Another difficulty in complying with the testing rules is the timing. The result of the ADP and ACP tests are not known until after the end of the plan year, usually the calendar year. If the test(s) are failed, there are several remedies available to adjust the results. The most common are refunds to the HCE's and/or additional contributions for the NHCEs. In an attempt to avoid these potentially costly and complex remedies, SBJPA allows current year ADP and ACP testing to be based on the prior year's results.

Beginning in 1997, the limit for the HCE group can be based on the prior year's NHCE ADP, or, optionally, the test can be done as in the past. As a result of this change, all plans can guarantee passing the ADP and ACP test by limiting the deferrals for each of the HCE's for the current year based on the NHCE average for the prior year. If, for example, it is known that the NHCE actual percentage for 1996 was 5.25%, a limit of 7.25% for each HCE will guarantee compliance. The disadvantage of this approach is that it results in underuse of the allowable HCE percentage. If some of the HCEs defer less than 7.25%, this would normally allow others to defer more than 7.25% because the test is based on an average. If all HCEs are individually limited to 7.25% to guarantee compliance, the average may be less than 7.25%, thereby not taking maximum advantage of the limits. For 1997 and 1998, either approach may be used in either year. Rules still must be issued by the IRS for plans that use current-year data for 1998, or a later year, and want to switch to prior-year data for a subsequent year.

One other item of importance regarding existing 401(k) plans deals with one of the corrective mechanisms for failed ADP or ACP tests. Under current law, if the ADP test is failed, a refund can be made to the HCEs to reduce their ADP to the allowable level. This process is sometimes referred to as "top down" approach. (See Table 2)

Under current law, the first step would be to reduce the highest percentage deferral to the next highest level and test again. If the test is not satisfied, the process is repeated until the test is met. In the example in Table 2, HCE #3 would be reduced to 9% and the HCE average recalculated to 7.979% ((5.938 + 9 + 9)/3). We still haven't passed the test so HCE #2 and HCE #3 are reduced further until the test is passed. If that reduction is below the 5.938% of HCE #1 then HCE #1 also is reduced. So step two is to reduce HCE #2 and HCE #3 to 8.281%, and the test is passed (((5.938 + 8.281+ 8.281)/3)= 7.500%).

As you can see, the lowest-paid HCE received the highest refund. The IRS has believed for some time that this result is not equitable, and SBJPA has corrected that inequality. Beginning in 1997, the refund will be calculated the same but will be allocated beginning with the highest deferral amount, not the highest deferral percentage. In our example, HCE #1 first received $500 to bring him to the level of HCE #2. Next, both HCE #1 and HCE #2 receive refunds of equal amounts ($797). Even though when the test is rerun it doesn't satisfy the ADP test guidelines, it is assumed to pass when this process is followed correctly

Another change now allows plans to include short-service, nonhighly compensated- employees (less than one year of service), yet exclude them from the ADP or ACP tests. This provision will allow companies to open the plan to all employees without the risk of failing the test because, generally short-service employees tend not to participate (see Table 3).