ALERT / AUGUST 12, 2026

IRS proposes rules on dependent care FSA discrimination testing, including 55% benefits test

9 MIN READ

Included with proposed regulations issued this week on the tax treatment of employer contributions to Trump Accounts, the IRS has issued first-of-its-kind proposed rules on the discrimination testing requirement for dependent care flexible spending accounts (FSAs).

Why the dual guidance? Because the rules that apply under the Tax Code to employer contributions to Trump Accounts indicates that nondiscrimination requirements apply similar to the requirements to dependent care assistance plans. So, the IRS guidance addresses both Trump Accounts and dependent care assistance plans.

Below we will discuss the new IRS proposed guidance and how it impacts dependent care FSAs.

Executive Summary

The IRS proposed guidance on dependent care assistance plans provides two significant clarifications for employers sponsoring dependent care FSA plans:

· Discusses the operation of the 55% average benefits test that applies to dependent care FSAs. Surprisingly, the IRS endorsed testing methodology will help many dependent care FSAs pass the test.

· Allows employers to correct 55% average benefits testing failures by including in income any excess benefit amounts of highly compensated employees up until Jan. 31 of the year following the testing year. For example, employers would have until Jan. 31, 2027, to fix testing failures of the 2026 calendar year.

Overview of IRS Nondiscrimination Rules

At a high level, the purpose of the dependent care FSA nondiscrimination testing rules is to ensure employers are not treating highly compensated employees (HCEs) and owners more favorably than non-highly compensated employees (non-HCEs). Although employers have offered dependent care FSAs for decades, until now the IRS had yet to issue guidance on the mechanics of nondiscrimination testing. Dependent care FSAs are subject to four separate nondiscrimination tests:

Contributions and benefits test (do the benefits provided under the plan favor HCEs?)
Eligibility test (is eligibility under the plan available to a nondiscriminatory group of employees?)
Owner concentration test (“25% Test”) (are owners receiving more than 25% of all benefits under the plan?)
55% average benefits test (are non-HCEs receiving at least 55% of the average benefits received by HCEs?)

Who qualifies as an HCE is the same definition as used in 401(k) plans. For 2026, employees who earned more than $160,000 in 2025 and 5% owners are HCEs. If a dependent care FSA fails the nondiscrimination rules, the tax consequences generally fall on the HCEs. To correct a testing failure, employers have historically needed to reduce the pretax elections of HCEs to get the plan to pass by year end.

Lockton comment: Employers are encouraged to test early in the plan year so there is sufficient time to make any corrections needed for the plan to satisfy the nondiscrimination rules by year-end.

55% average benefits test in operation

One of the challenges in applying the 55% average benefits test is determining how to calculate the "average benefit" provided to HCEs and non-HCEs. The Code requires that the average benefit provided to non-HCEs be at least 55% of the average benefit provided to HCEs, but prior to the recent proposed regulations, there was little IRS guidance on how to determine the average benefit for each group. As a result, testing vendors developed different methodologies, often producing significantly different results. Because the test is based on utilization, plans with low non-HCE participation rates have often struggled to satisfy the test.

In practice, the average benefit for each group is determined by dividing the total dependent care benefits provided to the group by a specified number of employees in that group. Historically, vendors have used one of three approaches for determining that denominator:

The total number of employees in that group;
The number of employees in that group who are eligible to participate in the program; or
The number of employees in that group who actually participate in (e.g., enrolled in and benefiting from) the program.

The choice of denominator can have a significant impact on the outcome of the test because it changes the average benefit calculated for both HCEs and non-HCEs. In the absence of IRS guidance, many vendors used the first approach, which generally produces the lowest average benefit amounts and often makes the 55% Average Benefits Test more difficult to pass. The proposed regulations endorse the third approach by providing that the average benefit should be determined using only employees who receive dependent care assistance during the year.

Take the following example as an illustration of the three options:

Company A (a single employer not within a controlled group) has 5,000 employees: 1,000 are HCEs and 4,000 are non-HCEs. Company A offers a dependent care FSA program where employees can contribute up to $7,500 per year. There are 800 HCEs who are eligible, but only 200 participate and they contribute a total of $1 million. For the non-HCEs, 2,000 are eligible but only 50 participate and they contribute a total of $225,000.

The average benefit for non-HCEs is determined by dividing $225,000 by one of the options below:

The total number of non-HCEs in the group (4,000), giving us an average benefit of $56.25;
The total number of eligible non-HCEs (2,000), giving us an average benefit of $112.50; or
The total number of participating non-HCEs (50), giving us an average benefit of $4,500.

Of course, the same issues apply when determining the average benefit for HCEs. The average benefit for HCEs is determined by dividing $1 million by one of the options below:

The total number of HCEs in the group (1,000), giving us an average benefit of $1,000;
The total number of eligible HCEs (800), giving us an average benefit of $1,250.
The total number of participating HCEs (200) which gives us an average benefit of $5,000.

The non-HCE figure is then divided by the HCE figure to determine the average benefits, and the plan passes only if the average benefit provided to non-HCEs is at least 55% of the average benefits provided to HCEs. In the three methods above, the plan passes using Option 3 but fails under Options 1 and 2.

In the absence of IRS guidance, many FSA vendors have used Option 1. As illustrated, this typically resulted in many plans failing this test.

Lockton comment: Given the varying methodologies, employers should understand how their testing vendor calculates average benefits and whether that methodology aligns with their interpretation of the applicable nondiscrimination requirements.

Pleasant surprises in proposed rules

Arguably, the most significant clarification in the proposed regulations for dependent care FSA plans is what HCEs and non-HCEs can be counted in the denominator for calculating the notoriously difficult-to-pass 55% average benefits test. The IRS has indicated in the proposed rules that plans divide “by the number of employees in that group [meaning HCE or non-HCE] to whom any dependent care assistance is provided during the plan year, via salary reduction or otherwise.” In other words, the IRS has endorsed Option 3 where we only look to employees who are participating in the dependent care FSA.

Lockton comment: Given the challenges of passing the 55% average benefits test, the administrative complexity of making corrections, and the employee frustration that can result when elections must be reduced, some employers have reconsidered whether to continue offering a dependent care FSA. The proposed rules provide good news for employers who have struggled to pass the 55% average benefits test and may open up opportunities to enhance the benefit for employers who were hesitant to increase the dependent care FSA max to $7,500 (as allowed under the One Big Beautiful Bill Act, due to testing concerns. The increase to $7,500 (if adopted) could potentially still result in a higher likelihood of failing the 55% average benefits test, especially if HCEs elect the larger maximum while non-HCEs continue to participate at lower rates.

Additionally, if a plan fails the 55% average benefits test, the IRS permits the plan to reduce pre-tax elections or tax HCEs on their otherwise pre-tax contributions to get the plan to pass. The proposed rules provide that any correction must be made by Jan. 31 of the year after the testing year (e.g., the IRS’s deadline for issuing employees W-2s for the prior year). For example, corrections for 2026 must be made no later than Jan. 31, 2027, and included on the 2026 W-2. The best practice is to test early in the year to allow plenty of time to adjust pre-tax elections.

Lockton comment: It is welcome news that employers have additional time to make necessary corrections to pass testing. If the corrections for 2026 are made in 2027, withholding taxes may prove challenging.

Other odds and ends

The proposed rules also address topics such as who can be excluded from the 55% average benefits test (collectively bargained employees, some short service employees, employees earning less than $25,000), the operation of the eligibility test (mirrors the ratio percentage test for retirement plans) and the owner concentration test (rarely an issue except for very small employers). There is nothing really new with these odds and ends, and most plans currently follow the proposed rules. Testing vendors will likely welcome the clarified regulatory framework for the existing tests.

Now What?

The IRS will take comments on the proposed rules until Sept. 25, 2026, and will issue final guidance at some point in the future. The final guidance will apply for plan years beginning after the final rules are issued.

Of note, the proposed regulations do not create new nondiscrimination requirements, rather they provide a clearer regulatory framework for applying the existing rules. The clarification regarding the 55% average benefits test is good news for employers and especially timely given the permissible increased dependent care exclusion from $5,000 to $7,500 (if adopted).

We anticipate that some testing vendors may begin to modify their testing processes and procedures in keeping with the IRS guidance, while others may wait until final regulations are issued.

Although the current rules are only proposed, employers who have not completed their 2026 testing may want to ensure their vendor’s approach to testing aligns with the IRS guidance. Employers who have tested and failed may want to ponder retesting based on the IRS guidance.