SOP 03-01 and UL policies

[mayreeh]

There seems to be a flurry of activity protesting the SOP as folks getting closer to implementing and realize what the SOP literally says with regards to the valuation of UL policies.

Just wondering what the folks out there are thinking?

The literal interpretation of the SOP (paragraph 26) says that if the mortality EGP is positive early on and negative later on, then you have to establish a reserve for "excess payments". Paragraph 25 explicitly defines "excess payments" as the excess of death benefit over fund value.

Hence, a block where there are no future losses in the mortality EGP would have no extra reserve established, but one with slight future losses would have to establish a reserve for the death on NAR - even on a policy with NO no lapse guarantees. (Even though the SOP explicitly states that it is intended to address new product features such as no lapse guarantees.)

JMO, but the literal interpretation makes no sense for UL. (OK for VA). I think that they just didn't think things through well and folks are so scared of SOX etc that they are reluctant to bend the rules in a logical direction.

However, will ACSEC step up and admit that - postponing or altering the SOP for UL? I've got no idea on the chances of that.

[mayreeh]

Is anyone out there even looking at SOP 03-01 yet for UL?

I can't believe that I am the only actuary in the country who is dumbfounded at the impact of this SOP....

[Double High C]

We're hout there (here?)

I have to refresh my memory of some of the deets, but based on my recollection / discussion with colleagues (some of whom attended the Valuation Actuary Symposium), I recall that there are differing opinions w.r.t. e.g. the interpretations of what you have quoted. (Perhaps I'll go into more deets in future posts; don't have time now, but I hope that this discussion continues.)

Certainly, the discontinuity to which you refer is an undesirable anomaly.

OTOH, something like this was needed to cause companies to reserve for future guarantees, not just ones that are very obvious, such as nontrivial NLGs, but also more inconspicuous ones, like e.g. moderately agressive COI or credited rate guarantees. For this purpose, I define "moderately agressive" in such a manner that it would include cases in which the mean of the distribution of the charge (credit) to be materially lower (higher) than the median. (My guess is that I am preaching to the choir on this.)

I am aware that some examples of what falls under I mention above can probably be captured by mechanisms already in FAS 97, while others can not.

Since the SOP has been finalized, the time for AcSEC to amend it has passed, though of course, I suppose that they could - and perhaps might - issue another SOP some time in the future.

[mayreeh]

I just can't believe that there isn't more discussion on this.

Have you guys talked to your auditors? Better yet, have you talked to the national partner at your auditor?

From what I here, all of the big four are interpreting this pretty much the same. If the ultimate COI's are less than the ultimate expected mortality, then you have to hold a reserve for the net amount at risk.

This would be the equivalent of the stat guys coming along and saying that if you met this criteria, you would have to hold a reserve for the difference between the death benefits and the stat reserves.

It is huge and no one is talking about it???????

ACLI has written a protest letter to ACSEC and FASB because this changes the fundamental accounting for a UL policy - even if there are no special guarantees and thus the policy isn't supposed to fall under the SOP.

Some companies have followed suit. I really don't understand why there isn't more discussion out there on this topic....

[BigEEE]

It's not as big for our company because we have proven to our auditor that we don't have future losses. We are close, but we never break the barrier. Some companies are probably holding FEL for this already, so it won't be a big hit.

We are expecting a bigger hit on the Annuity excess DB Liability.

We'll see what happens with the ACLI. our aduitor is skeptical it will change anything since it is so late in the process. If it is a big problem, I would work closer with your auditor to try to find a way out of it. ie(find some cohort group that doesn't fail and sell it to them)

[mayreeh]

I agree that the ACLI approach isn't likely to pass - but I'm still holding out hope.

The whole thing doesn't make sense to me. The SOP clearly says in the introduction that is there to define how to reserve for new things like GMDBs and no lapse guarantees.

However, paragraph 26 changes FAS 97 if there are future mortality losses.

Has anyone gotten their auditors to buy into the idea that part of the interest margin is to pay for mortality?

[JMO]

The Financial Reporting Section newsletter has an article this month (part 2 of 3) that discusses 03-1. See pages 16-17.

[mayreeh]

It is a good article.

It takes the position that only the mortality charges can be considered in this test. That is, if interest margin takes up some of the slack in later years and the mortality margin is negative, then you have to hold a reserve under the SOP.

I guess I'm just dreaming, but I'm still hoping that someone is going to wake up and realize how material this can be. I don't believe (based on all I've read and heard) that they really intended this to be that material. They make it very clear in the introduction that they aren't out to change FAS 97 - but then they go ahead and change it anyway.

Products that have been on the books for years with no future negative margins (in the aggregate) can end up holding a huge additional reserve for no apparent good reason.

From an actuarial point of view, I agree that the fund value isn't always a reasonable reserve given the wide range of products out there. However, reserving for the shortfall makes more sense than reserving for the entire net amount at risk.

[Old Timer]

Just got reference to this (http://www.actuary.org/pdf/practnotes/life_aicpa04.pdf) in my e-mail today. May be of interest.

[mayreeh]

Definitely of interest... I've been poring over it. However, our auditors have already told us that they disagree strongly with some sections of it.

They just won't tell us yet which sections they disagree with.....

[BigEEE]

A couple of questions:

1.) If you do have profits followed by losses, and your auditor says that you should hold a liability, does the SOP give guidance on what baseline mortality to use in determining your excess coi. Does it revert back to FAS97 UR methodology at this point?

2.) Has anyone seen the as a result of the reverse select and ultimate coi, that the Unearned Revenue Liability is actuall negative in the early durations? I understand this mecahnically becasue you are, in essence, amortizing FEL that you haven't really set up yet. What do you do with this negative balance? Do you zero it out or hold a negative liability. Also, because of this phenomenon, it suggests that the FEL is not a good mechanism to smooth profits because of the negative liability in early durations. I suppose a better approach is to have an anuual or monthly "bucket" for each FEL, and amortize each bucket separately.

Any comments would be helpful...Thanks

[BigEEE]

Another thing that I was finding is that the impact isn't quite so big because there is an offsetting DAC impact. Are you seeing this?

[mayreeh]

If you have profits followed by losses, paragraph 26 of the SOP defines a method for calculating a reserve that doesn't depend on defining an "excess COI" - so I'm not sure what you are referring to.

You have to calculate a benefit ratio and use this to calculate a retrospective reserve. Because this is constantly unlocked under FAS 97 methodology, the amount of "assessments" that are being deferred can change from period to period.

This can also lead to a negative liability (which would be zeroed out because the SOP says you can't hold a negative liability) if you have had more actual deaths than were projected.

The offsetting DAC impact does help... but how much depends on the design of the products. There are a number of companies that are seeing a huge impact due to this.

[BigEEE]

I guess my understanding is that there are two potential liabilities on the UL side. One is for the excess death benefits that would be paid when the policy has a no lapse provision and there account value is below zero.

The second liability is for the front ended COI charges which are disproportionate to the amount of mortality risk in each period. My understanding is that this was akin to the FAS97 treatment where you set up an unearned revenue liability for the amount of excess coi in each period. From reading paragraph 26, it looks like you set up the excess payments in each period as your liability, but how do you determine your excess payment? Excess over what?

For the first liability, we are generating projected excess benefits stochastically. If the second liability does not work like Fas97, where you select a baseline mortality rate and difference it from the COI charged to determine the excess, then I ahve no idea how it works under the SOP.

[Double High C]

I guess my understanding is that there are two potential liabilities on the UL side. One is for the excess death benefits that would be paid when the policy has a no lapse provision and there account value is below zero.

The second liability is for the front ended COI charges which are disproportionate to the amount of mortality risk in each period. My understanding is that this was akin to the FAS97 treatment where you set up an unearned revenue liability for the amount of excess coi in each period. From reading paragraph 26, it looks like you set up the excess payments in each period as your liability, but how do you determine your excess payment? Excess over what?

For the first liability, we are generating projected excess benefits stochastically. If the second liability does not work like Fas97, where you select a baseline mortality rate and difference it from the COI charged to determine the excess, then I ahve no idea how it works under the SOP.

Apparently the "two liabilities" are really not separate liabilities, but rather are really just different examples in which the (single "SOP-") liability is needed.

Note that in this post, what I am asserting is based on what appears to be the opinion of "mayreeh's auditors", which is also what the ACLI fears will become widely accepted unless AcSEC is persuaded to amend the language / scope of the SOP, or to delay its becoming effective until such time that a different interpretation gained wide acceptance. (The hope expressed by many is that things like the no lapse guarantee be treated on a stand-alone basis, and that in the case of front-loaded COIs, that other margins in the policy be considered in the determination of whether there are "profits followed by losses"; after all, we all know that there can be "margins" in any of the charges to cover items that are not thought of as being attributable to that item - such as COI loads covering some expenses.)

Note that in this post, I am trying to remain agnostic on which approach makes the most sense; the point is to illustrate the thinking of many (and help clarify it - including for myself).

In fact, in some cases, you might not need (what you call) the "first liability", e.g. if there are very high mortality loads in early years, and somewhat high mortality loads in the later years (but not as high proportionally as in the early years), and if the guarantee is sufficiently deep out of the money to cause the expected (as opposed to median) - based on e.g. your stochastic analysis - mortality margins in all years to be positive (and if all other expected margins are also positive).

In either case (the "first" or "second" liability), the excess payments would be the projected net amount of risk (i.e. benefit minus account balance), and expected assessments includes all charges.

Edit: Now, from rereading the Financial Reporter article (2/2004, p16-17, i.e. section II), it seems to me that the approach for your "first liability" might in fact be correct, and I am not entirely sure that it is out of step with the position of the AcSEC. I think that they (the AcSEC and the author of the article) might mean that you only have to reserve for the shortfall with respect to the particular insurance beneft feature (which in your case is the no lapse guarantee). Still, it is the projected net amount at risk in the "trigger scenarios" which is being reserved for (which happens to equal the entire DB, so the distinction is moot when the example of the NLG is being considered). The question is whether, in Paragraph 26, the inserted, italicized words are implied:

"26. For a contract determined to meet the definition of an insurance contract as described in paragraphs 24 and 25, if the amounts assessed against the contract holder each period for the insurance benefit feature are assessed in a manner that is expected to result in profits in earlier years and losses in subsequent years from the insurance benefit function, a liability should be established in addition to the account balance to recognize the portion of such assessments that compensates the insurance enterprise for benefits (???attributable to the particular insurance benefit feature???) to be provided in future periods."

Now, given that you are projecting the account balance forward in all of the cases, one can think of the projected net amount of risk as a term policy (generally decreasing, if we ignore things like corridors and level NAR policies). So, it seems that if the slope of all the assessments (combined) is fairly steep, then the reserve should be fairly small. (Of course, a major cause of the reserve - which is the case of your "second liability - is "front-ended" COI charges. However, if your other charges were steep enough to make the overall slope similar to that of "non-front-ended COI charges", then the reserve would disappear, as

(a. the "current benefit ratio multiplied by the cumulative assessments)
plus
(c. accreted interest)
would equal
(b. cumulative excess payments - including amounts reflected in claims payable liabilities).

Now note that the AAA paper has different interpretations (including what I believe is the approach to your first liability).

Also note that if there is a URL (Unearned Revenue Liability), then it appears that the process of setting up the "additional" (SOP) liability and the EGPs becomes iterative, as the URL is amortized in proportion to EGPs, while the EGPs are a function of the change in the "additional" liability, and the changes in the URL affect the assessments, which affect the emergence of the additional liability.

[mayreeh]

Has anyone out there had their auditors suggest that the pass/fail test for gains followed by losses had to be at the policy level?

For sales inducements, are persistency bonuses pretty much getting the go ahead for the asset or are auditors requiring another product which is identical except for the persistency bonus?

[BigEEE]

We were told that aggregating at the DAC Cohort level was good enough, but they weren't opposed to the policy level test either. We weren't allowed to aggregate at a higher level than our DAC cohorts.

Also, they were lenient on the situation where there were only a couple of negative durations in the future. As long as the PV of the future mortality gains at each point in time was positive, they were ok with negatives.

We didn't have anything that qualified as persistency bonuses and I haven't heard that it was an issue either...

[mayreeh]

We were told that we could test at the product level, but that it was preferable to test at the policy level and that they may want us to support the supposition that we would produce the same results under either scenario.