Assume in a certain company has a Retirement Preparation Benefit that gives x times the employee wages at n-1 retirement age. The company can approved or disapproved an employee to get this benefit. What is the best method to calculate the proportion of the approved benefit using historical data ? Is it with \frac{\text{ # total approved employee}}{\text{ # total eligible employee}} or \frac{\text{ # total approved benefit amount}}{\text{ # total eligible (?) benefit amount}} and which one is acceptable in market practice ? From my understanding it should be the proportion of the employee and not the amount, but I can’t have a strong reason why
Depends on local laws
1 Like
So, if there is no explicit law that regulates this, then both of the methods are acceptable ?
This is a US dominated site. “No rules” doesn’t really exist.
Obviously if there is a law specifying which one you must use, then do that.
But more generally this is a question that comes up all the time in every kind of product line any time you are looking at any type of experience data. Do you go by count or weight by dollar amount?
Where possible, I find it best to collect both pieces of data side by side and look at both quotients together. If they are materially different, that says to me that I should try to understand why, and ideally break the data down into sub-categories to where the quotients are pretty close.
This isn’t always possible due to data limitations, time available, or overcoming the inertia of “but that’s the way we’ve always done it!” But it can be enlightening.
1 Like