On the repeated purchase behaviors of auto insurance in Taiwan

This paper develops a discrete choice modeling framework with multinomial logit (ML) and nested logit (NL) specifications tested against the samples drawn from a non-life insurance company in Taiwan-each sample has revealed a consumer's actual repeated purchase behaviors of physical damage coverage (PDC) auto insurance policies for up to consecutive five years. The results of ML estimation indicate that car owner's age, vehicle brand and engine capacity are significant factors affecting the repeated purchase in the following years. In terms of modeling structures and statistical results, however, the NL models have shown superior to the ML model in elucidating the consumer's repeated purchase of PDC policies from the same insurance company. Our study provides valuable implications to the insurance companies as well as to the car dealers in developing effective marketing strategies to promote the written volumes of PDC auto insurance policies.