PDA

View Full Version : Cape Cod


John Connor
04-16-2012, 11:41 AM
In the 2011 Cape Cod problem, the solution derives the loss ratio by using all years excluding the most recent?

Should this always be the case?

Will we lose points for using all years?

Vorian Atreides
04-16-2012, 11:45 AM
You shouldn't lose points for excluding the latest year if you state why. In many cases, if the CDF for the latest year is highly leveraged, it makes sense to exclude it since very little of the premium is "used up" and will likely distort the ratio.

But whether or not to exclude the latest year (or two) isn't an automatic deal. Just look at the CDF for the year(s) and make a judgment call.

John Connor
04-16-2012, 11:50 AM
ok, essentially it probably would not be marked off, for exam purposes.

Dedicated111
04-16-2012, 12:00 PM
just state your reasoning for excluding a year (or two), and you're good to go.

Vorian Atreides
04-16-2012, 12:08 PM
ok, essentially it probably would not be marked off, for exam purposes.
If you're referring to using all years to determine the ECR . . . you might lose some credit (although, I doubt you'd lose much) if the latest year (or two) had highly leveraged CDFs. With the move to evaluating at higher levels of Bloom's Taxonomy, they may be assigning points (or partial points) to recognizing whether or not to include years with highly leveraged CDF.

I guess we'll see in Aug how this will play out.

John Connor
04-16-2012, 02:50 PM
ok cool, it seems reasonable to then assume that it is best to exclude based on CDF. I imagine this is also probably true if the loss ratio from the most recent year is substantially different from prior years? On the 2011 exam the most recent was 64%ish while the other three were 69/70%

Vorian Atreides
04-16-2012, 02:59 PM
ok cool, it seems reasonable to then assume that it is best to exclude based on CDF. I imagine this is also probably true if the loss ratio from the most recent year is substantially different from prior years? On the 2011 exam the most recent was 64%ish while the other three were 69/70%
I agree.

Aw Yeah
04-18-2012, 01:10 AM
It is worth noting that the 2010 Cape Cod problem, which has an Ult LDF of 5 for the most recent year, uses ALL years in calculating the expected loss rate. Thoughts?

Aw Yeah
04-18-2012, 01:11 AM
although i think the point about being clear regarding the assumptions is probably worthwhile, and I'm inclined to exclude the most recent year based on the fact that the 2011 solution excludes it, even though the 2010 includes it.

Joshua0317
05-08-2012, 09:55 AM
How high is highly leveraged? CDF>3, CDF>2. I think 4 is too much, but I'm not sure.

Also, does highly leveraged only refer to CDFs? If we have high trend factors for past years, would those be considered highly leveraged in some instances?

Vorian Atreides
05-08-2012, 10:19 AM
"Highly leveraged" generally applies to CDF's.

This is because you're applying a factor to an amount that may have random fluctuations in its value. For example, it may be the case that in most years, your largest claims are realized in the 12 months after the end of the AY, making your 12-24 ldf very large on average. But one year, several of these large claims were realized in the first 12 months of the AY. These combine to produce an unreasonably high estimate of ultimate claims.

We don't have control over when claims emerge, but we have control over what development factors to apply to those claims that have emerged.

TheZaha
05-08-2012, 10:32 AM
IF it's highly leveraged, and you are using a weighted average, it should matter to much right? Won't be getting a lot of weight anyway

Vorian Atreides
05-08-2012, 12:04 PM
IF it's highly leveraged, and you are using a weighted average, it should matter to much right? Won't be getting a lot of weight anyway
The problem is that your weighting with a lot of historical data that places those large losses in the 12 months after the end of the AY and applying that factor to a year (more specifically, the latest AY or the latest two AY) that has some of those large losses in the first 12 months.

TheZaha
05-08-2012, 12:35 PM
The problem is that your weighting with a lot of historical data that places those large losses in the 12 months after the end of the AY and applying that factor to a year (more specifically, the latest AY or the latest two AY) that has some of those large losses in the first 12 months.

Not sure I follow. If you have lets say 5 AYs with CDFs of 3, 2, 1.5, 1.2, 1.05

the last year is leveraged and if you have relatively stable exposure base, the losses in that last year should be dwarfed by the loses in the more mature years. even if the last AY has a LR of 90 compared to the others who have 80, the weighted average should be very close to 80 and it's not entirely unreasonable that you have a deteriorating LR. Guess i'm just not seeing what evidence we might have to exclude it, even if it is highly leveraged

Vorian Atreides
05-08-2012, 12:42 PM
Not sure I follow. If you have lets say 5 AYs with CDFs of 3, 2, 1.5, 1.2, 1.05

the last year is leveraged and if you have relatively stable exposure base, the losses in that last year should be dwarfed by the loses in the more mature years. even if the last AY has a LR of 90 compared to the others who have 80, the weighted average should be very close to 80 and it's not entirely unreasonable that you have a deteriorating LR. Guess i'm just not seeing what evidence we might have to exclude it, even if it is highly leveraged
Are the losses in the most recent AY developing similarly to the losses in the earlier AY?

The development technique assumes that this to be the case, but in reality, is it?

Does it make sense to apply a factor to develop a portfolio of losses that has already developed faster than expected?