richard

Tom, If I understand your January 9 post, is the following a correct application to the following situation. ADP test is failed for 2007. 6 HCEs will receive refunds by March 15. However, 1 HCE (whose refund would be $3,000 including interest) terminated employment in mid-2007 and rolled his entire account balance to his own IRA in late 2007. Since the $3,000 cannot be taken out of his 401(k) account and refunded to him now, that $3,000 is considered as a refund and is taxable to him for 2007. Is this correct? If this is correct, what happens to the $3,000 that is now sitting in his IRA? Richard

How about the following variation on a theme (new client which I just picked up). This is the third year of their safe harbor 401k. 5 owners, 15 employees. They have a safe harbor match (100% of the first 3% of pay plus 50% of the next 2% of pay) And they have an additional fixed match of 200% of the first 6% of pay! Does this plan design meet the safe harbor ADP and ACP rules? Also, are top-heavy minimums automatically satisfied? Other issues that I can handle --- all 5 owners are contributing the maximum and none of the 15 employees have ever contributed, despite having been provided with safe harbor notices. (Yeah, I know the questions this might raise.) And in year 1, they started this as a brand new plan 30 days before the end of the year, not 90+ days (this is not a new company, so I know year 1 was blown badly The client will enjoy hearing this). Richard

My instinct is that new elections would have to be offered, and that a QJSA would have to be one of the options. This is, after all, an additional benefit provided via plan amendment, and not part of the original accrued benefit. (My rationale is that when an employee elects a $10,000 lump sum instead of a $100 per month pension, he is not committing himself to an ultimate $11,000 lump sum instead of a $110 per month pension.) I suspect that the large plans that have offered ad hoc COLAs provide it to the employee in the form of benefit already in effect, without giving the retiree any choice. If so, either they have done it wrong (but as a practical matter, the retirees wouldn't care, since they would normally elect the same form anyway), or my instinct is wrong (which is always a distinct possibility!).

Another approach I've seen (particularly in the 1980's) was where companies philosophically adopted a policy of ad hoc COLAs every 3-5 years desired to replace X% (such as 75%) of the cost of living increase since the last ad hoc COLA (or the employee's retirement date, if later). Then, we (or they) calculated the percentage increase that applied to the different cohorts of retirees, based on retirement year, that approximated the desired X%. The trade-off was the simplicity of the formula vs. matching the desired X%. Often, a fixed percent per year since the last ad hoc COLA (or the employee's retirement date, if later) came close enough.

A fidelity bond equal to at least 10% of plan assets is generally required. The minimum required bond is $1,000, the maximum required bond is $500,000 (ERISA Section 412). 1. I thought the maximum required bond had been increased to $1 million. Is this correct (or am I dreaming)? What is the site? 2. If a plan sponsor maintains 2 plans, each with assets above $5 million, can a single $500,000 bond that covers both plans meet the bonding requirement, or must the plan sponsor obtain two separate $500,000 bonds. (This assumes that the maximum required bond is still $500,000.) I suspect that two bonds must be obtained. Thanks

I've always shown the contributions in Form 5500 as the contributions "for the plan year." In other words, I take the contributions actually made during the year, add the receivable as of the end of the year, and subtract the receivable as of the beginning of the year. Pretty standard stuff. (And of course, the contribution agrees with the tax deduction taken.) Recently, I've seen (and taken over) a number of plans where the contributions shown on the 5500 were the contributions actually made "during the year" --- and excluding receivables. 1. If I transition to the "for the plan year" basis, I will wind up including an extra contribution (the end-of-year receivable) in my first 5500. So, the contribution shown in the 5500 will not equal the deduction. (Of course, in future years, everything will be fairly normal.) Any problems with this approach. 2. Or, can I stay on the "during the year" basis? And, if the IRS questions the deduction vs. the 5500 contribution, explain that the difference is due to the receivables. (Assume that this is not a DB plan.) This has come up several times on 401(k) plans where the employee contributions and assets were taken from the mutual fund statements (there were no company matches), and the receivable was ignored. The receivable is simply the employee's deferral in the last week of December that wasn't sent to the mutual fund until early January. Shouldn't the ADP test be done based on contributions deferred for the year (and not contributions actually made during the year)? Shouldn't the 5500 include those late December deferrals?

Both the old allocation method and the new allocation method would be OK as far as being definitely determinable. In other words, once the company makes its contribution, it has no discretion on how it is allocated; it merely follows the terms of the document. More important is the question of what is accrued? (And its corollary, what cannot be taken away?) Does an employee accrue a right to an allocation method for a specific period of time during the year? Actually no. The employee only accrued a right to contributions once they have been allocated to him. And they are allocated to the employee only when the employer "declares" his contribution, which technically is done as of the end of the year. (By "technically," I mean that an employer should declare their profit sharing contribution with a Board Resolution. I realize that this is not always formally done.) So, since the contribution (and its allocation) aren't "official" until the end of the year, there is no accrual until that time. A sidebar on this. IRS rules are generally set up to be consistent internally. In other words, the IRS generally does not allow you to do "X" directly, but allow you to accomplish "X" indirectly. This discussion of whether a change in allocation formula during the year is an example. Note that you could not change the allocation formula after the end of the year retroactive to the beginning of the year. In other words, by December 31 (in a calendar year plan), the allocation formula is cast in stone. The alternative of setting up another plan (like I described in by previous comment) doesn't work, since the other plan would have to be established by December 31 to accomplish this change in allocation formula. Consider, as an alternative, a money purchase plan. Here, once the employee has met the requirements for a contribution (say 500 hours without a last day requirement), the employer cannot reduce the formula for the entire year. The approach of freezing the first money purchase plan and establishing a second money purchase plan with a reduced formula doesn't work, because even if you freeze the first plan, the employee still gets the contribution. As a final example, there was a debate about 3-5 years ago about whether the allocation formula in new comparabiliy profit sharing plans (i.e., different contribution percentages for different groups of employees) was definitedly determinable. The IRS originally objected, but later relented (and agreed that it is definitedly determinable as long as the plan document is carefully written) in part because one could accomplish the exact same thing by establishing several separate profit sharing plans, each covering the specified group of employees and excluding the other groups. I'm not aware of any examples whether the IRS didn't allow "X" directly but allowed "X" indirectly. Sorry for the lengthy response. Hope this helps.

I agree with Mo Agains's comments. #1 - one has to be careful with the wording of the amendment, but that really shouldn't be a problem for someone competant. #2 - If you're currently in a safe-harbor, you'd have to do the general nondiscimination test, but unless you've got some weird changes in payroll midyear, passing should be easy. If you're currently not in a safe-harbor and already using the general test, this design should probably not make the testing any more difficult to perform (you still test annual allocations vs. annual compensation even it the annual allocations are determined based on the sum of two semiannual allocations), nor more difficult to pass (unless again you've got some weird changes in payroll midyear).

My understanding is that you can amend a profit sharing plan's allocation formula at any time prior to the end of the plan year, even if the eligibility for allocation in that year has been met by any participant. The rationale here is that there is no obligation to make a contribution; therefore, no right to allocation is "accrued" until the contribution is determined. For those who don't like this answer, one could accomplish the same thing by adopting another profit sharing plan before the end of the year, retroactive to the beginning of the year, with the desired allocation formula. Then make the desired contribution to the second plan and no contribution to the first plan. (Note that with a money purchase plan, this doesn't work since the contribution is "accrued" once the eligibility requirements are met. Therefore, once cannot amend a money purchase plan to take away these contributions once accrued.)

Is a 5500 required for a premium-only cafeteria plan? If it has more than 100 participants, is an audit required?

Say we have a pension plan with age 65 normal retirement and a lump sum feature (using GATT rates). Also, let's say that for funding purposes, the actuary is assuming 7% preretirement interest and that all employees will take a lump sum at the assumed retirement age of 65. Currently, the GATT rates are about 5.2% and the PBGC interest rate for the variable premium is about 4.5%. For valuation purposes, the actuary uses the current GATT rate to determine the lump sum. Is the present value of vested benefits for PBGC Variable Premium purposes equal to the lump sum (calculated at 5.2%) discounted at 4.5%, or the value of the age-65 life annuity discounted at 4.5%. The former will obviously produce a lower PVVB, since the lump sum (per the actuary's assumptions) will be less than the age-65 value of the annuity at the PBGC's 4.5% discount. As a second question, I have seen actuaries assume lump sum will be paid, but do not assume the current GATT rates will remain constant. Rather, they assume a specific GATT rate in the future (for example, 6.5%) in the funding calculations. In this situation, would the PVVB for PBGC purposes be calculated as the lump sum (determined at 6.5%) discounted to the present at the PBGC's 4.5%?

And why did you fire them the first time? And why did you take them back?

Actually, the laws are complex with good reason. Otherwise, experienced practitioners would allow clients to deduct more than the IRS would allow. For example, consider a individual sole proprietorship (with no employees) who nets $400,000. He sets up a pension and profit sharing plan like yours, and contributes the legal maximum of $30,000. He then thinks, "hey, if I incorporate half of my business and keep the other half as a sole proprietorship, I can net $200,000 from each and have a $30,000 contribution for myself for from each. Gee this is nice." Or the person earning $2 million who sets up ten corporations, with net earnings of $200,000 in each, and a contribution of $30,000 from each. Or, why don't I set up 2 companies, one that employs me (with a pension plan) and one that employs my other employees (with no pension plan). Not a bad idea to avoid contributing for those other employees. Or, if that won't work, why don't I simply lease those other "employees" so they won't even be my employees. Gee, I could have an entire business with one employee (me) and 10 non-employees, a/k/a leased employees. And no pension contribution for any of them. None of these approaches work, because the law has evolved nearly 30 years to prevent them. I (along with most contributors to this thread) could give additional examples --- suffice it to say the complexity in the law is deliberate, and prevents unwanted consequences, i.e. abusing the system. As far as Schwab's people not being familiar with these nuances, this applies to most investment brokerage firms (or banks or insurance companies) involved in this business. They make their money by HOLDING AND INVESTING YOUR PLAN'S ASSETS. They set up the pension documents as a loss leader (either as a freebie or at a significant discount to market rates). They do not make any money on administrative or consulting questions. They also only provide the types of retirement plans that they can understand and can support. The questions you ask are actually quite routine for experienced practitioners --- but not routine for Schwab (or others). That's why the people on this thread exist, because there are a lot of complex rules that become simple only after 10-20 years experience.

In the situation at hand, there actually is an old determination letter (from 1986 --- yeah , I know, there will be other problems when we restate at the end of this year). And the client is willing to indicate that this is the current plan.

It is typically not the best practice for an actuary to sign a Schedule B if the client (or his advisor) provides only an unsigned plan document. But, it is a violation of ERISA, etc., if an actuary does so? On takeover cases, I've sometimes seen prototypes (either standardized or nonstandardized) that are unsigned. That doesn't necessarily mean that there isn't a copy somewhere that has been signed. And the actuarial valuation and Schedule B signed by the prior actuary reflects plan terms consistent with the unsigned document. And when I push, a signed document is generally sent to me. However, what happens if the original signed documents simply do not exist. Assuming (for the moment) that they were truly signed, there is obviously no direct evidence of the existance of a plan. The indirect evidence would be -- the client has been operating assuming he has a valid plan, the prior actuary has been operating assuming he has a valid plan, the investment firm has been operating assuming he has a valid plan, etc. Now, this would of course become an issue when obtaining a determination letter, but that is often delayed, courtesy of 401(B). Meanwhile, there is an ongoing valuation requirement and a 5500 requirement. What to do in this imperfect situation --- ideas?

We've successfully trained most of our clients to ask us before doing most things --- to avoid unpleasant surprises. Boy was I surprised to find out what this client did! This is a 1 person DB plan, covering a married owner, about 45 years old. About $160,000 in plan assets. Nonstandardized prototype. I find out that his business is doing poorly, so over the last 6 months, he paid about $80,000 from the plan to himself. Gee, that's nice, but now I'm not happy (and he's gonna be less happy when I tell him how deep he is in it). Let's see. There is a loan provision in the plan (it's a C Corp), but I'm not aware of any loan documents being prepared (let alone spousal consent). Besides, $80,000 is greater than $50,000. And, he hasn't repaid any of the payments, if it even was a loan. If he is considered to be actively employed, he's made a benefit payment before NRD. Oops! Maybe he terminated employment before the first payment. (he hasn't taken a salary for several years). Well, the plan does allow for installment payments (as well as a lump sum and QJSA), but the actual payments were in random amounts at random times --- doesn't sound like installments to me. (Again, there's that spouse consent issue.) So, let's talk about damage control. Disqualifying defects. Penalties. EPCRS. By the way, the brokerage firm is issuing 1099's for the $80,000 just about now. (How did this happen, you might ask. I figure he called his broker and said "I need $_____ from my pension plan. Please transfer it into my personal brokerage account that you also have. Did they tell him it's not that simple? Did they tell him they'd be issuing 1099's? What do you think?) So, before I break the good news about all this to him, what alternatives and ideas can we come up with. Yeah, I know they're all bad --- I'm just looking for the least objectionable ones available! Like I said, we've trained MOST of our clients!

Regarding Question 1 - In general, no problem (subject to Top Heavy, if applicable). Be careful, and don't get greedy. For example, if you have a $250 flat contribution for a young employee who earns $2,500 in a year, you've got an extremely large cross-tested benefit accrual for such a tiny contribution. It just doesn't seem right, and the IRS might fight you if, for example, the employee terminated with no or little vesting. Regarding Question 2 - I agree with your concern. It doesn't smell right. Now, do you mean that the plan would fail if she were benefiting at the same percentage level as her husband? Why not pay her a salary, and give her no contribution or a small contribution? (The small contribution could be the same percent as one of the other classes.) Now, I agree the client would like to save FICA tax. Assuming the husband is above $80,400, the actual after-tax cost for FICA would be about $1,000 if her pay were $10,000. Offsetting this would be the small after-tax savings on the contribution for her. (If 3% of $10,000, the after-tax savings would be about $120, depending on their marginal tax bracket.) Several other ideas. Caveat your correspondence to them indicating your concern about an IRA audit. They might not like it, but you've got to cover yourself -- especially if it's an area that's unclear. Another idea is have them pay her no salary but a FICA-taxable bonus before yearend of, say, $1,000, and include her in the plan at either 0% or 3% of pay. Now, the FICA cost is only $100. Now is a salary plus bonus of $1,000 reasonable? Could the IRS argue against this? No guarantees, but perhaps this mitigates the FICA expense. Good luck.

IRS has historically been lenient in this area, as long as you tell them about it (rather than them finding out upon audit). How the Department of Labor will be henceforth, any ideas?

I think the site for recognition of amendments in funding is Rev Proc 77-2. I think. For an already-closed window, one couldn't fault you for making a beginning-of-year assumption of early retirements equal to the actual experience. Technically, I guess, you could place yourself at the beginning of the year and make a different assumption; the difference between this assumptions and reality would be an experience gain or loss. This could be done, for example, to increase the contribution requirement this year, with a lower requirement in future years as the gain is amortized (arguably if the actual takers were truly fewer than expected). I'd probably be more comfortable with assuming the actual experience if it is known, however. (For a multi-year window, you will have gains or losses because you don't know the results two or three years from now.)

Thank you Pax and RCK. Unfortunately, it's not gonna be simple. Mandatory employee contributions stopped being made over 10 years ago. Even if we knew the contribution rate back then (documents from that time are not to be found, so far), I suspect that salary information that old is unlikely to be found. (The plan sponsor is a non-profit; payroll records aren't the best.) The immediately prior actuary (for about 5 years) never dealt with this issue. The actuary before him was an insurance company; the plan sponsor has contacted them and is hoping that they have some records. Even though we have the accumulated value (contributions plus interest) as of now, I think that trying to reconstruct the original contributions by backing off the historical interest rates will not work --- I suspect there was inconsistency in the interest rate crediting methodology in the past (translated, "nobody has a clue!") We actually did receive information on actual employee contributions from one participnat! I suspect it was a fluke, and at this time, the client does not want to ask the affected employees (most of whom are actually former employees who terminated vested) for their pay records from 10-20 years ago. Kinda embarassing. Also, if in fact the plan sponsor is ultimately responsible, what are the consequences? Penalties? Fines? IRS and/or DoL. Disqualification! (Alright, let's be reasonable.) Thanks.

In an old-style contributory DB plan, part of benefit payments to a participant are nontaxable. This is because the participant made after-tax contributions; the amount of these contributions (without interest) is the tax basis in the benefit. Let's assume the benefit is a lump sum; this avoids the complicated basis recovery rules that I once knew a long time ago. What happens if neither the employer, the plan sponsor, the plan administrator (who happen to the the same company) nor the trustee (who happens to be an external bank trustee) have the records that show the amount of employee contributions? Who is ultimately responsible if this information cannot be located? Any suggestions on what to do if the information is truly unavailable. To assume zero contributions would overstate the participant's tax liability. To arrive at any nonzero amount would essentially be a made up number. And salary records for the time that employee contributions were made (over 10 years ago) are unavailable. Any ideas?

Adding a 401(k) plan is often needed because the client wants it as a benefit for their employees A situation we often face is with an existing cross tested profit sharing plan, and the client asks "why can't I have a 401(k) plan? My employees want it. etc." Our solution is to suggest they adopt a 401(k) plan with no company match. That way, there is no cost to the company, other than for administration/recordkeeping. We'll do (if asked) the ADP test (since there is no match, it's easy), and the 5500 (for a fee). Since the motivating factor behind the 401(k) plan is to offer something for the rank-and-file employees, I usually can convince the HCEs not to push the ADP limits. (Besides, with the look-back rule for ADP testing, it's usually no problem.) And with 401(k) deferrals floating around, that can mean that the plan's are not top-heavy, and that the 401(a)(4) testing is easier to pass.

To Andy H. No, the boxer wasn't George Foreman. It was some middleweight about 20 years ago. (Far be it for me to err on Mr. Foreman --- he's bigger than I am!) To Sampat and the other readers --- this thread illustrates how complicated some situations can be, and how it pays to look into a lot of the what if's. To Sampat -- how old is your employee? When was he hired? How many hours per year does he work? And how much does he earn over the year? This information is crucial in determining some of your alternatives?

Consider an old style contributory DB plan. You know, after-tax contributions that were required to get a pension benefit. But there's a twist. I have a client with this type of plan, but they changed it so that no new contributions are required and that existing employees would henceforth receive their regular plan benefit (1% of pay per years of service) PLUS their contributions and interest. Interest is being credited to the employees "contribution account" as actual investment earnings earned by the overall plan assets. A 65-year-old retiring participant is now due a pension from the 1% per year of service formula. The lump sum value of this pension is $10,000 (lump sums are permitted in the plan). In addition, he contributed $3,000 in after-tax dollars to the plan; with $5,000 interest, this now totals $8,000. So the participant is due $18,000. My question is: what can be rolled over into an IRA. Clearly the $10,000 is taxable, but can be rolled over. Also, the $3,000 is not taxable, and cannot be rolled over. What about the $5,000. It would be taxable, but can it be rolled over? HELP!

This has been a good thread. Some good insights for Sampat, and some good discussion among practitioners. Side comment. I once used an assumed retirement age of 28 for a boxer. My justification (to my boss, it never came up with the IRS) --- "I saw his last fight!"