28.05.2013 Views

Etude des marchés d'assurance non-vie à l'aide d'équilibres de ...

Etude des marchés d'assurance non-vie à l'aide d'équilibres de ...

Etude des marchés d'assurance non-vie à l'aide d'équilibres de ...

SHOW MORE
SHOW LESS

You also want an ePaper? Increase the reach of your titles

YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.

tel-00703797, version 2 - 7 Jun 2012<br />

when k ∈ N. Using the Kronecker product δij, it can be rewritten as<br />

P (X = k) = qδ0,k + (1 − q)(1 − δ0,k)ρ(1 − ρ) k−1 .<br />

4.6. Appendix<br />

The expectation and variance are given by E(X) = (1 − q)/ρ and V ar(X) = (1 − q)(1 − ρ +<br />

q)/ρ 2 . We also have<br />

P (X > k) = (1 − q)(1 − ρ) k .<br />

We get back to the classic geometric distribution with ρ = q.<br />

Secondly, we present the 0,1-modified geometric distribution: X ∼ G(q, ρ, 1 − α)<br />

which is stricly equivalent to<br />

⎧<br />

⎨<br />

q if k = 0,<br />

P (X = k) =<br />

(1 − q)ρ<br />

⎩<br />

(1 − q)(1 − ρ)(1 − α)α<br />

if k = 1,<br />

k−2 otherwise.<br />

P (X = 0) = q = 1 − p and P (X = k/X > 0) = ρδk,1 + (1 − ρ)(1 − α)α k−2 (1 − δk,1).<br />

The mean and the variance are given by<br />

<br />

1 − ρ<br />

E(X) = p 1 +<br />

1 − α<br />

3 − α<br />

and V ar(X) = qρ + (1 − q)(1 − ρ)<br />

(1 − α)<br />

We get back to the 0-modified geometric distribution with α = 1 − ρ.<br />

4.6.4 Error function linked terms<br />

2 − p2<br />

We want to compute the following integral, linked to the error function<br />

J(a, b, x) =<br />

∞<br />

x<br />

e −ay2 −b/y 2<br />

dy,<br />

,<br />

2 1 − ρ<br />

1 + .<br />

1 − α<br />

where a, b, x > 0. The SAGE mathematical software (Stein et al. (2011)) suggests to do a<br />

change of variable in or<strong>de</strong>r to get e−t2dt. Since the equation t2 = ay2 + b/y2 does not have<br />

a unique solution, we consi<strong>de</strong>r<br />

t = ± √ ay + √ b/y.<br />

This leads to split the integral J(a, b, x). With algebric manipulations, we get<br />

Therefore,<br />

with ˜x1 = √ ax +<br />

J(a, b, x) =<br />

2 √ ady = √ ady +<br />

2 √ aJ(a, b, x) = e 2√ ∞<br />

ab<br />

√<br />

b<br />

−y2 dy + √ √<br />

b<br />

ady − dy.<br />

−y2 ˜x1<br />

√<br />

b<br />

x and ˜x2 = √ √<br />

b ax − x . Hence<br />

√ π<br />

4 √ a<br />

<br />

e 2√ ab erfc<br />

√ax +<br />

e −t2<br />

dt + e −2√ab √ <br />

b<br />

x<br />

∞<br />

˜x2<br />

+ e −2√ ab erfc<br />

e −t2<br />

dt,<br />

√ <br />

√ax b<br />

− .<br />

x<br />

209

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!