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.

290 <strong>Actuarial</strong> <strong>Modelling</strong> <strong>of</strong> <strong>Claim</strong> <strong>Counts</strong><br />

Table 7.7 Results for a deductible per claim varying according to the level occupied in the malus<br />

zone, combined with reduced relativities r l , for the scale −1/top.<br />

Level l r l r l with deductible d l if C 1 ∼ xp d l if C 1 ∼ or<br />

5 197.3 % 157.8 % 4610 4604<br />

4 170.9 % 136.7 % 4610 4604<br />

3 150.7 % 120.6 % 4610 4604<br />

2 134.8 % 107.8 % 4610 4604<br />

1 122.0 % 97.6 % 4610 4604<br />

0 54.7 % 54.7 % 0 0<br />

Table 7.8 Results for a deductible per claim varying according to the level occupied in the malus<br />

zone, combined with reduced relativities r l , for the scale −1/ + 2.<br />

Level l r l r l with deductible d l if C 1 ∼ xp d l if C 1 ∼ or<br />

5 309.1 % 247.3 % 4610 4604<br />

4 241.4 % 193.1 % 4610 4604<br />

3 207.7 % 166.2 % 4610 4604<br />

2 142.9 % 114.3 % 4610 4604<br />

1 130.2 % 104.2 % 4610 4604<br />

0 62.4 % 62.4 % 0 0<br />

As already mentioned, since this equation does not depend on the level l, the amount <strong>of</strong><br />

deductible will be the same for each level <strong>of</strong> the scale.<br />

Tables 7.7 and 7.8 display the numerical results. The third column gathers the relativities:<br />

those in the malus zone have been reduced by 20 % compared to column 2. The last two<br />

columns display the amounts <strong>of</strong> deductible. In this case, the amounts <strong>of</strong> deductible are<br />

reasonable and can be implemented in practice (about 150 % <strong>of</strong> the annual pure premium).<br />

Quite surprisingly, the LogNormal distribution now produces smaller deductibles than its<br />

Negative Exponential counterpart.<br />

7.5 Further Reading and Bibliographic Notes<br />

This chapter is based on Pitrebois, Denuit & Walhin (2005) for the most part.<br />

The Panjer family <strong>of</strong> counting distributions is known as the Katz family in applied<br />

probability. This family attracted a lot <strong>of</strong> attention in the literature, due to the fact that it<br />

contains underdispersed (Binomial), equidispersed (Poisson), and overdispersed (Negative<br />

Binomial) distributions. As a result <strong>of</strong> Panjer’s (1981) publication, a lot <strong>of</strong> other articles<br />

have appeared in the actuarial literature covering similar recursion relations. Multivariate<br />

versions <strong>of</strong> the Panjer algorithm will be used in Chapter 9.<br />

<strong>Claim</strong> amounts have also been taken into account by Bonsdorff (2005), who studied<br />

bonus-malus systems where the transitions between bonus levels in the entire interval a b

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

Saved successfully!

Ooh no, something went wrong!