Ulae

[iActuary]

This is usually not covered when we talk about reserving. But in practice, regulators normally require to reserve for this. So I thought you guys can give me some thoughts about it!

As far as I am aware, there are two commonly used approaches:
1) Classical (traditional) approach, or called "paid-to-paid ratio method";
2) Kittel approach (a refined approach based on (1)).

Now, I am trying to understand the logic of a newly seen (at least to me) approach used in the industry. This is described as below:

First step - Derive the ULAE ratio by dividing the actual claims-related expenses by the best estimate of claim liability held at the beginning of financial year.

Second step - Apply the ratio to the best estimate of claim liability held at the end of financial year to derive the indirect claims expense provision as at the valuation date.

In my understanding, this approach somehow projects only the claims expense expected to be incurred for the next financial period, not the total expense expected to be needed to settle all the outstanding claims. Am I missing something? Could someone help me please?

Thanks a lot.

BTW, I suggest this topic to be included in the ST7 notes in the future.

 

[Ian Senator]

You're right, none of this is covered by either ST7, ST8 or SA3 at the moment. If you can give me some links to material on these methods that you've seen elsewhere, I'll forward it on to the Profession so that they can consider whether they want to include it in the future syllabi.

 

[Calum]

There's an argument to be made that if you have secure access to capital, then you should only reserve for claims arising in the next period. At the end of that period, you reserve for the subsequent period and raise capital as required.

 

[iActuary]

This can be a useful start: http://www.casact.org/pubs/forum/03fforum/03ff093v2.pdf

Another comment on the scope of ST exams is that the topic has been quite focused on "claim cost", and we miss out on other areas that may need actuarial inputs as well, e.g. expense analysis and ALM (for CAT and liquidity risks in particular), though they are generally less applicable to GI actuaries than to Life actuaries.

Not sure if Capital Modelling was introduced to address this somehow?

 

[Ian Senator]

thanks for this - I'll forward on to the Profession.

I rather suspect the 'loadings' elements of premiums are covered less thoroughly as (arguably) they are covered more by CA1 these days. Expense analysis and ALM being a particularly good example!

 

[mario]

This is usually not covered when we talk about reserving. But in practice, regulators normally require to reserve for this. So I thought you guys can give me some thoughts about it!

As far as I am aware, there are two commonly used approaches:
1) Classical (traditional) approach, or called "paid-to-paid ratio method";
2) Kittel approach (a refined approach based on (1)).

Now, I am trying to understand the logic of a newly seen (at least to me) approach used in the industry. This is described as below:

First step - Derive the ULAE ratio by dividing the actual claims-related expenses by the best estimate of claim liability held at the beginning of financial year.

Second step - Apply the ratio to the best estimate of claim liability held at the end of financial year to derive the indirect claims expense provision as at the valuation date.

In my understanding, this approach somehow projects only the claims expense expected to be incurred for the next financial period, not the total expense expected to be needed to settle all the outstanding claims. Am I missing something? Could someone help me please?

Thanks a lot.

BTW, I suggest this topic to be included in the ST7 notes in the future.

Why would the method only give you a ULAE estimate for the coming year? Isn't it just using (the simplistic assumption) that ULAE as a proportion of indemnity remains constant?

As it's being applied to your outstanding case reserves, you're getting an estimate of the ULAE required to pay all outstanding claims, which is what you want?

If anything, to the extent that some of your ULAE will be fixed costs not rising with indemnity settlements, you could argue that ULAE is overstated when reserves are increasing (and similarly, there is a danger of underestimation when reserves are decreasing, if fixed costs won't be covered?)

I'm thinking as I type here but you should probably at least split ULAE into fixed and variable (with your fixed coming from future business plans, and your variable using this ratio approach), to get a ULAE estimate?

I've never heard of the other methods, but will take a quick look now. Thanks for the link

 

[iActuary]

Why would the method only give you a ULAE estimate for the coming year? Isn't it just using (the simplistic assumption) that ULAE as a proportion of indemnity remains constant?

As it's being applied to your outstanding case reserves, you're getting an estimate of the ULAE required to pay all outstanding claims, which is what you want?

In my understanding, the derivation is just to get the relationship between the opening balance of outstanding claims and the ULAE incurred during the past year. So when you apply it to the equivalent closing balance, you are somehow estimating the ULAE for the next period (e.g. 10*(1/10)=1). So, if the company is not able to settle all the outstanding claims as at the end of next year, it will still need additional ULAE to settle the remaining portion.. (I hope I am being clear enough, at least for the exam)

If anything, to the extent that some of your ULAE will be fixed costs not rising with indemnity settlements, you could argue that ULAE is overstated when reserves are increasing (and similarly, there is a danger of underestimation when reserves are decreasing, if fixed costs won't be covered?)

I'm thinking as I type here but you should probably at least split ULAE into fixed and variable (with your fixed coming from future business plans, and your variable using this ratio approach), to get a ULAE estimate?

I've never heard of the other methods, but will take a quick look now. Thanks for the link

Yes I agree. In this case, I would opt for a more detailed expense analysis. But for a stable past experience, the above-mentioned method probably will not work..

 

[td290]

First, huge thanks to iActuary for raising this issue. It wasn't something I'd thought about, precisely because it's not really discussed in the Core Reading.

I'm with iActuary on this. The method seems to only reserve for ULAE incurred over the next year. But I would be interested to know where you saw it. It's possible to think of scenarios where the approximation might not be too unreasonable, e.g. a stable book of short-tailed business.

Of course, as mario points out, a more detailed analysis might look at the split between fixed and variable expenses. But fixed expenses would need to be allocated by line of business and pro-rating them by outstanding reserves might not be too unreasonable a way of doing this.

Another thing to bear in mind is that, rightly or wrongly, reserving actuaries are very often not in possession of reliable information on fixed expenses. They may well not regard them as particularly material in comparison to the sum of claims costs, ALAE and variable ULAE. This might particularly be the case for a portfolio of large heterogeneous risks where individual claims costs are significant and specialist claims handling is required with the majority of expenses being easily attributable to a particular claim. The fixed ULAE would be relatively small and would cover the more trivial elements of the process such as the initial logging of the claim on the system.

I would take issue with Calum's comment though. A company with secure access to capital may take the view that it only needs to hold sufficient capital to see the business through one year. This is not true of reserves though (and of course, by reserves, I mean technical provisions.) This is because, in the event that the company was wound up, it would have to pay an amount equivalent to the reserves up front in order to transfer its liabilities to a third party and it would have no way of raising this money if it didn't already have it.

 

[mario]

In my understanding, the derivation is just to get the relationship between the opening balance of outstanding claims and the ULAE incurred during the past year. So when you apply it to the equivalent closing balance, you are somehow estimating the ULAE for the next period (e.g. 10*(1/10)=1). So, if the company is not able to settle all the outstanding claims as at the end of next year, it will still need additional ULAE to settle the remaining portion.. (I hope I am being clear enough, at least for the exam)

It took me a while, but I get what you're saying now - as you're using only one year's ULAE in the numerator, the ratio can only give you an approximation for next year's ULAE, not the ULAE reserve.

What you need in addition to the ULAE cost per year assumption is an assumption of how long it takes claims to settle. I guess... do some kind of exposed to risk analysis, where you count how many months/days/weeks/whatever have claims remained opened so you can estimate how much longer you need a ULAE allowance. That'd take you to ultimate.

Of course, if someone has done that once, you can get a rolled-forward ULAE reserve really quickly by applying the ratio of ULAE : best estimate reserves to your new best estimate reserves.

I hope ULAE comes up in the exam now

 

[iActuary]

I'm with iActuary on this. The method seems to only reserve for ULAE incurred over the next year. But I would be interested to know where you saw it. It's possible to think of scenarios where the approximation might not be too unreasonable, e.g. a stable book of short-tailed business.

One of my clients did it this way. And I struggled to persuade him.

They have a medium term portfolio (about 3-5 years of full run-off). Experience is rather stable.

 

[buddingactuary]

I found this while looking for answers about ULAE estimation. Not sure if this topic is still under discussion. Looking forward to some more views on this.

Admin edit: unfortunately the original attachment was lost in a server upgrade. Here are a couple of links to related papers. We're not sure if either is the one originally referenced:

https://www.casact.org/sites/default/files/old/forum_03fforum_03ff093v2.pdf
https://www.actuaries.org.uk/documents/unallocated-loss-adjustment-expense-provisions