Minimum allocation rate for age-based schedule

[MWeddell]

This is kind of a picky, technical question about an example in the comparability regulations.

Under the comparability regulations, one can test a DC plan with a gradual age schedule using a minimum allocation rate on a benefits basis if the schedule satisfies one of two conditions. The second condition is Treas. Reg. 1.401(a)(4)-8(b)(1)(iv)(D)(2). There could be an employee in each age band with an allocation rate greater than the minimum allocation rate who has an equivalent accrual rate that is less than or equal to the equivalent accrual rate that would apply to an employee whose age is the highest age in the band receiving the minimum allocation rate. (I'm paraphrasing a bit.)

This provision is illustrated by an example in Treas. Reg. 1.401(a)(4)-8(b)(1)(viii)(Example 4)(vi). It states that the steepness condition is not satisfied because the equivalent accrual rate for an age 39 employee (the oldest employee receiving the minimum allocation) is 2.81% but that the lowest equivalent accrual rate for the oldest employee in the next band is 3.74%. Since 3.74% is not <= 2.81%, the steepness condition fails.

I think I understand the provision and the example, but I can't figure out how the IRS came up with exactly those figures (to make sure I completely understand the example). Has anyone figured out the standard interest rate and standard mortality table used to compute the 2.81% and 3.74% equivalent accrual rates in this example? I already checked The ERISA Outline Book and it didn't address this detail.

[Tom Poje]

without looking at that example, I had the same problem (issue?) with another of their examples - for converting the 1% DB benefit into an equivalent allocation rate.

it appears they use

Mortality Table is 1983 GAM 50/50 Blended (The old GATT table)

Interest rate is 8.5%

APR = 106.662

that is a wild stab in the dark.

(I have grown to detest examples that don't disclose such things)

tried very hard to put those things in the Coverage/Nondiscrimination Answer Book when I use an example.

[MWeddell]

Thanks, Tom. I'll try it out.

[Tom Poje]

I took the trouble (like a fool at home, no time at work) to try this

(But then again I guess that is no different than studying and reading up in the field I work in, always hope it makes me better)

anyway

using

Mortality Table is 1983 GAM 50/50 Blended (The old GATT table)

Interest rate is 8.5%

APR = 106.662

I get the following in that example:

Age 39 (26 yrs to retire) 3% contribution

Age 44 (21 yrs to retire) 6% contribution

for simplicity, assume both make $10,000 (though it doesn't matter what comps you use)

Age 39 = $300 contribution

Age 44 = $600 contribution

Age 39 E Bar = [300 * (1.085 ^ 26) / 106.662 * 12] / 10000 = 2.81

age 44 = [600 * (1.085 ^ 19) / 106.662 * 12] / 10000 = 3.74

maybe my rounding is off a little, but this seems to work.

[MWeddell]

Tom, thanks for taking an interest in my arcane question. Two observations, although the first one is trivial:

1. In your last equation, you meant 21, not 19. This is clear from your earlier text that the age 44-year-old has 21 years until reaching age 65 (which is your testing age).
   
2. There is a good reason that the IRS didn't show all of its steps in the example in the regulation: it doesn't look like they used an annuity conversion factor from a permissible mortality table. 1.401(a)(4)-8(b)(1)(iv)(D)(2) refers to "equivalent accrual rates." The definition of that term is in 1.401(a)(4)-8(b)(2)(i) , and it includes "normalizing" the increase in the account balance. "Normalizing" is defined in 1.401(a)(4)-12 and it requires reasonable actuarial assumptions for which standard interest rates and standard mortality tables are deemed to be reasonable. "Standard mortality table" is also defined in 1.401(a)(4)-12 as one of 9 tables, unless the Commissioner changes the list. The list of 9 tables includes 1983 GAM Male and 1983 GAM Female but doesn't include 1983 GAM 50/50 Blended, which is the conversion factor that you figured they were using. In practice, the example still should work -- "Normalizing" allows for mortality tables other than those listed as "standard" as long as the assumptions are reasonable, and I can't imagine the IRS wanting to argue that either the male or female tables are reasonable but blending them, which of course produces a intermediate result, is not reasonable -- but it sure would have been tidier if the example used one of the 9 standard mortality tables instead of using a different one.
   

[Tom Poje]

don't you love it when the IRS uses a table that is not on the list? I actually have thought about that.

I think that table might have been allowed under a rev proc. or some similar bulletin. I think the same applies to the 1994 GAR table, but maybe I have dreamed all that. if someone knows for sure, I'd like a clarification.

I think the APRs at the extreme ends (UP84 = 95.xx or 98.xx - I dont recall off the top of my head) and 1983 IAF at 115.xx.

since the blended table APR falls between the two extremes it makes sense it would be permissible.

while you may call this process 'arcane', I actually don't find it so.

I have to work on an ASPPA talk on Cross Testing for July and was using the DB/DC example, so I really needed to know what table was apparently being used. plus goin over the example you cited is just reinforcement of the material. keeps me on my toes.

[MWeddell]

while you may call this process 'arcane', I actually don't find it so.

"arcane adj : requiring secret or mysterious knowledge"

Given the difficulty I had deciphering how the IRS example worked, "arcane" may have been the right description! I find it interesting too.

I took a quick look using an electronic utility and didn't find any IRS guidance adding any tables to the 9 deemed to be reasonable in the 401(a)(4) regulations.

[Tom Poje]

ah, if that is the correct definition then I agree.

the only other time I have seen that word was in a context that implied it meant something similar to archaic.

of course, my curiosity would be why you want to know where the calculations came from in the first place, or was it more from a standpoint of just wanting to know where the heck the numbers came from.

[MWeddell]

I wrote a spreadsheet to check whether or gradual age or service or age+service schedule met that portion of the comparability regulations. One of the things the spreadsheet is checking for is whether minimum allocations for a gradual age schedule are permitted, so I wanted to make sure I understood the IRS' calculation. I wanted to try to recreate the IRS' results with its example of part of checking whether the spreadsheet worked correctly.

[Tom Poje]

neat. had someone (jusy a few weeks ago) in the office here who was working on a spreadsheet for combo DB and DC and I told him 'he had to duplicate the numbers in the regs' to prove his spreadsheet worked - that is how he figured out what mortality was being used (sort of trial and error - there were enough mortality tables built into the spreadsheet he simply tried different ones at 8.5% and lo and behold it worked)

I suppose if 8.5% hadn't of been used it would have been more difficult.

[AndyH]

Just scanning this.

The 2001 cross testing regulations added "the applicable mortality table", i.e the 417(e) table, to 1.401(a)(4)-12's list of standard mortality tables.

Does that fill in a gap?

[Tom Poje]

ah, you knew it was there all along. that was what I was thinking of but couldn't remember.

thanks Andy!

[MWeddell]

Yes, thanks Andy. I don't know why I didn't see it -- it's the last sentence in the "standard mortality table" definition of 1.401(a)(4)-12, although it requires looking up some cross-references to discover that it means the 1983 GAM 50/50 Blended mortality table.

[Tom Poje]

actually it probably means 1994 GAR at this point in time, though I am not sure if that eliminates 1983 GAM 50/50 or not.

[AndyH]

Yeah, I agree-the 83 Blended table is probably invalid for testing now-kind of weird.