Hello, dear friend, you can consult us at any time if you have any questions, add WeChat: daixieit

MATH 374

FINAL EXAM 2022

STATISTICAL METHODS IN INSURANCE FINANCE

1.   Consider the following loss function

L(θ, d) =  - log ╱  、 - 1.

Let X1 , . . . , Xn  be exchangeable so that the Xi   are conditionally independent given parameter θ . Suppose that Xi  | θ ~ Poisson(θ) and θ ~ Gamma(α, 1) with α > 1.

Find the Bayes rule of an immediate decision when π(θ) is the prior distribution. [10 marks]

2.   The number of claims in a year has a Poisson distribution with mean λ . The parameter λ has the uniform distribution over the interval (1, 10).

(i). If an insured had one claim during the rst year, estimate the expected number of claims for the second year using B¨uhlmann credibility.   [5 marks]

(ii). If an insured had one claim during the rst year, estimate the expected

number of claims for the second year using Bayesian credibility.   [5 marks]

3.   Let {Xi  | θ}i1 be normally distributed with mean µ (known) and variance ϑ .

(i).  Show that f (xi  | ϑ) belongs to the exponential family and for X = (X1 , . . . , Xn ) state the sufficient statistics for learning about ϑ .                      [10 marks]

(ii). Write down the likelihood for X = (X1 , . . . , Xn ). By viewing the likelihood

as a function of ϑ , which generic family of distributions (over ϑ) is the likelihood a kernel of?                                                                   [10 marks]

(iii). Let ϑ be inverse Gamma with parameters a and b, i.e.  ϑ ~ InvG(a, b).

By rst nding the corresponding posterior distribution for ϑ given x = (x1 , . . . , xn ), show that this family of distributions is conjugate with respect

to the likelihood f (x | ϑ).                                                             [10 marks]