Abstract
We systematically explore the time-series properties of life insurance demand using a novel statistical procedure that allows multiple unobservable (but interpretable) components to be extracted. This methodology allows the data to be modelled in new and innovative ways. We find univariate series decomposition allows us to more easily explain the behaviour of life insurance demand over the sample period (1981–2003), than would otherwise be possible. A multivariate model (including a number of variables thought to influence demand) produces quite pleasing results overall. A SUTSE model involving demand and each of the explanatory variables in turn shows evidence of common components in all cases but one. Finally, an out-of-sample forecast comparison shows the univariate model to outperform the multivariate model for accuracy.
Get full access to this article
View all access options for this article.
