Handout on a compound Poisson process and ruin probability WITH ANSWERS.

 

For a particular group of clients, a flow of claims arriving at an insurance company may be represented as a compound Poisson process, that is, we deal with the classical scheme. Let the amount of a particular claim be equal to either 2 or 3 or 4 with probabilities 1/4, 1/2, and 1/4, respectively. Let the mean number of claims the company receives per day is 10. Assume that the company chooses for its activity a relative loading coefficient \theta =0.1.

 

1.      Assume that the third claim has just arrived. What is the expected waiting time for the fourth claim? 1/10.

     What is the probability that the fourth claim will come in less than three hours? 1-exp{-(3/24)*10}.

2.       What is the probability that during two days less than 25 claims will come?        SUM_{k=0}^24 [(exp{-20}}(20)^k/k!]

3.      What is the expected value of an individual claim?      3.

4.      What is the expected value of the aggregate claim the company receives per day, per hour? 30, 30/24. What premium does the company get per day, per hour? 33, 33/24.

5.      Write an equation for the adjacent coefficient R. (1/4)exp{2R}+(1/2)exp{3R}+(1/4)exp{4R}=1+3.3R

6.      Does this equation involve \lambda ? NO.

7.      Solve the equation numerically, that is, find an approximate solution (in any way you like). I got around  .055

8.      Find an approximate value of the initial capital for which the ruin probability for the company will be less than 0.03.  ln(100/3)/(.055). About 64.

9.      Think how to answer all questions above, if a particular claim is, say, uniformly distributed on [2,4]. (You are not required to calculate everything.

          The first four answers are the same. In Question 5 you should replace      

                       (1/4)exp{2R}+(1/2)exp{3R}+(1/4)exp{4R} by [exp{4R}-exp{2R}]/2R.

10.    Think how to answer all questions above, if the number of claims during a day is exactly equal to 10, that is, we consider the discrete time scheme, and day is a unit of time.

                     The first four answers are the same. In Question 5 the equation is

                                 (1/4)exp{2R}+(1/2)exp{3R}+(1/4)exp{4R}=exp{3.3R}.