Short answer
ADP/ACP automation is mostly a data problem. Validate the census, determine HCEs with recorded inputs, compute each eligible employee’s deferral and contribution ratios using the plan’s compensation definition, compare group averages with the statutory limit, then calculate corrections and their deadlines. Store every input so results can be reproduced and reviewed.
What the tests measure
The Actual Deferral Percentage (ADP) test compares the average elective deferral rates of highly compensated employees (HCEs) with those of non-highly compensated employees (NHCEs). The Actual Contribution Percentage (ACP) test does the same for matching and employee after-tax contributions. Both are set out in Treasury regulations (26 CFR 1.401(k)-2 and 1.401(m)-2). Safe harbor plans are generally exempt from the ADP test, and from the ACP test if the match meets the safe harbor conditions.
Data preparation is most of the work
Automated tests are only as good as the census. Before testing, check every row for:
- Missing or implausible dates of birth, hire, termination and rehire.
- Hours and service needed for eligibility and entry dates.
- Compensation under the plan’s testing definition, capped at the 401(a)(17) limit for the year.
- Deferrals by type (pre-tax, Roth, catch-up) and employer contributions by source.
- Duplicate or changed SSNs.
Our payroll file check example shows this kind of row-level validation with sample data.
HCE determination
An employee is generally an HCE for a year if they owned more than 5% of the employer in that year or the preceding year, or received compensation above the IRS threshold in the preceding year (optionally limited to the top-paid group). Ownership includes attribution from family members. Automation needs ownership data from the sponsor, family relationships and prior-year compensation, and should record which test made each person an HCE. Thresholds come from the IRS cost-of-living table and should be stored as dated configuration.
The calculation
- For each eligible employee, compute the actual deferral ratio: elective deferrals for the year divided by testing compensation, including zero for eligible employees who did not defer.
- Average the ratios for HCEs and for NHCEs. Under prior-year testing, the NHCE average comes from the preceding year; under current-year testing, from the same year.
- The HCE average passes if it does not exceed the greater of 1.25 × the NHCE average, or the lesser of the NHCE average + 2 percentage points and 2 × the NHCE average.
| NHCE average | 1.25 × NHCE | Lesser of +2 and ×2 | Maximum HCE average |
|---|---|---|---|
| 1.0% | 1.25% | 2.0% | 2.0% |
| 3.0% | 3.75% | 5.0% | 5.0% |
| 4.0% | 5.0% | 6.0% | 6.0% |
| 8.0% | 10.0% | 10.0% | 10.0% |
| 10.0% | 12.5% | 12.0% | 12.5% |
Try the interactive ADP example on our home page or the ADP/ACP estimator on kinetiplan.net.
When a test fails
The common corrections are distributing excess contributions to HCEs (the total excess is found by leveling the highest HCE percentages, then distributed starting with the HCEs who deferred the highest dollar amounts), recharacterizing excess deferrals as catch-up contributions where eligible, or making qualified nonelective or matching contributions (QNECs or QMACs) for NHCEs. Corrective distributions made within 2½ months after the plan year ends generally avoid a 10% excise tax on the employer (6 months for certain plans with an eligible automatic contribution arrangement), and the failure must be corrected by the end of the following plan year. The IRS Fix-It Guide describes these options. Software should date each plan’s deadline and track the correction to completion.
Make every result traceable
Store, with each run: the census version, plan elections (testing method, safe harbor status, compensation definition), limits and thresholds used, the HCE list with reasons, each employee’s ratio, group averages and the limit. A reviewer should be able to open any result and reproduce it. When rules or data change, a new run is created rather than overwriting the old one.
Automation checklist
- Census validated with an exception queue before testing
- Limits and thresholds as dated configuration
- HCE and key employee determination with recorded inputs
- Current-year and prior-year methods supported
- Corrections calculated with deadlines in a calendar
- Results stored with inputs, versioned and reviewable
- Regression suite of worked examples agreed with compliance staff
This article is general information for plan professionals, not legal or tax advice. Apply the plan document and current IRS guidance.
References
- eCFR: 26 CFR 1.401(k)-2, ADP test
- IRS 401(k) Plan Fix-It Guide: the plan failed the ADP and ACP nondiscrimination tests
- IRS: retirement plan definitions (including highly compensated employee)
- IRS: COLA increases for dollar limitations on benefits and contributions
- IRS: EPCRS overview
Written by the Software Home engineering team, which builds 401(k) software and the KinetiPlan platform; we have an interest in some of the choices discussed. General information only, not legal, tax or investment advice.