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

Create successful ePaper yourself

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

3<br />

<strong>Credibility</strong> Models for <strong>Claim</strong><br />

<strong>Counts</strong><br />

3.1 Introduction<br />

3.1.1 From <strong>Risk</strong> <strong>Classification</strong> to Experience Rating<br />

We have seen in Chapter 2 how to partition a heterogeneous portfolio into more homogeneous<br />

classes with all policyholders belonging to the same class paying the same premium.<br />

However, tariff cells are still quite heterogeneous despite the use <strong>of</strong> many a priori variables.<br />

The expected claim frequency for the tariff cell is designed to reflect the average experience<br />

<strong>of</strong> the entire group. If the experience <strong>of</strong> a policy is consistently better (or worse) than the<br />

average experience <strong>of</strong> the group, the insurance company may consider adapting the amount<br />

<strong>of</strong> premium to be charged for this policy. Of course, this requires a model which can separate<br />

random variation from signal in the historical data to indicate whether this policy is <strong>of</strong> better<br />

(or worse) quality compared to the group average.<br />

It is reasonable to believe that the hidden features (unobserved risk characteristics that<br />

have been modelled by a random effect in the mixed Poisson regression model) are revealed<br />

by the number <strong>of</strong> claims reported by the policyholders over the successive insurance periods.<br />

Hence the adjustment <strong>of</strong> the premium based on the individual claims experience in order<br />

to restore fairness among policyholders. The allowance for the history <strong>of</strong> the policyholder<br />

in a rating model thus derives from interpretation <strong>of</strong> serial correlation for longitudinal data<br />

resulting from hidden features in the risk distribution.<br />

3.1.2 <strong>Credibility</strong> Theory<br />

<strong>Credibility</strong> theory is the art <strong>of</strong> combining different collections <strong>of</strong> data to obtain an accurate<br />

overall estimate. It provides actuaries with techniques to determine insurance premiums for<br />

<strong>Actuarial</strong> <strong>Modelling</strong> <strong>of</strong> <strong>Claim</strong> <strong>Counts</strong>: <strong>Risk</strong> <strong>Classification</strong>, <strong>Credibility</strong> and Bonus-Malus Systems<br />

S. Pitrebois and J.-F. Walhin © 2007 John Wiley & Sons, Ltd<br />

M. Denuit, X. Maréchal,

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

Saved successfully!

Ooh no, something went wrong!