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.

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

most cases). Actuaries use regression techniques to predict the expected number <strong>of</strong> claims<br />

knowing some information about the policyholders, vehicles and types <strong>of</strong> contract. It is worth<br />

mentioning that even with all the covariates included here, there still remain substantial risk<br />

differentials between individual drivers (due to hidden characteristics, like temper and skill,<br />

aggressiveness behind the wheel, knowledge <strong>of</strong> the highway code, etc.). Random effects<br />

are added to the linear predictor on the score scale to take this residual heterogeneity into<br />

account, reconciling the two approaches (i)–(ii) mentioned above.<br />

In nonlife business, the pure premium is the expected cost <strong>of</strong> all the claims that<br />

policyholders will file during the coverage period (under the assumption <strong>of</strong> the Law <strong>of</strong> Large<br />

Numbers). The computation <strong>of</strong> this premium relies on a statistical model incorporating all<br />

the available information about the risk. The technical tariff aims to evaluate as accurately as<br />

possible the pure premium for each policyholder via regression techniques. It is well-known<br />

that market premiums may differ from those computed by actuaries; see, e.g., Coutts (1984)<br />

for a discussion. In that respect, the overall market position <strong>of</strong> the company compared to its<br />

competitors with regard to growth and pricing is crucial.<br />

Sometimes, motor ratemaking is performed on panel data. Bringing several observation<br />

periods together has some advantages: it increases the sample size and avoids granting too<br />

much importance to a single calendar year (during which the particular weather conditions<br />

could have increased or decreased the number <strong>of</strong> traffic accidents, for instance). However,<br />

this induces some dependence in the data, since observations relating to the same policyholder<br />

across time are expected to be correlated. The analysis <strong>of</strong> correlated data with Poisson<br />

marginals arising from repeated measurements can be performed with the help <strong>of</strong> Generalized<br />

Estimating Equations (GEEs). GEEs provide a practical method with reasonable statistical<br />

efficiency to analyse such panel data. GEEs also give initial values for maximum likelihood<br />

procedures in models for longitudinal data.<br />

2.1.2 <strong>Risk</strong> Sharing in Segmented Tariffs<br />

The following discussion is inspired by the paper by De Wit & Van Eeghen (1984).<br />

Consider a portfolio <strong>of</strong> n policies from motor third party liability insurance. The random<br />

variable Y i models a quantity <strong>of</strong> actuarial interest for policy i (for instance the amount <strong>of</strong> a<br />

claim, the aggregate claim amount in one period or the number <strong>of</strong> accidents at fault reported<br />

by policyholder i during one period). In order to explain the outcomes <strong>of</strong> Y i , the actuary has<br />

observable covariates Xi<br />

T = X i1 X i2 at his disposal (e.g., age, gender and occupation<br />

<strong>of</strong> policyholder i, the place where he resides, type and use <strong>of</strong> his car). However, Y i also<br />

depends on a sequence <strong>of</strong> unknown characteristics Zi<br />

T = Z i1 Z i2 (e.g., annual mileage,<br />

accuracy <strong>of</strong> judgment, aggressiveness behind the wheel, drinking behaviour, etc.). Some <strong>of</strong><br />

these quantities are unobservable, others cannot be measured in a cost efficient way.<br />

The ‘true’ premium for policyholder i is EY i X i Z i . It is the function g <strong>of</strong> X i and Z i<br />

that is the ‘closest’ to Y i , in the sense that EY i − gX i Z i 2 is minimum for gX i Z i =<br />

EY i X i Z i . If the insurer charges EY i X i Z i to policyholder i, then the policyholders pay<br />

premiums that absorb the inter-individual variations (that is, the variations <strong>of</strong> the premiums<br />

due to the modifications in personal characteristics X i and Z i , the magnitude <strong>of</strong> which are<br />

quantified by V [ EY i X i Z i ] ). The company covers the purely random intra-individual risks<br />

(that is, the random fluctuations <strong>of</strong> Y i , which are quantified by the variance E [ VY i X i Z i ]<br />

<strong>of</strong> the outcomes <strong>of</strong> Y i once the personal characteristics X i and Z i have been fixed). <strong>Risk</strong>

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

Saved successfully!

Ooh no, something went wrong!