Tom Poje

try IX B 1 of notice 98-52 .....a plan...covers both collectively baragined employees....treated as two separate plans....and the ADP safe harbor need not be satisfied with respect to both plans....

taking a stab at it: I don't think it is truly clear how to handle forfeitures in a class allocated plan - one of those fairly new animals. however, suppose in your example the company had set up two plans - one for each class. (I honestly believe that is why class plans are allowed - you could accomplish the same thing setting up separate plans) now, if you had forfeitures, what would happen? They would simply remain within each class itself. I don't know if that is possible to run easily on Relius. But another way of looking at it, would simply to say forfeitures reduce contributions. In other words, forfeitures get tossed into suspense rather than showing them allocated to participants. mentally my mind can handle that, though I know some clients like to see forfeitures 'reallocated' so to speak. I suppose if the document says forfeitures increase contributions you may conceivably have a problem, but I would argue the following (example only): class 1 got 9%, class 2 got 3%. without forfeitures I would only hav run 8% and 2.67% so I have indeed increased my contribution by forfeitures. I am simply not showing people as receiving 'forfeitures', but rather as a contribution

at the 2004 ASPPA conference this was similar to question #16 of the IRS Q and A's. response: failure to provide notice is an operational defect which should be corrected under EPCRS. [ooooohhhh. I like that. its operational, the formula is in the document. that should give some leg to stand on] it was discussed additionally from the podium, basically that there are no guidelines. Take your chance with whatever correction you choose. With a SHNEC simply provide the notice (since the SHNEC was already made and it doesn't really effect if someone defers.) If it involved a match, solution would be different. [if I recall, no solution was provided] now, of course Q and A's are not written in concrete, an opinion expressed by one or more IRS agents. The example at the conference involved a 2004 plan year. In this particluar instance it was indicated the error has occurred over a number of years...so that doesn't help. one of the points of EPCRS is there are practices in place that prevent such things from happening. still, the argument would be 'who was hurt by the failure not to provide the notice when the contribution was made anyway'?

safe harbor matches (for purposes of the ACP safe harbor) must be limited to amounts up to 6% of deferrals. The discretionary match must further be limited to 4% of comp. Thus, you could have a discretionary match of 66.6% up to 6% deferred, because ee would end up with 4% of comp

I dont think you can simply declare or treat an amount of deferral as a catch-up and therefore not use it in cross testing. Catch up was for hitting a limit (402(g)) or failing ADP test, or even 415 limit. Failing nondiscrimination was not one of the options listed. Now, the plan could have a cap on deferrals, but at that point you are really guessing - what happens the following year with a different census population. Now you have a plan imposed limit that might make no sense.

Andy- only 'cuz she moved. I liked KateSmithMD when everyone thought she was a doctor who was also dabbling in pensions. Now she is only a lowly PA.

my understanding is that catch-ups would not be included in testing (I suppose prior year catch ups would be included if one was using accrued to date) I believe it is buried in the regs or one of the notices, but I am too lazy to research at this time. as for an example of a plan failing testing and whether to refund excess deferrals first.... suppose HCE 1 14,000 def 200,000 comp 7% HCE 2 10,000 def 100,000 comp 10% Avg = 8.5% NHCE avg = 5.8% therefore plan fails If I first treated excess deferrals as catch up before running test I would have HCE 1 12,000 def 200,000 comp 6% HCE2 10,000 def 100,000 comp 10% avg = 8%, plan fails, must return additional excess contrib However if I run test first, my refund to bring average to 7.8% I must reduce HCE 2 to 8.6% HCE 1 14,000 def 7% HCE 2 8,600 def 8.6% avg 7.8% plan passes, refund =$1400 so 1400 = excess contribution for HCE 1 this leaves him at 12,600, so still has 600 in excess deferral. so it looks like I leave all 2000 in the plan as catch up, as oppossed to having to refund an additional amount.

Fred, I don't know if the answer is real clear. In the old days you included excess deferrals in the test and then calculates the refund. Interestingly enough, there is a section in the ERISA Outline book that discusses what to do when you have both excess deferrals and excess contributions involving the same person, and the discussion talked about the results if you did one refund first or did the other one first in regards to testing. I'd like to think that still holds true. Consider a plan that fails testing if excess deferrals are immediately counted as catch up. this would require an additional refund. On the other hand, it is entirely possible for a plan to fail testing, and the refund amount to pass is, for example less than $3000 in 2004. Plan failed because a diferrent HCE had a higher percentage, but the refund went to the HCE who was catch up eligible. Thus the refund could be treated as an excess contribution. Now that HCE was still over the deferral limit, so an additional amount was treated as an excess deferral to bring him down to 13,000 for the year. Logically to me it would seem you could still do this.

see http://www.irs.gov/formspubs/article/0,,id=109875,00.html you have to scroll about 2/3 of the way down to find the item that is applicable to 1099R looks like you would use a code 8,D or P to indicate the year

if plan was to test using SSRA then it is strongly recomended you get an approval letter and get the language in the document for that testing assumption. otherwise you have to use NRD. a plan should have language for the gateway - you can't simply bump someone up to the 1/3 or 5% allocation rate. the document doesn't have to say 'we will only use 5% or we will only use 1/3

The following formula could be added to Part1pg (or a variety of certificates) to project out a cash at retirement. This particular formula, since it uses '1' would be for projecting out the deferral account on that report. It assumes the current year deferral will be constant for all future years. This is written for 6%, easily modified for other interest rates. You would have to add PLANEE table to make it work. you could modify it to project out on other sources. e.g ContrForf2 would be match. the first half of the formula uses the current year contribution. The second half is the ending balance in that source. Round ((Sum ({@ContrForf1}, {RPTEE.SSNUM})*((1.06^{PLANEE.FUTRSVCYRS})-1)/.06 +Sum ({@EndBal1}, {RPTEE.SSNUM})*1.06^{PLANEE.FUTRSVCYRS}))

insufficient info to even attempt to answer #2. read 1.401(a)(4)-9(b)(2)(v) you can not test a DB/DC plan on a benefits basis unless 1. plan is primarily DB in character OR 2. consists of broadly available separate plans (You have indicated this is not true) OR 3. provides min aggregate allocation gateway since some ees are receiving DB and not DC you might actually be primarily DB in character. if you only had a ps plan, and a last day rule, AND AND AND it is not a 401(k) plan then normally you would not have to provide a top heavy (though there are a few documents with less stringent top heavy requirements)

possibly, it is unclear from the original question if that would apply in this case. the particular reg cited is applicable if allocations are made at the same rate for someone who is past NRD as someone who is at age 65. (age weighted plans are often written to provide the same rate for someone who is past NRD as someone who is at NRD.) I guess technically this doesn't really mean you use APR at 65 for someone who is past age 65, but rather those people won't cause the plan to fail if they receive the same rate. I suppose if you needed 'proof' the plan passes you would 'fake the system out' using the same APR

the regs 1.401(a)(4)-12 "testing age": 1. if all have same uniform ret age, then use normal ret age (note 65 and 5 is considered uniform) 2. if different ret ages for diff groups, use the latest ret age 3. if not uniform use age 65 (this can happen if you have something like plan val date nearest) 4. if beyond testing age, the employees testing age is the employees current age. (There is an exception if the plan is a defined benefit) but if it is a DC plan I dont see how you can use APR other than current age, as indicated in #4

the comp limit (in this particular case = 205,000) is used for 1.testing purposes. eg ee defers 13,000, so for ADP test his percentage is 13,000 / 205,000 not 13,000 / 1,000,000 if that is how much comp he actually had. 2. if plan makes profit sharing contribution that will also be based on 205,000 not something larger. if it helps, pretend you get one paycheck at the end of the year, and you defer only once. at that point in time you limit the comp to 205,000. by the way, and I dont think it was mentioned, or maybe not clearly, the plan could allocate a match on a payroll basis. further suppose the individual defers 13,000 the first payroll with comp of 26,000, simply because he wants the earnings for the whole year. A whopping 50% of pay. if match was limited to 100% of the first 6% deferred, he would receive a small match, and there would be no true up of the match during the rest of the year because he had no more payroll deferrals..

at 9.1.3 no major grief, aside from the usual of having to update the customized crystal reports. I suppose getting used to the error message while running transaction. you have to open up a 'tree'

since no one has accrued for 2005 you can always amend to take out match. (or at least make it discretionary, capped at 4% of comp. you can also add the safe harbor feature, need to give notice by Dec 1 (at least preferably by that date to be on the safe side)

looked it up, and the following was in the preamble to the regs: if a plan benefits employees who have not met the min age and svc requirements, the plan may be treated as 2 separate plans, one for otherwise excludables and one for the other benefiting ees. Thus, if treated as two separate plans, cross testing the portion of the plan benefiting the nonexcludable ees will not result in minimum required allocations under the gateway for the ees who have not met the min age and service. guess I was wrong on my assumption you pass gateway first. split into two plans, then worry about the possible gateway

thanks for making me look this one up. I looked it up under -11(g) and actually it looks like the corrective amendment 'is not part of a pattern of amendments being used to correct repeated failures with respect to a particular benefit, right, or feature.' Thus, I guess in this case it would not apply as the corrective amendment is used for nondiscrim purposes (3)(v) and not brf (3)(vi). still, sounds like a pain in the rear if you have a corrective amendment every year.

I would disagree that the item mentioned is a 'new retirement plan' coming for 2006 EGTRRA simply states If an applicable retirement plan includes a qualified Roth contribution program.... In other words, begining in 2006 a retirement plan can ADD a Roth 401 k feature. so instead of going out and setting up your own Roth you simply now have it as part of the plan. That of course would require separate accounting, requires administrator to file returns indicating Roth contributions, etc. so extra work at that end. advantage to the employee: individual Roth probably has an annual fee in a plan, the $ are pooled so maybe no fee???? or does it get 'spread' across the board and therefore other participants pay part of it???that is unclear. since $ are pooled, supposedly better rate of return since there is a larger investment involved so to speak.

My logic (which could certainly be wrong) says the following: 1. Plan has smoothly increasing schedule (Fact) 2. by regulations that satisfies one of the three requirements permitting me to cross test. 3. I now cross test and plan fails. 4. therefore, I must take corrective action to pass. this corrective action does not require me to go back to step 1 and change my formula or take me out of having smoothly increasing schedule. it is merely some type of corrective action after I have met the requirement which allows me to cross test in the first place. now, if I have to do this every year, then I have poor plan design, but also you start falling into the 'series of corrective amendments' which is frowned upon.

Andy raises an interesting point about the gateway minimum. my understanding is that before I even get to cross testing I have to 'pass through the gateway'. once an ee has received the top heavy he would have to be bumped up. now I can cross test, and one of my options is to test separately. e.g. until I pass GO and give each person '$200' I can not cross test under any way shape or form. the only exception to the rule is under the broadly available option - two groups could be treated as a single rate if each could pass 410b by itself. e.g. one group gets 10% and another gets 3%. this could happen if you have people in classes and each class consists of some HCEs as well as some NHCEs, enough in each class so as to pass BRF.

of course personal opinion don't account for much so the following could only be taken for what its worth. supposedly the final regs will be out soon, I would think they really want to get them out so things are in effect by 1/1/2005, but who knows. until that time, I think there is little that can be done to stop the use of bottom up QNECs. never did a bottom up QNEC, never had that written into a document, and certainly that would be a requirement, you can't simply make one if the document simply calls for a QNEC. (e.g. most say to all NHCE eligible) as for the memorandum, I'm not sure it would apply in this case. The memorandum talks about hiring practices that excludes a number of NHCEs, or providing large benefits to a select few NHCEs who won't vest, or were short time hirees. if the plan you are talking about has been failing the ADP test every year, and has excluded a number of NHCEs from deferring, as well as including some short time ees, then I could see there being problems. but again, my opinion only

ah, my mistake. when you said smoothly increasing, I assumed age weighted. I have never seen someone take the trouble to set up a plan with the ranges you noted. If a plan is smoothly increasing then that is suppossed to satisfy the gateway minimum. if I recall, any top heavy allocations would not toss the plan out of being smoothly increasing, but I haven't looked it up for awhile.- yes it is in there at 1.401(a)(4)-8(b)(iv)(2)(D) or something like that. it doesnt surprise me avg ben % test fails if owners kids (youngsters) are deferring. that will blkow that test out of the water. if age 56 receives 19% and age 57 receives 21%, I dont see how the rate group test would pass since 19 * 1.085 = 20.62. which is less than 21. unless there are a lot of 3 % people age difference of 24 that will have an ebar that will be greater than the hce.

One easy answer would be to test otherwise excludables separately. In that case all ees with less than a year would be tested separately anyway. I am not wild about the term 'ignored' because I can't tell from that term if you mean 'includable and not benefiting' or simply 'doesn't show up at all' well maybe its not that easy, because maybe I need the guy to pass avg ben % test. now, what if you don't want to test otherwise excludables separately. I think it might be a gray area. I lean toward the following logic. I have 2 profit sharing plans - one with a 1 year wait, and another with (in your example) a 3 month wait - that plan provides the top heavy. given those conditions, I would lean toward testing all employees, using the requirement that you have to test using the plan with the least stringent requirements e.g. 3 months. but I would also say I think it isn't 100% clear.