quick question about age band compression

[stevepax]

Suppose I have a base rate that I then multiply by a factor for a member's age band to get final premium. Now, suppose I want to limit the spread of the age bands - like now, the largest factor is three times as big as the smallest, and I would like to have the largest factor be only 2 times as large as the smallest.

What's my procedure for getting the new age factors in order?

[Zee]

Originally Posted by stevepax

Suppose I have a base rate that I then multiply by a factor for a member's age band to get final premium. Now, suppose I want to limit the spread of the age bands - like now, the largest factor is three times as big as the smallest, and I would like to have the largest factor be only 2 times as large as the smallest.

What's my procedure for getting the new age factors in order?

Depends on a lot of things... there is no set answer to this question.

You could cap rates at the top end, floor them at the bottom, flatten the slope around the pivot point of the age slope, some combo of those approached, etc. Many different approaches. Each will drive a different expected distribution by age. You should make sure that in the end you have an aggregate per member average that you can live with, since most likely the individual ages will not be rated to stand alone.

If your expected distribution by age is already very small at one of the endpoints, you might want to start by cutting off the slope on that end. You'll be taking a step toward your goal without much associated cost.

[G.O.A.T.]

Originally Posted by Zee

Depends on a lot of things... there is no set answer to this question.

You could cap rates at the top end, floor them at the bottom, flatten the slope around the pivot point of the age slope, some combo of those approached, etc. Many different approaches. Each will drive a different expected distribution by age. You should make sure that in the end you have an aggregate per member average that you can live with, since most likely the individual ages will not be rated to stand alone.

If your expected distribution by age is already very small at one of the endpoints, you might want to start by cutting off the slope on that end. You'll be taking a step toward your goal without much associated cost.

In addtion, look at your competitors age slopes for ideas. Rate filings will have them for individual insurance. For group insurance, it is much more difficult.

[G.O.A.T.]

Originally Posted by Zee

Depends on a lot of things... there is no set answer to this question.

You could cap rates at the top end, floor them at the bottom, flatten the slope around the pivot point of the age slope, some combo of those approached, etc. Many different approaches. Each will drive a different expected distribution by age. You should make sure that in the end you have an aggregate per member average that you can live with, since most likely the individual ages will not be rated to stand alone.

If your expected distribution by age is already very small at one of the endpoints, you might want to start by cutting off the slope on that end. You'll be taking a step toward your goal without much associated cost.

In addition, look at your competitors age slopes for ideas. Rate filings will have them for individual insurance. For group insurance, it is much more difficult.

[Random Walk]

Originally Posted by G.O.A.T.

In addition, look at your competitors age slopes for ideas. Rate filings will have them for individual insurance. For group insurance, it is much more difficult.

This is worth mentioning twice. You don't want to have much cheaper rates than your competition at, say, age 60-65 if those rates aren't self-supporting, 'cause you'll attract a lot of business that doesn't pay for itself.