01.06.2015 Views

Actuarial Modelling of Claim Counts Risk Classification, Credibility ...

Actuarial Modelling of Claim Counts Risk Classification, Credibility ...

Actuarial Modelling of Claim Counts Risk Classification, Credibility ...

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.

Bonus-Malus Systems with Varying Deductibles 281<br />

7.2.3 Panjer Algorithm<br />

Direct Convolution Approach<br />

The Panjer algorithm then allows us to compute the distribution function <strong>of</strong> the discretized<br />

version <strong>of</strong> S, defined as<br />

S =<br />

N∑<br />

C k <br />

k=1<br />

To realize the merit <strong>of</strong> the Panjer approach, let us first write the formulas in the direct<br />

approach. Let us denote as f i = PrC1<br />

= i the probability mass function <strong>of</strong> C 1 . The<br />

probability mass function <strong>of</strong> C1 +···+C k<br />

is<br />

[ ]<br />

k∑<br />

f ⋆k<br />

i = Pr C j<br />

= i <br />

j=1<br />

In the applications we have in mind, the distribution function for the claim costs satisfies<br />

F0 = 0. Clearly, with Discretization Method (7.3), we then have f 0 = 0 so that f ⋆k<br />

i = 0if<br />

i ≤ k − 1 (since C1 ⋆k<br />

≥ with probability 1). The fi<br />

s satisfy<br />

f ⋆k<br />

i−k+1<br />

∑<br />

i = f ⋆k−1<br />

i−j f j if i ≥ k (7.5)<br />

Then, the probability mass function g i = PrS = i <strong>of</strong> S satisfies<br />

g i =<br />

i∑<br />

k=0<br />

j=1<br />

PrN = kf ⋆k<br />

i i∈ (7.6)<br />

This direct computation <strong>of</strong> the probability mass function <strong>of</strong> S requires a lot <strong>of</strong> computation<br />

time. Things are even worse in the Discretization (7.4) for which f 0 > 0.<br />

Panjer Family<br />

The Panjer formula holds for a class <strong>of</strong> probability distributions referred to as the Katz<br />

family in statistical circles, and as the Panjer family in the actuarial literature. This family<br />

contains all the counting distributions such that the relation<br />

(<br />

p k = a + b )<br />

p<br />

k k−1 k= 1 2 (7.7)<br />

is fulfilled for some a and b. The Panjer family contains three elements. Specifically, the<br />

probability distributions satisfying (7.7) are<br />

(i) the Poisson distribution, obtained with a = 0 and b>0;<br />

(ii) the Negative Binomial distribution, obtained with 0

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

Saved successfully!

Ooh no, something went wrong!