Ruin Probability and distribution fit for different business lines

I believe we need calculate ruin probability for a company as a whole to determine the solvency?

Suppose a company deals with multiple business lines such as Motor, Fire, Health, Marine, WC, Aviation and etc, the claim paid amount (severity) which we are suppose to use to arrive at the compound distribution and the experience of each business line which decides which distribution fits best would be different for each one of them.

Considering the above scenario

1) Are we suppose segregate the premium, claims and surplus for each business line separately and decide which distribution fits each business line separately and there by arrive at ruin probability separately?

or

Can we use the entire company's claims data ignoring above issues to fit distribution?

2) Generally how many years historical claims data is required to fit distribution?

3) I am planning to ignore reinsurance and use only gross data.

4) I am also planning to use entire data to fit distribution and would ignore fits based on sample data for better results. A

Any suggestions?


Ruin Probability and distribution fit for different business lines