Testing of 15 Separate Plans in a Controlled Group

Holding company has 15 wholly owned subsidiaries which all have separate qualified plans (mostly 401k plans). 

Does anybody have any tips/tricks for simplifying the process of performing (or avoiding?) coverage (and other?) testing across the controlled group? 

I presume you mean beyond "make all the plans have identical provisions" and are interested in how you're going to tabulate results. 

Maybe a big spreadsheet with rows for each plan, columns for excludable/nonexcludable benefiting/nonbenefiting.  At least that way you'll have your denominators ready to go when you copy formulas in making each ratio.  Might also help if you're going to need to permissively aggregate any of the plans.

@Bri is on the right track.  You will need, for example, to track meticulously a lot of information about:

• tests to be performed by plan (Deferrals, Match, NEC), and across the controlled group or testing groups (410 coverage, ADP/ACP, 401(a)(4), 415, comp limits...)
• each plan's provisions (eligibility, entry dates, service accrual (hours/elapsed time), compensation, testing compensation, birth/hire/term dates, classifications...)
• plan years (hopefully all plans have the same year)
• any mandatory disaggregations (ESOPs?, current vs prior year ADP/ACP testing?, Safe Harbor vs Non-Safe Harbor?, unions??)
• any permissive aggregations that may be needed to pass testing
• census data for all employees of each employer
• census data for employees who worked for more than one employer within the controlled group
• testing strategy (ratio vs average benefits test, benefit accruals vs allocations...)
• any mergers, spinoff, or acquisitions within any permissible transition time period
• ownership at all levels.

If there are Defined Benefit or Cash Balance plans in the mix, then there needs to be close cooperation between the actuaries and any others involved in testing.

Look out for testing quicksand/tar pits:

• Keep in mind that the determination of HCEs is done across the controlled group and, generally, if one of the plans is not using top paid group rules, none of the plans can use the top paid group rules.
• 5% ownership rules applies at the employer level, ownership of a subsidiary can trigger unexpected HCEs,
• Plans with very liberal eligibility provisions.
• Union employees.
• Leased employees (who may need to be included in testing even if they are excluded as a classification).

This is in many ways the tip of the iceberg and is not a comprehensive list. 

Testing this many plans within a controlled group is very labor intensive, and the complexity increases almost exponentially with the number of plans and employers involved.  Be sure to price the effort properly and make sure you consider the fees that may be charged by other professional like outside legal counsel or independent actuaries.

If this is your first time performing testing at this level of complexity, you will have lifetime memories from this project.  You may want seek out someone who has done it before to provide some guidance and be a resource as your engagement progresses.  

Good Luck!

Agreed - this is definitely a compliance consulting project. The first year set up and identification of all relevant issues noted above will be the most involved, then ongoing it's mostly data collection and manipulation. This is not trivial, nor are the ramifications if not done properly, so you should charge commensurately. This is where we as consultants earn our money.

Thanks, obviously a daunting and expensive task.

Since the 15 employers are widely separated geographically, no employees work for more than 1 of the employers and all the plans have TPAs that "test" the plans as if they're not part of a CG could we reasonably reach a conclusion that the CG is "in compliance" if we gathered enough data to confirm that all plans pass coverage when tested on the entire CG basis?

I seem to remember that if all plans can pass coverage you don't have to perform any group ADP/ACP testing or worry about different BRFs?

Short answer - No.

One way to look at it is there are two different levels of 410 coverage testing.  One is done at the plan level (or aggregated plan level).  The other is done at the controlled group level, and must pass before you get to the plan level test.

For example, if you are using the ratio test at the pan level, the denominators are the nonexcludable HCEs and NHCEs in the plan, and an employee is excludable when applying the provisions of the plan.  (The numerators are the HCEs and NHCEs that are benefiting from the plan.)

If you using the ratio test at the controlled group level, the denominators are the nonexcludable HCEs and NHCEs across the entire controlled group, and an employee is excludable when applying the most lenient provisions of any plan in the controlled group.  This is a greatly oversimplified comment meant primarily to illustrate how the 410 coverage test at the controlled group level and plan level may look the same, but they are not the same.

FYI, there is a possibility of using QSLOBs where based on geographic locations (among a wide range of other things), but the QSLOBs must be elected by 9/15 following the close of the plan year and QSLOBs have a whole other set of arcane rules.

When a one-employer group has about 60 distinct plans, is it feasible to use (after data entry) a commercially-developed software product, or must the service provider do the work by custom-crafting?

And hey, as someone who does controlled group tests....I suppose you're going to have to assume that everyone on the census reports you receive actually was offered the relevant plan when appropriate.  I can't even imagine testing this where people who "actually were eligible" weren't picked up by the plan administrator.  It's a completely different issue, but in theory could affect your results if corrective QNECs aren't made. 

(Insert scariest face emoji here.  🤬)

@Peter Gulia, commercially-developed software has limitations simply because attempting to program all possible testing scenarios would be cost prohibitive, be a nightmare to maintain, and still not be able to cover the universe of all possible testing scenarios.  it also would consume a lot of computer resources and likely not be able to create reports that describe how the results were arrived at.  Some software can be used to help if the input required to do the test can be accommodated reliably.

Controlled group testing easily can expand rapidly, particularly when a plan actually is considered three different plans for purposes of coverage testing (deferrals, match and NECs).  It would be not surprising that a group with 60 401(k) plans would result in needing over 150 separate tests.

Big controlled groups exist, and compliance testing is being done, but it takes a lot of discipline, careful engagement management, broad knowledge of testing quirks and available software tools, and meticulous tracking of plans and census data.

As @CuseFan said, "This is where we as consultants earn our money."

I think Excel was commercially developed. That should do the trick, plus at least a 50 hours of billable time.

2 hours ago, John Feldt ERPA CPC QPA said:

I think Excel was commercially developed. That should do the trick, plus at least a 50 hours of billable time.

50?! i.e 3 hours and 20 minutes per plan?  assuming no data issues, all provisions are standardized, everything is passed on the first try and zero time spent on documentation and notes?  double that.  Here is another tip - in old good times we would put on a wall a large map with multipe colors highlighting the permissive aggregations.  Might still be an efficient and relevant tool inspite of all the AI advancement.