130 likes | 494 Vues
1/12. ? Nomura International plc 7 July 2005. Correlation Skew and TRIFs. Outline of talk. No theoremsModelling issues linked to actual trade typeDescription of trade type and its characteristicsCorrelation and skew issuesIndividual solutionsCombining both correlation and skewWork for academic
E N D
1. Correlation, Skew and Target Redemption Inverse Floaters Martin Baxter Fixed Income Quantitative Research
Developments in Quantitative FinanceIsaac Newton Institute, 7 July 2005, 4.10pm
2. 1/12 Nomura International plc 7 July 2005 Correlation Skew and TRIFs Outline of talk No theorems
Modelling issues linked to actual trade type
Description of trade type and its characteristics
Correlation and skew issues
Individual solutions
Combining both correlation and skew
Work for academics
3. 2/12 Nomura International plc 7 July 2005 Correlation Skew and TRIFs Practitioner vocabulary Issue : Problem
Digital : Coupon is indicator function of some event
Inverse Floater : Coupon is (K-L)+, L is Libor rate
Target Redemption : trade terminates when total of coupons paid so far reaches or exceeds a threshold
4. 3/12 Nomura International plc 7 July 2005 Correlation Skew and TRIFs Target Redemption Inverse Floater We pay (K-L)+ to investor and receive Libor in swap structure
Trade terminates when total indexed coupon paid so far reaches threshold
Investor wants rates to stay low
Trade is good with steep yield curve because forwards are higher than customer expects
5. 4/12 Nomura International plc 7 July 2005 Correlation Skew and TRIFs Correlation issue For us, we want the Libors to be highly correlated
increases the chance they are all high
Like being long swaptions and short caplets
Essential feature to include in the model
6. 5/12 Nomura International plc 7 July 2005 Correlation Skew and TRIFs Skew issues Coupon is floor struck at K
when K is not at-the-money, skew is relevant
Trade often looks like a digital on a swap floating leg
Swap strike is lower than K and digital in nature
Implied BS digital vol <> Implied BS call/put vol
Digital strike is also dynamic
We dont know where it is
7. 6/12 Nomura International plc 7 July 2005 Correlation Skew and TRIFs Correlation solutions Term structure models
Single-factor models
Multi-factor models
instantaneous (local) vol structure
Vanilla case
Single-factor driving all the Libor rates
8. 7/12 Nomura International plc 7 July 2005 Correlation Skew and TRIFs Term structure models Multi-factor model
Multi-dimensional driving factor
Instantaneous vol model
Volatility is time dependent
9. 8/12 Nomura International plc 7 July 2005 Correlation Skew and TRIFs Skew solutions Lookup table works badly need actual model
Stochastic volatility (for example)
Good for matching individual marginals
Handles unknown strikes
10. 9/12 Nomura International plc 7 July 2005 Correlation Skew and TRIFs Combining correlation and skew Term structure model with stochastic vol
Evolution of rates and vols
Correlation structure
Forward skew
11. 10/12 Nomura International plc 7 July 2005 Correlation Skew and TRIFs Example model Very simple two-factor model
Drive all rates with one factor
Drive all vols with the other factor
Drawbacks
Lack of correlation control
Problems if Rho varies between Libor rates
12. 11/12 Nomura International plc 7 July 2005 Correlation Skew and TRIFs Bigger example model Multi-factor model
Drive rates and vols from m-dim Brownian motion
Where ei, hi are unit m-vectors, with
The choice of cross-correlations is open
13. 12/12 Nomura International plc 7 July 2005 Correlation Skew and TRIFs Things to do Choice of dimension and correlation vectors
Keep it sensible and implementable
Match the market prices and dynamics
Local stochastic volatility model
Forward skew
Extend to jointly calibrate to
Caplets and swaptions
Caplet skew and swaption skew
14. 13/12 Nomura International plc 7 July 2005 Correlation Skew and TRIFs