Understanding the Poisson Log-Bilinear Regression Approach

Mortality tables play a very important role in planning for health care systems and in computing life insurance premiums. For the past few decades, there has been an increase in the number of studies about estimating and forecasting mortality tables as a response to mortality improvements. In the early 1990 s, the Lee-Carter model was developed and has been widely used for studies on mortality rate and has been subjected to some modifications for improvements. This paper utilized one of these modifications, which was proposed by Brouhns, et.al. (2002). The number of deaths, a count random variable, is said to be well-suited to the Poisson distribution. Thus, this makes the Poisson log-bilinear regression model appropriate in modeling and forecasting mortality. This book aims to be able to fit the Poisson log-bilinear model to the number of deaths for some northern European countries. A maximum likelihood estimation technique is used to estimate the parameters of the model with the aid of LEM program and SAS. A chi-squared goodness of fit test was performed to test significance and usability of model. Lastly, the deviance residuals for the estimated number of deaths were taken.

Autorentext

Denzel Justin S. Chua and Stefanni S. De Guzman both graduated with a Bachelor's Degree in Statistics major in Actuarial Science from De La Salle University in the Philippines. They are both currently working as Actuarial Analysts.