Coverage Testing

Controlled Group

Company A sponsors a traditional 401(k)

Company B sponsors a traditional 401(k)

1. Plan design is identical - same service, match etc.

When completing the coverage test each plan on its own will pass coverage

Company A   25 HCE all benefiting / 100 NHCEs all benefiting.

Company A   75 HCE all benefiting / 1250 NHCEs all benefiting.

Each entity passes coverage on their own.  It there a requirement that the numerator in the coverage test include the total for each entity?

It looks like you are hoping to pass the Ratio Test.  Keep in mind that when you are testing a plan for coverage within a controlled group, the numerator is the count of individuals benefiting in the plan, and the denominator is the number of nonexcludable individuals in the controlled group.

For A, you have 25 out of 100 HCEs and 100 out of 1350 NHCEs.  The ratio is (100/1350) / (25/100) = 7.41% / 25% = 29.63% < 70% = fails.

For B, you have 75 out of 100 HCEs and 1250 out of 1250 NHCEs.  The ratio is (1250/1350) / (75/100) = 92.59% / 75% = 123.46% > 70% = passes.

(Please double check the math).

You can test the plan together (permissibly aggregate) the plans and get a ratio of 100% = passes, or try using average benefits testing on A.

Also keep in mind, when it comes to coverage testing, Elective Deferrals are a "plan", Match Contributions are a "plan" and Nonelective Employer Contributions are a "plan".

I am told by the TPA since the plans are identical,  they pass coverage and no need to permissibly aggregate the plans.   This does not follow the above example ( which agrees with our findings).

Not sure if the TPA's opinion is that the plans are identical so the the numerator is the # benefiting for both employers and the denominator is the total for both employers. Runs counter to  our findings, but Controlled Group testing is cumbersome and very confusing.  

Both plans being identical, does not give them a pass on 410(b).  The process is the same, the denominator is all the excludable employees of the controlled group, the numerator are the participants benefiting for the respective employer. 

Bottom line the only way for Company A and B to pass coverage would be to permissibly aggregate the plans.

Thoughts???

kinda feel like we want to give the TPA the benefit of the doubt that they meant they pass by permissively aggregating, figuring everybody who mattered was in at least one of the plans. 

Getting the right result but the wrong reason, as anyone who likes pension credential exams may be used to seeing as a multiple choice option....

If everything is identical, then permissive aggregation very, very likely should not pose a problem for either plan.  Under permissive aggregation, the more vulnerable test may be testing the NECs if Company B has a disproportionately large number of employees who, on applying allocation conditions like 1000 hours and a last day rule, are not benefiting but are considered nonexcludable (think for example terminated with more than 500 hours).  Again, not likely.

Looking forward to the 2023 5500s, there is a new compliance question asking if the plan used permissive aggregation.

When aggregating for coverage, how "identical" do the plans need to be?  Does it matter if the investments options are not the same, for example?

Differing benefits, rights, and features are also subject to nondiscrimination testing. See 1.401(a)(4)-4. You have “current availability” that you can test, and “effective availability” that you can smell.

Note if the TPA aggregated the plans for coverage, that they would also need to aggregate the plan for ADP/ACP.