Association Rules for Understanding Policyholder Lapses

For automobile insurance, it has long been implied that when a policyholder made at least one claim in the prior year, the subsequent premium is likely to increase. When this happens, the policyholder may seek to switch to another insurance company to possibly avoid paying for a higher premium. In such situations, insurers may be faced with the challenges of policyholder retention by keeping premiums low in the face of competition. In this paper, we seek to find empirical evidence of possible association between policyholder switching after a claim and the associated change in premium. In accomplishing this goal, we employ the method of association rule learning, a data mining technique that has its origins in marketing for analyzing and understanding consumer purchase behavior. We apply this unique technique in two stages. In the first stage, we identify policyholder and vehicle characteristics that affect the size of the claim and resulting change in premium regardless of policy switch. In the second stage, together with policyholder and vehicle characteristics, we identify the association among the size of the claim, the level of premium increase and policy switch. This empirical process is often challenging to insurers because they are unable to observe the new premium for those policyholders who switched. However, we used a 9-year claims data for the entire Singapore automobile insurance market that allowed us to track information before and after the switch. Our results provide evidence of a strong association among the size of the claim, the level of premium increase and policy switch. We attribute this to the possible inefficiency of the insurance market because of the lack of sharing and exchange of claims history among the companies.