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

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

Stochastic Modelling in Finance - Solutions to sheet 8

4.6 One Dimensional Kinematics and Integration

Differential Equations

Chapter 2. First Order Scalar Equations

Unit Root Time Series. Univariate random walk

6. Stochastic calculus with jump processes

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

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

Elements of Stochastic Processes Lecture II Hamid R. Rabiee

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.

Some Basic Information about M-S-D Systems

EXERCISES FOR SECTION 1.5

I. Return Calculations (20 pts, 4 points each)

Loss of martingality in asset price models with lognormal stochastic volatility

( ) ( ) if t = t. It must satisfy the identity. So, bulkiness of the unit impulse (hyper)function is equal to 1. The defining characteristic is

Solutions to Assignment 1

Diebold, Chapter 7. Francis X. Diebold, Elements of Forecasting, 4th Edition (Mason, Ohio: Cengage Learning, 2006). Chapter 7. Characterizing Cycles

dy dx = xey (a) y(0) = 2 (b) y(1) = 2.5 SOLUTION: See next page

d 1 = c 1 b 2 - b 1 c 2 d 2 = c 1 b 3 - b 1 c 3

Nature Neuroscience: doi: /nn Supplementary Figure 1. Spike-count autocorrelations in time.

An Introduction to Malliavin calculus and its applications

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.

GEM4 Summer School OpenCourseWare

Challenge Problems. DIS 203 and 210. March 6, (e 2) k. k(k + 2). k=1. f(x) = k(k + 2) = 1 x k

Module 2 F c i k c s la l w a s o s f dif di fusi s o i n

2.3 SCHRÖDINGER AND HEISENBERG REPRESENTATIONS

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

Physics 235 Chapter 2. Chapter 2 Newtonian Mechanics Single Particle

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...

ψ(t) = V x (0)V x (t)

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

7 The Itô/Stratonovich dilemma

System of Linear Differential Equations

CHAPTER 2: Mathematics for Microeconomics

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

Online Appendix to Solution Methods for Models with Rare Disasters

Course Notes for EE227C (Spring 2018): Convex Optimization and Approximation

ACE 562 Fall Lecture 8: The Simple Linear Regression Model: R 2, Reporting the Results and Prediction. by Professor Scott H.

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

KEY. Math 334 Midterm I Fall 2008 sections 001 and 003 Instructor: Scott Glasgow

Option pricing and implied volatilities in a 2-hypergeometric stochastic volatility model

Vehicle Arrival Models : Headway

13.3 Term structure models

Section 3.5 Nonhomogeneous Equations; Method of Undetermined Coefficients

WEEK-3 Recitation PHYS 131. of the projectile s velocity remains constant throughout the motion, since the acceleration a x

Module 3: The Damped Oscillator-II Lecture 3: The Damped Oscillator-II

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).

Chapter 5. Heterocedastic Models. Introduction to time series (2008) 1

Chapter 6. Systems of First Order Linear Differential Equations

2.1: What is physics? Ch02: Motion along a straight line. 2.2: Motion. 2.3: Position, Displacement, Distance

HOMEWORK # 2: MATH 211, SPRING Note: This is the last solution set where I will describe the MATLAB I used to make my pictures.

DEPARTMENT OF ECONOMICS AND FINANCE COLLEGE OF BUSINESS AND ECONOMICS UNIVERSITY OF CANTERBURY CHRISTCHURCH, NEW ZEALAND

Excel-Based Solution Method For The Optimal Policy Of The Hadley And Whittin s Exact Model With Arma Demand

Solutionbank Edexcel AS and A Level Modular Mathematics

u(x) = e x 2 y + 2 ) Integrate and solve for x (1 + x)y + y = cos x Answer: Divide both sides by 1 + x and solve for y. y = x y + cos x

IMA Preprint Series # 2056

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

14 Autoregressive Moving Average Models

Predator - Prey Model Trajectories and the nonlinear conservation law

R t. C t P t. + u t. C t = αp t + βr t + v t. + β + w t

ME 391 Mechanical Engineering Analysis

Announcements: Warm-up Exercise:

ADVANCED MATHEMATICS FOR ECONOMICS /2013 Sheet 3: Di erential equations

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

Financial Econometrics Introduction to Realized Variance

A general continuous auction system in presence of insiders

Lecture 4 Kinetics of a particle Part 3: Impulse and Momentum

Basic Circuit Elements Professor J R Lucas November 2001

Laplace transfom: t-translation rule , Haynes Miller and Jeremy Orloff

From Particles to Rigid Bodies

4.1 - Logarithms and Their Properties

10. State Space Methods

MEI STRUCTURED MATHEMATICS 4758

SOLUTIONS TO ECE 3084

LAPLACE TRANSFORM AND TRANSFER FUNCTION

Solutions Problem Set 3 Macro II (14.452)

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

2. Nonlinear Conservation Law Equations

5.1 - Logarithms and Their Properties

Y 0.4Y 0.45Y Y to a proper ARMA specification.

Math 2214 Solution Test 1B Fall 2017

ACE 562 Fall Lecture 4: Simple Linear Regression Model: Specification and Estimation. by Professor Scott H. Irwin

Section 7.4 Modeling Changing Amplitude and Midline

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

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

1 st order ODE Initial Condition

CHEMICAL KINETICS: 1. Rate Order Rate law Rate constant Half-life Temperature Dependence

Linear Response Theory: The connection between QFT and experiments

Empirical Process Theory

k B 2 Radiofrequency pulses and hardware

!!"#"$%&#'()!"#&'(*%)+,&',-)./0)1-*23)

Non-uniform circular motion *

THE MYSTERY OF STOCHASTIC MECHANICS. Edward Nelson Department of Mathematics Princeton University

Class Meeting # 10: Introduction to the Wave Equation

= ( ) ) or a system of differential equations with continuous parametrization (T = R

Lecture 4: Processes with independent increments

References are appeared in the last slide. Last update: (1393/08/19)

Designing Information Devices and Systems I Spring 2019 Lecture Notes Note 17

Licenciatura de ADE y Licenciatura conjunta Derecho y ADE. Hoja de ejercicios 2 PARTE A

Transcription:

Saisics 441 (Fall 214) November 19, 21, 214 Prof Michael Kozdron Lecure #31, 32: The Ornsein-Uhlenbeck Process as a Model of Volailiy The Ornsein-Uhlenbeck process is a di usion process ha was inroduced as a model of he velociy of a paricle undergoing Brownian moion We know from Newonian physics ha he velociy of a (classical) paricle in moion is given by he ime derivaive of is posiion However, if he posiion of a paricle is described by Brownian moion, hen he ime derivaive does no exis The Ornsein-Uhlenbeck process is an aemp o overcome his di culy by modelling he velociy direcly Furhermore, jus as Brownian moion is he scaling limi of simple random walk, he Ornsein-Uhlenbeck process is he scaling limi of he Ehrenfes urn model which describes he di usion of paricles hrough a permeable membrane In recen years, however, he Ornsein-Uhlenbeck process has appeared in finance as a model of he volailiy of he underlying asse price process Suppose ha he price of a sock {S, } is modelled by geomeric Brownian moion wih volailiy and drif µ so ha S saisfies he SDE ds = S db + µs d However, marke daa indicaes ha implied volailiies for di eren srike prices and expiry daes of opions are no consan Insead, hey appear o be smile shaped (or frown shaped) Perhaps he mos naural approach is o allow for he volailiy funcion of ime so ha S saisfies he SDE ds = ()S db + µs d () o be a deerminisic This was already suggesed by Meron in 1973 Alhough i does explain he di eren implied volailiy levels for di eren expiry daes, i does no explain he smile shape for di eren srike prices Insead, Hull and Whie in 1987 proposed o use a sochasic volailiy model where he underlying sock price {S, } saisfies he SDE and he variance process {v, ds = p v S db + µs d } is given by geomeric Brownian moion dv = c 1 v db + c 2 v d wih c 1 and c 2 known consans The problem wih his model is ha geomeric Brownian moion ends o increase exponenially which is an undesirable propery for volailiy Marke daa also indicaes ha volailiy exhibis mean-revering behaviour This lead Sein and Sein in 1991 o inroduce he mean-revering Ornsein-Uhlenbeck process saisfying dv = db + a(b v )d where a, b, and are known consans This process, however, allows negaive values of v 2 1

In 1993 Heson overcame his di culy by considering a more complex sochasic volailiy model Before invesigaing he Heson model, however, we will consider he Ornsein- Uhlenbeck process separaely and prove ha negaive volailiies are allowed hereby verifying ha he Sein and Sein sock price model is flawed We say ha he process {X, } is an Ornsein-Uhlenbeck process if X saisfies he Ornsein-Uhlenbeck sochasic di erenial equaion given by dx = db + ax d (21) where and a are consans and {B, } is a sandard Brownian moion Remark Someimes (21) is called he Langevin equaion, especially in physics conexs Remark The Ornsein-Uhlenbeck SDE is very similar o he SDE for geomeric Brownian moion; he only di erence is he absence of X in he db erm of (21) However, his sligh change makes (21) more challenging o solve The rick for solving (21) is o muliply boh sides by he inegraing facor e a and o compare wih d(e a X ) The chain rule ells us ha d(e a X )=e a dx + X d(e a )=e a dx ae a X d (22) and muliplying (21) by e a gives so ha subsiuing (23) ino (22) gives e a dx = e a db + ae a X d (23) d(e a X )= e a db + ae a X d ae a X d = e a db Since d(e a X )= e a db,wecannowinegraeoconcludeha and so e a X X = X = e a X + e as db s e a( s) db s (24) Observe ha he inegral in (24) is a Wiener inegral Definiion 81 ells us ha e a( s) db s N, e ( s) ds = N, e 1 In paricular, choosing X = x o be consan implies ha X = e a x + e a( s) db s N xe a, 2 (e 1) 2 2

Acually, we can generalize his slighly If we choose X N(x, 2 )independenlyof {B, }, henexercise312ellsusha X = e a X + e a( s) db s N xe a, 2 e 2 (e 1) + = N xe a, 2 2 + e 2 Exercise 21 Suppose ha {X, } is an Ornsein-Uhlenbeck process given by (24) wih X = Ifs<,compueCov(X s,x ) We say ha he process {X, saisfies he SDE } is a mean-revering Ornsein-Uhlenbeck process if X dx = db +(b X )d (25) where and b are consans and {B, } is a sandard Brownian moion The rick for solving he mean-revering Ornsein-Uhlenbeck process is similar Tha is, we muliply by e and compare wih d(e (b X )) The chain rule ells us ha and muliplying (25) by e gives so ha subsiuing (27) ino (26) gives d(e (b X )) = e dx + e (b X )d (26) e dx = e db + e (b X )d (27) d(e (b X )) = e db e (b X )d + e (b X )d = e db Since d(e (b X )) = e db,wecannowinegraeoconcludeha and so e (b X ) (b X )= X =(1 e )b + e X + e s db s Exercise 22 Suppose ha X N(x, 2 )isindependenof{b, disribuion of X given by (28) e s db s (28) } Deerminehe Exercise 23 Use an appropriae inegraing facor o solve he mean-revering Ornsein- Uhlenbeck SDE considered by Sein and Sein, namely dx = db + a(b X )d Assuming ha X = x is consan, deermine he disribuion of X and conclude ha P{X < } > forevery> Hin: X has a normal disribuion This hen explains our earlier claim ha he Sein and Sein model is flawed 2 3

As previous noed, Heson inroduced a sochasic volailiy model in 1993 ha overcame his di culy Assume ha he asse price process {S, } saisfies he SDE ds = p v S db (1) + µs d where he variance process {v, } saisfies dv = p v db (2) + a(b v )d (29) and he wo driving Brownian moions {B (1), } and {B (2), } are correlaed wih rae, ie, dhb (1),B (2) i = d The p v erm in (29) is needed o guaranee posiive volailiy when he process ouches zero he sochasic par becomes zero and he non-sochasic par will push i up The parameer a measures he speed of he mean-reversion, b is he average level of volailiy, and is he volailiy of volailiy Marke daa suggess ha he correlaion rae is ypically negaive The negaive dependence beween reurns and volailiy is someimes called he leverage e ec Heson s model involves a sysem of sochasic di erenial equaions The key ool for analyzing such a sysem is he mulidimensional version of Iô s formula Theorem 24 (Version V) Suppose ha {X, } and {Y, } are di usions defined by he sochasic di erenial equaions and dx = a 1 (, X,Y )db (1) dy = a 2 (, X,Y )db (2) + b 1 (, X,Y )d + b 2 (, X,Y )d, respecively, where {B (1), } and {B (2), } are each sandard one-dimensional Brownian moions If f 2 C 1 ([, 1)) C 2 (R 2 ), hen df(, X,Y )= f(, X,Y )d + f 1 (, X,Y )dx + 1 2 f 11(, X,Y )dhxi where he parial derivaives are defined as + f 2 (, X,Y )dy + 1 2 f 22(, X,Y )dhy i + f 12 (, X,Y )dhx, Y i f(, x, y) = @ f(, x, y), @ f 1(, x, y) = @ @x f(, x, y), f 11(, x, y) = @2 f(, x, y) @x2 f 2 (, x, y) = @ @y f(, x, y), f 22(, x, y) = @2 @y f(, x, y), f 12(, x, y) = @2 f(, x, y), 2 @x@y and dhx, Y i is compued according o he rule dhx, Y i =(dx )(dy )=a 1 (, X,Y )a 2 (, X,Y )dhb (1),B (2) i 2 4

Remark In a ypical problem involving he mulidimensional version of Iô s formula, he quadraic covariaion process hb (1),B (2) i will be specified However, wo paricular examples are worh menioning If B (1) = B (2),hendhB (1),B (2) i =d, whereasifb (1) and B (2) are independen, hen dhb (1),B (2) i = Exercise 25 Suppose ha f(, x, y) = xy Using Version V of Iô s formula (Theorem 24), verify ha he produc rule for di usions is given by d(x Y )=X dy + Y dx +dhx, Y i Thus, our goal in he nex few lecures is o price a European call opion assuming ha he underlying sock price follows Heson s model of geomeric Brownian moion wih a sochasic volailiy, namely 8 >< ds = p v S db (1) + µs d, dv = p v db (2) + a(b v )d, >: dhb (1),B (2) i = d 2 5