Average Benefit Test - Controlled Group - Tested Separately

We have 2 plans (A and B - they are part of a Controlled Group) that we are trying to test separately.

They fail Ratio Percentage Test when tested separately.

My question is how should the Average Benefit Test be done? Based on my research, the correct approach is one of these 2 ways:

1) Average Benefit Test is done on an aggregate basis (so, the contributions of all employees in both plans are taken into account). If the ABT is passed this way, then the plans can be tested separately.

2) Average Benefit Test is done by considering all contributions of Company A and $0 and 0% for Company B's participants (since none of the participants of Company B are benefiting under Company A). And, the test is also done for Company B (in which case, all contributions of Company A participants are counted as $0 or 0%).

Which way is the correct way to test Average Benefit Test for a Controlled Group where we would like to test each plan separately for compliance testing?

Thank you!

The average benefit test is only satisfied if the plan satisfies both a) the nondiscriminatory classification test, and b) the average benefit percentage test.

The nondiscriminatory classification test looks at the ratio percentage for the plan, and compares it to either the safe harbor or unsafe harbor ratio (depending whether you are using the safe harbor test or the facts-and-circumstances test). Only the employees benefiting under the plan being tested are treated as benefiting when performing this test.

The average benefit percentage test is performed for the entire testing group; in other words, all benefits under all plans of the employer are aggregated and there is only a single average benefit percentage for all plans in the group.