IBNR

[josie]

The company I work for is looking to shorten the time it takes to close the books on a monthly basis. They are looking to receive the IBNR from us by the 4th calendar day of the month. Considering that we do not get our monthly data until the 4th business day of the month, it's impossible to do this unless we report using a 1-month data lag.

Does anyone else close in under 5 days and if so, how do you handle it? (That is, do you get the data early enough or do you report using the data lagged one month. If you are using a 1-month lag, do you use loss ratios or something like that to calculate the current month's IBNR?)

Thanks for your help.

[Patience]

What product are you working with??

If it is fairly predictable based on seasonality & monthly fluctuations you should be able to get a reasonable estimate based on three weeks of claims data.

You will need to monitor actuals vs. these estimates to get a truer feel have how you are doing. But it is possible.

[josie]

I work with a medical health company (HMO to indemnity, prescription, and dental). Our files are only updated monthly as of now, so 3 weeks worth of data is not an option at this point. But it is something worth considering if we can update our files more frequently. Thanks.

[I_actuate™]

Funny this should come up. My company is changing as well. We are currently on a 14 day closing for quarters. By the year 2003, we are going to close in 3 days.

We too deal with TPA's only and get our data like day 3 or 4. I just don't see how it's going to be possible.

[Dr T Non-Fan]

1. Try getting claim data on a daily basis. That way a giant extract isn't waiting in line with the other month-end jobs. It would only be 1/23 or so of that. It can go in a quicker-job queue.
2. The day before the month-end should be good enough to sweep enrollment files for figuring out who was covered in the most recent month, and any restatements of previous months. This assumes that your enrollment files have a start date and a cancel date.
3. Tell them to shove off. This is extremely sensitive actuarial analysis that requires the utmost of attention, blah, blah. Then threaten them with your wand.

Seriously, though, you should try to get your data by the first of the month. You'll have to convince those who are pressing to press the systems area to put your extracts at the top of the queue.

[CJL]

We've used the one-month-lag for several years now, and it's worked out pretty well. To estimate the last month, we use a loss ratio and some judgement, and have established pretty good communication with Finance on what weird things have gone on (e.g., if there is an entry that is going to go in, but isn't in what we have to make the loss ratio off of, they're good about letting us know to adjust our number).

[CJL]

Oh, yeah - for December, the close schedule is delayed a few days and there is a big rush to do everything with current data.

[Dr T Non-Fan]

A month lag? That's amazing. I'll have to carbon-date your company one of these days.

[CJL]

We used to come up with IBNR in 2 or 3 days and do it all current. Then other people spent the next 3 weeks criticizing what we booked. This way, there's more time to investigate and be thorough, and we look much better in front of the rest of the company we we've checked into most of the questions they come up with.

Also, being a fairly large and rapidly growing company has caused us to get some pushback from some systems folks. We have one enormous data warehouse, and it's relatively new, so they don't have all of the kinks worked out. The loading is slow, but the big kink is run time, primarily due to insufficient indexing. Even if the data were ready on the 2nd or 3rd day (which it usually isn't), we'd be lucky to turn it around into lag triangles and such by the 4th day, much less look at it.

There's always tomorrow (or next year, or 3 years from now). We are making progress, so maybe there's still hope.

[Dr T Non-Fan]

I know the type:
1. "How come you were wrong last month about IBNR?"
2. "Why can't you be more accurate?"
3. "What's this thing attached to my shoulders doing up my ..."

Your remark about December is interesting: we take MORE time for the year-end close. We do a normal close, but the books will stay open for an extra month while we wait for an extra month's worth of claims in order to be more accurate, given the audience.

[CJL]

We used to do something close to that, but going public has pushed up the Finance timeframes, naturally without pushing up the data timeframes.

[Rockhound]

If you are dealing with a managed care product that has authorizations or referrals associated with it, you can use that information--its much more timely than payments.