401k plan with multiple discretionary match formulas

A plan came to our attention that has been making the same discretionary matching contributions for the past three years at least. The first formula is 100% of the first 3%, but they also pay 100% of the first 6% if a participant is employed longer than 10 years. I've never seen this before. I thought the maximum service requirement for an employer allocation is 1000 hours. Would they need to pass ACP testing on each formula separately? 

These are a pain but otherwise ok.  Don't look at the 10 years as an eligibility, think of it as a points-based allocation.  The ACP testing still only looks at the match on an aggregate basis, it doesn't care about the underlying formula(s), it just compares the HCE amounts to the NHCE amounts.  Each formula will need to be tested under 401(a)(4) to ensure the group benefitting isn't discriminatory.

Can you explain what the difference is between ACP testing the aggregate contributions, and testing each formula under 401(a)(4)? Passing the ACP test is supposed to be the exclusive means of showing employer contributions satisfy 401(a)(4). Unless by "each formula" you mean the combination of contributions produced from each formula.

I'm always confused what 401(a)(4) is referring to unless it's ADP/ACP, and I'm taking the QKA..

You know the ACP test, run that as usual.

The rate of match itself is a BRF (benefit, right, or feature). See Treasury Regulation 1.401(a)(4)-4. The plan’s BRFs pass 401(a)(4) if they are available to nondiscriminatory groups.

There’s both current availability and effective availability to satisfy.

To satisfy current availability, look at the ratio of NHCEs available to receive that rate of match (or better) divided by all NHCEs who can get any match, regardless of what they deferred. Do the same for the HCEs at that same rate of match. Now divide the NHCE fraction by the HCE fraction. Look up the safe harbor percent that would apply from 1.410(b)-4 (don’t worry about the average benefits test for BRF testing, that does not apply here). Now do it again for the next rate of match for the HCEs and NHCEs.

For example, suppose all the HCEs have over ten years of service so the higher rate is available to all of them, whether they defer or not. The HCE ratio is 100%. Suppose only 1 of ten of the NHCEs have ten years so their fraction at this rate of match is only 10%. You have a test result of 10%/100%, or 10%. There is no safe harbor rate under 20.75% under 1.410(b)-4, so the higher match rate fails.

If you’re within 9.5 months after the plan year end, look at 1.401(a)(4)-11(g)(3)(vi) and (vii) to fix the failure.

For effective availability, subjectively decide if you think it’s nondiscriminatory and if the IRS would agree with you. No test of numbers. Good luck there. Alright, alright, if you pass current availability for multiple rates of match, then I would say it sure points us to a passing effective availability result as well.

Hopefully that helps. 

Ahh, that makes sense. The availability tests aren't covered in QKA, so I was missing that piece. Appreciate the help all.