Mike Preston

Darn, I should have recognized that at the beginning. That *MUST* be it.

You and I have a fundamental disagreement. I believe the regulations mean what they say and you believe in the Chinese way of government (see your section above which I added the bold to). That is, you believe that it is not necessary to be precise because then, if you want, you can change what you meant not by re-issuing the rules with clarifications, but instead by just declaring that what you meant was different from what you wrote. Sorry, it doesn't work that way in the US. At least, not yet. Look, I'm sure you are a fine, upstanding citizen in some respects. So I will credit you with actually believing the drivel you are spouting about "intent". I think you need to look at Q&A 29 again and tell us why you think it is different from the case at hand.

OK, I lied. I'm back. The above needs clarification, lest somebody stop by and actually believe it means something. "There is safe harbor and unsafe harbor." Uh....ok....I think I can get on board with that sentence. "Anything above the safe harbor is fine in the avg. benefits test." If that sentence is meant to convey meaning, it hasn't done its job very well. The average benefits test is a simple, straightforward comparison of the average, uh, benefits. If the average for the NHCE's is at least 70% of the average for the HCE's the avg. benefits test is satisfied. It is only once someone has satisfied the average benefits test that a ratio-percentage of less than 70% can be used in other tests that need to be done. Consider it this way: 1) Do I pass the average benefits test? If no, when performing a 410(b) test I am stuck with 70%. If yes, I can use the safe-harbor percentage. 2) Do I pass the average benefits test (same one as above)? If no, when performing rate group testing under 401(a)(4) I am stuck with 70%. If yes, I can use the mid-point in my testing. We aren't talking about (2). We are talking about having the "plan" satisfy 410(b). If it does, then there is absolutely no need to test under 401(a)(4) because the formula is a safe-harbor formula.

What exception are you talking about? I'm not talking about an exception. I'm saying that the gateway requirements do not apply. You don't need an exception when the gateway rules don't even apply. Remember, I'm not running my rate group testing on the basis of crosstesting. I'm just using plain contributions rate testing. I don't know, are we? At this point, I'll not endeavor to educate you any further, as I believe the issues have been exposed for all to see and people can make their own judgments. If others want to chime in, that is fine with me.

Here's another: (IRS Q&A's 2002 ASPA Annual Conference) 29. An employer maintains only a profit sharing plan. The contributions under the plan do not satisfy any of the gateways under the cross-testing regulations. The plan passes the average benefit percentage test on a benefits (cross-tested) basis, but not on a contributions basis. In applying the general test to the amount of contributions, may the rate groups be tested by comparing their ratio percentages to the mid-point of the safe and unsafe harbors? Yes. In other words, may the plan use cross testing in the average benefits percentage test, even though it may not use cross testing in the 401(a)(4) general test itself? Yes.

Here's one: QUESTION 20 (IRS Q&A's Grey Book 2003) Nondiscrimination: New Comparability Regulations Do the new comparability regulations apply for purposes of the average benefits percentage test? Assume an employer sponsors both a defined benefit and defined contribution plan, each of which is tested for coverage and nondiscrimination separately, but uses the average benefits test for demonstrating compliance with the coverage requirements for one or both plans. Do the new comparability regulations affect the employer’s ability to do the average benefits percentage test on a benefits basis? RESPONSE No. Although Treas. Reg. 1.410(b)-5(d)(5) may be interpreted to suggest that the new comparability regulations do apply for purposes of the average benefits percentage test, the preamble to the final new comparability regulations clearly states, "These rules do not apply ...to the situation in which plans are aggregated solely for purposes of satisfying the average benefit percentage test of section 1.410(b)-5."

Yes, it should have been 8% + 13.7% I will stipulate that the coverage ratio test (the 70% test) is not satisfied. I am presuming that the plan covers 100% of the employee population - that is all the HCE's and all the NHCE's. I am presuming that 100% of the HCE's benefit at the highest rate. I am presuming that the only participants that do not benefit (either because they actually do not benefit or that they received only the top heavy minimum) are NHCE's. So, forget the body counts, and go with the percentages. I am saying that 67% of the NHCE's are receiving an allocation under the 8% + 13.7%. I am saying that the average benefits test is satisfied (pretend for now that it is passed using a methodology you approve of). I am saying that if the average benefits test is satisfied, then a plan that covers 67% of the entire NHCE population is guaranteed to satisfy 410(b). All that is left is to convince you that use of crosstesting in the average benefits test does not give rise to a requirement to satisfy the gateway rules. I'll go look up at least one quote from the IRS on this issue. You are not alone in your misunderstanding of the rules in this area.

I remain supremely unconvinced that the above is a true statement. There is a difference between "benefiting" under the plan and having the plan formula satisfy the safe-harbor requirements of a uniform allocation formula. For instance, if you have a formula that provides a benefit in excess of the TH minimum, then that formula must be tested under 410(b) while treating the participants who received only the TH minimum as not benefiting under that formula. If not, then you have to perform more testing to prove nondiscrimination. I don't disagree with anything you have said. Are you disagreeing with something I said? Perhaps a set of numbers will convince you? As I said earlier, we need body counts to give a better idea as to where things should really go from here. Assume there are 70 participants that "count" for 410(b) purposes and 23 of them are not benefitting because they either are not receiving an allocation at all because they terminated before the end of the year or they are getting only the top-heavy minimum. If I run the whole thing through the average benefits test (using any methodology allowed, including crosstesting) and find that the average benefits test is satisfied, then the plan's regular allocation formula (5.7% up to wage base + 13.7% above) satisfies 410(b) because it covers 47/70, which is 67%. 67% is greater than the safe-harbor percentage, since the safe-harbor percentage can't be greater than 50%. What part of my calculations do you disagree with?

Tom, I was suggesting using cross-testing to confirm that the average benefits test is satisfied. I was not suggesting the use of cross-testing for satisfaction of rate group testing under 401(a)(4). Hence, there would be no gateway requirement. Not exactly correct. You can apply the average benefits testing without cross-testing. But when you cross-test, the gateway automatically applies (unless a broadly available exception or primarily DB in Nature exceptions applies). No exceptions to this would seem to be applicable here. ERISAnut, please, please, please review prior threads on this issue. It is a settled issue and your understanding is incorrect. The use of crosstesting in the average benefits test does not, in any way, shape or form, cause the gateway requirements to come into play. The regs say it. The IRS has confirmed it at many, many conferences. If you think otherwise, please provide a citation.

We would have to disagree on this one. I would normally take the approach of maintaining prototype status. I think the amendment that you are proposing would take the plan out of prototype status by creating separate class allocations. Also, I do stand by the my comment (but would admit that it would depend on the level of contributions being provided to these lower paid employees). I would not agree with any notion of directing large contributions to the a few lower paid people to get the test to pass. Also, the point that Tom Poje pointed out that they are likely failing the gateway into cross-testing. There may be an amendment to increase this group to pass the gateway. Realize that nothing I say is absolute. They are all things to consider with a set of "facts and circumstances". We've all performed corrective amendments and have even written such amendments (unauthorized practice in law). These are thought process which must be considered when determining how to be correct. For the reasons mentioned, I think it is better to amend to bring employees into the formula already in place than to create an entirely different formula. I suppose agree to disagree is the correct phrase. 1) I searched for "prototype" in the prior messages and couldn't come up with that word except in your messages. Why are you attempting to retain a status that isn't evident from the discussion at hand? However, even if you are correct (from some prior knowledge not shared with the rest of us) I vehemently disagree with your approach. The "cost" of elimination of prototype status is a few thousand dollars, at most; and that is if you submit as an individually designed plan (wouldn't it be smarter just to arrange for a volume submitter plan that has the language you want?). If we are talking about just 1% of pay being allocated across the board, even with only 10 participants with average compensation of $25,000, we are talking about a $2,500 "cost". And from what I can gather we are talking about much, much, much larger numbers. And, even if we are talking about a prototype, there really isn't much to say, is there? Don't prototypes have fail-safe language in them which mandates what can/must be done to cure this type of problem? Basically, my responses were not based on prototype status either being initially in play or being something worth saving when dealing with this sort of situation. I note that you waffle by saying that you want to retain prototype status but then saying that it would depend on the numbers. So maybe we are actually in agreement and you just think the numbers are a lot smaller than I think they are. Fair enough. That is why I said we need, at the least, body counts. 2) "I would not agree with any notion of directing large contributions to the a few lower paid people to get the test to pass." Why not? If it makes the test pass, and it is otherwise not problematic to the company in terms of employee relations, why is this particular methodology something that you feel so strongly about? Are the thousands of plans that have used this approach over the last 10 years or so somehow in danger of being disqualified? Why? Citation, please. I am rather vehement about this because there are many in the industry who have gone to great lengths to see where it is appropriate to draw the lines and it is bad enough when rogue IRS agents begin to regulate from the hip by stating that their own regulations don't mean what they say. When I hear it from somebody within the industry, it rankles. Now, if you want to tell your client that you think the methodology is "unfair" and that you don't think it is appropriate, from an employee relations perspective, I have absolutely no problem with that. When you talk about it in a way that makes it seem like it is contrary to all that is good and right, I see red.

I remain supremely unconvinced that the above is a true statement.

Tom, I was suggesting using cross-testing to confirm that the average benefits test is satisfied. I was not suggesting the use of cross-testing for satisfaction of rate group testing under 401(a)(4). Hence, there would be no gateway requirement.

ERISAnut, thanks for the clarification. We all cramp from time to time. But I've got another one..... Emphasis added by me. I completely disagree with the bolded statement. Making an amendment which benefits a group of NHCE's is fundamentally impossible to ring the discrimination bell, except in the rarest and most abusive situations. The Gold "short service" memo is my citation. There is absolutely no evidence in this case that providing the individuals that worked less than 1000 hours with a "top-up" benefit of any kind (through an amendment) would be "majorly aggressive". And there certainly is no evidence that it would do more harm than good. It could very well be just what is required to make the average benefits test pass, which, in turn, then allows the plan to satisfy 410(b) without breaking a sweat.

Jim, stop being logical. IAWWDIK&JN

Well, that's a new one on me. Care to share with us a citation for your assertion that "the class of participants brought into the allocation by the corrective amendment must remain thoughout the end of 2006.."?

I am so confused.... I guess that is to be expected at times. You can not increase "the top-heavy contribution" above 3%. "The top-heavy contribution" is defined in the plan. End of story. As indicated, without amending the plan, about the only thing you can do is put more money in. More money going in will be allocated in accordance with the allocation rules defined in the plan. End of story. But I'm having trouble with this whole thing. I am finding it difficult to believe that you are failing 410(b). Have you run the average benefits test? Have you run the average benefits test on a cross-tested basis? We need numbers here (probably lots of them - at least start with body counts), or there really isn't any way to help.

I read the language as being even more strange than you do. I see the court's position as actually punishing Xerox for trying to determine the offset using theoretical earnings between date of earlier distribution and later calculation. I think Xerox was wrong to do what it tried. And I think the court is wrong in its interpretation. Hopefully, the Supreme Court will set the record straight. The way I read the ruling, the following is what Xerox wanted to do: DB Benefit on initial date of termination: $300/month DC Account balance on initial date of termination: $X (the exact amount isn't relevant) Converted value of $X on initial date of termination: >$300/month (the exact amount isn't relevant) DB Benefit on later date of termination: $1,500/month DC Account balance on later date of termination : $Y Converted value of $Y on later date of termination: $Y' The court seems to be arguing for a final DB benefit equal to: $1,500 less $300 less $Y'. What Xerox appears to have wanted to do is to determine what $X would have grown to between initial date of termination and later date of termination and then convert that value as an offset as of the later date of termination. Let's say that amount turns out to be $900/month, they would have the participant get $1,500 less $900 less $Y'. Note that the $900 offset comes from the initial $X, only a percentage of which was actually used as an offset at the time. Let's say, for argument's sake, that $X would have been an offset of $450/month had the DB accrued benefit been $450 or more. This means that 1/3 of the original account balance (that wasn't used as an offset the first time around) becomes an offset the second time around. I would have thought the appropriate determination at the later date would have been $1,500 less $450 less $Y'. So, the choices seem to be: a) $1,500 less $300 less $Y' b) $1,500 less $450 less $Y' c) $1,500 less $900 less $Y' The Court decided on (a). I would have chosen (b). Xerox, based on good returns between initial date of termination and final date of termination, decided on ©. As I said, I think the Court punished them, by choosing the worst possible outcome for Xerox they could come up with. What the Court seems to be saying, to me, is that the offset, when first applied is cast in concrete. That the PS account balance, whatever it is, can only be taken into account to the extent it is actually used to reduce a true "accrued benefit" in the defined benefit plan. Once it is used, it is gone. In essense, the employer forfeits the right to use it as an offset against future benefits that might accrue in the defined benefit plan. I actually hope that my understanding of the decision is wrong and that, if it is right, that the Supreme Court makes the Ninth circuit abandon this interpretation.

Capitulate. Surely it satisfies the definition of an inconsequential error that can be self-corrected, notwithstanding the number of years since it took place. This is the sort of thing that degrades the entire industry, in my opinion. The IRS agent should be ashamed of himself/herself.

Andy, have you looked at the law and done your own analysis?

I think what mjb is saying, is that you should do your own analysis and then decide what to do for yourself. It took me many messages before I decided to whip out the code and do my own analysis. To my amazement, there is no question that mjb's analysis is technically correct. In fact, it is trivial to confirm it, if you decide to do so. However, I'm equally convinced that the IRS has decided that his analysis is incorrect for pension purposes and will be hard pressed to change what they have consistently communicated since 1999/2000, the first year when the threshold increased. I think the best resolution would be a release from the IRS that says a plan sponsor is/was free to do it either way until plan years beginning a certain date XXX days after publication of the release, whereupon it switches to the proper methodology.

How about if the plan is part of a permissively aggregated group with a non-safe harbor plan? Think DB/DC combo here.

So what is the practical position at this point in time? That either is ok?

Aw, heck yeah! We ARE talking Monopoly money, right? How many sets do I have to supply you with to get up to 100 mil? And we are talking about what the IRS thinks the right answer is (notwithstanding what the technically correct answer is). You asked why the IRS threshold is not used until the next year, in that no other limitation is published in this manner. This is because the compensation is a "lookback" year compensation, not a current year compensation. Look at it this way, the whole purpose of the lookback year compensation was so that employers would know and be able to determine who is an HCE without surprises. My guess is that the Blue Book has something to that effect. although I don't have time to check. When push came to shove, the IRS decided that the only way to live up to the spirit of the rules was to make the lookback year compensation known at the beginning of the lookback year. In order to do that, we need to know that the $100,000 threshold just recently published is the lookback compensation for the 2006 year (that is, for the 2007 determination year). I think, by the way, that if you look at the law itself, without this rationale, you do come up with your conclusion. In fact, it is somewhat trivial - if the base period was 9/30/96 and that applied to $80,000 for the 1997 detemination year, then crawling all dates forward by precisely 9 years results in a base period ending 9/30/2005, a published figure of $100,000 for the 2006 determination year. I would need to do the actual calculations (from 9/30/1996 to 9/30/2005) under 415 to see if it results in a figure of just over $100,000 to see if this actually holds true. But even if it does, as I said, the IRS has interpreted the law to mean that an employer should know the compensation that needs to be paid in the lookback year prior to the beginning of the lookback year. My guess is that if a given employer chose to follow your line of reasoning and use, for example, $100,000 as the 2005 lookback year compensation for determining who is an HCE in 2006 that the IRS would have a hard time challenging it. Especially given the language of Pub 560.

While I agree that your simpler solution would be best, if I get a letter saying that Pub 560 is wrong and that the threshold of compensation to be used in a lookback year (say, 2004) is the threshold as published for that year ($90,000) for the determination as to whether an individual is an HCE in the determination year (say, 2005) then I will be a happy camper, whether you are or not. Actually, I'm already a happy camper, because I am 99.99999% sure that the above is accurate.

mjb, I agree with stephen that the ASPPA Q&A's are usually pretty well thought out. Frankly, the wording of both the Code and the IRS guidance on this issue is so dramatically imprecise. I guess this sort of thing is inevitable when the change is foisted upon us and then the threshold is not modified for a number of years. The initial $80,000 threshold applies for three years, so the newness of the rule sort of evaporated before the uptick became important. As I understand mjb's opinion the threshold for determining whether an individual is an HCE incremented to $85,000 "for 2000" such that compensation in 1999 of $85,000.01 was required to be an HCE "in 2000". I can tell you that my experience is that plans used $80,000.01 as the threshold of compensation needed in 1999 to be an HCE in 2000. Calendar year plans in all cases. I have fired off an general information request to the IRS and, in about 1 or 2 milleniums, I'll post what the response was.