"Rate Banding"

[BG5150]

Any good write-ups on using rate banding when allocating a profit sharing.

I'm kinda hazy on the topic. One of my plans magically passes testing when I check that little box in Relius. Otherwise it fails.

[FAPInJax]

Relius uses an 'aggressive' approach when the little box is checked and has always been explained in those terms. I would recommend using the manual banding option and then reviewing the results carefully. This is because the regulations just refer to 'significant' differences between HCEs and NHCEs with no definition of the terms. What is significant - I would argue that having the HCEs have a larger number than the NHCEs (regardless of the size of the difference)??

[Tom Poje]

the issue hinges on how you interpret the regs

Within a specified measurement range, it is permissible to simplify rate group testing by treating all participants as having the midpoint accrual rate—as long as the HCEs do not fall significantly higher than the NHCEs within the range. [Treas. Reg. §§ 1.401(a)(4)-2©(2)(v), 1.401(a)(4)-3(d)(3)(ii)]

My understanding of that is illustrated as follows (I pulled this from the book I have worked on, Coverage and Nondiscrimination Answer Book (Chapter 8 Q 8-23) (As best as I can paste in this format)

(and much of this from comments from Larry Deutsch as well)

Nadia (HCE) 11.2%

Orson (NHCE) 11.0%

Paul (NHCE) 10.9%

At first glance, the plan appears to fail nondiscrimination testing because there are no NHCEs with an EBAR equal to or greater than that of Nadia, the HCE. If the accrual rates are banded, however, the following results will be obtained. With a midpoint of 11.0 percentage points, using the 5 percent rule results in a measurement range of 10.45 (11.0 − 0.55) to 11.55 (11.0 + 0.55). Thus, all participants with EBARs between 10.45 and 11.55 are treated as having an EBAR of 11.0. The plan passes the general test because the testing results are now as follows:

was now

Nadia 11.2 11.0

Orson 11.0 11.0

Paul 10.9 11.0

The ratio percentage is now 100 percent, because both Orson and Paul are now in Nadia's rate group.

Caution. The IRS could interpret the banding described in Example 8-6 as significantly favoring Nadia, the HCE, because her EBAR is being brought down to the EBAR of Orson, who is an NHCE, by a factor larger than the factor used to raise Paul's EBAR. Therefore, such adjustments should be made only for mathematical convenience—and should not be used as a corrective measure to enable the plan to pass a failed rate group test.

Example 8-7. The facts are the same as those in Example 8-6.

________________________________________ ________________________________________ ________________________________________

Nadia (HCE) 11.0% 11.0%

Orson (NHCE) 11.2% 11.0%

Paul (NHCE) 10.9% 11.0%

In this example, one NHCE's (Paul's) allocation rate is increased to that of Nadia, while the other NHCE's (Orson's) allocation rate is reduced to that of Nadia. In this case, there is a balance taking place in the grouping of the rates (unlike Example 8-6, in which no NHCE's EBAR is decreased to the HCE's EBAR).

.........................

The logic being as follows, the regs say you can use 5%, but then turn around and say you 'can't' if it significantly favors the HCE. well, if the HCE is at the high point and you bring him down to a mid point and bring NHCEs up to a midpoint that would indeed seem to fail the 'significantly' favor the HCE requirement. (as opposed to bringing some NHCEs down to the midpoint as well) how far can you push it? Obviously there is no clear guidance.

[Tom Poje]

by the way, if you are that close to passing, it is possible that using

age definition nearest (if the HCE is born in the first half of the plan year) would help

age definition last birthday if the HCE was born in the second half of the plan year.

or using interest rates of 8.5% pre and 7.5% post

using 1983 IAF as the mortality and impute disparity

[John Feldt ERPA CPC QPA]

Derrin Watson recently did an advanced topic in cross-testing session at the SunGard Advanced Pension Conference in Chicago where he had about 7 or 8 slides which included a good explanation and a good example. He credits Tom Finnegan of The Savitz Organization for his assistance. Not sure if this topic might be in any of the ERISA workshops SunGard is doing, but if you have any SunGard product, perhaps you could ask Derrin for a copy of those slides and see how it goes.

Or maybe test using highest consecutive 3-year average compensation, or highest 4-year average, or 5-year, etc.

Or comp from date of entry.

Or if you need testing fixed for just a year (or a few years), perhaps accrued-to-date testing might help.