f(s)dw Solution 1. Approximate f by piece-wise constant left-continuous non-random functions f n such that (f(s) f n (s)) 2 ds 0.

Similar documents
Stochastic Modelling in Finance - Solutions to sheet 8

An Introduction to Backward Stochastic Differential Equations (BSDEs) PIMS Summer School 2016 in Mathematical Finance.

6. Stochastic calculus with jump processes

Utility maximization in incomplete markets

THE WAVE EQUATION. part hand-in for week 9 b. Any dilation v(x, t) = u(λx, λt) of u(x, t) is also a solution (where λ is constant).

Homework 10 (Stats 620, Winter 2017) Due Tuesday April 18, in class Questions are derived from problems in Stochastic Processes by S. Ross.

Section 3.5 Nonhomogeneous Equations; Method of Undetermined Coefficients

An Introduction to Malliavin calculus and its applications

A general continuous auction system in presence of insiders

arxiv: v1 [math.pr] 21 May 2010

Backward stochastic dynamics on a filtered probability space

1 1 + x 2 dx. tan 1 (2) = ] ] x 3. Solution: Recall that the given integral is improper because. x 3. 1 x 3. dx = lim dx.

Lecture #31, 32: The Ornstein-Uhlenbeck Process as a Model of Volatility

Optimal Investment Strategy Insurance Company

Hamilton- J acobi Equation: Weak S olution We continue the study of the Hamilton-Jacobi equation:

Chapter 2. First Order Scalar Equations

Uniqueness of solutions to quadratic BSDEs. BSDEs with convex generators and unbounded terminal conditions

Physics 127b: Statistical Mechanics. Fokker-Planck Equation. Time Evolution

EXERCISES FOR SECTION 1.5

Cash Flow Valuation Mode Lin Discrete Time

14 Autoregressive Moving Average Models

Transform Techniques. Moment Generating Function

(MS, ) Problem 1

On a Fractional Stochastic Landau-Ginzburg Equation

Hamilton- J acobi Equation: Explicit Formulas In this lecture we try to apply the method of characteristics to the Hamilton-Jacobi equation: u t

Macroeconomic Theory Ph.D. Qualifying Examination Fall 2005 ANSWER EACH PART IN A SEPARATE BLUE BOOK. PART ONE: ANSWER IN BOOK 1 WEIGHT 1/3

Hamilton Jacobi equations

1. Consider a pure-exchange economy with stochastic endowments. The state of the economy

Unit Root Time Series. Univariate random walk

Inventory Analysis and Management. Multi-Period Stochastic Models: Optimality of (s, S) Policy for K-Convex Objective Functions

Optimal Investment, Consumption and Retirement Decision with Disutility and Borrowing Constraints

Solutions to Assignment 1

MA 214 Calculus IV (Spring 2016) Section 2. Homework Assignment 1 Solutions

Numerical Approximation of Partial Differential Equations Arising in Financial Option Pricing

Math 2142 Exam 1 Review Problems. x 2 + f (0) 3! for the 3rd Taylor polynomial at x = 0. To calculate the various quantities:

t is a basis for the solution space to this system, then the matrix having these solutions as columns, t x 1 t, x 2 t,... x n t x 2 t...

Question 1: Question 2: Topology Exercise Sheet 3

Online Appendix to Solution Methods for Models with Rare Disasters

CHAPTER 2: Mathematics for Microeconomics

Chapter 14 Wiener Processes and Itô s Lemma. Options, Futures, and Other Derivatives, 9th Edition, Copyright John C. Hull

Example on p. 157

f t te e = possesses a Laplace transform. Exercises for Module-III (Transform Calculus)

t 2 B F x,t n dsdt t u x,t dxdt

A proof of Ito's formula using a di Title formula. Author(s) Fujita, Takahiko; Kawanishi, Yasuhi. Studia scientiarum mathematicarum H Citation

1 Solutions to selected problems

The Strong Law of Large Numbers

Lecture 20: Riccati Equations and Least Squares Feedback Control

Vanishing Viscosity Method. There are another instructive and perhaps more natural discontinuous solutions of the conservation law

Predator - Prey Model Trajectories and the nonlinear conservation law

The Optimal Stopping Time for Selling an Asset When It Is Uncertain Whether the Price Process Is Increasing or Decreasing When the Horizon Is Infinite

Essential Microeconomics : OPTIMAL CONTROL 1. Consider the following class of optimization problems

Loss of martingality in asset price models with lognormal stochastic volatility

Chapter 6. Systems of First Order Linear Differential Equations

MODULE 3 FUNCTION OF A RANDOM VARIABLE AND ITS DISTRIBUTION LECTURES PROBABILITY DISTRIBUTION OF A FUNCTION OF A RANDOM VARIABLE

Math 10B: Mock Mid II. April 13, 2016

MATH 4330/5330, Fourier Analysis Section 6, Proof of Fourier s Theorem for Pointwise Convergence

4 Sequences of measurable functions

Math Final Exam Solutions

An random variable is a quantity that assumes different values with certain probabilities.

Optimal Investment under Dynamic Risk Constraints and Partial Information

Differential Equations

Finish reading Chapter 2 of Spivak, rereading earlier sections as necessary. handout and fill in some missing details!

Comparison between the Discrete and Continuous Time Models

Representation of Stochastic Process by Means of Stochastic Integrals

E β t log (C t ) + M t M t 1. = Y t + B t 1 P t. B t 0 (3) v t = P tc t M t Question 1. Find the FOC s for an optimum in the agent s problem.

KINEMATICS IN ONE DIMENSION

Lecture 6: Wiener Process

Algorithmic Trading: Optimal Control PIMS Summer School

MATH 128A, SUMMER 2009, FINAL EXAM SOLUTION

LECTURE 1: GENERALIZED RAY KNIGHT THEOREM FOR FINITE MARKOV CHAINS

Properties Of Solutions To A Generalized Liénard Equation With Forcing Term

t + t sin t t cos t sin t. t cos t sin t dt t 2 = exp 2 log t log(t cos t sin t) = Multiplying by this factor and then integrating, we conclude that

Stochastic models and their distributions

Quadratic and Superquadratic BSDEs and Related PDEs

IMPLICIT AND INVERSE FUNCTION THEOREMS PAUL SCHRIMPF 1 OCTOBER 25, 2013

dt = C exp (3 ln t 4 ). t 4 W = C exp ( ln(4 t) 3) = C(4 t) 3.

Homework 4 (Stats 620, Winter 2017) Due Tuesday Feb 14, in class Questions are derived from problems in Stochastic Processes by S. Ross.

Random Walk on Circle Imagine a Markov process governing the random motion of a particle on a circular

ANSWERS TO EVEN NUMBERED EXERCISES IN CHAPTER 6 SECTION 6.1: LIFE CYCLE CONSUMPTION AND WEALTH T 1. . Let ct. ) is a strictly concave function of c

THE 2-BODY PROBLEM. FIGURE 1. A pair of ellipses sharing a common focus. (c,b) c+a ROBERT J. VANDERBEI

Echocardiography Project and Finite Fourier Series

2. Nonlinear Conservation Law Equations

Mean-Variance Hedging for General Claims

Math 334 Test 1 KEY Spring 2010 Section: 001. Instructor: Scott Glasgow Dates: May 10 and 11.

Local risk minimizing strategy in a market driven by time-changed Lévy noises. Lotti Meijer Master s Thesis, Autumn 2016

BU Macro BU Macro Fall 2008, Lecture 4

Class Meeting # 10: Introduction to the Wave Equation

and Applications Alexander Steinicke University of Graz Vienna Seminar in Mathematical Finance and Probability,

The Structure of General Mean-Variance Hedging Strategies

L p -L q -Time decay estimate for solution of the Cauchy problem for hyperbolic partial differential equations of linear thermoelasticity

ENGI 9420 Engineering Analysis Assignment 2 Solutions

Optimal Consumption and Investment Portfolio in Jump markets. Optimal Consumption and Portfolio of Investment in a Financial Market with Jumps

23.2. Representing Periodic Functions by Fourier Series. Introduction. Prerequisites. Learning Outcomes

where the coordinate X (t) describes the system motion. X has its origin at the system static equilibrium position (SEP).

On the Timing Option in a Futures Contract

Solutions from Chapter 9.1 and 9.2

GENERALIZATION OF THE DYBVIG INGERSOLL ROSS THEOREM AND ASYMPTOTIC MINIMALITY

1 st order ODE Initial Condition

Math Week 14 April 16-20: sections first order systems of linear differential equations; 7.4 mass-spring systems.

GMM - Generalized Method of Moments

Transcription:

Advanced Financial Models Example shee 3 - Michaelmas 217 Michael Tehranchi Problem 1. Le f : [, R be a coninuous (non-random funcion and W a Brownian moion, and le σ 2 = f(s 2 ds and assume σ 2 <. By considering he L 2 consrucion of he sochasic inegral, show ha f(sdw s is a normal random variable wih mean zero and variance σ 2. Soluion 1. Approximae f by piece-wise consan lef-coninuous non-random funcions f n such ha (f(s f n (s 2 ds. The exisence of such a sequence is a consequence of he densiy of sep funcions in L 2 (Leb. Bu for a concree consrucion, le MN f M,N = f( N i 11 ( N i 1, N i ] i=1 where N i = i/n. Noe (f(s f M,N (s 2 ds = M f(s 2 ds + M sup{ f(s f( : s, M, s 1 N }. For each n 1, by square-inegrabiliy of f we can find M n such ha M n f(s 2 ds < 1 n and by uniform coninuiy of f on he compac [, M n ] we can find N n such ha sup{ f(s f( : s, M n, s 1 N n } 1 nm n. Seing f n = f Mn,Nn does he job since Now f n (dw = i f( i 1 (W i W i 1 is he sum of independen mean-zero normal random variables, and hence is mean-zero normal wih variance f( i 1 2 ( i i 1 f( 2 d. i Since f n (dw f(dw in L 2 (P by he definiion of he sochasic inegral, we will be done once we appeal o he following sandard fac: Fac. Suppose X n N(µ n, σn 2 and µ n µ and σn 2 σ 2. If X n X in L 2 hen X N(µ, σ 2. 1

Proof. Since X n X in L 2, hen X n X in disribuion. Le F n and F be he disribuion funcions of X n and X, respecively. A he poins of coninuiy of coninuiy of F we have F (x = lim F n (x n ( x µn = lim Φ n σ n ( x µ = Φ σ by he coninuiy of he sandard normal disribuion funcion Φ. Problem 2. * (Ornsein Uhlenbeck process Le W be a Brownian moion, and le X = e a x + b e a( s dw s for some a, b, x R. (a Verify ha (X saisfies he following sochasic differenial equaion: dx = ax d + b dw, X = x. (b Show ha X N (e a x, b2 2a (e2a 1. (c Wha is he disribuion of he random variable X d? Soluion 2. (a Since we can apply Iô s formula (b Since X = e a (x + b dx = e a ( be a dw + (x + b = b dw + ax d e as dw s (e a( s 2 ds = e2a 1 2a e as dw s ae a d his par follows from Problem 1. (c Mehod 1: Noe ha by rearranging he sochasic differenial equaion we have X d = 1 a (X T x bw T 2

and hence X d is normally disribued wih mean (e at 1x/a. To compue he variance, firs noe ha ( Cov(X T, W T = Cov b e a(t, dw dw = b e a(t d = b a (eat 1 by Iô s isomery. Hence ( Var X d = 1 ( Var(XT 2b Cov(X a 2 T, W T + b 2 Var(W T Mehod 2: X d = = b2 2a 3 (e2at 4e at + 3 + 2aT. = e at x d + e at x d + = eat 1 x + a s e a( s b dw s d e a( s b d dw s e a(t s 1 b dw s a Hence X d is normally disribued wih mean (e at 1x/a and variance b 2 a 2 (e a(t s 1 2 ds = b2 2a 3 (e2at 4e at + 3 + 2aT This calculaion is useful in he sudy of he Vasicek ineres rae model. Problem 3. Le W be a Brownian moion. Show ha if Y = W 3 maringale (1 by hand, and (2 by Iô s formula. 3W hen Y is a Soluion 3. (1 By hand: Since Gaussian random variables have finie momens of all orders, Y is inegrable. Indeed, we have E( Y E( W 3 + 3E( W = C 3/2 < where C = 5 2/π. Therefore, using he independence of he incremens of W we have E(Y F s =E(W 3 3W F s =E[(W W s + W s 3 3(W W s + W s F s ] =E[(W W s 3 ] + 3E[(W W s 2 ]W s + 3E(W W s W 2 s + W 3 s 3E(W W s W s = + 3( sw s + + W 3 s + W s for s <. =Y s 3

(2 By Iô s rule: dy = d(w 3 3W = (3W 2 dw + 3W d 3( dw + W d = 3(W 2 dw ( and hence Y is a local maringale. Recall ha if E α2 s ds < for all hen ( he process α s dw s is a maringale. Again, i s clear ha he inegrand is square inegrable in his case since Guassian random variables have finie momens of all orders. Bu, jus o be explici, E [3(W 2 s s] 2 ds = 9 and hence (Y is a maringale. E(W 4 s 2W 2 s s + s 2 ds = 9 2s 2 ds = 6 3 < Problem 4. (Hea equaion Le W be a scalar Brownian moion, and le g : [, T ] R R be a smooh funcion ha saisfy he parial differenial equaion wih erminal condiion g + 1 2 g 2 x = 2 g(t, x = G(x. (a Show ha (g(, W [,T ] is a local maringale. (b If he funcion g is bounded, deduce he formula g(, x = G(x + T z e z2 /2 dz. (c Use Problem 2 o find explicily he unique bounded soluion o he PDE wih erminal condiion h + x h x + 1 2 2 h x = 2 h(t, x = cos x. Useful fac: If Z N(µ, σ 2 hen E(cos Z = e σ2 /2 cos µ. Soluion 4. (a By Iô s formula: ( g dg(, W = + 1 2 g d + g 2 x 2 x dw = g x dw and hence (g(, W [,T ] is a local maringale. 4

(b Recall a bounded local maringale is a rue maringale. In paricular, by he independence of he incremens of Brownian moion, we have g(, W = E[g(T, W T F ] = E[G(W T F ] = E[G(W + W T W F ] = G(W + T z e z2 /2 dz since W T W N(, T. Since he above formula holds idenically, we have he desired inegral represenaion of he soluion of he hea equaion. (c Le X be he Ornsein Uhlenbeck process dx = X d + dw. By Iô s formula, he process (h(, X [,T ] is a local maringale if and only if h is soluion o he PDE. Now if h is bounded, hen (h(, X [,T ] is a rue maringale. Using problem 2, we have h(, X = E[cos X T F ] [ ( = E cos e T X + e T s dw s F ] = cos(e T X exp( (e 2(T 1/4 since he condiional disribuion of X T given F is normal wih mean e T X and variance (e 2(T 1/2. In paricular, he unique soluion is h(, x = cos(e T x exp( 1 4 (e2(t 1 Problem 5. (Sricly local maringale This is a echnical exercise o exhibi a local maringale ha is no a rue maringale. Le W = (W 1, W 2, W 3 be a hree-dimensional Brownian moion and le u = (1,,. I is a fac ha P(W u for all = 1. (a Le X = W u 1. Use Iô s formula and Lévy s characerisaion of Brownian moion o show ha dx = X 2 dz, X = 1 where Z is a Brownian moion. In paricular, show ha X is a posiive local maringale. (b By direcly evaluaing he inegral or oherwise, show ha E(X = 2Φ( 1/2 1 for all >, where Φ is he disribuion funcion of a sandard normal random variable. Why does his imply ha X is a sricly local maringale? Soluion 5. (a Le f(x 1, x 2, x 3 = ((x 1 1 2 + x 2 2 + x 2 3 1/2 so ha ( f, f, f = [f(x 1, x 2, x 3 ] 3 (x 1 1, x 2, x 3 x 1 x 2 x 3 5

2 f x 2 1 + 2 f x 2 2 + 2 f x 2 3 = f 3 + 3f 5 (x 1 1 2 + f 3 + 3f 5 x 2 2 f 3 + 3f 5 x 2 3 =. In paricular, Iô s formula yields dx = X 3 [(W 1 1dW 1 + W 2 dw 2 + W 3 dw 3 ]. Since X can be wrien as a sochasic inegral of a hree dimensional Brownian moion, i is a local maringale. Now le Z be he local maringale such ha Z = and Since dz = X [(W 1 1dW 1 + W 2 dw 2 + W 3 dw 3 ]. d Z = X 2 [(W 1 1 2 + (W 2 2 + (W 3 2 ]d = d by consrucion, he process Z is a Brownian moion by Lévy s characerisaion heorem. (b Swich o spherical coordinaes: E(X = ( 3/2 e x2 1 /2 x2 2 /2 x2 3 /2 ( dx 1 dx 2 dx 3 x 1 1 2 + x 2 2 + x 2 3 = ( 3/2 = ( 1/2 π r= θ= π r= = ( 1/2 θ= φ= re r2 /2 r= r 2 sin θe r2 /2 r2 2 dφ dθ dr cos θ + 1 r 2 sin θe r2 /2 r2 2 dθ dr cos θ + 1 π r 2 2 cos θ + 1 θ= = ( 1/2 2(r1 {r> 1/2 } + r 2 1 {r }e 1/2 r2 /2 dr r= 1/2 e r2 /2 = 2 dr Noe ha E(X < X for all >, so X is a sricly local maringale. Problem 6. (sricly local maringales again (a Suppose ha X is posiive maringale wih X = 1. Fix T > and le dq dp = X T. Le Y = 1/X for all. Use Girsanov s heorem o show ha (Y T is a posiive maringale under Q. (b Coninuing from par (a, now suppose ha X has dynamics dx = X σ dw where W is a Brownian moion under P. Show ha here exiss a Q-Brownian moion Ŵ such ha dy = Y σ dŵ 6 dr

(c Le X be a posiive local maringale wih X = 1 and dynamics dx = X 2 dw. Our goal is o show ha X is a sricly local maringale. For he sake of finding a conradicion, suppose X is a rue maringale. In he noaion of pars (a and (b, show ha P(Y > = 1 bu Q(Y > = Φ( 1/2. Why does his conradic he assumpion ha X is a rue maringale? Soluion 6. (a Since P and Q are equivalen and P(X > for all = 1 hen Q(Y > for all = 1. Now o show ha Y is a Q-maringale, noe ha (b By Iô s formula, E Q (Y T F = EP (X T Y T F E Q (X T F = 1 X dy = dx 1 = X 2 dx + X 3 d X = Y σ (dw σ d Now by Girsanov s heorem, he process d ˇW = dw σ d defines a Q Brownian moion. And of course Ŵ = ˇW is a Brownian moion also. (c Now assuming X is a rue maringale, hen Girsanov s heorem applies and hence dy = Y σ dŵ = dŵ since σ = X. Hence Q(Y > = Q(Ŵ > 1 = Φ( 1/2 < 1. Bu since P(Y > = P(X > = 1. Therefore P and Q are no equivalen afer all. Problem 7. Consider a hree asse marke wih prices given by db B = 2 d ds (1 S (1 ds (2 S (2 Consruc an absolue arbirage. = 3 d + dw (1 2 dw (2 = 5 d 2 dw (1 + 4 dw (2. Soluion 7. If he pure invesmen sraegy is decomposed as (φ, π a good choice for he holding is sock is given by ( 2 1 π =, S (1 7 S (2

bu, of course, i is no unique. I remains o find he holding in he bank accoun φ. Noe ha he wealh X evolves as so he unique soluion wih X = is dx = r(x π S d + π ds = (2X + 5d X = 5 2 (e2 1. Since X > a.s. for >, his is an arbirage wih he holding in he bank accoun given by φ = X π S B = 5 11e 2 2B. Problem 8. * Consider a Black Scholes marke wih wo asses wih dynamics given by db = B r d ds = S (µ d + σdw Find a replicaing sraegy H and he associaed wealh process for a claim wih payou (1 ξ T = S p T for some p R (2 ξ T = (log S T 2 (3 ξ T = S sds Show ha your answer o par (3 is unchanged if we only assume ha S is a posiive Iô process. Soluion 8. Noe ha S T = S e (r σ2 /2(T +σ(ŵt Ŵ where Ŵ = W + (µ r/σ defines a Brownian moion for he equivalen maringale measure Q. (1 We could solve he Black Scholes PDE wih boundary condiion V (T, S = S p, bu i is easier o compue condiional expecions. so ha The hedging porfolio is (2 Similarly, since ξ = E Q [e r(t S p T F ] = S p e (p 1(r+pσ2 /2(T V (, S = S p e (p 1(r+pσ2 /2(T. π = V S (, S = ps p 1 e (p 1(r+pσ2 /2(T. ξ = E Q [e r(t (log S T 2 F ] = e r(t {[log S + (r σ 2 /2(T ] 2 + σ 2 (T } we have (3 Finally, ξ = E Q [ e r(t π = 2e r(t [log S + (r σ 2 /2(T ]/S. ] S s ds F = e r(t 8 r(t 1 e S s ds + S. r

Unforunaely, he payou is no of he form g(s T so our heorem for calculaing he replicaion porfolio doesn apply. Sill, le and π = φ = ξ π S B 1 e r(t r = 1 B e rt S s ds. Noice ha φ B + π S = ξ by consrucion and ha by a calculaion dξ = φ db + π ds, so (φ, π is a self-financing replicaion sraegy. To check he self-financing condiion we did no need o use he specific dynamics of S. Indeed, we need only assume ha we can apply Iô s formula. Problem 9. (Black Scholes formula Le X N(, 1 be a sandard normal random variable, and v and m be posiive consans. Express he expecaion in erms of Φ, he disribuion funcion of X. Soluion 9. E[(e v/2+ vx m + ] = = = = F (v, m = E[(e v/2+ vx m + ] (e v/2+vx m + /2 e x2 dx log m/ v+ v/2 log m/ v+ v/2 log m/ v+ v/2 = Φ ( log m v + (e v/2+vx /2 m e x2 dx e v/2+ vx x 2 /2 dx m e s2 /2 ds m v 2 log m/ v+ v/2 log m/ v v/2 ( m Φ log m v v 2 e x2 /2 dx e 2 /2 d Problem 1. (sricly local maringale in finance Consider a marke wih zero ineres rae r = and sock price wih dynamics ds = S 2 dw. Consider a European claim wih payou ξ T = S T. (a Show ha here exiss a rading sraegy which replicaes he claim wih corresponding wealh ξ = V (, S where V (, S = S [ ( 2Φ 1 S T ] 1. (b Consider he sraegy of buying S claims and selling ξ shares. The ime wealh is V = and he ime T wealh is V T = (S ξ S T >. Is his sraegy an absolue arbirage? 9

Soluion 1. (a I is sraigh-forward, if a bi edious, o verify V + 1 2 V 2 S4 S = 2 and lim T V (, S = S. The replicaion sraegy is given by π = V (, S S as usual. (b This candidae arbirage is no admissible. Indeed, le V = S ξ ξ S. Noe ha V is a local maringale, as i is he linear combinaion of wo local maringales. Now if he sraegy were admissible, he wealh V would be bounded from below and hence a supermaringale. Therefore we would have o conclude = V E(V T = E(S ξ T ξ S T = (S ξ E(S T Bu his conradics S > ξ. So in his marke i is impossible o lock in he sure fuure profi a zero iniial cos, because doing so leaves open he possibiliy ha he wealh is goes negaive beween imes = and = T. Noe, however, ha here is an admissible arbirage relaive o he asse wih price S. Indeed, he sraegy of holding one share of he claim is a relaive arbirage, since he iniial discouned wealh is ξ /S < 1 and he erminal discouned wealh is ξ T /S T = 1. This is only a relaive arbirage since you do no shor S, and hence mus sar wih posiive iniial wealh. 1