Koc2(dba)

1. Lectures on Lévy Processes and Stochastic Calculus (Koc University) Lecture 2: Lévy Processes David Applebaum School of Mathematics and Statistics, University of Shefﬁeld, UK 6th December 2011 Dave Applebaum (Shefﬁeld UK) Lecture 2 December 2011 1 / 56

2. Definition: Lévy Process Let X = (X (t), t ≥ 0) be a stochastic process defined on a probability space (Ω, F, P). We say that it has independent increments if for each n ∈ N and each 0 ≤ t1 < t2 < · · · < tn+1 < ∞, the random variables (X (tj+1 ) − X (tj ), 1 ≤ j ≤ n) are independent and it has stationary increments if each d X (tj+1 ) − X (tj ) = X (tj+1 − tj ) − X (0). Dave Applebaum (Sheffield UK) Lecture 2 December 2011 2 / 56

5. We say that X is a Lévy process if (L1) Each X (0) = 0 (a.s), (L2) X has independent and stationary increments, (L3) X is stochastically continuous i.e. for all a > 0 and for all s ≥ 0, lim P(|X (t) − X (s)| > a) = 0. t→s Note that in the presence of (L1) and (L2), (L3) is equivalent to the condition lim P(|X (t)| > a) = 0. t↓0 Dave Applebaum (Shefﬁeld UK) Lecture 2 December 2011 3 / 56

9. The sample paths of a process are the maps t → X (t)(ω) from R+ to Rd , for each ω ∈ Ω. We are now going to explore the relationship between Lévy processes and infinite divisibility. Theorem If X is a Lévy process, then X (t) is infinitely divisible for each t ≥ 0. Dave Applebaum (Sheffield UK) Lecture 2 December 2011 4 / 56

12. Proof. For each n ∈ N, we can write (n) (n) X (t) = Y1 (t) + · · · + Yn (t) where each Yk (t) = X ( kt ) − X ( (k −1)t ). (n) n n (n) The Yk (t)’s are i.i.d. by (L2). 2 From Lecture 1 we can write φX (t) (u) = eη(t,u) for each t ≥ 0, u ∈ Rd , where each η(t, ·) is a Lévy symbol. Dave Applebaum (Shefﬁeld UK) Lecture 2 December 2011 5 / 56

15. Theorem If X is a Lévy process, then φX (t) (u) = etη(u) , for each u ∈ Rd , t ≥ 0, where η is the Lévy symbol of X (1). Proof. Suppose that X is a Lévy process and for each u ∈ Rd , t ≥ 0, deﬁne φu (t) = φX (t) (u) then by (L2) we have for all s ≥ 0, φu (t + s) = E(ei(u,X (t+s)) ) = E(ei(u,X (t+s)−X (s)) ei(u,X (s)) ) = E(ei(u,X (t+s)−X (s)) )E(ei(u,X (s)) ) = φu (t)φu (s) . . . (i) Dave Applebaum (Shefﬁeld UK) Lecture 2 December 2011 6 / 56

21. x Now φu (0) = 1 . . . (ii) by (L1), and the map t → φu (t) is continuous. However the unique continuous solution to (i) and (ii) is given by φu (t) = etα(u) , where α : Rd → C. Now by Theorem 1, X (1) is inﬁnitely divisible, hence α is a Lévy symbol and the result follows. 2 Dave Applebaum (Shefﬁeld UK) Lecture 2 December 2011 7 / 56

26. We now have the Lévy-Khinchine formula for a Lévy process X = (X (t), t ≥ 0):- 1 E(ei(u,X (t)) ) = exp{ t i(b, u) − (u, Au) 2 + (ei(u,y ) − 1 − i(u, y )1B (y ))ν(dy ) ˆ },(2.1) Rd −{0} for each t ≥ 0, u ∈ Rd , where (b, A, ν) are the characteristics of X (1). We will deﬁne the Lévy symbol and the characteristics of a Lévy process X to be those of the random variable X (1). We will sometimes write the former as ηX when we want to emphasise that it belongs to the process X . Dave Applebaum (Shefﬁeld UK) Lecture 2 December 2011 8 / 56

29. Let pt be the law of X (t), for each t ≥ 0. By (L2), we have for all s, t ≥ 0 that: pt+s = pt ∗ ps . w By (L3), we have pt → δ0 as t → 0, i.e. limt→0 f (x)pt (dx) = f (0). (pt , t ≥ 0) is a weakly continuous convolution semigroup of probability measures on Rd . Conversely, given any such semigroup, we can always construct a Lévy process on path space via Kolmogorov’s construction. Dave Applebaum (Shefﬁeld UK) Lecture 2 December 2011 9 / 56

34. Informally, we have the following asymptotic relationship between the law of a Lévy process and its Lévy measure: pt ν = lim . t↓0 t More precisely 1 lim f (x)pt (dx) = f (x)ν(dx), (2.2) t↓0 t Rd Rd for bounded, continuous functions f which vanish in some neighborhood of the origin. Dave Applebaum (Shefﬁeld UK) Lecture 2 December 2011 10 / 56

37. Examples of Lévy Processes Example 1, Brownian Motion and Gaussian Processes A (standard) Brownian motion in Rd is a Lévy process B = (B(t), t ≥ 0) for which (B1) B(t) ∼ N(0, tI) for each t ≥ 0, (B2) B has continuous sample paths. It follows immediately from (B1) that if B is a standard Brownian motion, then its characteristic function is given by 1 φB(t) (u) = exp{− t|u|2 }, 2 for each u ∈ Rd , t ≥ 0. Dave Applebaum (Shefﬁeld UK) Lecture 2 December 2011 11 / 56

41. We introduce the marginal processes Bi = (Bi (t), t ≥ 0) where each Bi (t) is the ith component of B(t), then it is not difﬁcult to verify that the Bi ’s are mutually independent Brownian motions in R. We will call these one-dimensional Brownian motions in the sequel. Brownian motion has been the most intensively studied Lévy process. In the early years of the twentieth century, it was introduced as a model for the physical phenomenon of Brownian motion by Einstein and Smoluchowski and as a description of the dynamical evolution of stock prices by Bachelier. Dave Applebaum (Shefﬁeld UK) Lecture 2 December 2011 12 / 56

47. The theory was placed on a rigorous mathematical basis by Norbert Wiener in the 1920’s. We could try to use the Kolmogorov existence theorem to construct one-dimensional Brownian motion from the following prescription on cylinder sets of the form ItH,...,tn = {ω ∈ Ω; ω(t1 ) ∈ [a1 , b1 ], . . . , ω(tn ) ∈ [an , bn ]} where 1 H = [a1 , b1 ] × · · · [an , bn ] and we have taken Ω to be the set of all mappings from R+ to R: P(ItH,...,tn ) 1 2 1 1 x1 (x2 − x1 )2 = n exp − + + ·· H (2π) 2 t1 (t2 − t1 ) . . . (tn − tn−1 ) 2 t1 t2 − t1 (xn − xn−1 )2 + dx1 · · · dxn . tn − tn−1 However there there is then no guarantee that the paths are continuous. Dave Applebaum (Shefﬁeld UK) Lecture 2 December 2011 13 / 56

52. The literature contains a number of ingenious methods for constructing Brownian motion. One of the most delightful of these (originally due to Paley and Wiener) obtains this, in the case d = 1, as a random Fourier series for 0 ≤ t ≤ 1: √ ∞ 2 sin(πt(n + 1 )) 2 B(t) = 1 ξ(n), π n+ 2 n=0 for each t ≥ 0, where (ξ(n), n ∈ N) is a sequence of i.i.d. N(0, 1) random variables. Dave Applebaum (Shefﬁeld UK) Lecture 2 December 2011 14 / 56

55. We list a number of useful properties of Brownian motion in the case d = 1. Brownian motion is locally Hölder continuous with exponent α for every 0 < α < 1 i.e. for every T > 0, ω ∈ Ω there exists 2 K = K (T , ω) such that |B(t)(ω) − B(s)(ω)| ≤ K |t − s|α , for all 0 ≤ s < t ≤ T . The sample paths t → B(t)(ω) are almost surely nowhere differentiable. For any sequence, (tn , n ∈ N) in R+ with tn ↑ ∞, lim inf B(tn ) = −∞ a.s. lim sup B(tn ) = ∞ a.s. n→∞ n→∞ The law of the iterated logarithm:- B(t) P lim sup 1 =1 = 1. t↓0 (2t log(log( 1 ))) 2 t Dave Applebaum (Shefﬁeld UK) Lecture 2 December 2011 15 / 56

60. 1.2 0.8 0.4 0 −0.4 −0.8 −1.2 −1.6 −2.0 0 1 2 3 4 5 Simulation of standard Brownian motion Dave Applebaum (Shefﬁeld UK) Lecture 2 December 2011 16 / 56

61. Let A be a non-negative symmetric d × d matrix and let σ be a square root of A so that σ is a d × m matrix for which σσ T = A. Now let b ∈ Rd and let B be a Brownian motion in R m . We construct a process C = (C(t), t ≥ 0) in Rd by C(t) = bt + σB(t), (2.3) then C is a Lévy process with each C(t) ∼ N(tb, tA). It is not difficult to verify that C is also a Gaussian process, i.e. all its finite dimensional distributions are Gaussian. It is sometimes called Brownian motion with drift. The Lévy symbol of C is 1 ηC (u) = i(b, u) − (u, Au). 2 In fact a Lévy process has continuous sample paths if and only if it is of the form (2.3). Dave Applebaum (Sheffield UK) Lecture 2 December 2011 17 / 56

69. Example 2 - The Poisson Process The Poisson process of intensity λ > 0 is a Lévy process N taking values in N ∪ {0} wherein each N(t) ∼ π(λt) so we have (λt)n −λt P(N(t) = n) = e , n! for each n = 0, 1, 2, . . .. The Poisson process is widely used in applications and there is a wealth of literature concerning it and its generalisations. Dave Applebaum (Shefﬁeld UK) Lecture 2 December 2011 18 / 56

72. We deﬁne non-negative random variables (Tn , N ∪ {0}) (usually called waiting times) by T0 = 0 and for n ∈ N, Tn = inf{t ≥ 0; N(t) = n}, then it is well known that the Tn ’s are gamma distributed. Moreover, the inter-arrival times Tn − Tn−1 for n ∈ N are i.i.d. and each has 1 exponential distribution with mean λ . The sample paths of N are clearly piecewise constant with “jump” discontinuities of size 1 at each of the random times (Tn , n ∈ N). Dave Applebaum (Shefﬁeld UK) Lecture 2 December 2011 19 / 56

76. 10 8 6 Number 4 2 0 0 5 10 15 20 Time Simulation of a Poisson process (λ = 0.5) Dave Applebaum (Shefﬁeld UK) Lecture 2 December 2011 20 / 56

77. For later work it is useful to introduce the compensated Poisson ˜ ˜ ˜ process N = (N(t), t ≥ 0) where each N(t) = N(t) − λt. Note that ˜ E(N(t)) = 0 and E(N(t)˜ 2 ) = λt for each t ≥ 0 . Dave Applebaum (Shefﬁeld UK) Lecture 2 December 2011 21 / 56

78. For later work it is useful to introduce the compensated Poisson ˜ ˜ ˜ process N = (N(t), t ≥ 0) where each N(t) = N(t) − λt. Note that ˜ E(N(t)) = 0 and E(N(t)˜ 2 ) = λt for each t ≥ 0 . Dave Applebaum (Shefﬁeld UK) Lecture 2 December 2011 21 / 56

79. Example 3 - The Compound Poisson Process Let (Z (n), n ∈ N) be a sequence of i.i.d. random variables taking values in Rd with common law µZ and let N be a Poisson process of intensity λ which is independent of all the Z (n)’s. The compound Poisson process Y is deﬁned as follows:- 0 if N(t) = 0 Y (t) := Z (1) + · · · + Z (N(t)) if N(t) > 0, for each t ≥ 0, so each Y (t) ∼ π(λt, µZ ). Dave Applebaum (Shefﬁeld UK) Lecture 2 December 2011 22 / 56

82. From the work of Lecture 1, Y has Lévy symbol ηY (u) = (ei(u,y ) − 1)λµZ (dy ) . Again the sample paths of Y are piecewise constant with “jump discontinuities” at the random times (T (n), n ∈ N), however this time the size of the jumps is itself random, and the jump at T (n) can be any value in the range of the random variable Z (n). Dave Applebaum (Shefﬁeld UK) Lecture 2 December 2011 23 / 56

85. 0 -1 Path -2 -3 0 10 20 30 Time Simulation of a compound Poisson process with N(0, 1) summands(λ = 1). Dave Applebaum (Shefﬁeld UK) Lecture 2 December 2011 24 / 56

86. Example 4 - Interlacing Processes Let C be a Gaussian Lévy process as in Example 1 and Y be a compound Poisson process as in Example 3, which is independent of C. Define a new process X by X (t) = C(t) + Y (t), for all t ≥ 0, then it is not difficult to verify that X is a Lévy process with Lévy symbol 1 ηX (u) = i(b, u) − (u, Au) + (ei(u,y ) − 1)λµZ (dy ) . 2 Using the notation of Examples 2 and 3, we see that the paths of X have jumps of random size occurring at random times. Dave Applebaum (Sheffield UK) Lecture 2 December 2011 25 / 56

91. X (t) = C(t) for 0 ≤ t < T1 , = C(T1 ) + Z1 when t = T1 , = X (T1 ) + C(t) − C(T1 ) for T1 < t < T2 , = X (T2 −) + Z2 when t = T2 , and so on recursively. We call this procedure an interlacing as a continuous path process is “interlaced” with random jumps. It seems reasonable that the most general Lévy process might arise as the limit of a sequence of such interlacings, and this can be established rigorously. Dave Applebaum (Shefﬁeld UK) Lecture 2 December 2011 26 / 56

97. Example 5 - Stable Lévy Processes A stable Lévy process is a Lévy process X in which the Lévy symbol is that of a given stable law. So, in particular, each X (t) is a stable random variable. For example, we have the rotationally invariant case whose Lévy symbol is given by η(u) = −σ α |u|α , where α is the index of stability (0 < α ≤ 2). One of the reasons why these are important in applications is that they display self-similarity. Dave Applebaum (Shefﬁeld UK) Lecture 2 December 2011 27 / 56

101. In general, a stochastic process Y = (Y (t), t ≥ 0) is self-similar with Hurst index H > 0 if the two processes (Y (at), t ≥ 0) and (aH Y (t), t ≥ 0) have the same finite-dimensional distributions for all a ≥ 0. By examining characteristic functions, it is easily verified that a rotationally invariant stable Lévy process is self-similar with Hurst index H = α , so that e.g. Brownian motion is self-similar with H = 1 . A 1 2 Lévy process X is self-similar if and only if each X (t) is strictly stable. Dave Applebaum (Sheffield UK) Lecture 2 December 2011 28 / 56

102. In general, a stochastic process Y = (Y (t), t ≥ 0) is self-similar with Hurst index H > 0 if the two processes (Y (at), t ≥ 0) and (aH Y (t), t ≥ 0) have the same finite-dimensional distributions for all a ≥ 0. By examining characteristic functions, it is easily verified that a rotationally invariant stable Lévy process is self-similar with Hurst 1 1 index H = α , so that e.g. Brownian motion is self-similar with H = 2 . A Lévy process X is self-similar if and only if each X (t) is strictly stable. Dave Applebaum (Sheffield UK) Lecture 2 December 2011 28 / 56

105. 80 40 0 −40 −80 −120 −160 −200 −240 0 1 2 3 4 5 Simulation of the Cauchy process. Dave Applebaum (Shefﬁeld UK) Lecture 2 December 2011 29 / 56

106. Densities of Lévy Processes Question: When does a Lévy process have a density ft for all t > 0 so that for all Borel sets B: P(Xt ∈ B) = pt (B) = ft (x)dx? B In general, a random variable has a continuous density if its characteristic function is integrable and in this case, the density is the Fourier transform of the characteristic function. So for Lévy processes, if for all t > 0, |etη(u) |du = et (η(u)) du < ∞ Rd Rd we then have ft (x) = (2π)−d etη(u)−i(u,x) du. Rd Dave Applebaum (Shefﬁeld UK) Lecture 2 December 2011 30 / 56

110. Every Lévy process with a non-degenerate Gaussian component has a density. In this case 1 (η(u)) = − (u, Au) + (cos(u, y ) − 1)ν(dy ), 2 Rd −{0} and so t et (η(u)) du ≤ e− 2 (u,Au) < ∞, Rd Rd using (u, Au) ≥ λ|u|2 where λ > 0 is smallest eigenvalue of A. Dave Applebaum (Shefﬁeld UK) Lecture 2 December 2011 31 / 56

113. For examples where densities exist for A = 0 with d = 1: if X is α-stable, it has a density since for all 1 ≤ α ≤ 2: α e−t|u| du ≤ e−t|u| du < ∞, |u|≥1 |u|≥1 and for 0 ≤ α < 1: ∞ α 2 1 e−t|u| du = e−ty y α −1 dy < ∞. R α 0 Dave Applebaum (Shefﬁeld UK) Lecture 2 December 2011 32 / 56

116. In general, a sufﬁcient condition for a density is ν(Rd ) = ∞ ν ∗m is absolutely continuous with respect to Lebesgue measure ˜ for some m ∈ N where ν (A) = ˜ (|x|2 ∧ 1)ν(dx). A Dave Applebaum (Shefﬁeld UK) Lecture 2 December 2011 33 / 56

119. A Lévy process has a Lévy density gν if its Lévy measure ν is absolutely continuous with respect to Lebesgue measure, then gν is dν deﬁned to be the Radon-Nikodym derivative . dx A process may have a Lévy density but not have a density. Example. Let X be a compound Poisson process with each X (t) = Y1 + Y2 + · · · + YN(t) wherein each Yj has a density fY , then gν = λfY is the Lévy density. But P(Y (t) = 0) ≥ P(N(t) = 0) = e−λt > 0, so pt has an atom at {0}. Dave Applebaum (Shefﬁeld UK) Lecture 2 December 2011 34 / 56

124. We have pt (A) = e−λt δ0 (A) + ac A ft (x)dx, where for x = 0 ∞ −λt (λt)n ∗n ftac (x) =e f (x). n! Y n=1 ftac (x) is the conditional density of X (t) given that it jumps at least once between 0 and t. In this case, (2.2) takes the precise form (for x = 0) ftac (x) gν (x) = lim . t↓0 t Dave Applebaum (Shefﬁeld UK) Lecture 2 December 2011 35 / 56

128. Subordinators A subordinator is a one-dimensional Lévy process which is increasing a.s. Such processes can be thought of as a random model of time evolution, since if T = (T (t), t ≥ 0) is a subordinator we have T (t) ≥ 0 for each t > 0 a.s. and T (t1 ) ≤ T (t2 ) whenever t1 ≤ t2 a.s. Now since for X (t) ∼ N(0, At) we have P(X (t) ≥ 0) = P(X (t) ≤ 0) = 1 , it is clear that such a process cannot 2 be a subordinator. Dave Applebaum (Shefﬁeld UK) Lecture 2 December 2011 36 / 56

129. Subordinators A subordinator is a one-dimensional Lévy process which is increasing a.s. Such processes can be thought of as a random model of time evolution, since if T = (T (t), t ≥ 0) is a subordinator we have T (t) ≥ 0 for each t > 0 a.s. and T (t1 ) ≤ T (t2 ) whenever t1 ≤ t2 a.s. Now since for X (t) ∼ N(0, At) we have 1 P(X (t) ≥ 0) = P(X (t) ≤ 0) = 2 , it is clear that such a process cannot be a subordinator. Dave Applebaum (Shefﬁeld UK) Lecture 2 December 2011 36 / 56

130. Subordinators A subordinator is a one-dimensional Lévy process which is increasing a.s. Such processes can be thought of as a random model of time evolution, since if T = (T (t), t ≥ 0) is a subordinator we have T (t) ≥ 0 for each t > 0 a.s. and T (t1 ) ≤ T (t2 ) whenever t1 ≤ t2 a.s. Now since for X (t) ∼ N(0, At) we have 1 P(X (t) ≥ 0) = P(X (t) ≤ 0) = 2 , it is clear that such a process cannot be a subordinator. Dave Applebaum (Shefﬁeld UK) Lecture 2 December 2011 36 / 56

131. Theorem If T is a subordinator then its Lévy symbol takes the form η(u) = ibu + (eiuy − 1)λ(dy ), (2.4) (0,∞) where b ≥ 0, and the Lévy measure λ satisﬁes the additional requirements λ(−∞, 0) = 0 and (y ∧ 1)λ(dy ) < ∞. (0,∞) Conversely, any mapping from Rd → C of the form (2.4) is the Lévy symbol of a subordinator. We call the pair (b, λ), the characteristics of the subordinator T . Dave Applebaum (Shefﬁeld UK) Lecture 2 December 2011 37 / 56

135. For each t ≥ 0, the map u → E(eiuT (t) ) can be analytically continued to the region {iu, u > 0} and we then obtain the following expression for the Laplace transform of the distribution E(e−uT (t) ) = e−tψ(u) , where ψ(u) = −η(iu) = bu + (1 − e−uy )λ(dy ) (2.5) (0,∞) for each t, u ≥ 0. This is much more useful for both theoretical and practical application than the characteristic function. The function ψ is usually called the Laplace exponent of the subordinator. Dave Applebaum (Shefﬁeld UK) Lecture 2 December 2011 38 / 56

140. Examples of Subordinators (1) The Poisson Case Poisson processes are clearly subordinators. More generally a compound Poisson process will be a subordinator if and only if the Z (n)’s are all R+ valued. Dave Applebaum (Shefﬁeld UK) Lecture 2 December 2011 39 / 56

141. Examples of Subordinators (1) The Poisson Case Poisson processes are clearly subordinators. More generally a compound Poisson process will be a subordinator if and only if the Z (n)’s are all R+ valued. Dave Applebaum (Shefﬁeld UK) Lecture 2 December 2011 39 / 56

142. (2) α-Stable Subordinators Using straightforward calculus, we ﬁnd that for 0 < α < 1, u ≥ 0, ∞ α dx uα = (1 − e−ux ) . Γ(1 − α) 0 x 1+α Hence for each 0 < α < 1 there exists an α-stable subordinator T with Laplace exponent ψ(u) = u α . and the characteristics of T are (0, λ) where λ(dx) = Γ(1−α) x dx . α 1+α Note that when we analytically continue this to obtain the Lévy symbol we obtain the form given in Lecture 1 for stable laws with µ = 0, β = 1 and σ α = cos απ . 2 Dave Applebaum (Shefﬁeld UK) Lecture 2 December 2011 40 / 56

146. (3) The Lévy Subordinator The 1 -stable subordinator has a density given by the Lévy distribution 2 2 (with µ = 0 and σ = t2 ) t 3 t2 fT (t) (s) = √ s− 2 e− 4s , 2 π for s ≥ 0. The Lévy subordinator has a nice probabilistic interpretation as a ﬁrst hitting time for one-dimensional standard Brownian motion (B(t), t ≥ 0), t T (t) = inf s > 0; B(s) = √ . (2.6) 2 Dave Applebaum (Shefﬁeld UK) Lecture 2 December 2011 41 / 56

150. To show directly that for each t ≥ 0, ∞ 1 E(e−uT (t) ) = e−us fT (t) (s)ds = e−tu 2 , 0 write gt (u) = E(e−uT (t) ). Differentiate with respect to u and make the t2 substitution x = 4us to obtain the differential equation t t gt (u) = − 2√u gt (u). Via the substitution y = 2√s we see that gt (0) = 1 and the result follows. Dave Applebaum (Shefﬁeld UK) Lecture 2 December 2011 42 / 56

153. (4) Inverse Gaussian Subordinators We generalise the Lévy subordinator by replacing Brownian motion by the Gaussian process C = (C(t), t ≥ 0) where each C(t) = B(t) + µt and µ ∈ R. The inverse Gaussian subordinator is deﬁned by T (t) = inf{s > 0; C(s) = δt} where δ > 0 and is so-called since t → T (t) is the generalised inverse of a Gaussian process. Using martingale methods, we can show that for each t, u > 0, √ 2 E(e−uT (t) ) = e−tδ( 2u+µ −µ) , (2.7) In fact each T (t) has a density:- δt 3 1 fT (t) (s) = √ eδtµ s− 2 exp − (t 2 δ 2 s−1 + µ2 s) , (2.8) 2π 2 for each s, t ≥ 0. In general any random variable with density fT (1) is called an inverse Gaussian and denoted as IG(δ, µ). Dave Applebaum (Shefﬁeld UK) Lecture 2 December 2011 43 / 56

158. (5) Gamma Subordinators Let (T (t), t ≥ 0) be a gamma process with parameters a, b > 0 so that each T (t) has density bat at−1 −bx fT (t) (x) = x e , Γ(at) for x ≥ 0; then it is easy to verify that for each u ≥ 0, ∞ −at u u e−ux fT (t) (x)dx = 1 + = exp −ta log 1 + . 0 b b From here it is a straightforward exercise in calculus to show that ∞ ∞ e−ux fT (t) (x)dx = exp −t (1 − e−ux )ax −1 e−bx dx . 0 0 Dave Applebaum (Shefﬁeld UK) Lecture 2 December 2011 44 / 56

161. From this we see that (T (t), t ≥ 0) is a subordinator with b = 0 and λ(dx) = ax −1 e−bx dx. Moreover ψ(u) = a log 1 + u is the associated b Bernstein function (see below). Dave Applebaum (Shefﬁeld UK) Lecture 2 December 2011 45 / 56

162. From this we see that (T (t), t ≥ 0) is a subordinator with b = 0 and λ(dx) = ax −1 e−bx dx. Moreover ψ(u) = a log 1 + u is the associated b Bernstein function (see below). Dave Applebaum (Shefﬁeld UK) Lecture 2 December 2011 45 / 56

163. 5 4 3 2 1 0 0 1 2 3 4 5 Simulation of a gamma subordinator. Dave Applebaum (Shefﬁeld UK) Lecture 2 December 2011 46 / 56

164. Before we go further into the probabilistic properties of subordinators we’ll make a quick diversion into analysis. Let f ∈ C ∞ ((0, ∞)). We say it is completely monotone if (−1)n f (n) ≥ 0 for all n ∈ N, and a Bernstein function if f ≥ 0 and (−1)n f (n) ≤ 0 for all n ∈ N. Dave Applebaum (Shefﬁeld UK) Lecture 2 December 2011 47 / 56

168. Theorem 1 f is a Bernstein function if and only if the mapping x → e−tf (x) is completely monotone for all t ≥ 0. 2 f is a Bernstein function if and only if it has the representation ∞ f (x) = a + bx + (1 − e−yx )λ(dy ), 0 ∞ for all x > 0 where a, b ≥ 0 and 0 (y ∧ 1)λ(dy ) < ∞. 3 g is completely monotone if and only if there exists a measure µ on [0, ∞) for which ∞ g(x) = e−xy µ(dy ). 0 Dave Applebaum (Shefﬁeld UK) Lecture 2 December 2011 48 / 56

171. To interpret this theorem, ﬁrst consider the case a = 0. In this case, if we compare the statement of Theorem 4 with equation (2.5), we see that there is a one to one correspondence between Bernstein functions for which limx→0 f (x) = 0 and Laplace exponents of subordinators. The Laplace transforms of the laws of subordinators are always completely monotone functions and a subclass of all possible measures µ appearing in Theorem 4 (3) is given by all possible laws pT (t) associated to subordinators. A general Bernstein function with a > 0 can be given a probabilistic interpretation by means of “killing”. Dave Applebaum (Shefﬁeld UK) Lecture 2 December 2011 49 / 56

175. One of the most important probabilistic applications of subordinators is to “time change”. Let X be an arbitrary Lévy process and let T be a subordinator defined on the same probability space as X such that X and T are independent. We define a new stochastic process Z = (Z (t), t ≥ 0) by the prescription Z (t) = X (T (t)), for each t ≥ 0 so that for each ω ∈ Ω, Z (t)(ω) = X (T (t)(ω))(ω). The key result is then the following. Theorem Z is a Lévy process. Dave Applebaum (Sheffield UK) Lecture 2 December 2011 50 / 56

180. We compute the Lévy symbol of the subordinated process Z . Theorem ηZ = −ψT ◦ (−ηX ). Proof. For each u ∈ Rd , t ≥ 0, eiηZ (t) (u) = E(ei(u,Z (t)) ) = E(ei(u,X (T (t))) ) = E(ei(u,X (s)) )pT (t) (ds) = esηX (u) pT (t) (ds) = E(e−(−ηX (u))T (t) ) = e−tψT (−ηX (u)) . 2 Dave Applebaum (Shefﬁeld UK) Lecture 2 December 2011 51 / 56

188. Example : From Brownian Motion to 2α-stable Processes Let T be an α-stable subordinator (with 0 < α < 1) and X be a d-dimensional Brownian motion with covariance A = 2I, which is independent of T . Then for each s ≥ 0, u ∈ Rd , ψT (s) = sα and ηX (u) = −|u|2 , and hence ηZ (u) = −|u|2α , i.e. Z is a rotationally invariant 2α-stable process. In particular, if d = 1 and T is the Lévy subordinator, then Z is the Cauchy process, so each Z (t) has a symmetric Cauchy distribution with parameters µ = 0 and σ = 1. It is interesting to observe from (2.6) that Z is constructed from two independent standard Brownian motions. Dave Applebaum (Shefﬁeld UK) Lecture 2 December 2011 52 / 56

192. Examples of subordinated processes have recently found useful applications in mathematical finance. We briefly mention two interesting cases:- (i) The Variance Gamma Process In this case Z (t) = B(T (t)), for each t ≥ 0, where B is a standard Brownian motion and T is an independent gamma subordinator. The name derives from the fact that, in a formal sense, each Z (t) arises by replacing the variance of a normal random variable by a gamma random variable. Using Theorem 6, a simple calculation yields −at u2 ΦZ (t) (u) = 1+ , 2b for each t ≥ 0, u ∈ R, where a and b are the usual parameters which determine the gamma process. It is an easy exercise in manipulating characteristic functions to compute the alternative representation: Z (t) = G(t) − L(t), Dave Applebaum (Sheffield UK) Lecture 2 December 2011 53 / 56

197. where G and L are independent gamma subordinators each with √ parameters 2b and a. This yields a nice ﬁnancial representation of Z as a difference of independent “gains” and “losses”. From this representation, we can compute that Z has a Lévy density a √2bx √ gν (x) = (e 1(−∞,0) (x) + e− 2bx 1(0,∞) (x)). |x| Dave Applebaum (Shefﬁeld UK) Lecture 2 December 2011 54 / 56

200. The CGMY processes are a generalisation of the variance-gamma processes due to Carr, Geman, Madan and Yor. They are characterised by their Lévy density: a gν (x) = (eb1 x 1(−∞,0) (x) + e−b2 x 1(0,∞) (x)), |x|1+α where a > 0, 0 ≤ α < 2 and b1 , b2 ≥ 0. We obtain stable Lévy processes when b1 = b2 = 0. They can also be obatined by subordinating Brownian motion with drift. The CGMY processes are a subclass of the tempered stable processes. Note how the exponential dampens the effects of large jumps. Dave Applebaum (Shefﬁeld UK) Lecture 2 December 2011 55 / 56

207. (ii) The Normal Inverse Gaussian Process In this case Z (t) = C(T (t)) + µt for each t ≥ 0 where each C(t) = B(t) + βt, with β ∈ R. Here T is an inverse Gaussian subordinator, which is independent of B, and in which we write the parameter γ = α2 − β 2 , where α ∈ R with α2 ≥ β 2 . Z depends on four parameters and has characteristic function ΦZ (t) (α, β, δ, µ)(u) = exp {δt( α2 − β 2 − α2 − (β + iu)2 ) + iµtu} for each u ∈ R, t ≥ 0. Here δ > 0 is as in (2.7). Each Z (t) has a density given by −1 x − µt x − µt fZ (t) (x) = C(α, β, δ, µ; t)q K1 δtαq eβx , δt δt for each x ∈ R, where √ √ 2 2 q(x) = 1 + x 2 , C(α, β, δ, µ; t) = π −1 αeδt α −β −βµt and K1 is a Bessel function of the third kind. Dave Applebaum (Shefﬁeld UK) Lecture 2 December 2011 56 / 56

Koc2(dba)

More Related Content

What's hot (20)

Similar to Koc2(dba) (20)

Recently uploaded (20)

Koc2(dba)