Testing where self-correction involved

Monday brain cramp. Suppose you have a DC plan utilizing cross testing. The rate groups either pass the ratio test, or the plan passes the ABT test. Fast forward a year, and it is discovered that census data was incorrect, and a participant or participants were improperly excluded. So the employer has to do make-up contributions, matches, whatever.

Do you have to go back and re-run the nondiscrimination testing taking into account the corrections? I don't think you do, but I'm not putting my finger on the guidance to back that up.

Thanks.

Why would you not? If you include this person and testing would then fail, do you think you then don't have to correct? If IRS audited do you think they would give the employer a free pass on testing with incorrect data? 

I dunno. I was probably thinking about the correction for failure to implement an employee election, etc., where determining an ADP/ACP test failure in addition to the missed implementation,  the plan could rely on a test performed with respect to those employees not impacted by the failure, and disregard those who do require the make-up contribution.

P.S. - finally found what I was looking for which of course agrees with you - RP 2019-19, Section 6.06, .02(2)(d). Gotta fix a 401(a)(4) failure if created as a result of the correction.

Hopefully your allocation method is an "each in own" so that you can just give this person(s) enough to pass testing.  If it's a "class" allocation it certainly has the potential to be a bigger problem.