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.

Bonus-Malus Systems with Varying Deductibles 287<br />

monetary units. The distribution <strong>of</strong> S is then determined by Panjer’s algorithm. <strong>Claim</strong><br />

amounts are discretized according to Method (7.4).<br />

7.4.3 Annual Deductible<br />

The relativities computed for the scale −1/top are displayed in the second column <strong>of</strong> Table 7.1<br />

and for the scale −1/ + 2 in Table 7.2. In the pure bonus-malus case, it is thus clear that the<br />

r l s associated with the upper levels are considerable (more than 300 % for level 5 in case <strong>of</strong><br />

the scale −1/ + 2). Now, let us replace the r l s in the malus zone (i.e. levels 1 to 5) with an<br />

annual deductible d l . In order to obtain the d l s from equation (7.10), we first discretize the<br />

claim sizes using Method (7.4). Finally, (7.10) is solved numerically using routines available<br />

from the SAS R /IML package.<br />

The third column <strong>of</strong> each table displays the new relativities. In this case, the maluses<br />

disappear (r l = 100 % for l = 15) and are compensated for by the deductibles listed<br />

in the two last columns. The fourth column shows the deductible to be applied if the loss<br />

amounts are Negative Exponentially distributed and the last column shows the deductible to<br />

be applied if the loss amounts are LogNormally distributed. Since the LogNormal distribution<br />

has a thicker tail than the Negative Exponential one, we expect larger amounts <strong>of</strong> deductible<br />

for the former. This is indeed the case, as can be seen from Tables 7.1 and 7.2.<br />

The very high r l s in the second column induce high amounts <strong>of</strong> deductible, even in the<br />

Negative Exponential case. Therefore, this solution seems to be difficult (if not impossible)<br />

to implement in practice.<br />

Table 7.1 Results for an annual deductible varying according to the level occupied in the malus zone<br />

and 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 1973 % 100 % 18 367 35 253<br />

4 1709 % 100 % 14 028 29 303<br />

3 1507 % 100 % 10 472 23 544<br />

2 1348 % 100 % 7470 17 934<br />

1 1220 % 100 % 4897 12 452<br />

0 547 % 54.7 % 0 0<br />

Table 7.2 Results for an annual deductible varying according to the level occupied in the malus zone<br />

and 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 3091 % 100 % 34 576 50 794<br />

4 2414 % 100 % 25 076 42 704<br />

3 2077 % 100 % 20 004 37 241<br />

2 1429 % 100 % 9027 20 933<br />

1 1302 % 100 % 6563 16 079<br />

0 624 % 62.4 % 0 0

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

Saved successfully!

Ooh no, something went wrong!