Background: Used the statutory exclusion option to pass the 401(k) and 401(m) nondiscrimination test. Understand that my coverage testing for these components must reflect this as well. Do I have the ability to test the profit sharing component of the plan w/o carving out the statutory exclusions for my 401(a)(4) nondiscrimination test (rate groups) and coverage test. The profit sharing component passes the coverage test using the ratio percentage test, but the issue is the rate group testing. Can I test each component for nondiscrimination and coverage differently? Believe me including the employees in my nondiscrimination testing that have only been employed for 6 months would definetly help my rate group testing.

Any clarrification would be greatly appreciated

I think your second sentence is incorrect. I think your consistency requirement is between the discrimination and coverage testing of each plan, of which you have three.

I believe we are talking about the same thing: ie. component plan. So you are saying I can test each component plan differently so long as I'm consistent in my application of the coverage and nondiscrimination testing for that specific component plan.

That is my understanding. Tom et al feel free to chime in. Not Blinky though; he is still hibernating.

This leads me to a related question. It's been agreed here in earlier posts that the mandatory disaggregation rules allow for the separate treatment of the otherwise excludibles. The OEs are also subject to mandatory disaggregation, but unlike the deferrals and match, are not thrown back into the Average Benefit % test. Would I then have 2 ab%t, one for the deferrals and match of the OE only (probably irrelevant), and the other based on the (k), (m), and nec for all nonexcludables?

Andy - I don't think I can chime in, probably closer to being a dull thud.

Certainly all agree that each component of the plan must be mandatorily disagregated for testing
1.410(b)-7©


1.410(b)-7©(1)
the wording in the regs says "..the portion of the plan ...is treated as a separate plan' thus ...a plan...is treated as three separate plans for purposes of coverage"

therefore the conclusion by most is that you could test the 401k 'plan' using the otherwise excludable option and test the nonelective portion testing all


1.410(b)-7(e) says all plans in the tesing group must be taken into account. and the examples that follow show that means everything (deferrals, match and nonelectives)

the problem is the regs give no clear cut examples of the situation in question.

so, one is left hanging...

I prefer the argument along these lines
when testing coverage you split the plan into 3 'plans'. the regs say you do.
so now you are looking at a 'plan'. what conditions are you using for that 'plan'? apply the same conditions to the avg ben % test. there seems to be consistency in this line of thought.

another argument says if one of the 'plans' includes all ees, then the avg ben % test includes all despite the fact what assumptions were used for each separate plan. this interprets the regs to say the avg ben % test applies to the plan as a whole rather than separate plans.

I have seen a further argument that says there is only one avg ben % test - that consists of all employees, even if otherwise excludable option was used anywhere. this would seem to fail based on 1.410(b)-5(d)(3) which says all plans included in the testing group under 1.410(b)-7(e)(1) AND ONLY THOSE PLANS are taken into account. Personally I think this statement also supports the idea you could have an avg ben % test for the 401(k) portion that is different than the nonelective portion.

long winded enough for you Andy? Did you 'get smart' on cross testing?

Wo, a warm welcome back to Merlin. Been hibernating like the Blinkster? Is that like "sleeping with the fishes"?

I think the consensus is that there is only one test. Somewhere on one of these boards EF Hutton expressed that one-test opinion adamantly and if I recall correctly added "even the IRS now agrees to that". But there still may be differing opinions.

Back to the original question, the a(4)/410(b) regs say that the testing group must be the same between nondisdrim and coverage. And elsewhere it says that ADP and ACP are the exclusive nondiscrim tests for K and M, so logic might apply the consistency principal to 410(b) and ADP and 410(b) and ACP while not requiring any coordination between the separate dissagregated plans. But, I agree with Tom that we could use this in print. How bout the next edition? The book could even quote "an Elvis impersonator in Las Vegas (moderated by a yellow blinking 3 eyed fish?) speaking at an ASPA session opined that ........." Conclusive!

well, guess I can write up a Q and A and toss it to the folks at the IRS for that session.


gee whiz Andy, the Vegas talk has a different theme. when you get a free moment take a look at what I sent you.

So far, so good, although I think it is more complicated than what you are implying.


Oops. Now, re-read 1.410(b)-7(e)(1)(iii) and see if that doesn't change your mind.

In English, I believe that -7(e)(1) says something like the following:

Plans in the testing group include the plan being tested and any plan that can be permissively aggregated. You use the rules of (d) to determine what plans can be permissively aggregated. However, as you careen through (d), you do so with the knowledge that (d)(4) is to be ignored (QSLOB stuff); with the knowledge that (d)(5) is to be ignored (you aggregate a plan with a different plan year for this purpose (the ABPT) even though you can't normally aggregate such a plan for 410(b)/401(a)(4) testing); with the knowledge that the general rules of disaggregation ((d)(2)) are to be applied ignoring the fact that k/m/ESOP plans are mandatorily disaggregated (in other words, k/m/ESOPs are not mandatorilly disaggregated for this purpose).

If a k/m/ESOP is not mandatorily disaggregated then it is, by definition, part of the testing group because the testing group includes all plans of the employer that can be permissively aggregated.

That is, it includes all plans of the employer other than plans that are mandatorilly disaggregated.

If a k/m/ESOP is not mandatorilly disaggregated, then, by definition, it is a plan that can be permissively aggregated, at least for this purpose.

So, for ABT purposes, you absolutely must include k/m/ESOP's.

By extension, since 1.410(b)-7(e)(1)(iii) only references ©(1) through ©(2) (that is, it does not reference ©(1) through ©(3)), we absolutely know that when doing this test, we treat the portion of a plan benefitting otherwise excludables as mandatorilly disaggregated from the portion of the plan that benefits those who are not otherwise excludable (and vice-versa). To get here is not straightforward, though, because it requires reading the last sentence of (d)(2). That sentence says to go back to yet another section for rules on how to test otherwise excludable employees. That is 1.410(b)-6(b)(3)(ii).

So, the sequence is to look at 1.410(b)-7(e)(1)(iii) which back references 1.410(b)-7©(1) and -7©(2) and note that -7©(3) is not back referenced.

I'm not sure whether all of this is a triple-negative or a quadruple-negative. It is surely not as simple as a mere double-negative.

can you put all that in english?

you said

"we treat the portion of a plan benefitting otherwise excludables as mandatorilly disaggregated from the portion of the plan that benefits those who are not otherwise excludable"

ok, so running the avg ben % test it sounds like you say I have 2 tests.

now, when I ran coverage I tested 401k using otherwise excludable options, but when I tested nonelective I didn't.
so when I run the avg ben % test do I have one number for the 401k portion of coverage (should I fail ratio % test) and a different value for the nonelective portion.

It's all Greek to me.

If I have elected to test a plan for compliance with 410(b) by separating the plan into two distinct testing populations, then I have two distinct ABTests.

We are not limited to two, however. While I grant that in most cases we won't have more than two. In any event, an ABT is "run" for each separate plan of the employer. In the case of an employer having 9 plans, the 9 tests may be identical. But they don't need to be.

The rule is that you run a test including the plan being tested and each and every single other plan that can be permissively aggregated with that plan - taking into account the rules previously described to determine which plans are, in fact, permissively aggregated.

Think of the ABT as being spoke-and-hub kind of thing. Each plan takes its turn as the hub. Then the spokes are inserted to the extent they are required to be inserted for that plan. You test the combination of the hub and spokes. Now, if we merely move one of the spokes to the hub position and move what used to be in the hub position out to where the spoke used to be, we end up testing the exact same group. That is, the tests are identical.

It is only when the spokes (permissively aggregated plans) and hub (plan being tested) looked at together comprise different populations do we end up with a distinct test.


Since I'm not sure what the above means (in English) I shall conjecture and you can tell me whether my conjecture is close to what you intended. After I get through conjecturing, let's apply the description above and see what we get. You have 3 separate plans for 410(b) purposes. The first is 401(k) taking into consideration only those who were otherwise excludable. The second is the 401(k) taking into considration only those who are statutorily eligible. The third is the non-401(k) portion which has only one testing population: everybody.

But I'll stop there and let you tell me:

1) With respect to the first plan - the 401(k) taking into consideration those who were otherwise excludable, what other "plans" of the employer are capable of being permissively aggregated?

2) With respect to the second plan - the 401(k) taking into consideration those who were statutorily eligible, what other "plans" of the employer are capable of being permissively aggregated?

3) With respect to the third plan - the non-401(k), what other "plans" of the employer are capable of being permissively aggregated?

I'll answer (3): All three plans.

Hey Zeús, how about the original question?

if I understand the original question correctly, and trying to keep the numbers really simple

I have a plan with immediate eligibility

1 hce has 2 years service
1 nhce has 2 years of service
1 nhce is so new that he doesn't even have dust and he doesn't even know where the supplies are kept.

so I run the 401k using otherwise excludable option - 1 HCE 1 NHCE in the stautory includable group.
I have a second test which consists of only oytherwise excludables. - 1 nhce

I run the nonelective 'plan' including all ees - 1 HCE and 2 NHCEs.

pretending that I fail ratio % test, I need to run the avg ben % test

when I look at the 401k 'plan' who is included.

when I look at the nonelective 'plan who is included.

Well, to some extent, it looks like this thread has gotten a bit off-track. At least, if adherence to the OP's question is the benchmark.

As I understand the question, my paraphrase is: Can one test 410(b)/401(a)(4) compliance of an employer's profit sharing contribution on the basis of "all employees" (that is, not testing the otherwise excludables and the statutorily includables separately) even though one is testing 410(b)/401(a)(4) compliance of the 401(k) and 401(m) components of a plan by testing those groups separately?

As far as I know, there is no consistency requirement that requires any plan to ever consider the testing methodology of another plan.

However, this is not the first time I've seen this question in the very recent past (I forget where I saw it the first time). It makes me wonder whether the IRS is challenging this at some level and these questions are the result. If it is true that the IRS is challenging this then I would appreciate knowing the specifics. If it is true that the IRS is challenging this, my guess is that the Government Affairs Committee of ASPPA would also be interested in it. To the point where they would likely be willing to help clarify the issue with the IRS.

By the way, the original question in this thread did not venture into the average benefit test. It was strictly limited to rate group testing. As I think the very first response by AndyH indicated, there is nothing precluding the rate group testing from considering all employees.

This thread then branched off into the average benefit test (see Merlin's post).

Doesn't take much to imagine what you are pretending. If the rate group for the one HCE includes one of the two NHCE's, then the rate group fails the ratio % test.

Mike, it seems we agree on the separate plan, no consistency issue. I am unaware of any contrary opinions.

Am I imagining it, or hasn't your thinking on issue #2 (merlin's question) come full circle?

Tom, you are Darth. Not Elvis and not Quint and not Zeus. Kudo's and Congrats. Well done. You do need to post images at some point.

I don't recall ever having an alternate position. Anything is possible, I suppose, but I thought I had posted on the -7(e) issues long ago in this same vein. Remind me.

By the way, the EA Grey book for the 2006 meeting has a question which bears on this thread.

Q&A17.

Maybe somebody can post it, along with an appropriate copyright notice.