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
Do Leveraged Credit Derivatives Modify Credit
Allocation ?
Introduction
5 000 10 000 15 000 20 000
25 000 M arché G lobal des D érivés du C rédit en M ds $ (source: B B A )
O bligations G lobales C orporate en souffrance en M ds $ (indice M L global broad corporate)
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
Introduction
0% 10% 20% 30% 40% 50% 60% Single-name CDSs Full index trades Tranched index Synthetic CDO's Credit linked notes Basket products 2004 2006 A strong product innovationInitial growth began with single names CDS market
Development of synthetic CDO, index tranches, CDO² (CDO of CDO), CPPI, CPDO…
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 ?
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
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)
Methodology
This was true before the Subprime crisis !
CDS monthly returns (iTraxx, source JPMorgan)
Corporate bond – swap rate +3M cash rate monthly returns
-0.8% -0.4% 0.0% 0.4% 0.8% 1.2% 1.6% 02 /07/20 04 02 /09/20 04 02 /11/20 04 02 /01/20 05 02 /03/20 05 02 /05/20 05 02 /07/20 05 02 /09/20 05 02 /11/20 05 02 /01/20 06 02 /03/20 06 02 /05/20 06 02 /07/20 06 02 /09/20 06 02 /11/20 06
itraxx unfunded Merrill 5-7Y IG - sw ap 5Y
-3.0% -2.0% -1.0% 0.0% 1.0% 2.0% 3.0% 02/ 07/ 2004 02/ 10/ 2004 02/ 01/ 2005 02/ 04/ 2005 02/ 07/ 2005 02/ 10/ 2005 02/ 01/ 2006 02/ 04/ 2006 02/ 07/ 2006 02/ 10/ 2006 02/ 01/ 2007 02/ 04/ 2007 02/ 07/ 2007 02/ 10/ 2007 02/ 01/ 2008 02/ 04/ 2008
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
100 150 200 250 300 350 400 450 27 /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
Data
Credit derivatives returns
IG and HY CDS indices approximation Swap rates from Datastream
50 100 150 200 250 300 350 400 450 27 /0 3 /1 9 9 5 27 /0 3 /19 96 27 /0 3 /19 9 7 27 /0 3 /19 98 27 /0 3 /19 9 9 27 /0 3 /20 00 27 /0 3 /20 0 1 27 /0 3 /20 02 27 /0 3 /20 0 3 27 /0 3 /20 04 27 /0 3 /20 0 5 2 7 /0 3/20 06 27 /0 3 /20 0 7
Historical risk / return tradeoff
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 0.18 0.49 0.22 0.69 0.26 0.41
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
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
0% 1% 2% 3% 4% 5% 6% 7% 8% 0% 5% 10% 15% 20% volatility e xcess r e tu rn HY bonds IG bonds Gvt bonds Equities 0% 1% 2% 3% 4% 5% 6% 7% 8% 0% 5% 10% 15% 20% volatility excess r e tu rn HY spreads IG spreads Gvt bonds Equities
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 bonds IG bonds HY bonds Equities Gvt bonds 100.0% 94% 16% -7% IG bonds -33.8% 100.00% 38% 2% HY bonds -62.8% 37.90% 100.00% 30% Equities -6.6% 8.4% 25.7% 100.0% Gvt bonds IG spreads HY spreads Equities 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%
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 0.22-0.26
This makes portfolio composition depend only on the risk profile Optimal weights independant of the level of the Sharpe ratio
Results
Efficient frontiers : traditional credit compared to credit derivatives (no
leverage) 0.0% 0.5% 1.0% 1.5% 2.0% 2.5% 3.0% 3.5% 4.0% 1.0% 2.0% 3.0% 4.0% 5.0% 6.0% 7.0% 8.0% Risk E xce ss Re tu rn
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%
Traditional credit compared to leveraged credit derivatives
Influence of leverage
No change in the Sharpe ratio
0.0% 0.5% 1.0% 1.5% 2.0% 2.5% 3.0% 3.5% 4.0% 4.5% 5.0% 2.0% 3.0% 4.0% 5.0% 6.0% 7.0% 8.0% 9.0% Risk Ex c e s s R e tu rn
leverage=1 leverage=2 leverage=3 ) ( * ) (L rf L r rf r = + − σ σ(L) = L* σ σ σ ) ( * ) ( * ) ( ) ( ) ( f f r rf L r r L L r L r L SR = − = − = −
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%
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
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