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.

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

=<br />

∑k w k<br />

∫ +<br />

0<br />

l k dF <br />

∑k w k<br />

∫ +<br />

0<br />

l k dF (4.13)<br />

It is easily seen that<br />

[ ]<br />

Er L = E EL = E = 1<br />

resulting in financial equilibrium once steady state is reached.<br />

Remark 4.2 If the insurance company does not enforce any a priori ratemaking system,<br />

all the k s are equal to E = and (4.13) reduces to the formula<br />

r l =<br />

∫ +<br />

0<br />

l dF <br />

∫ +<br />

0<br />

l dF <br />

(4.14)<br />

that has been derived in Norberg (1976).<br />

The way a priori and a posteriori ratemakings interact is described by<br />

EL = l = ∑ k<br />

k Pr = k L = l<br />

= ∑ k<br />

=<br />

k<br />

PrL = l = k w k<br />

PrL = l<br />

∑k k w k<br />

∫ +<br />

0<br />

l k dF <br />

∑k w k<br />

∫ +<br />

0<br />

l k dF (4.15)<br />

If EL = l is indeed increasing in the level l, those policyholders who have been granted<br />

premium discounts at policy issuance (on the basis <strong>of</strong> their observable characteristics) will<br />

also be rewarded a posteriori (because they occupy the lowest levels <strong>of</strong> the bonus-malus<br />

scale). Conversely, the policyholders who have been penalized at policy issuance (because<br />

<strong>of</strong> their observable characteristics) will cluster in the highest bonus-malus levels and will<br />

consequently be penalized again.<br />

Example 4.11 (−1/Top Scale, Portfolio A) The results for the bonus-malus scale −1/top<br />

are displayed in Table 4.3. Specifically, the values in the third column are computed with<br />

the help <strong>of</strong> (4.14) with â = 0889 and ̂ = 01474. Those values were obtained in Section 1.6<br />

by fitting a Negative Binomial distribution to the portfolio observed claim frequencies given<br />

in Table 1.1. Integrations have been performed numerically with the QUAD procedure <strong>of</strong><br />

SAS R /IML. The fourth column is based on (4.13) with â = 1065 and the ̂ k s listed in<br />

Table 2.7.<br />

Once the steady state has been reached, the majority <strong>of</strong> the policies (58.5 % <strong>of</strong> the<br />

portfolio) occupy level 0 and enjoy the maximum discount. The remaining 41.5 % <strong>of</strong> the<br />

portfolio are distributed over levels 1–5, with about 13 % in level 5 (those policyholders who<br />

just claimed). Concerning the relativities, the minimum percentage <strong>of</strong> 54.7 % when the a<br />

priori ratemaking is not recognized becomes 61.2 % when the relativities are adapted to the

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

Saved successfully!

Ooh no, something went wrong!