Mauricio Bedoya javierma36@gmail.com August 2014 To understand the title of this blog, you must know: 1. What is a convex function ? 2. Jensens Inequality From Wikipedia, we have that a function f(x) is convex if: x 1 , x 2 X, [0, 1] : f((1 ) x 1 + x 2 ) (1 ) f(x 1 ) + f(x 2 ) (1) for = 0,3, x 1 = 2, x 2 = 3 and f(x) = x 2 ; you can prove that the previous inequality remains. When we plot the pay-o of a European Call option at maturity, we found that the function Pay-o at maturity is convex. Then, we can dene: e r(Tt) = (2) x 1 = 0 (3) g(S (T) ) = (S (T) K) + = (S (T) K) I (S (T) K) 1 (4) Replacing equation 2, 3 and 4 in equation 1, and taking expectations 2 , we found 1 Option pay-o at maturity of a Call. 2 Risk neutral world. 1 E[g(e r(Tt) S (T) )|F (t) ]
Here use Jensens Inequality E[e r(Tt) g(S (T) )|F (t) ] (5) Now, lets remember what does Jensens Inequality state (Wikipedia) Jensens Inequality If S (T) is random and g(S (T) ) is convex (option pay-o), then: g(E[S (T) ]) E[g(S (T) )] (6) replacing 6 in 5 E[g(e r(Tt) S (T) )|F (t) ] g(E[e r(Tt) S (T) )|F (t) ]) g(e rt E[e rT S (T) |F (t) ]) g(e rt E[e rT S (t) e (r0,5 2 )(Tt)+(w(T)w(t)) |F (t) ]) g(e rt S (t) e rt E[e 0,5 2 (Tt)+(w(T)w(t)) |F (t) ]) g(S (t) ) (7) I hope that you dont get lost with the implementation of the previous equation. In step number 4 of the previous equation we have the expectation of the exponential Martingala which is equal to 1 (ask if any doubt). Replacing equation 7 in 5 g(S (t) ) E[e r(Tt) g(S (T) )|F (t) ] t [0, T] (8) operating (taking what is known till t); e rt g(S (t) )
American option Pay-o E[e r(T) g(S (T) )|F (t) ]
European Option Pay-o t [0, T] (9) This means that the European Option Pay-o dominate the American Option Pay-o. 2