5/7/2012
1. Executive summary
The purpose of this study is to investigate the appropriateness of the existing uniform pricing scheme. Hypothesis tests are performed to test the mean difference in claim rates between insured properties that are located at different states (NSW or QLD), belong to a different housing type (HOUSE or UNIT) or made up of different materials (BRICK or OTHER). The mean differences in claim sizes are also tested in the same fashion. From the results, it is concluded that the uniform pricing scheme is inappropriate and adjustments are required to improve the performance of the insurance product. Recommended improvements include the adjustments to the policy pricing according to the location of the insured properties and the housing type of the properties. Important factors in setting up the internal model are also considered and they include the modelling of claim rate of storm claims separately from other claims.
3. Data analysis
3.1 Claim rates
Claim rate equals to the proportion of the number of claims out of the number of contracts. A hypothesis testing is carried out to test whether there is a significant difference between the claim rate in NSW and the claim rate in QLD. Several assumptions were made during the test. The claim rate is assumed to follow a binomial distribution and each claim is independent and identically distributed. While these assumptions are valid for claim rates of the claims other than storm claims (OTHER claims), it might not hold true for STORM claims as the data suggests that most storm claims seem to happen only one a few dates. Therefore it was decided to exclude STORM claims in the analysis. Similar methods and assumptions are used to test the difference in claim rates between the properties that fall into the categories BRICK and OTHER, and HOUSE and UNIT.
independent of each other and is identically distributed. The variances of the claims data in each group are also assumed to be different and this can be verified by the available data. Same methods are applied to test the difference in average claim size between the properties that belong to the categories BRICK and OTHER, and HOUSE and UNIT.
4. Results of analysis
4.1 Claim rates
There is no observed difference between the claim rates of OTHER claims in NSW and the claim rates in QLD. Similarly, there is no evidence suggesting that there exists a difference when comparing the claim rates of OTHER claims in properties which are built of BRICKS and properties which are built of OTHER materials. There is also no difference observed between the claim rates of a HOUSE and a UNIT. Overall, no significant relationship is observed between the claim rates of an OTHER claim and the different characteristics of an insured property.
5. Recommendations
5.1 Improvements to pricing scheme
There are strong evidence showing that the claim size has some correlation with the location of the insured property and the housing type of the insured property. This suggests that the existing uniform pricing scheme is inappropriate. Several adjustments to the pricing scheme are suggested: The pricing of the policy should be adjusted according to the location of the insured property. As shown in the data analysis section, the average claim size in NSW is higher than in QLD, therefore the premium charged on customers in NSW should be on average higher in order to cover the higher costs brought by the higher claim sizes. Extra costs will be
incurred as more technical analysis; marketing and administrative procedures will be required for a non-uniform pricing scheme. However they are most likely to be offset by the extra profit that will be generated by differentiating the pricing of policy according to their location. This is suggested by the huge difference between the average claim size in NSW and in QLD. The numerical figures are shown in the Appendix section. Apart from the location of the insured property, the pricing of policy should probably be adjusted according to the housing type of the property as well. However the financial feasibility will need to be confirmed using more in-depth analysis. Even though the claim size of a HOUSE is shown to be greater on average than the claim size of a UNIT, the numerical figures suggest that the difference is not substantially large. As mentioned previously, changing the pricing system of policies usually incur most costs on technical analysis, as well as the marketing and administrative procedures. Some cost-benefits analysis will be required to evaluate whether the extra profit generated by adjusting policy price according to housing type would offset those extra costs incurred.
6. Conclusion
After several hypothesis testings, the uniform pricing scheme is determined to be inappropriate. In order to improve the performance of the home insurance product, the pricing scheme needs to be improved by adjusting the policy price according to the location. There are also observed difference in average claims sizes in properties of different housing types. The development of an internal model is also highly important as it projects the future cash flows and tells us information on the minimum required level of reserve capitals.
Appendix
Table 1 Test statistics of OTH claims sizes of NSW vs QLD
NSW Mean Variance Observations Hypothesized Mean Difference df t Stat P(T<=t) one-tail t Critical one-tail P(T<=t) two-tail t Critical two-tail
QLD
38235.38 11267.79 2.94E+09 1.21E+08 540 0 607 11.20232 6.59E-27 1.647368 1.32E-26 1.96388 341
HOUSE Mean Variance Observations Hypothesized Mean Difference df t Stat P(T<=t) one-tail t Critical one-tail P(T<=t) two-tail t Critical two-tail
UNIT
28978.51 21994.35 2.21E+09 1.06E+09 732 0 290 2.195149 0.014473 1.650125 0.028945 1.968178 149
Self reflection
What Ive learnt in this assignment
During the process of doing this assignment, I have learnt a lot on how to apply the skills Ive learnt in the actuarial studies course to real life application in the work field. This assignment good requires critical thinking skill as the problem encountered in the assignment is not as straight forward as a typical tutorial question. There are no definite solutions to the problems. It is more about finding a good solution and understands the advantages, disadvantages and implications of it. In the process of finding the solution, good technical skills are important as it directly affects the accuracy and reliability of the results. Interpretation of results is also highly important as I need to relate numerical results into real life applications. However, finding the solution is only one aspect. The biggest thing I have learnt is how to communicate your ideas to the audience. Different audiences have different knowledge and background. Therefore it is highly important to adjust the choice of words and languages specifically to different groups of audience. During the peer review process, I have again learnt the important of communication skill. Its important to know how to evaluate someones work so that it is beneficial for them. It is often difficult to make criticism to people but keep them encouraged at the same time. I have learnt how to balance positive feedbacks with constructive criticism so that my colleagues can further improve their work.