Do you include CWPs in Frequency?

Hello world! Didn’t realize an AO replacement was made, glad to find this place. I normally frequented Political Issues but for my first post, I have an actual actuarial question which I should probably know the answer to by now.

Do you include Closed Without Payment claims in your frequency calculation? I think I’ve seen it both ways before so perhaps either is acceptable. But excluding them seems correct to me.

You can go either way, you just have to make sure your severity calculations mirrors your selection. I.E. if frequency is going to be increased due to the inclusion, severity should be decreased due to payments of amount 0. That way pure premiums should theoretically work out to be the same in either case.

1 Like


Can be anything you want as long as you define it consistently.

1 Like

CWP is a horrible abbreviation for closed without payment, though.


I’d definitely exclude them as a general rule of thumb, that said it can depend. Maybe you’re short on claims but have a long history of data and closed without payment can give you insight into the true frequency of non-zero claims, in which case it can be very useful.

And then of course either way need to be careful you’re consistent on the severity side.


I prefer CNP (closed no pay) or CWOP (closed without pay). I’ve seen CWP used for both closed with pay and closed without pay. Ugh.

I sometimes show frequency with the CNPs, and sometime without, and sometimes I show just the frequency of CNPs, especially if it’s changing and that might indicate something interesting. I can’t answer “which do you use” because I use the one that makes sense in the context I am using it.


You might also look at what are the goals of your analysis.

Do these CNP claims incur material DCC expenses? If so, you might want to be sure to include that into your calculations/decisions.

Bottom line, you might look at your results when you include CNP (with consistent definitions) in the process vs. when you exclude them and compare results. This exercise might provide insight on whether or not it’s worth the effort to include/exclude them.

I agree with other comments that it often doesn’t make much difference with most lines as long as you are consistent.

In general I prefer to exclude closed without pay. In conversations with claims departments and executives, average severity and case reserve conversations can make more sense to them if the CNP are removed if they are a meaningful proportion of counts.

With any lines where the predominant portion of claim counts are CNP, I would highly recommend excluding them from frequency/severity reporting. Umbrella/Excess lines in particular can have the vast majority of claims as CNP.

1 Like

Related question that’s not really worth starting a new thread for: does anybody remember the old paper on estimating ultimate claim count? It’s really pretty straight forward, IIRC.

There’s a particular paper on that? I don’t recall. In a lot of lines, you can just take the reported as ultimate for every period more than a couple months old. If that’s not good enough, and if there are a lot of claims and relatively stable reporting patterns, I often just develop them. If there are fewer claims or the reporting patterns are more variable, I may do a BF looking at exposure, applying some trend to the a priori frequency.

1 Like