Average number of allocation groups or classes?

[AndyH]

I've been asked to respond to a comment from a software vendor that

"most documents don't provide for more than 2 groups resulting in substantial extra charges for those clients who end up with more than 2 groups"

Are there major document providers that limit plans to 2 allocation groups or discretionary classes?

Is having 2 groups the most common?

It would surprise me if either were true. Comments please?

[Guest Tbrown]

I can't imagine that many plan documents would limit the allocations to just 2 groups. What I have seen is that just about any plan that has more than 30 or so participants, uses more than 2 groups. The only ones we have that only use 2 are small medical or dental practices where there might only be 5-10 employees.

[Guest merlin]

Corbel's pre-printed volume submitter checklist allows for 5 allocation classes, but we took over a client at the end of 2001 that had 7 classes and we used the Corbel as plan without any problem. I don't know that they have a limit. I could ask if you'd like.

But in any event I think that the only limit should be the client's desires,tempered by practicality.The 12/31/01 allocation for my 7-class client above reduced to only 3 classes.We could't slice and dice the allocation any further. But again,that should be between me and the client. My document or software shouldn't be the governing factors.

[actuarysmith]

I would venture a guess that nearly half of our cross-tested plans have 3 (or more) allocation groups.

It seems that there is nearly always some reason to treat at least one participant (outside of the key execs group) differently.

For example, an office manager, HR person, or even a son or daughter of one of the HCE's covered in the "A" group. All of the other non-"A" group employees are typically treated the same.

Therefore, a very typical design would be Group A - Target Execs

Group B- Someone or position that needs different treatment, and Group c - all other employees.

As a practical note, we have some plans that have up to 6 or 7 groups (somewhat rare). I always encourage clients to select the absolute minumum number of classes that will get the job done. I explain that if you have too many groups (i.e. resulting in a small number of participants in each group) a relatively small change in the demographics from one year to the next can cause your testing to "Blow Up".

Maybe this is more information than you wanted - in which case I apologize. Bottom line - I don't know where the software vendor derived the notion that most plans used only 2 groups. It sounds like they did not have a very sophisticated system and were trying to justify not having the "goods" to deliver to you.

Good Luck!

[AndyH]

Thank you all for your comments. I agree completely; just wanted a little reinforcement.

[Fred Payne]

My experience has been different from that of actuarysmith's: the larger the number of allocation groups the better the ability to accomodate changes in demographics.

One client of mine illustrates this. There are two groups, 12 in Group A and 80 in Group B. A new doc is scheduled next year to move out of Group B into Group A. But because of his relative young age, the resulting allocation to Group B must increase from 7.7% to 11% if Group A is to continue to receive its current 20% allocation. The new doc will get an increase of about $24K in contribution at the expense of a $50K+ increase in funding costs of the NHCs! Alternatively, I could keep Group B's funding the same but drop Group A's contribution to accomodate this "shift" in demographics.

Adding a Safe Harbor 401k and having all of Group A defer the max drops Group A's allocation to 14.5%. I can consequently keep Group B's allocation with the shifting demographics to about 8%. But if I could isolate the new Doc in his own allocation group, I could skip the 401(k) and keep the allocation rates of Group A and B the same. Group C members will get an allocation somewhere in between.

I think more groups give you the flexibility to maximize contributions for the greatest number of HCEs while minimizing funding costs to the NHCs. By isolating the HCEs with the potentially highest EBARs in a separate group, you can better avoid a "blow up."

However, this begs the question as to the IRS' tolerance for such manipulation of contributions to large numbers of allocation groups comprised primarily of HCEs. Several attorneys I have spoken to warn against a potential characterization of the allocation to numerous groups that essentially have only one or two members as a "closet" CODA--even if the 401k feature exists. I see in another current post Tom Poje notes his having seen allocation groups identified by participant name. A Corbel attorney told me last week not to do that (although I would love to do this with impunity!)

Ideally, I think the math of the cross test works best if there is significant latitude in identifying larger numbers of allocation groups for the HCEs. What are people's thoughts about this, particularly identification by name?

[Tom Poje]

Fred:

A couple of thoughts.

Always request a determination letter for whatever classes you name.

Since minimum participation for DC plans was eliminated years ago, the following possibility was created. You could conceivable set up 10 plans and name 10 individuals, each with his own plan. As long as you pass nondiscrimination you are ok. Obviously you are going to aggregate the plans for testing purposes. Now, except for the expense of maintaining multiple plans, what is the difference between that and simply having one plan with named classes?

[actuarysmith]

Fred-

I should have clarified my comments a little better. I do not disagree with you about having more rates for HCE's.

The problem is too many rate groups (spreading the groups thin) with NHCE's.

I have had clients that want to have target HCE's in 1, 2, or 3 different groups. Fine. But then, they want office managers, sales managers, and who knows who else in then next group, followed by wharehouse managers, clerks, etc. in the next group, followed by the receptionist, etc. etc. etc. in the next group.

Then you show them this illustration in which the relative order between the groups works out fine (I.e A%>B%>C%>D% etc.)

The problem with these lower groups, is that sometimes there can be one young NHCE that is allowing your testing to pass. They end up quiting, and the entire order of the universe (the plan) is upset.

So to clarify what I really intended to say is limit the number of rate for NHCE's to the absolute minimum possible that gets the job done.

[Mike Preston]

I think I need a fruit salad. ;-)

I'm going to weigh in on what I think is the opposite side. More groups equal more flexibility. Whatever level of flexibility exists for n groups, that flexibility is enhanced with n+1 groups. Whether the groups selected result in the optimum flexibility is subject to the demographic shifts you mention.

[AndyH]

What is the acronym for an acronym error? Did you mean :-) ?

[actuarysmith]

I am almost to the bottom of my fruit salad. I'll try to say what I am trying to say one more time (did that make sense to any of you? it didn't really to me either).

At the risk of sounding like a complete bowl of fruit myself (Ok, I agree - we are past that point.............) I agree with Mike Preston.

n+1 groups is more flexible than n groups no matter what the groups are defined as.

What I mean, is that when you are setting up your plan, and you show your client the following scenario-

Group A - Physicians 20% of pay

Group B - Clinic Directors 10%

Group C - All others 5% of pay

Let's assume that they reply with , "WOW that's great, that is exactly what we wanted" . If group C is reasonably large (relative term) then it is not too likely that you will have to come back next year and tell them that Group C requires a 8% of pay allocation to pass ABT.

On the other hand, if you have a plan that is set up as-

Group A - Physicians 20% of pay

Group B - Clinic Directors 15% of Pay

Group C - Registered Nurses 12%

Group D - Orderlies who like Fruit Salad 7%

Group E - Orderlies who hate Fruit Salad 6.5%

Group F - Receptionist 6.0%

Group G - all Others 5.0%

I would agree you have a much more flexible plan. However, I would advise extreme caution in how this plan design is communicated to the sponsor.

When that very young orderly who hates fruit salad gets mad and quits, it may result in Group E requring a 9% of pay allocation rate to get your testing to work. (He was one of only two orderlies who hate fruit salad. The other orderly is 50 years old)

We would probably all think that the plan still worked pretty well. But we all have clients who feel that there is a certain pecking order and they do not want to violate this in the amounts of retirement plan contributions. Neither do they want to be surprised by large changes in the contribution amounts. They might be upset that Group E had to be treated better than Group D.

I know this is a silly example, and quite trivial. I know that many clients would be fine. I am merely trying to explain why I still feel that it is best to recommend to a client to think of the minimum number of rate groups to get the job done - YES the plan is a little less flexible, but the numbers tend to be more stable from year-to-year.

Alas, I have come to the bottom of my fruit salad............ I will take leaf of you all and peel out of here, it's the pits .......................(Whew, it's either not early enough or not late enough for such a bad pun to be funny. Sorry

;) )

[Mike Preston]

AHA! (How do you spell that expression?) Having just finished lunch (I'll let you guess!), I might now understand your position.

In most plans (certainly not all) the definition is more along the lines of:

Group A - Physicians

Group B - Clinic Directors

Group C - Registered Nurses

Group D - Orderlies who like Fruit Salad

Group E - Orderlies who hate Fruit Salad

Group F - Receptionist

Group G - all Others

where the percentages are decided after we have given advice to the client. In the case where the percentages aren't cast in concrete, you have more flexibility.

The basic point you are trying to make, which I think is that the volatility of the results is inversely proportional to the body counts in each group is pretty close to the mark. I probably wouldn't word it quite that way, but it is a valid point.

What I think is more relevant is the specific bodies being used as foundation. I've seen a case with 400 participants, but there is one HCE (youngish) that requires exactly two NHCE's in the highest contribution category be below age 25. The rest of the test doesn't matter. You need those two specific bodies to make the test work. And it really doesn't matter whehter those folks like fruit salad or not, or whether they are part of a very large group.

[actuarysmith]

Mike-

I am pretty sure that "AHA" is spelled "AHA" !

Yeah, I see your point. BTW, we do not use the old hard-coded formulas that specify percentages in our documents. I was just showing the allocation %'s for illustration.

We put together proposal illustrations during the sales process and work through those with the clients near the end of year. Often, they become quite focused on a specific dollar amount or percentage of pay that they will recieve (i.e. $35,000 or 20% etc.)

They also have a sense of how the other allocation groups will stack up.

And you are right, frequently there are a couple of key young participants that provide the basis for passing the testing - it is not just any youngen' (I don't know how to spell that........)

So we now have the Preston - Smith Theorem that states that "The volatility of the allocation rate groups is inversly proportional to the number of participants in each rate group". I want to see this on actuarial exams soon!

Wow, we could become famous or something!

Anyway, I am all out of fruit salad, licked the bowl clean. I guess I'll have to go have a couple of beers now instead. At least those lame jokes I tried to pull earlier will start to sound a little funnier.

[Guest wlank]

"Are there major document providers that limit plans to 2 allocation groups or discretionary classes?"

Mine does NOT!

"Is having 2 groups the most common?"

Most plans I administer have more - three of four groups being most common.

Bill