Processus de Lévy en Finance - Laboratoire de Probabilités et ...
Processus de Lévy en Finance - Laboratoire de Probabilités et ...
Processus de Lévy en Finance - Laboratoire de Probabilités et ...
Create successful ePaper yourself
Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.
174 CHAPTER 5. APPLICATIONS OF LEVY COPULAS<br />
Example 5.2. L<strong>et</strong><br />
⎧<br />
⎨<br />
g (α 1,...,α d ) 1 for α 1 = · · · = α d<br />
(u) =<br />
⎩<br />
0 otherwise.<br />
Th<strong>en</strong> the Lévy copula F in Theorem 5.3 satisfies F (u 1 , . . . , u d ) = 0 if u i u j < 0 for some i, j.<br />
This means that the Lévy measure is supported by the positive and the negative orthant: either<br />
all compon<strong>en</strong>ts of the process jump up or all compon<strong>en</strong>ts jump down.<br />
5.2 Simulation of multidim<strong>en</strong>sional <strong>de</strong>p<strong>en</strong><strong>de</strong>nt Lévy processes<br />
Lévy copulas turn out to be a conv<strong>en</strong>i<strong>en</strong>t tool for simulating multidim<strong>en</strong>sional Lévy processes<br />
with specified <strong>de</strong>p<strong>en</strong><strong>de</strong>nce. In this section we first give the necessary <strong>de</strong>finitions and two auxiliary<br />
lemmas and th<strong>en</strong> prove two theorems which show how multidim<strong>en</strong>sional Lévy processes<br />
with <strong>de</strong>p<strong>en</strong><strong>de</strong>nce structures giv<strong>en</strong> by Lévy copulas can be simulated in the finite variation case<br />
(Theorem 5.6) and in the infinite variation case (Theorem 5.7).<br />
To simulate a Lévy process {X t } 0≤t≤1 on R d with Lévy measure ν, we will first simulate a<br />
Poisson random measure on [0, 1] × R d with int<strong>en</strong>sity measure λ [0,1] ⊗ ν, where λ <strong>de</strong>notes the<br />
Lebesgue measure. The Lévy process can th<strong>en</strong> be constructed via the Lévy-Itô <strong>de</strong>composition.<br />
L<strong>et</strong> F be a Lévy copula on (−∞, ∞] d such that for every I ∈ {1, . . . , d} nonempty,<br />
lim F (x 1, . . . , x d ) = F (x 1 , . . . , x d )| (xi ) i∈I =∞. (5.5)<br />
(x i ) i∈I →∞<br />
This Lévy copula <strong>de</strong>fines a positive measure µ on R d with Lebesgue margins such that for each<br />
a, b ∈ R d with a ≤ b,<br />
V F (|a, b|) = µ((a, b]). (5.6)<br />
For a one-dim<strong>en</strong>sional tail integral U, the inverse tail integral U (−1) was <strong>de</strong>fined in Equation<br />
(4.25). In the sequel we will need the following technical lemma.<br />
Lemma 5.4. L<strong>et</strong> ν be a Lévy measure on R d with marginal tail integrals U i , i = 1, . . . , d, and<br />
Lévy copula F on (−∞, ∞] d , satisfying (5.5), l<strong>et</strong> µ be <strong>de</strong>fined by (5.6) and l<strong>et</strong><br />
Th<strong>en</strong> ν is the image measure of µ by f.<br />
f : (u 1 , . . . , u d ) ↦→ (U (−1)<br />
1 (u 1 ), . . . , U (−1)<br />
d<br />
(u d )).