Page 225 - Contributed Paper Session (CPS) - Volume 1

CPS1283 Kelvin H.C.Y et al.
The Generalized Linear Model output is relatively easy to explain to
stakeholders. The riskiness of a specific rating factor’s characteristics is
expressed through the fitted and computed parameters.
Based on the hypothetical examples above
• The higher relativity factor of 1.250 for Construction Class 2
compared to Construction Class 1A implies that, all else being
equal, Construction Class 2 are 25% more risky compared to
Construction Class 1A. Hence a 25% loading is justified for
Construction Class 2 relative to Construction Class 1A.
• The lower relativity factor of 0.450 for Residential properties
compared to Hotel/Offices properties implies that, all else
being equal, Residential properties are 55% less risky compared
to Hotel/Offices properties. Hence a 55% discount is justified
for Residential properties relative to Hotel/Offices properties.

4. Discussion and Conclusion
Generalised Linear Models
One-way analyses are simple to understand and help summarize key
explanatory metrics such as average premiums, loss ratios, claim frequencies,
and average claims sizes. However, one-way analyses can be distorted by
correlations between rating factors. For example, young drivers may tend to
drive more affordable car make/models. Hence, a one-way analysis of car
make/models may suggest that the affordable car make/models have higher
claims experience. However, the underlying reason for the higher claims
experience may not be car make/model specific. Rather, the underlying reason
for the higher claims experience for the affordable car make/models may be
the driver age/experience.
Using Generalised Linear Models would help adjust for correlations and
allow for investigation into interaction effects.

Actuaries’ role during the Phased Liberalisation of the Malaysian Motor
and Fire Tariffs
As the Malaysian general insurance industry transitions into a phased
liberalised environment, the board and senior management are expected to
leverage on the advice of a professionally qualified actuary in fulfilling their
governance requirements and the introduction of new products.
The actuaries have widely adopted Generalised Linear Models as their
individual risk pricing model for Motor and Fire products. The outputs of the
Generalised Linear Model are easy to explain and implement into information
technology systems.

214 | I S I W S C 2 0 1 9