Exam MAS-II Practice Question 5
Study guide: Exam MAS-II
An actuary fits an AR(1) model to 120 months of claim counts for a homeowners book. The correlogram of the model's residuals shows every autocorrelation for lags 1 through 24 inside the approximate 95% bounds \pm 2/\sqrt{120} = \pm 0.18, except for a single autocorrelation of 0.35 at lag 12.
Which of the following conclusions is best supported?
- A) The residuals still carry dependence at a twelve-month lag, so the model has missed a seasonal component
- B) The residuals are white noise, since one exceedance among 24 lags is what chance alone produces
- C) The autoregressive order is too low, so the model should be refitted as an AR(2)
- D) The series still contains a trend, so it should be differenced before an AR model is fitted
- E) The residuals are not normally distributed, so the claim counts should be log-transformed
Answer this question and see the worked solution