Applying grey models to categorical data- policyholder’s characteristics and purchase decision of unit linked insurance

Unit linked insurance is a form of insurance where the policyholder bears the investment risk. The purchase decision here denotes only whether to buy unit linked insurance or not. It is a binary variable and the response variable in this paper. The consumers´ characteristics have a strong impact on their purchase decision. The consumers´ characteristics include cultural, social, individual and psychological factors. Policyholders´ residence area is proxy for cultural factors; policyholders´ marital status is proxy for social factors; policyholders´ total market value of assets and balance of other investments and savings are proxy for individual factors; policyholders´ risk attitude is proxy for psychological factors. They are explanatory variables in this study which are either categorical data or binary data. We intend to test the applicability of grey models to categorical data and binary data, and attempt to figure out the relationship between the above-mentioned variables. Based on results of GM(1,N) and GM(0,N), policyholders´ risk attitude is the most important explanatory variable for their purchase decision; policyholders´ residence area and their total market value of assets influence their purchase decision indirectly by way of risk attitude; policyholders´ marital status might influence their purchase decision indirectly by way of other unknown factors; policyholders´ balance of other investments and savings have no significant influence on their purchase decision. Finally we justify the applicability of grey models to categorical and binary data.