<strong>Extension</strong> <strong>of</strong> <strong>stochastic</strong> <strong>volatility</strong> equity <strong>models</strong> 990.060.055Cliquet prices from SZHW and HHW <strong>with</strong> r x,r= −0.7 and the HestonSchöbel−Zhu−<strong>Hull</strong>−<strong>White</strong>Heston−<strong>Hull</strong>−<strong>White</strong>Heston0.060.055Cliquet prices from SZHW and HHW <strong>with</strong> r x,r= 0.7 and the HestonSchöbel−Zhu−<strong>Hull</strong>−<strong>White</strong>Heston−<strong>Hull</strong>−<strong>White</strong>Heston0.050.05Value <strong>of</strong> Cliquet at t=00.0450.040.0350.03Value <strong>of</strong> Cliquet at t=00.0450.040.0350.030.0250.0250.020.020.015−0.1 −0.08 −0.06 −0.04 −0.02 0 0.02 0.040.015−0.1 −0.08 −0.06 −0.04 −0.02 0 0.02 0.04MinCouponMinCouponDownloaded by [Lech A. Grzelak] at 10:55 24 January 2012Figure 1. Pricing a cliquet product under the SZHW, the HHW and the Heston <strong>models</strong>. Both figures present the price <strong>of</strong> a globallyfloored cliquet as a function <strong>of</strong> MinCoupon given by (52) for T ¼ 3 years and M ¼ 36. The remaining parameters are as in table 4.Left: Pricing <strong>with</strong> x,r ¼ 0.7. Right: Pricing <strong>with</strong> x,r ¼ 0.7.Value <strong>of</strong> Diversification product at t = 01.31.251.21.151.11.051Diversification product prices from SZHW and HHW <strong>with</strong> r x,r= −0.7 and the HestonSchöbel−Zhu−<strong>Hull</strong>−<strong>White</strong>Heston−<strong>Hull</strong>−<strong>White</strong>Heston0.950 50% 100% 150% 200% 250%ωValue <strong>of</strong> Diversification product at t = 01.31.251.21.151.11.05Diversification product prices from SZHW and HHW <strong>with</strong> r x,r= 0.7 and the Heston1Schöbel−Zhu−<strong>Hull</strong>−<strong>White</strong>Heston−<strong>Hull</strong>−<strong>White</strong>Heston0.950 50% 100% 150% 200% 250%ωFigure 2. Pricing <strong>of</strong> a diversification hybrid product under different <strong>models</strong>. The simulations were performed <strong>with</strong> ¼ 10. Theremaining parameters are as in table 4. Left: Pricing <strong>with</strong> x,r ¼ 0.7. Right: Pricing <strong>with</strong> x,r ¼ 0.7.shows that the Heston model gene<strong>rate</strong>s a significantlyhigher price, whereas the HHW and SZHW prices arerelatively close. The absolute difference between the<strong>models</strong> increases <strong>with</strong> percentage !.4.4. St<strong>rate</strong>gic investment hybrid (best-<strong>of</strong>-st<strong>rate</strong>gy)Suppose that an investor believes that if the price <strong>of</strong> anasset, S 1 t , goes up, then the equity markets under-performrelative to the <strong>interest</strong> <strong>rate</strong> yields, whereas if S 1 t goesdown, the equity markets over-perform relative to the<strong>interest</strong> <strong>rate</strong> (Hunter and Picot 2005/2006). If the prices <strong>of</strong>S 1 t are high, the market may expect an increase in inflationand hence in <strong>interest</strong> <strong>rate</strong>s and low S 1 t prices could havethe opposite effect. In order to include such a feature in ahybrid product we define a contract in which an investoris allowed to buy a weighted performance coupondepending on the performance <strong>of</strong> another underlying.Such a product can be defined as follows:<strong>with</strong>ðt 0 ¼ 0, T Þ¼E Q ðeV T ¼ max 0, ! L 0þð1L TR T0 r sds V T jF 0 Þ, ð56Þ!Þ S TS 0þ max 0, ð1 !Þ L 0L Tþ ! S TS 01 S 1T >S 1 01 S 1T 5S 1 0,where ! 0 is a weighting factor related to a percentage,and L T ¼ P Mi¼1 PðT, t iÞ <strong>with</strong> t 1 ¼ T is the T-value <strong>of</strong> theprojected liabilities for certain time t M , <strong>with</strong>!4100% !.Figure 3 shows the prices obtained from Monte Carlosimulation <strong>of</strong> the contract at time t 0 ¼ 0 for maturityT ¼ t 1 ¼ 3 and time horizon t M ¼ 12 <strong>with</strong> one year spacing.
100 L. A. Grzelak et al.10.9St<strong>rate</strong>gic Investment SZHW, HHW <strong>with</strong> r x,r= −0.7 and the Heston modelSchöbel−Zhu−<strong>Hull</strong>−<strong>White</strong>Heston−<strong>Hull</strong>−<strong>White</strong>Heston10.9St<strong>rate</strong>gic Investment SZHW, HHW <strong>with</strong> r = 0.7 and the Heston modelx,rSchöbel−Zhu−<strong>Hull</strong>−<strong>White</strong>Heston−<strong>Hull</strong>−<strong>White</strong>HestonValue <strong>of</strong> St<strong>rate</strong>gic Investment at t = 00.80.70.60.5Value <strong>of</strong> St<strong>rate</strong>gic Investment at t = 00.80.70.60.50.40.40% 50% 100% 150% 200% 250%ω0% 50% 100% 150% 200% 250%ωFigure 3. Discounted pay<strong>of</strong>fs <strong>of</strong> the st<strong>rate</strong>gic investment hybrid priced <strong>with</strong> the SZHW, the HHW and the Heston <strong>models</strong> as afunction <strong>of</strong> !. The remaining parameters are as in table 4. Left: Pricing <strong>with</strong> x,r ¼ 0.7. Right: Pricing <strong>with</strong> x,r ¼ 0.7.Downloaded by [Lech A. Grzelak] at 10:55 24 January 2012Since we did not model the second underlying process, S 1 T ,we assume that S 1 T > S1 0 . We see that, for ! 2 [0%, 100%],the max over the sum <strong>of</strong> performances disappears andthe hybrid can be relatively easily priced, i.e. sepa<strong>rate</strong>lyfor both underlyings (L 0 /L T and S T /S 0 ). The differencebetween the <strong>stochastic</strong> <strong>models</strong> becomes more pronouncedfor !40% since, then, the correlation plays a moreimportant role. The simulations performed for x,r ¼ 70% and x,r ¼ 70% show that the absolutedifference between the SZHW and HHW <strong>models</strong> becomessignificant for !4200%. The figure shows that, for small!, the prices <strong>of</strong> the SZHW and HHW <strong>models</strong> arerelatively close, whereas the Heston model gives lowerprices for !450%.5. ConclusionsIn this paper we have presented an extension <strong>of</strong> theScho¨bel–Zhu <strong>stochastic</strong> <strong>volatility</strong> model <strong>with</strong> a <strong>Hull</strong>–<strong>White</strong> <strong>interest</strong> <strong>rate</strong> process and evaluated it by means <strong>of</strong>pricing structured hybrid derivative products.The aim was to define a hybrid <strong>stochastic</strong> process thatbelongs to the class <strong>of</strong> affine jump-diffusion <strong>models</strong>, asthis may lead to efficient calibration <strong>of</strong> the model. Wehave shown that the so-called Scho¨bel–Zhu–<strong>Hull</strong>–<strong>White</strong>model belongs to the category <strong>of</strong> affine jump-diffusionprocesses. No restrictions regarding the choice <strong>of</strong>correlation structure between the different Wiener processesappearing need to be made.We also compared the model <strong>with</strong> the Heston–<strong>Hull</strong>–<strong>White</strong> hybrid model <strong>with</strong> an indirectly implied correlationbetween the equity and the <strong>interest</strong> <strong>rate</strong>. We found thatalthough the model is very attractive because <strong>of</strong> its squareroot <strong>volatility</strong> structure, it is unable to gene<strong>rate</strong> extremecorrelations.Due to the resulting semi-closed (for Scho¨bel–Zhu–<strong>Hull</strong>–<strong>White</strong>) and closed (Heston–<strong>Hull</strong>–<strong>White</strong>)characteristic functions we were able to calib<strong>rate</strong> the<strong>models</strong> in an efficient way by means <strong>of</strong> the Fourier-cosineexpansion pricing technique, adapted to a <strong>stochastic</strong><strong>interest</strong> <strong>rate</strong>.It has been shown by numerical experiments fordifferent hybrid products that under the same plainvanilla prices the extended <strong>stochastic</strong> <strong>volatility</strong> <strong>models</strong>give different prices than the Heston model.The present hybrid model cannot model a skew in the<strong>interest</strong> <strong>rate</strong>s, which will form part <strong>of</strong> our future work.AcknowledgementsThe authors would like to thank two anonymous refereesfor valuable suggestions that have significantly improvedthe paper. Moreover, the authors thank NataliaBorovykh and Roger Lord from RabobankInternational and Bin Chen and Fang Fang from theDelft University <strong>of</strong> Technology for helpful comments.ReferencesAlbanese, C. and Lawi, S., Laplace transforms for integrals <strong>of</strong>Markow processes. Working Paper, 2007. Available onlineat: http://arxiv.org/PS_cache/arxiv/pdf/0710/0710.1599v1.pdf(accessed 2008).Andersen, L.B.G., Efficient simulation <strong>of</strong> the Heston <strong>stochastic</strong><strong>volatility</strong> model. J. Comput. Finance, 2007, 11, 1–42.Arnold, L., Stochastic Differential Equations, Theory andApplications, 1973 (Wiley: New York).Brigo, D. and Mercurio, F., Interest Rate Models—Theory andPractice: With Smile, Inflation and Credit, 2nd ed., 2007(Springer: Berlin).Broadie, M. and Kaya, O¨ ., Exact simulation <strong>of</strong> <strong>stochastic</strong><strong>volatility</strong> and other affine jump diffusion processes. Oper.Res., 2006, 54, 217–231.Carr, P.P. and Madan, D.B., Option valuation using the fastFourier transform. J. Comput. Finance, 1999, 2, 61–73.