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

Create successful ePaper yourself

Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.

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

By successive conditioning, we get<br />

P (Card(It) > 1) = P (Card(I0) > 1)<br />

t<br />

P (Card(Is) > 1|Card(Is−1) > 1) <<br />

So, the probability P (Card(It) > 1) <strong>de</strong>creases geometrically as t increases.<br />

s=1<br />

2.4. Dynamic framework<br />

<br />

1 − ˜ t ξ .<br />

Proposition 2.4.2. For the repeated I − player insurance game <strong>de</strong>fined in the previous subsection,<br />

if for all k = j, xj ≤ xk and xj(1 − ej) ≤ xk(1 − ek), then the un<strong>de</strong>rwriting result by<br />

policy is or<strong>de</strong>red UWj ≤icx UWk where UWj is the random variable<br />

UWj = xj(1 − ej) − 1<br />

Nj(x)<br />

<br />

Yi.<br />

Nj(x)<br />

Proof. Let us consi<strong>de</strong>r a price vector x such that xj < xk for all k = j. Since the change<br />

probability pk→j (for k = j) is a <strong>de</strong>creasing function (see Appendix 2.6.1), pk→j(x) > pk→l(x)<br />

for l = j given the initial portfolio sizes nj’s are constant.<br />

Below we use the stochastic or<strong>de</strong>rs (≤st, ≤cx) and the majorization or<strong>de</strong>r (≤m) whose<br />

<strong>de</strong>finitions and main properties are recalled in the Appendices 2.6.2 and 2.6.2 respectively.<br />

Using the convolution property of the stochastic or<strong>de</strong>r I times, we can show a stochastic or<strong>de</strong>r<br />

of the portfolio size<br />

Nk(x) ≤st Nj(x), ∀k = j.<br />

Let us consi<strong>de</strong>r the un<strong>de</strong>rwriting result per policy<br />

uwj(x, n) = 1<br />

<br />

n<br />

nxj(1 − ej) −<br />

n<br />

i=1<br />

Yi<br />

<br />

i=1<br />

= xj(1 − ej) −<br />

n<br />

i=1<br />

1<br />

n Yi,<br />

for insurer j having n policies, where Yi <strong>de</strong>notes the total claim amount per policy.<br />

Let n < ñ be two policy numbers and añ, an ∈ Rñ be <strong>de</strong>fined as<br />

⎛<br />

⎞<br />

<br />

1 1<br />

⎜<br />

añ = , . . . , and an = ⎜<br />

1 1 ⎟<br />

ñ ñ<br />

⎝ , . . . , , 0, . . . , 0 ⎟<br />

n n<br />

⎠<br />

size ñ−n<br />

size n<br />

.<br />

Since añ ≤m an and (Yi)i’s are i.i.d. random variables, we have <br />

i añ,iYi<br />

<br />

≤cx i an,iYi i.e.<br />

ñ<br />

i=1<br />

1<br />

ñ Yi ≤cx<br />

n<br />

i=1<br />

1<br />

n Yi.<br />

For all increasing convex functions φ, the function x ↦→ φ(x + a) is still increasing and<br />

convex. Thus for all random variables X, Y such that X ≤icx Y and real numbers a, b, a ≤ b,<br />

we have<br />

E(φ(X + a)) ≤ E(φ(X + b)) ≤ E(φ(Y + b)),<br />

i.e. a + X ≤icx b + Y .<br />

As xj(1 − ej) ≤ xk(1 − ek) and using the fact that X ≤cx Y is equivalent to −X ≤cx −Y ,<br />

we have<br />

uwj(x, ñ) ≤icx uwk(x, n), ∀k = j.<br />

119

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

Saved successfully!

Ooh no, something went wrong!