Deep dive tax wonkery: the DCFSA vs. the new child care credit

Time for another round of tax wonkery. This one starts with a question I keep hearing from parents: "my dependent care FSA went up to $7,500 this year, that's good, right?"

Yes. Kinda. For some of you.

For a lot of families, the smartest DCFSA election going forward will be zero dollars, and the reason is a different provision that got a bigger upgrade with far less press. And that other upgrade creates a strange problem for the companies running these plans, which we'll get to, because it's the fun part.

Three provisions changed at once

The 2025 tax law changed three child care provisions at once, all effective starting this year:

1. The dependent care FSA cap went from $5,000 to $7,500. The old cap had been sitting there since 1985, so this got the headlines.
2. The child and dependent care credit, the one you claim on your own tax return, jumped from a 35% maximum rate to 50%, with a new income schedule. This got much less press.
3. The employer child care credit roughly tripled. Hold that thought; it's the answer to the problem the first two create.

I've written before about what a DCFSA actually is: your own wages, redirected before tax, spent on care. Its value is how much you put in times your marginal tax rate. Federal bracket, plus 7.65% payroll tax (less if you make over the Social Security limit), plus your state's rate. For a family in the 12% federal bracket with a 5% state tax, that's about 25 cents per dollar. For a high earner, closer to 40.

The credit works differently. It's a percentage of up to $3,000 of care expenses for one child, $6,000 for two or more, taken straight off your tax bill. The $3,000 and $6,000 bases haven't changed, but the percentages are now much more generous. The new schedule: 50% at the bottom, sliding down one point per $2,000 of income until it holds at 35% just above $43,000, staying there until $75,000 for single filers or $150,000 for joint filers, then sliding again to a floor of 20% (which it reaches above $103,000, or $206,000 joint).

The two benefits step on each other's feet

Here's the rule that makes this a fight instead of a stack: every dollar you run through a DCFSA is subtracted from the credit's expense cap. Elect $7,500 with two kids and your $6,000 credit cap becomes zero. It's one or the other.

So the choice is: marginal-rate cents per dollar on up to $7,500 of your own money, or schedule-rate cents per dollar on up to $3,000 or $6,000. Which is bigger depends on your income (it drives both sides), how many kids you have (the credit cap), what care actually costs you (whether the FSA's larger capacity matters), your state tax (FSA side), whether your state pays its own dependent care credit as a percentage of the federal one (credit side; New York does), and whether you owe enough tax to use a credit that isn't refundable.

Three families, all with care bills bigger than any of these caps:

Married, $95,000 income, two kids. The FSA saves their marginal stack, about 24.65% with a 5% state tax, on $7,500: roughly $1,850. The credit pays 35% of $6,000: $2,100. The credit wins by about $250, before any state piggyback credit widens the gap. Notice the trap: the FSA's first dollar costs this family 35 cents of credit and saves them under 25. Contributing a little is worse than contributing nothing.

Married, $185,000 income, two kids. Now the credit rate has slid to 26%, worth $1,560, while the FSA's value has climbed to about 34.65% of $7,500, almost $2,600. The FSA wins by over $1,000, and the gap keeps growing with income.

Single parent, $80,000, one kid. Credit rate: 32%, but one child caps the eligible expenses at $3,000, so the credit maxes out at $960. The FSA holds $7,500 at about 24.65%: roughly $1,850. The FSA wins on sheer capacity. But give that same parent a part-day arrangement costing $3,000 a year and the answer flips: $960 from the credit beats $740 from the FSA. Same person, same income, opposite answer, purely on the size of the care bill.

At the top of the income range, with a 5% state tax, the crossover point for a married two-kid household lands between roughly $130,000 and $150,000 of income, for two reasons. Around $133,000 the family crosses into the 22% federal bracket, and if their care bill is big, the FSA's $7,500 of capacity starts beating the credit's $6,000 cap even while the rates are nearly tied. At $150,000 the credit rate itself starts fading toward 20%, which settles the argument for everyone else. Single parents with two kids lose the credit's edge around $85,000, and the one-kid expense cap drags the answer back toward the FSA for anyone with a full-price care bill.

That totally makes sense

So, DCFSA is better at higher incomes, CDCTC at lower, right? Long acronyms but easy enough to remember once you get them down, right?

Oh no. Not so fast. The CDCTC is a nonrefundable tax credit. If you don't owe any federal income tax in the first place, the CDCTC does absolutely nothing for you. And quite a lot of lower to middle income families owe nothing, because of two other credits. These are the child credit and the earned income tax credit. The child credit is a flat $2,200 per child under 17 for all families making under $400,000 (joint), up to $1,700 of which can be refundable. The EITC is a "negative income tax" designed to incentivize poor people to work; people with at least one kid get from 34 to 45 cents paid for every dollar they earn up to a maximum of around $7,500 for incomes around $20,000. Beyond that range, it phases out again, at a 21% rate.

The EITC and child credit are both worthy programs, but they make the CDCTC useless for most families making less than around $60,000, who receive money on net at tax time. In this income range, the DCFSA wins again: it still shields income from the 7.65% payroll tax, and by lowering reported income, it increases the size of the EITC. Until income gets too low, below roughly $25,000, when the EITC phases in. Then a DCFSA becomes actively harmful, as it makes the EITC get (much) smaller. A pre-tax benefit with negative value. Tax policy is fun.

In practice, very few jobs paying below $30,000 offer a DCFSA in the first place, but it's still something to keep in mind, particularly for part-time workers.

A New York note, since I live here. The state pays its own dependent care credit as a percentage of the federal one: 110% at low incomes, 100% through $50,000 of state AGI, sliding to 20% above $65,000. Two mechanics make it stronger than a simple percentage suggests. It is refundable, unlike the federal credit. And it's computed from the federal credit you were allowed on paper, before the liability cap and before the child-credit displacement above, because the state form recomputes the credit from your expenses and never asks about your federal tax. The $40,000 family whose federal care credit nets to zero still collects the full $2,220 from Albany, in cash. In the calculator's New York mode, that keeps the credit path winning up to roughly $37,000, and within a few hundred dollars of the FSA's EITC-boosted total around $40,000, which is exactly the range where guessing is most expensive.

Technical takeaway #1: for two-or-more-kid married households, the credit's winning zone now runs from roughly $70,000 to $135,000 of income. Below it, the EITC makes the FSA the better tool for most families, and the care credit's actual value can be zero. Above it, brackets and the FSA's bigger cap take over. Wherever you sit, do not default into last year's election. Run the numbers; the gap is often a few hundred to a couple thousand dollars.

What does this mean for employers?

Dependent care plans have to pass fairness tests, and the hardest one is usually the 55% average benefits test: the average benefit going to non-highly-compensated employees has to be at least 55% of the average going to highly compensated ones. These are computed over everyone eligible, not just participants.

An example: A company has 110 eligible employees: 100 non-highly-compensated, 10 highly compensated. Twelve of the hundred participate, averaging $5,000 each. That's $60,000 spread over 100 people: an average benefit of $600. The test then caps the highly compensated average at $600 divided by 0.55, about $1,090, which across ten people is $10,900 of total elections. If three executives each elect the full $7,500, that's $22,500, and the plan fails by a mile. Failing means the highly compensated participants lose the exclusion on everything they put in: it lands back on the W-2 as taxable wages. This test is why your TPA cuts executive elections mid-year and why DCFSA administration email has a certain haunted quality.

So the plan lives and dies by non-highly-compensated-employee uptake. And the new credit is, mechanically, a targeted offer to exactly those employees: across a wide middle band of household incomes, the credit now beats the FSA for families with 2 or more kids, so the rational move is to stop participating. Every one of them who correctly runs the math lowers the non-highly-compensated average, which lowers the ceiling on what anyone highly compensated can elect. In the example above, if half of those twelve participants rationally walk, the executive ceiling drops from about $1,090 average to about $545. Your most financially knowledgeable mid-income employees are now being told, by the tax code, to stop propping up your compliance test. Benefits consultancies have started flagging the same dynamic, Mercer among them.

Technical takeaway #2: if mid-income employees respond rationally to the new credit, DCFSA average benefits testing gets harder every year, and it was not easy to begin with. Model it before open enrollment, not in the March true-up.

The way out is employer money

You can patch this. Seed non-highly-compensated accounts with employer contributions (which count towards the average, but also cost you money). Cap executive elections early instead of clawing them back late. Those are patches.

The structural answer is to stop building the child care benefit out of employees' own salary. I've written about why employer credits play nice with DCAPs: a dependent care benefit funded by the employer keeps the same tax-free treatment for the employee, up to the same $7,500, and because it's employer spending, it can qualify for the employer child care credit. Under the expanded rules, the federal government reimburses 40% of what employers spend on employees' licensed child care, 50% for smaller companies, up to $500,000 a year ($600,000 for the smaller ones). After the credit and the deduction math, a dollar of qualified spend costs a profitable company roughly 40 to 50 cents federally. Some states add their own employer credit on top; New York has one, and in the right situations the combined recovery runs to roughly 70 cents on the dollar, sometimes more. There are guardrails: the care must be licensed, the benefit has to be available across the workforce rather than an executive perk, and the contracts and paperwork are finicky. That fairness requirement is a feature here: a flat employer-funded benefit passes the 55% arithmetic almost by construction, because both averages contain the same number.

One last wonky detail, because it decides how the fix should be built. Employer dollars dropped into FSA-style reimbursement accounts do not earn the employer credit; the credit attaches to money paid for care itself, through contracts with licensed providers. So an employer writing checks to fix a wobbling DCFSA test is spending the money in a way that earns nothing back. If employer money is going into child care anyway, and the test math says it increasingly must, spend it the way that gets 40 to 50 percent of it returned.

The takeaways:

• Families between roughly $70,000 and $135,000 with two or more kids in care: check the credit before you re-elect the FSA. Below that band the EITC usually hands the win back to the FSA; either way, partial elections are usually the worst of both worlds.
• Higher earners: the FSA is still clearly your tool, assuming your plan still lets you use it.
• Employers: assume rational behavior thins your non-highly-compensated participation, model the 55% test now, and price the seeding you'd need to keep the plan upright.
• Then compare that seeding cost against an employer-funded program that passes its own fairness test and earns a 40 to 50 percent federal credit. That comparison is not close.

I built a calculator that runs both sides of this: your own FSA-or-credit math as a family, child tax credit and EITC interactions included, with a New York mode (city tax and all), and the 55% stress test plus program economics as an employer. It's here.

Illustrative, not tax or legal advice.