Exam MAS-I Practice Question 104

Study guide: Exam MAS-I

An actuary fits five GLMs and, for each, estimates the dispersion parameter as the Pearson chi-square statistic divided by its degrees of freedom. Every model has about 2,000 residual degrees of freedom. The binomial model's response is, for each agency and year, the number of policies renewed out of the number offered renewal.

For which model does the dispersion estimate indicate that the assumed response distribution does not describe the variability in the data?

• A) A Normal GLM with identity link for a loss ratio: \hat\phi = 0.0225
• B) A Gamma GLM with log link for claim severity: \hat\phi = 1.45
• C) A Poisson GLM with log link for claim counts: \hat\phi = 1.04
• D) A Tweedie GLM with log link for pure premium: \hat\phi = 85
• E) A binomial GLM with logit link for agency renewal counts: \hat\phi = 3.1

Answer this question and see the worked solution

Concepts tested

• Dispersion Parameter
• Exponential Family
• Variance Function
• Binomial Distribution
• Gamma
• Tweedie Distribution

More questions from this set

• Exam MAS-I Practice Question 103
• Exam MAS-I Practice Question 105