If you’re like me, your ASC 715 valuations around 2013 and earlier often used for the mortality assumption what I’ll call the “IRS funding tables” – specifically, the separate or combined static annuitant/nonannuitant tables in effect for a given year published in IRS Notices. (And, if you’re like me, you still call ASC 715 “FAS 87,” but that is a separate matter.)
Use of the IRS funding tables as an ASC 715 mortality assumption in 2008 through 2013 seemed reasonable for our clientele, which tends to be small plans. These tables were derived from the Society of Actuaries’ RP-2000 Mortality Tables Report. Given the exhaustive research performed by the SOA for this project, we had confidence that there was no better data available for use in setting mortality assumptions for the small plans we serve. Our clients, tasked with making the final call on actuarial assumptions but relying heavily on the recommendations of their actuaries and auditors, were readily agreeable to using the IRS funding tables. If this assumption was “IRS-endorsed” and was good enough for the funding valuation, they reasoned, then why not use it for the ASC 715?
This all changed around October 2014 when the SOA released its long-awaited RP-2014 Mortality Tables Report. RP-2014 represented the latest and the greatest; think of it as the iPhone X of mortality data. It was obvious that we actuaries would want to incorporate RP-2014 into our ASC 715 valuations posthaste, starting with the 12/31/2014 valuation for calendar fiscal years.
Or was it so obvious? In fact, actuaries might feel they had some very good reasons not to implement RP-2014 into their recommended mortality assumptions right away, some of which can best be described as, “It’s too hard!”
More Calculations, More Datasets
First, RP-2014 is intended to be a generational table, not a static table. Under a static mortality table, the probability of an individual dying at, say, age 65, is the same whether that individual is now age 4 or age 64. With a generational table, the probability of dying at a given age depends on what age the individual is now. The probability of dying at age 65 will be higher for a 64-year-old than it would be for a 4-year-old, assuming that longevity is expected to increase over time.
A typical static mortality table used by small plan actuaries, such as UP 84 or GAR 94, takes up three columns on a spreadsheet: age, male q rate, and female q rate. A generational table will have separate columns of q rates for the current year, each future year up to some point, and possibly some past years. To add further spice to the soup, the MP-2014 projection scale released concurrently with RP-2014 is also two-dimensional, meaning lots more spreadsheet columns. While actuaries in the 21st Century should have enough computing power at their disposal to handle generational tables with relative ease, it is not hard to imagine someone used to working solely with static tables wanting to avoid the bleeding edge of generational mortality tables with two-dimensional projection scales.
As if that weren’t enough, there is not just one RP-2014 mortality table. There are a total of 12 RP-2014 tables. We have seen some of the variations before, such as healthy vs. disabled and employee vs. annuitant. However, RP-2014 also has separate tables for blue collar vs. white collar employees, low income quartile vs. high income quartile for employees, and low benefit quartile vs. high benefit quartile for annuitants. In addition to having to contend with a generational table, the actuary would have to decide whether the RP-2014 total dataset best reflects the plan demographics or if one of the collar or quartile tables was more appropriate.
Surely, the IRS would come to the rescue, wouldn’t they? The IRS had been publishing static mortality tables for years. Wouldn’t the IRS’ crack team of mortality table constructors sift through all that RP-2014 actuarialese and give us some easy-to-use (preferably static) mortality table to use in our 12/31/2014 ASC 715 valuations? Uh, no. I hope you weren’t holding your breath. For one thing, the static funding tables for 2014 and 2015 had already been published more than a year earlier. There was no way the IRS could turn on a dime and give us new static tables in just a few short months. (We know now that the first funding tables based on RP-2014 the IRS was able to produce would be for 2018.)
Despite these challenges, there were at least a couple of very good reasons to go ahead and start using RP-2014 mortality for the 12/31/2014 ASC 715 valuation:
- Some auditors were insisting on it. Some had been instructed to ask for justification if an actuary did not want to use RP-2014 mortality.
- As actuaries, we had a professional obligation to use assumptions we judged to be reasonable. If the IRS funding tables no longer represented our best estimate of future mortality experience, we should not continue to use those tables just because they were “easier.”
So, like many of you, I rolled up my sleeves and began talking to my clients about using RP-2014. Some clients changed to RP-2014 immediately. Others waited a year or two. One client said something along the lines of, “We are a conservative firm and are not going to make this change until it is required!” My response that ASC 715 did not require any specific mortality assumption to be used was not compelling. I think he meant his firm would change when the IRS funding tables incorporated RP-2014.
Two things are worth noting. First, because of anticipated longer life expectancy, the RP-2014 tables were expected to increase pension liabilities compared to whatever tables they were replacing, including the then-current IRS funding tables. Second, despite the widespread acceptance of RP-2014 as the new standard for mortality assumptions, it was still the actuary’s right not to accept RP-2014 as the most reasonable assumption. For example, some felt RP-2014 with Scale MP-2014 went too far in projecting future mortality improvement. If the actuary did not believe RP-2014 was the most reasonable assumption, he or she was free to recommend a different assumption. The fact that RP-2014 made the benefit obligation increase was not a good reason to recommend a different assumption, though.
You’ve Decided RP-2014 Is Best. Now What?
Let’s turn back the clock and pretend it’s late 2014 or early 2015 when you were doing your 12/31/2014 ASC 715 valuations. If you decide to utilize RP-2014 in your mortality assumption, you must then decide which set of rates to use. Will you use the total dataset? Blue collar? White collar? Top quartile? Bottom quartile? Your pension software may have all these tables. If you are a do-it-yourselfer, you can find what you need online rather easily by typing “RP-2014 mortality” or “projection scale MP-2014” in your search engine and going to the SOA site result that comes up. In addition, there are links to some of the SOA resources available in the “Resources” section at the end of this article.
The first things you will need are the RP-2014 base mortality rates and Scale MP-2014. With these, you can construct your RP-2014 generational table. I did this in a spreadsheet with separate tabs for male employee, male annuitant, female employee, and female annuitant. Examples using the total dataset are provided in the “Resources” section at the end of this article, in separate Excel files. Let’s work through a quick example. The base q rate for a 65-year-old male healthy annuitant in the total dataset is 0.011013. This is the probability of a male annuitant dying while age 65 in 2014.
If we know that the 2015 value from Scale MP-2014 for a 65-year-old male is 0.0105, we can determine that the q value for a 65-year-old male annuitant in 2015 is 0.011013 × (1-0.0105) = 0.010897. Likewise, we can determine that the q value for a 65-year-old male annuitant in 2016 is 0.011013 × (1-0.0105) × (1-0.0103) = 0.010785, where 0.0103 is the 2016 value from Scale MP-2014 for a 65-year-old male annuitant. As you can see, under Scale MP-2014 the probability of dying while age 65 is lower the later one attains that age.
An actuary could simply choose to use as the mortality assumption the static base RP-2014 rates without applying the projection scale. By doing so, the q value for any age will remain constant in future years, just as they do when using GAR 94 or any of the other traditional static pension tables. It could be said that one is not getting the “full RP-2014 experience” by not applying a projection scale. On the other hand, the base rates may indeed reflect the actuary’s best estimate of future mortality experience. Perhaps the actuary feels the projection scale represents an overly optimistic outlook of future longevity. The actuary will wish to be specific in his disclosures exactly what rates are being used, whether a projection scale is being applied, etc.
Scale MP-2015 Updates Scale MP-2014
In 2015, the SOA issued Scale MP-2015 to replace MP-2014. Pretending now we are back in late 2015 or early 2016, if we thought the RP-2014 table with Scale MP-2014 was a reasonable assumption for the 12/31/2014 ASC 715, then there is a good chance we will simply recommend a substitution of Scale MP-2015 for MP-2014 at 12/31/2015 now that MP-2015 is available. Like MP-2014, Scale MP-2015 can be downloaded from the SOA website.
There is something very important to be aware of as you construct your generational RP-2014 tables using Scale MP-2015. You do not simply derive your future q values by projecting the base RP-2014 rates from 2014 using Scale MP-2015. The central year of the RP-2014 dataset is 2006. The base RP-2014 rates were derived by projecting the 2006 rates to 2014 using Scale MP-2014. If you want to use the new Scale MP-2015 fully in the way it was intended, you will need back out the mortality improvements from 2006 through 2014 that came from Scale MP-2014, then apply Scale MP-2015 from 2006 forward. (Or, you can just go on the SOA website and download these adjusted 2006 rates which they call “RP-2006,” then project forward from there.)
When disclosing assumptions in our ASC 715 reports, we have used the rather clunky name of “RP-2014 static employee and healthy annuitant mortality (total dataset) published by the Society of Actuaries adjusted backward to 2006 with Scale MP-2014 and projected with Scale MP-2015” to describe this assumption. (And you wonder why there is a robust assortment of actuary jokes out there.)
It is technically not “wrong” for an actuary to use as the mortality assumption the base RP-2014 rates projected forward from 2014 using Scale MP-2015. The actuary should be sure that this really represents the best estimate of future experience and that he or she hasn’t simply applied Scale MP-2015 in a way different than what was intended by the SOA. The assumption should be described in the ASC 715 report in such a way that there is no confusion about which rates are being used at every age.
Projection Scales Updated Annually Since 2014
The SOA’s stated original goal was to update the projection scale every three years. In fact, a new projection scale has been issued every year since 2014. The new projection scales have been named, appropriately, MP-2015, MP-2016, MP-2017, and MP-2018. New generational tables may be produced each year using the updated projection scale for that year. Since the adjustment of the original RP-2014 rates to 2006 has already been done once, it does not need to be done again each year. You may construct the current-year generational table by projecting the so-called RP-2006 rates forward using the current-year projection scale.
The mortality improvement rates have declined from one year to the next in all four years the updated projection scales have been issued. This means that future q values are higher and liabilities have declined from year to year (all else being equal). This suggests that those who thought in 2014 that Scale MP-2014 was overstating future mortality improvement may have had a valid point.
Resources
SOA data:
- RP-2014 Mortality Tables, including report and rates, plus RP-2006 adjustment factors and rates
- Scale MP-2014, including report and rates
- Scale MP-2015, including report and rates
Reports and rates for projection scales beyond 2015 may be found on the SOA website with similar links that include the year you are looking for.
Author’s derived example spreadsheets (click to download):
- Derivation of the RP-2014 Generational Annuitant Table (Female) using Projection Scale MP-2014
- Derivation of the RP-2014 Generational Employee Table (Male) using Projection Scale MP-2015 with adjustments for use of MP-2014 from 2006-2014
- Derivation of the RP-2014 Generational Employee Table (Female) using Projection Scale MP-2016 with adjustments for use of MP-2014 from 2006-2014
- Derivation of the RP-2014 Generational Annuitant Table (Male) using Projection Scale MP-2017 with adjustments for use of MP-2014 from 2006-2014
- Derivation of the RP-2014 Generational Employee Table (Male) using Projection Scale MP-2018 with adjustments for use of MP-2014 from 2006-2014
- Log in to post comments