The Nondiscrimination Testing Part I article discussed what tests are required for different plan types and what information is needed to prepare for testing—including the definitions of highly compensated employees (HCEs) and key employees.

Plan sponsors also need to understand what the tests compare and how to correct failed tests.

Test methods

Internal Revenue Code Section 410(b) minimum coverage testing compares the proportion of non-highly compensated employees (NHCEs) who are covered by the plan against the proportion of HCEs who are covered by the plan, explains Mark Carolan, senior associate with Groom Law Group in Washington, D.C. If the percentage of NHCEs is at least 70% of the proportion of HCEs covered, the test is satisfied.

Alternatively, Carolan says, this test can be performed by weighing the average benefits that are earned by the NHCEs versus the HCEs (although certain gateway requirements must be met initially in order to use this alternative). He explains that the gateway requirements are: a reasonable classification test, and a nondiscriminatory classification test. “Essentially, these are designed to ensure that an employer doesn’t just randomly pick specific employees just to get over average benefit percentage testing threshold,” he says. The “reasonable classification test” is just a facts-and-circumstances look at whether the group of employees who are covered by the plan is based on bona fide business criteria (i.e. salaried, union, management, etc.). The “nondiscriminatory classification test” is a technical test that considers a number of different ratios under the plan, to evaluate if there are enough NHCEs covered. If both of these are satisfied, the plan can use the average benefit levels to satisfy the coverage testing under 410(b).

Amy Ouellette, director of retirement services at Betterment for Business in New York City, explains that the average deferral percentage (ADP) test compares the 401(k) contributions, as a percentage of income, of the HCE group against the NHCE group each year, and limits how much more the HCE group can defer on average. Employee Roth and traditional deferrals, but not age-50 catch-up contributions are considered. The average for each group (HCEs and NHCEs) is the sum of all of the group’s deferral rates divided by the number of all eligible participants in that group, whether they deferred or not.

There are two methods for performing the ADP tests: The first provides that the ratio of the contribution average of HCEs to that of the NHCEs may be no more than 125%, and under the second method, the average contribution for the HCEs may not exceed the lesser of the average contribution of the NHCEs plus 2% or the average contribution of the NHCEs times two.

However, Ouellette explains the testing this way:

• NHCE average rate 0% to 2% – the HCE rate cannot exceed by more than double (NHCE % x 2). For example, if the NHCE average rate is 1.7%, the test passes if the HCE rate is less than or equal to 3.4%;
• NHCE average rate 2% to 8% – the HCE rate cannot exceed by more than 2 points (NHCE % + 2%). For example, if the NHCE average rate is 4.7%, the test passes if the HCE rate is less than or equal to 6.7%;
• NHCE average rate 8% or greater – the HCE rate cannot exceed by more than 125% (NHCE % x 1.25). For example, if the NHCE average rate is 9.2%, the test passes if the HCE rate is less than or equal to 11.5%.

For 401(k) and 403(b) plans, the average contribution percentage (ACP) test on employer matching contributions and after-tax employee deferrals (not Roth) uses the same methods.

Plan sponsors may base the ADP and ACP percentages for NHCEs on either the current or prior year contributions. The election to use current or prior year data is in the plan document.

In certain cases—for example, when plans may offer different matching contribution formulas for different employee locations—a benefits rights and features test may need to be performed. Carolan explains that generally, the amount of benefits (and other plan rights and features) provided by a plan cannot disproportionately discriminate in favor of HCEs. The testing compares the benefit accrual under a plan for HCEs with the benefit accrual for NHCEs. Essentially, there must be sufficient NHCEs receiving benefit accruals in the same range as the HCEs for the plan to pass the test.

Ouellette says top heavy testing utilizes the account balances, with some adjustments to ignore certain rollovers and add back certain recent distributions, as of the end of the prior plan year. If key employees are determined to hold more than 60% of plan assets, the plan is considered top heavy for the next plan year.

According to Ouellette, Employee Retirement Income Security Act (ERISA) Section 401(a)(4)—or general nondiscrimination—testing measures the fairness of non-elective contributions made to HCEs versus NHCEs. It groups HCEs by rate of allocations, and looks at how many NHCEs received an equal or greater rate (rate groups).

The ‘rate’ can be determined in one of two ways—allocation based (the amount provided today) or benefits based (the present value of the expected future value of such contribution). Ouellette says benefits based testing is often referred to as ‘cross-testing’ because it crosses over into testing like a defined benefit (DB) plan, rather than a DC plan (which typically only looks at contributions in what’s allocated today, not a potential future value). Cross-tested allocations are typically used in plans in which the HCE group skews significantly older than the NHCE group, e.g., owners and higher-paid employees are older than rank-and-file staff. This allows a greater portion of the profit sharing allocation to go to those closer to retirement.

Many plans bypass this test by using uniform profit sharing allocation formulas, such as pro-rata, flat dollar or Social-Security integrated formulas, Ouellette says.

Internal Revenue Code Section 401(a)(26) minimum participation testing means the plan must benefit a minimum number of employees. The minimum is equal to the lesser of 50 employees or 40% of total employees. According to Carolan, this is required even if the plan would otherwise pass the coverage test described above. In addition, the benefits provided (or that have already accrued) to these employees must be “meaningful.” Whether or not benefits are “meaningful” is determined from all the relevant facts and circumstances, he says.

Corrections for failed tests

If a plan initially fails the coverage test, many plan sponsors look to other options for performing the test, Carolan says. This can include aggregating the plan with other plans sponsored by the employer, making a qualified separate line of business designation, or a number of other alternatives. “These approaches can be complex and should be carefully reviewed by counsel and other providers, but they may provide a more cost-effective way to pass this test,” he says.

If none of those alternatives work, specific corrective measures must usually be taken and a close review of the options for correction is sometimes beneficial, according to Carolan. For example, a retroactive amendment must usually be adopted by the 15th day of the 10th month after the end of the plan year in which the plan did not satisfy the coverage test. Generally, the amendment must retroactively extend coverage, to the beginning of the year that included the coverage testing failure, to excluded employees (or increase benefits to NHCEs) such that the plan would pass the coverage testing after taking the retroactive benefits into account. If the corrective amendment is not adopted by the deadline, then a plan usually files a voluntary correction program (VCP) request and receives IRS approval of the correction.

For ADP and ACP tests, the correction can be a return of excess contributions plus interest to HCEs by 2-1/2 months after plan year end, according to Ouellette. She explains that the tests must be corrected within 12 months, but anything past 2-1/2 months results in an excise tax for late correction. For the ACP test, vesting schedules are considered and some or all of the excess may be forfeited.

Employers may also correct the tests by making a qualified non-elective contribution (QNEC) to eligible NHCEs. Ouellette says the methodology varies, but the result is to bring the NHCE rate high enough that the HCE rate is considered passing. Finally, she says a combination of refunds to HCEs and QNECs to NHCEs may also have favorable results. The employer may make some level of employer contribution to reduce the refunds to HCEs.

Ouellette points out that there are nuances to testing, including the definition of compensation used, and whether all eligible employees are tested together, or if you can group them into those who have met statutory eligibility requirements versus those considered “otherwise excludable.” “What this means is that there is more than one way to skin a cat, and depending on the flexibility of a plan’s governing document and consistent practices, the failure and remedy can be chosen based on goal and best result—one testing method may mean a less expensive correction than another,” she says.

Correction options for a benefits rights and features test are essentially the same as those for minimum coverage testing, according to Carolan. There are several tools available that can help a plan pass testing if it initially does not pass, such as aggregating the plan with other plans, bifurcating the plan into smaller plans, and others. If those tools still do not make the plan pass testing, specific remedial action must usually be taken. A retroactive amendment must usually be adopted by the 15^th day of the 10th month after the end of the plan year in which the plan did not satisfy the benefit amount testing. Generally, the amendment must retroactively increase benefits such that the plan would pass testing after taking into account the new retroactive benefit increases. If the corrective amendment is not adopted by the deadline, then a plan usually files a VCP request and receives IRS approval of the correction.

According to Ouellette, in a plan year for which a plan is considered top heavy, a minimum contribution is due to eligible non-key employees. The minimum contribution is the lesser of 3% of compensation or the highest contributions made to any key employee for the year (including 401(k) deferrals).

Failure of ERISA Section 401(a)(4)—or general nondiscrimination—testing would require additional contributions to eligible NHCE such that the test passes. This means either a higher budget for the profit sharing contribution or a decrease in what was originally allocated to the HCEs, Ouellette explains.

Carolan adds that for DB plans, the minimum benefit generally equals 2% of the average compensation received over the last five years. Additional rules apply for non-key employees who participate in both types of plan, and the plan documents should be reviewed carefully to properly provide their minimum benefits. These benefits must also vest on a two- to six-year graded schedule, with 20% additional vesting provided each year.

Carolan says there are a few options to correct for a failed Internal Revenue Code Section 401(a)(26) minimum participation test. First, if the plan is merged with another plan before the 15th day of the 10th month after the end of the plan year with the failure, and the resulting merged plan passes the test, it will be deemed to retroactively satisfy the rule. Unlike other tests, though, two plans cannot just be deemed to be aggregated for this test; they must be formally merged together.

If a merger is not available, a retroactive amendment must usually be adopted by the 15th day of the 10th month after the end of the plan year in which the plan did not satisfy the test. The amendment must usually either extend coverage to more employees, or provide more employees with “meaningful” benefits, such that the plan would pass testing after taking those retroactive provisions into account. If the corrective amendment is not adopted by the deadline, then a plan usually files a VCP request and receives IRS approval of the correction.