CAS Exam 8 Fall 2016, Question 4
Study guide: Exam 8
An actuary is conducting a generalized linear model (GLM) analysis on historical personal automobile data in order to develop a rating plan.
Part a (1.5 points)
Argue against the following factors being included as predictors in the actuary's GLM analysis:
- i. Limit of liability.
- ii. Number of coverage changes during the current policy period.
- iii. ZIP code of the garaging location of the automobile.
Part b (1 point)
The actuary is modeling pure premium with a log-link function and a Tweedie error distribution (1 < p < 2). Provide two arguments against the inclusion of deductible as a predictor in the actuary's GLM analysis.
Part c (0.5 points)
Other than including deductible as a predictor in the GLM, describe how to determine deductible relativities and how such relativities can be incorporated in a GLM.
Answer this question and see the worked solution
Concepts tested
- Generalized Linear Model
- Offset Variable
- Tweedie Distribution
- Loss Elimination Ratio
- Rating Factors
- High Dimensional Variables