MSM 3G11/4G11 Exercise Sheet 4. Solutions 1. The Black Scholes equation is given by ∂V 1 ∂ 2V ∂V + σ 2 S 2 2 + rS − rV = 0. ∂t 2 ∂S ∂S We wish to change from independent variables (S, t) to (x, τ ) with S = Eex , t = T − σ2τ2 , 2 2 ∂x ∂τ = 0, ∂S = S1 , ∂τ = − σ2 and ∂S = 0. i.e., x = log(S/E), τ = σ2 (T − t). It follows that ∂x ∂t ∂t Using the chain rule ∂V ∂x ∂V ∂τ 1 ∂V ∂V = + = − σ2 ∂t ∂x ∂t ∂τ ∂t 2 ∂τ and ∂V ∂x ∂V ∂τ 1 ∂V ∂V = + = . ∂S ∂x ∂S ∂τ ∂S S ∂x Hence, ∂ 2V 1 ∂V 1 ∂V ∂ 1 ∂ 2V = − = + . ∂S 2 ∂S S ∂x S 2 ∂x S 2 ∂x2 This gives 1 ∂V 1 ∂V 1 2 2 1 ∂ 2V 1 2 ∂V + σ S − 2 + 2 2 + rS − rV = 0 − σ 2 ∂τ 2 S ∂x S ∂x S ∂x or ∂V ∂V ∂ 2V ∂V + − −k + kV = 0. 2 ∂τ ∂x ∂x ∂x If (k − 1)x (k + 1)2 τ V = Eu(x, τ ) exp − − = u(x, τ )F (x, τ ), 2 4 (k−1)x (k+1)2 τ with F = E exp − 2 − 4 , then ∂V ∂u (k + 1)2 = F+ − uF , ∂τ ∂τ 4 ∂u (k − 1) ∂V = F− uF ∂x ∂x 2 and ∂ 2u ∂u (k − 1)2 ∂ 2V = F − (k − 1) F + uF. ∂x2 ∂x2 ∂x 4 Now divide through by F to give ∂u (k + 1)2 (k − 1) (k − 1)2 k(k − 1) ∂u ∂ 2u +u − − − + + k + (1+(k −1)−k)− 2 = 0 ∂τ 4 2 4 2 ∂x ∂x or ∂ 2u ∂u = . ∂τ ∂x2 1 2. If the pay-off function of an option is V (S, T ) = g(S) we can use the transformation (k − 1)x (k + 1)2 τ V (S, t) = Eu(x, τ ) exp − − 2 4 to calculate V (S, T ) by noting τ = 0 when t = T : (k − 1)x . g(S) = Eu(x, 0) exp − 2 Denoting u0 (x) = u(x, 0) we have Hence, g(S = Eex ) exp u0 (x) = E 1 (k − 1)x . 2 g (S = Ees ) u0 (s) = exp E 1 (k − 1)s . 2 (a) Vanilla call: g(S) = max(S − E, 0) and u0 (s) = (es − 1, 0) exp = max exp 1 (k 2 1 (k 2 − 1)s + 1)s − exp 1 (k 2 − 1)s , 0 . (b) Vanilla put: g(S) = max(E − S, 0) and 1 1 (k − 1)s − exp (k + 1)s , 0 . u0 (s) = max exp 2 2 (c) Bullish vertical spread: g(S) = max(S − E1 , 0) − max(S − E2 , 0) Set E = E1 . Then, 1 1 u0 (s) = max exp (k + 1)s − exp (k − 1)s , 0 2 2 E2 1 1 (k + 1)s − (k − 1)s , 0 . exp − max exp 2 E1 2 (d) Cash or nothing call: g(S) = BH(S − E) and B u0 (s) = H (Ees − E) exp E 1 (k − 1)s . 2 (e) Supershare: g(S) = d1 (H(S − E) − H(S − E − d)) and 1 (H (Ees − E) − H (Ees − E − d)) exp u0 (s) = Ed 2 1 (k − 1)s . 2 (f) Asset or nothing call: g(S) = SH(S − E) and s s u0 (s) = e H (Ee − E) exp 1 (k − 1)s . 2 3. This is an optional question. The full solution is available on pages 78-79 of the recommended text and involves tedious but elementary integration and algebraic manipulation. R∞ 4. The delta function can be defined by δ(x) = 0 if x 6= 0 and −∞ δ(x)dx = 1. Other R∞ properties of the delta function are −∞ δ(x)F (x)dx = F (0) and if the solutions of the equation h(x) = 0 are x = x1 , x2 , . . . , xk and h′ (xi ) 6= 0, then Z Here we have g(S) = 1 d ∞ δ(h(x))F (x)dx = −∞ k X i=1 1 |h′ (x i )| F (xi ). (1) (H(S − E) − H(S − E − d)). Graphically we have V 1/d S E E+d R∞ For all d it is clear that −∞ g(S)ds = 1. Also it is intuitive that in the limit as d → 0+ , g(S) = 0 except at S = E, where it is infinity. Hence, g(S) = δ(S − E). 5. We calculate the value of the supershare in the limit as d → 0 by writing g(S) = δ(S −E). This gives u0 (s) = E1 δ(S − E) exp 12 (k − 1)s = 1 δ(E E (es − 1)) exp 1 (k 2 − 1)s . Using the solution to the diffusion equation we have Z ∞ 1 1 1 1 s 2 δ(E (e − 1)) exp (k − 1)s exp − (x − s) ds. u(x, τ ) = √ 2 4τ 2 τ π −∞ E 3 We use (1). The only root of E(es − 1) is s = 0. Also u(x, τ ) = d (E(es ds − 1))|s=0 = E. Hence, 2 1 1 − x4τ √ e . E2 2 τ π 6. As derived during lectures the appropriate form of the Black-Scholes equation for a European vanilla call option where the underlying S has a constant dividend yield D0 is ∂C 1 2 2 ∂ 2 C ∂C σ S + − rC + (r − D0 )S = 0. 2 2 ∂S ∂t ∂S (2) Our goal is to use the substitution C(S, t) = e−D0 (T −t) C1 (S, t) to transform the above into an equation which we already know the answer to. Note that at the exercise time t = T we have C(S, T ) = C1 (S, T ) = max(S − E, 0) (i.e C and C1 both have the same pay-off function as a standard European vanilla call). We now calculate ∂C ∂C1 = e−D0 (T −t) , ∂S ∂S 2 ∂ 2C −D0 (T −t) ∂ C1 = e ∂S 2 ∂S 2 and ∂C1 ∂C = D0 e−D0 (T −t) C1 + e−D0 (T −t) . ∂t ∂t Substituting into equation (2) we obtain ∂C1 1 2 2 −D0 (T −t) ∂ 2 C1 + σ S e ∂t 2 ∂S 2 ∂C1 − re−D0 (T −t) C1 = 0 +(r − D0 )Se−D0 (T −t) ∂S D0 e−D0 (T −t) C1 + e−D0 (T −t) Dividing through by e−D0 (T −t) and combining like terms gives ∂C1 1 2 2 ∂ 2 C1 ∂C1 σ S + − (r − D0 )C1 + (r − D0 )S = 0. 2 2 ∂S ∂t ∂S or 1 2 2 ∂ 2 C1 ∂C1 ∂C1 σ S − r ∗ C1 + r ∗ S = 0. + 2 2 ∂S ∂t ∂S where r∗ = r − D0 . This equation must be solved subject to the pay-off function C1 (S, T ) = max(S − E, 0). Note that is identically our standard Black Scholes equation for a European vanilla call except that r has been replaced by r∗ = r − D0 . Thus we can immediately write C1 = SN (d1 ) − E exp(−r∗ (T − t))N (d2 ) 4 where 1 N (q) = √ 2π Z q 1 2 exp − s ds, 2 −∞ 2 ln(S/E) + r∗ + σ2 (T − t) √ d1 = σ T −t and σ2 ∗ ln(S/E) + r − 2 (T − t) √ d2 = . σ T −t Our final answer is then C(S, t) = e−D0 (T −t) C1 = e−D0 (T −t) (SN (d1 ) − E exp(−r∗ (T − t))N (d2 )) 7. Substitute numbers directly into the formula from question 6 to give r∗ = 0.1, d1 = ln(1.3) + (0.1 + 0.08)(4) √ = 1.227959..., 0.4 4 d2 = 0.4279553306 and C(S, t) = exp(−D0 (T − t))(SN (d1 ) − E exp(−r∗ (T − t))N (d2 )) = 0.4766944475. 8. (a) By arbitrage we have and − − S t+ d = S td − dy S td (3) + − + (4) Cd S t− d , td = Cd S td , td + − − as re= Cd S t− Combining (3) and (4) gives Cd S t− d − dy S td , td d , td quired. (b) Since there is no dividend payment between t+ d ≤ t ≤ T then there will be no dividend jump condition and our lognormal random variable behaves in the usual way. It follows that Cd (S, t) = C(S, t; E) for all t+ d ≤ t ≤ T (i.e., Cd (S, t) and C(S, t; E) satisfies the same PDE and boundary conditions for t+ d ≤ t ≤ T ). (c) This is just the jump condition from (a). (d) The pay-off function for C(S − dy S, t; E) is simply E g(S) = max((S − dy S − E), 0) = (1 − dy ) max S − ,0 1 − dy by factoring out the term 1 − dy . 5 (e) We now combine the above results to give − − Cd S td , td = max S − E , 0 (1 − dy ). 1 − dy We treat this as a new pay-off function for Cd evaluated at time t = t− d , so for − 0 ≤ t ≤ td we value Cd by determining the standard value of a call option with E strike price 1−d and expiry date t− d and then multiply by a factor 1 − dy . y 9. If F = S(t) exp(r(T − t)) and V (S, t′ ) = S(t′ ) − F exp(−r(T − t′ )), then V (S, t′ ) = S(t′ ) − S(t) exp(r(T − t)) exp(−r(T − t′ )) = S(t′ ) − S(t) exp(r(t′ − t)). Thus (i) V (S, t) = 0, so it costs nothing to enter a forward contract. (ii) V (S, T ) = S(T ) − S(t) exp(r(T − t)) = S(T ) − F . The holder of the forward contact is obliged to purchase S at price F . If S(T ) > F the holder makes a profit. If S(T ) < F the holder makes a loss. 10. We begin by noting that the value of an option V (S, t) which only depends on the value of the underlying stock S and time t satisfies the Black Scholes equation ∂V ∂V 1 2 2 ∂ 2V σ S + − rV + rS = 0. 2 2 ∂S ∂t ∂S Our goal is to transform the Black-Scholes equation for the value of an option V (S, t) into a suitable form to determine the value of a call option C(F, t) which is a function of the forward price F and time t. To answer this question we must use the chain rule for a function of two variables. Suppose V (S, t) = C(F (S, t), τ (S, t)). We then use the chain rule to give ∂V ∂C ∂F ∂C ∂τ = + ∂S ∂F ∂S ∂τ ∂S and ∂C ∂F ∂C ∂τ ∂V = + . ∂t ∂F ∂t ∂τ ∂t In our case we have F (S, t) = Ser(T −t) and τ (S, t) = t which gives ∂F = er(T −t) ∂S ∂F = −rSer(T −t) ∂t ∂τ =0 ∂S 6 ∂τ = 1. ∂t This gives ∂V ∂C = er(T −t) ∂S ∂F and We now calculate ∂C ∂C ∂V = −rSer(T −t) + . ∂t ∂F ∂τ ∂ ∂V ∂ ∂ 2V r(T −t) ∂C = = e ∂S 2 ∂S ∂S ∂S ∂F 2 2 ∂ C ∂F ∂ 2 C ∂τ 2r(T −t) ∂ C = er(T −t) = e + ∂F 2 ∂S ∂F ∂τ ∂S ∂F 2 Substituting for all of the derivatives calculated using the chain rule, and using S = F e−r(T −t) the Black-Scholes equation becomes 2 ∂ 2C ∂C 1 2 σ F e−r(T −t) e2r(T −t) 2 + −rF e−r(T −t) er(T −t) 2 ∂F ∂F ∂C ∂C − rC + rF e−e(T −t) er(T −t) = 0. + ∂τ ∂F Combining all like terms and noting that τ = t we arrive at the desired result 1 2 2 ∂ 2 C ∂C σ F + − rC = 0. 2 ∂F 2 ∂t Recall that the Black-Scholes equation with continuous dividends is ∂V ∂V 1 2 2 ∂ 2V σ S + − rV + (r − D0 )S = 0. 2 2 ∂S ∂t ∂S Thus our partial differential equation for a Euorpean Call option on a future with a forward price F is identical to the Black-Scholes equations for an option on a stock S which pays continuous dividends for the case when r = D0 (i.e. in the above replace V with C, F with S and let r = D0 ). From question 6 we know how to determine the value of a European Call option for a stock which pays continuous dividends. We then use this result to determine the value of our call option on a futures contract to give C(F, t) = e−r(T −t) (F N (d1 ) − EN (d2 )) where 1 N (q) = √ 2π Z q 1 2 exp − s ds, 2 −∞ 2 ln(F/E) + σ2 (T − t) √ d1 = σ T −t 7 and 2 ln(F/E) − σ2 (T − t) √ d2 = . σ T −t 8