I'd like to hear from people experienced in discretionary class plans and client communications regarding same.

I've worked mostly with cross tested plans with hard wired formulas and am trying to get a large block changed to discretionary class plans. One downside would seem to be that more back and forth might be required with clients before the desired allocations are finalized and tested.

How is this typically approached when there are multiple groups? Is it efficient to have written guidelines in terms of how relative contributions are determined, i.e. class two gets 1/2 of class 1 and class 3 gets 1/3 of class 1?

The context is not a small, local-only operation where regular client meetings are held. I'm talking about a wide geographical spread of clients.

What are people's experiences with how to make this process most efficient? All opinions are welcome.

Moving to the discretionary class approach is definitely the way to go. Our experience has been with smaller clients where the main intent is to maximize the contribution for the principals while holding down costs for rank and file employees. We determined the appropriate groups first in design of the plan. Usually fall into two groups: favored and unfavored, with a possible third group if necessary for any younger HCEs who may blow up the allocation. We then determine the optimal contribution to pass the General Test and then communicate this to the client. If total cost is acceptable, we then prepare the written instructions from Employer to the Trustee which specify how contribution of $x (including forfeitures - need to take these into account also) will be allocated among various groups (ie. $Y to Class A and $Z to Class b). We're usually the ones making the decision given our small (doctor) clients.

In summary, clients are looking to us for the best solution; once we find it, we tell them the way to go.

On a different subject, I've heard and read of a few people promoting the idea of bringing spouses into plans for 2002 with high 401k deferrals under a Safe Harbor 401k plan with New Comparability discretionary PS allocation. Looks great and can be done under new 404 and 415 guidelines; however, may want to consider impact of an HCE getting 100% of comp before presenting to client (great if doc's wife can put away $11,000; not so great if this same $11,000 causes the rank and file amount to go to the moon in order to pass the General Test;) ).

Not sure exactly what the question is.....

However, we deal with exclusively discretionary class plans (new comp, cross-tested). I don't know why anyone would hard code an allocation formula in a document.

Anyway, our SPD's just define the groups, not any sort of percentage or ratio compared to the other allocation groups in the plan.

Here's the question posed a different way. How do you allocate a contribution in a time effective manner if there are few rules?

Assume for a moment you were doing new comparability allocations/testing for a mutual fund company. Where do you start in terms of allocations if there are no rules other than the gateways, the groups, and the 415 limits?

What questions do you ask?

And, actuarysmith, there are advantages to having formulas in the document. The number of unknowns are less. Once you have either a dollar contribution or a clear objective, the allocations are easy. There is no back and forth. I am trying to argue that the flexibility of discretionary class plans outweighs the increased time expenditure that might be caused by more back and forth, and I'm looking for supporting arguments.

I agree this is no big deal with small closely held plans, in the manner that mwyatt suggests. But not all plans fit this model. I have to deal with lots of plans with 4 or 5 classes that don't fit the "give me the max and minimize everybody else" mold, so I'm looking for suggestions.

Ah, so _that's_ the question!

I don't think I have a magic bullet for you, but I strongly believe that the flexibility of pure class-based allocations is worth its weight in gold. There are certain circumstances where it is unneceesary, such as the one HCE plan or the multiple HCE plan where everybody knows, in advance, that the HCE's want to max out. In those situations a formula based solution, like a super-integrated plan, makes a lot of sense. Or, where it is impossible, such as where a money purchase plan is required in order to allow a contribution in excess of 15% of pay. Of course, this last reason disappears in 2002 for all states that have conformed to EGTRRA.

However, even in those cases, things get a little sticky if the business model changes.

I think the back-and-forth is limited, although not eliminated, if the allocation for the year begins with an estimate that is performed based on the client's preference. In the absence of guidance, I use the allocation parameters from the prior year.

But, with guidance, if it works, there is no back and forth.

Sort of puts a little meat behind the expression: "Well, how much do you want it to be?"

Assuming the client says: "25% to Groups A and B, 10% to group C, 3% to groups D & E" and the test passes, there is nothing left to do. Assuming the client says: "$50,000 into the plan, maximize group A, minimize group E, and make Groups B, C and D equal" we can put tht together, too.

In the event that the test fails at the desired client contribution levels, we typically can identify two methods of making it work: increase the allocation to the groups necessary to make it pass, or design a point-and-shoot -11g amendment. Here is where the judgement comes in. One can spend quite a bit of time trying to figure out the best mechanism for lifting a failed test into passing territory.

I wish there was a good way to have the client determine, in advance, whether they want us to contemplate all the basic testing methods, such as accrued to date with permitted disparity, or not. And whether they want us to move into restructuring (amazing how that helps sometimes). But all of that takes extra time.

Does that help?

Thanks, all for the comments. I happen to agree with each of you, and these experiences will be helpful. I'm trying to convince others.

One other point alluded to by Mike Preston (even in the small business world) is what happens if the circumstances change (terminations, additions, etc. to your population affecting Testing results)? If your document specifically details exact allocations/relations among groups, what do you when/if you fail the General Test? In the discretionary among class approach, you really haven't "failed" anything yet, as you just adjust how the dollars are allocated until you pass, neatly sidestepping all the issues of corrective amendments, etc.

BTW, that was one of the problems I had with the Super Integrated approach. The "fixed" New Comp plans I've seen usually started with a "bucket" approach where you first fill 3% of everyone's bucket, then fill your favored groups' bucket up to 415, and then filled the other buckets with what's left over (usually amount to satisfy General Test). This is all well and good, but what happens when that third "bucket" overfills the 404 limit? Or say your client wants to contribute a lower amount?

Good point.

We had a solution for that, which was an amazingly efficient failsafe provision, which was in almost all of our (non-takeover) plans.

But, now with the GUST restatement process, the IRS has required that we remove the failsafe language, so your point is a good argument in favor of discretionary class plans. Thanks.

PS,

By the way Sal Tripodi sells a package of cross tested language and commentary, which we purchased. I think it is outstanding. Lots of great insights. For example, it has suggested design for super integrated plans under the gateway rules using discretionary class plans. And much more stuff I found very insightful. I recommend it highly. Just came out a few weeks ago.