Do Leveraged Credit Derivatives Modify Credit Allocation?

Transcription

1 Do Leveraged Credit Derivatives Modify Credit Allocation? J.F. Boulier, M. Brière & J.R. Viala Crédit Agricole Asset Management, Université Libre de Bruxelles EDHEC Symposium «Risk and Asset Management», Nice, April 19, 2008 August 16, 2006 Page 1 Diversified Bond Meeting For internal use only

2 Introduction Exponential expansion of credit derivatives markets The fastest growing part of the global financial derivatives Driven by : Need of higher yielding investments in a context of low yields Need of hedging or gaining exposure to credit without funding Strong interest by some market participants in leveraging Marché Global des Dérivés du Crédit en Mds $ (source: BBA) Obligations Globales Corporate en souffrance en Mds $ (indice ML global broad corporate) April 2008 Page 2

3 Introduction A strong product innovation Initial growth began with single names CDS market Development of synthetic CDO, index tranches, CDO² (CDO of CDO), CPPI, CPDO 60% 50% 40% % 20% 10% 0% Single-name CDSs Full index trades Tranched index Synthetic CDO's Credit linked notes Basket products 19 April 2008 Page 3

4 Introduction Our questions : Does this new asset class change the strategic global allocation for an asset manager? What happens if we replace the investment in traditional corporate bonds in credit derivatives? How much should be invested in this new asset class? How much should we optimally leverage this asset class? 19 April 2008 Page 4

5 Methodology This work: Builds efficient frontiers in a mean/variance framework based on long term assumptions on expected returns and covarariances Compares efficient frontiers traditional corporate bonds (IG and HY) or credit derivatives with different degrees of leverage The analysis requires long samples on all asset classes Credit derivatives : analysis of CDS market (deepest) CDS indices begin in 2004 Approximation to recover a simulated longer history 19 April 2008 Page 5

6 Methodology Approximation of CDS returns A CDS is an agreement between 2 parties to exchange credit risk of a reference entity The seller of a CDS sells protection He receives periodic fee if the credit of the reference stable or improves He pays a compensation to the buyer in case of credit event CDS premium approximated by the bond credit spread over swap Theoretically, funding at libor, buying protection through CDS and entering an asset swap is fully hedged in any state of the world (Hjort et al. (2002)) In practice, difference between the 2 : the basis (De Wit (2006)) Relatively good approximation to use the credit spread cf empirical tests : Blanco, Brennan and Marsh (2003), Houweling and Vorst (2002), Hull, Predescu and White (2004) 19 April 2008 Page 6

7 Methodology This was true before the Subprime crisis! CDS monthly returns (itraxx, source JPMorgan) Corporate bond swap rate +3M cash rate monthly returns 1.6% 1.2% 0.8% 0.4% 0.0% -0.4% -0.8% 3.0% 2.0% 1.0% 0.0% -1.0% 02/07/ /09/ /11/ /01/ /03/ /05/ /07/ /09/ /11/ /01/ /03/ /05/ /07/ /09/ /11/2006 itraxx unfunded Merrill 5-7Y IG - sw ap 5Y -2.0% -3.0% 02/07/ /10/ /01/ /04/ /07/ /10/ /01/ /04/ /07/ /10/ /01/ /04/ /07/ /10/ /01/ /04/2008 itraxx unfunded Merrill 5-7Y IG - sw ap 5Y 19 April 2008 Page 7

8 Data Traditional asset classes returns Weekly returns of gvt bonds, IG and HY bonds, equities in USD April 1995-June 2007 Data source : Datastream for equity and govies (10Y benchmark) indices, Merrill Lynch for corporate bonds indices /03/95 27/03/96 27/03/97 27/03/98 27/03/99 27/03/00 27/03/01 27/03/02 27/03/03 27/03/04 27/03/05 27/03/06 27/03/07 Gvt Bonds IG Bonds HY Bonds Equities 19 April 2008 Page 8

9 Data Credit derivatives returns IG and HY CDS indices approximation Swap rates from Datastream /03/ /03/ /03/ /03/ /03/ /03/ /03/ /03/ /03/ /03/ /03/ /03/ /03/2007 IG credit spread HY credit spread Equities Gvt Bonds 19 April 2008 Page 9

10 Historical risk / return tradeoff Historical volatility Low level of volatility of corporate bond indices (especially HY) compared to govies Smaller vol for credit derivatives than corporate bonds for IG Higher vol for credit derivatives than corporate bonds for HY Historical returns 120 bp credit spread for IG over govies, 70 bp over swap 210 bp credit spread for HY over govies, 170 bp over swap Gvt Bonds IG bonds IG spreads HY bonds HY spreads Equities Excess return 1.3% 2.5% 0.7% 3.4% 1.7% 7.2% Volatility 7.2% 5.1% 3.1% 4.9% 6.4% 17.4% Sharpe ratio April 2008 Page 10

11 Historical risk / return tradeoff Discrepancy in the Sharpe ratios for all asset classes Attractive Sharpe ratios for IG (0.49) and HY bonds (0.69) Less attractive picture for credit derivatives (0.22 and 0.26), but still more interesting than govies excess return 8% 7% 6% Equities 5% 4% HY bonds 3% IG bonds 2% 1% Gvt bonds 0% 0% 5% 10% 15% 20% volatility excess return 8% 7% 6% Equities 5% 4% 3% HY spreads 2% 1% IG spreads Gvt bonds 0% 0% 5% 10% 15% 20% volatility 19 April 2008 Page 11

12 Correlations Much weaker (even negative) correlation of credit spreads with Treasuries Similar correlation with equities Credit derivatives have a strong diversifying power in a global portfolio Gvt IG HY Equities bonds bonds bonds Gvt bonds 100.0% 94% 16% -7% IG bonds -33.8% % 38% 2% HY bonds -62.8% 37.90% % 30% Equities -6.6% 8.4% 25.7% 100.0% Gvt IG HY Equities bonds spreads spreads Gvt bonds 100.0% -34% -63% -7% IG spreads -33.8% 100.0% 64% 8% HY spreads -62.8% 63.8% 100.0% 26% Equities -6.6% 8.4% 25.7% 100.0% 19 April 2008 Page 12

13 Portfolio construction Classic mean variance optimization with no short selling constraint Historical VCV matrices on the studied period Historical expected returns lead to inconsistent efficient frontiers We suppose a constant Sharpe ratio at 0.3 for each asset class Intermediate level between historical levels for Treasuries 0.18 and equities 0.41, close to credit spreads This makes portfolio composition depend only on the risk profile Optimal weights independant of the level of the Sharpe ratio 19 April 2008 Page 13

14 Results Efficient frontiers : traditional credit compared to credit derivatives (no leverage) 4.0% 3.5% 3.0% Excess Return 2.5% 2.0% 1.5% 1.0% 0.5% 0.0% 1.0% 2.0% 3.0% 4.0% 5.0% 6.0% 7.0% 8.0% Risk Investment in Corporate Bonds Investment in Credit Derivatives, leverage = 1 19 April 2008 Page 14

15 Results Improvement in the efficient frontier by investing in credit derivatives Including IG credit derivatives, we can achieve much less volatile portfolios than with traditional corporate bonds At higher risk level, IG spreads disappear in favor of HY spreads Credit derivatives offer strong decorrelation and allow to introduce more risky assets (equities) for same level of portfolio risk Portfolio Risk 3.0% 5.0% 7.0% Excess Return - 2.3% 3.1% Optimal Weights Gvt bonds - 43% 59% IG bonds - 0% 0% HY bonds - 42% 8% Equities - 15% 32% Portfolio Risk 3.0% 5.0% 7.0% Excess Return 2.1% 2.8% 3.2% Optimal Weights Gvt bonds 45% 45% 44% IG spreads 4% 0% 0% HY spreads 47% 32% 20% Equities 5% 23% 37% 19 April 2008 Page 15

16 Traditional credit compared to leveraged credit derivatives Influence of leverage No change in the Sharpe ratio r( L) = rf + L *( r rf σ ( L ) = L *σ ) r( L) r SR( L) = σ ( L) f = L *( r r L * σ f ) = ( r r σ f ) Excess Return 5.0% 4.5% 4.0% 3.5% 3.0% 2.5% 2.0% 1.5% 1.0% 0.5% 0.0% 2.0% 3.0% 4.0% 5.0% 6.0% 7.0% 8.0% 9.0% Risk leverage=1 leverage=2 leverage=3 19 April 2008 Page 16

17 Traditional credit compared to leveraged credit derivatives Influence of leverage Leveraging increases portfolio risk and allows higher returns At identical risk level, the higher the leverage, the more we can reduce the share of risky assets in the portfolio in favour of Treasuries Optimal allocations contain majority of Treasuries, HY and then equities Portfolio Risk (L=2) 3.0% 5.0% 7.0% Excess Return - 3.2% 3.7% Optimal Weights Gvt bonds - 50% 35% IG bonds - 0% 0% HY bonds - 35% 36% Equities - 16% 29% Portfolio Risk (L=3) 3.0% 5.0% 7.0% Excess Return - 3.5% 4.0% Optimal Weights Gvt bonds - 62% 46% IG spreads - 0% 0% HY spreads - 28% 34% Equities - 10% 20% 19 April 2008 Page 17

18 Conclusion We examine how credit derivatives change the construction of an efficient portfolio We compare 2 types of credit instruments included in a US global portfolio (including gvt bonds and equities) conventional corporate bonds credit derivatives Credit risk component has : very low risk for IG, medium risk (smaller than govies) for HY strong diversifying power relative to traditional asset classes (negative correlation with govies) Efficient frontiers in a mean variance framework show the advantage of credit derivatives for portfolio diversification usefulness of leveraging to allow flexible risk modulation 19 April 2008 Page 18

19 Conclusion Directions for future research Expected returns hypothesis Strong hypothesis of constant Sharpe ratio Asymetric nature of credit spreads distribution short spikes and long periods of low values Mean variance framework problematic for credit spreads : use VaR, conditional VaR? Analysis of crisis episodes Credit derivatives performances / risk in times of stress Consequence for asset allocation 19 April 2008 Page 19