testing/early warning signals More timely indicators by incorporating multiple markets - Risk decomposition Equity factors can drive spread dynamics

48 48 Questions and comments?

49 Thank you For further information please contact: Axioma U.S.: / Europe: +44 (0) Asia:

50 Upcoming GARP Webcasts Certification and Careers in Energy Risk : Becoming an Energy Risk Professional (ERP) Thursday, Dec 04, 2014 Time: 11:00 am EST 4:00 pm GMT 11:00 pm HKT Cybersecurity A Critical Element to Enterprise Risk Management Tuesday, Dec 16, 2014 Time: 11:00 am EST 4:00 pm GMT 11:00 pm HKT To learn more and register, visit 50

51 About GARP The Global Association of Risk Professionals (GARP) is a not-for-profit organization dedicated to the risk management profession through education, training and the promotion of best practices globally. With a membership of over 150,000 individuals, GARP is the only worldwide organization offering comprehensive risk management certification, training and educational programs from board-level to entry-level. To learn more about GARP, please visit Creating a culture of risk awareness Global Association of Risk Professionals 111 Town Square Place, 14 th Floor Jersey City, New Jersey 07310, USA nd Floor, Bengal Wing 9a,Devonshire Square London EC2M 4YN +44 (0) Global Association of Risk Professionals. All rights reserved.