SOA Exam ASTAM Sample Question 18
Study guide: Exam ASTAM
A claims study examines the frequency of automobile claims among 400 policies. An actuarial assistant has fit several frequency models (including Zero-Modified distributions) to the claims data using Maximum Likelihood Estimation. Summarized results are shown below.
| Poisson | ZM Poisson | Geometric | ZM Geometric | Neg Binomial | ZM Neg Binomial | |
|---|---|---|---|---|---|---|
| Parameters | 1 | 2 | 1 | 2 | 2 | 3 |
p_0 |
0.5625 | 0.5625 | 0.5625 | |||
\lambda |
0.6175 | 0.7339 | ||||
r |
2.1894 | 0.4963 | ||||
\beta |
0.6175 | 0.4114 | 0.2853 | 0.5715 | ||
| Log-Likelihood function | -427.8 | -426.2 | -430.2 | -423.2 | -424.5 | -423.1 |
| SBC Statistic | -430.8 | -432.2 | -433.2 | -429.2 | -430.5 | -432.0 |
\chi^2 Statistic |
6.07 | 3.29 | 14.53 | 1.26 | 5.15 | 1.14 |
\chi^2 Degrees of Freedom |
2 | 1 | 3 | 1 | 1 | 1 |
\chi^2 P-value |
Part a
Describe how the SBC statistic is calculated and verify the calculation of the SBC for the Negative Binomial model.
Part b
Perform a Likelihood ratio test to compare the ZM Poisson to the Poisson.
H_0: population is a Poisson
H_1: population is a ZM Poison (as printed)
Part c
Give ranges for the \chi^2 test p-values for each model.
Part d
What other criteria could be used to select a potential model?
Part e
Using the above scores and any other criteria, pick a preferred model and justify your choice.
Answer this question and see the worked solution
Concepts tested
- BIC
- Likelihood Ratio Test
- Chi-Square Goodness-of-Fit Test
- Hypothesis Testing
- Frequency
- Maximum Likelihood Estimation