Dependent Care FSA vs. the Child Care Credit for Self-Employed Parents (2026): Both Rules Just Changed

Published: September 21, 2026 · Reading time: 13 min

TL;DR: Two different tax breaks pay for the same daycare invoice, and both of them were rewritten by the same statute for tax years beginning after December 31, 2025 — meaning 2026 is the first year either number applies. The §129 exclusion for employer-provided dependent care went from $5,000 to $7,500 (Pub. L. 119-21 §70404), and the §21 credit's top applicable percentage went from 35% to 50%, with a new second phase-down (Pub. L. 119-21 §70405). They cannot be stacked: §21(c) reduces the credit's $3,000/$6,000 expense base dollar for dollar by whatever you exclude. For a solo freelancer with no employees, the "Dependent Care FSA" side is effectively closed — not because a self-employed person isn't an "employee" (§129(e)(3) says they are), but because §129(d)(4) caps owners at 25% of total plan benefits, which a one-person plan fails by construction. In the worked 2026 example below, a freelancer married to a W-2 earner who elects the full $7,500 through the spouse's plan ends up $626.25 worse off than simply claiming the credit. The decision rule that falls out: elect the plan only if your combined marginal rate exceeds 0.8 × your credit percentage.

Child care is the largest deductible-looking expense in most freelance households, and it is the one the tax code handles least intuitively. There are two mechanisms, they draw on the same pool of expenses, and the one everybody has heard of — the "Dependent Care FSA" your friends with corporate jobs sign up for every November — is the one a self-employed person usually cannot use. Worse, the conventional wisdom about which is better was written against the old numbers. Both numbers changed for 2026, and they changed in opposite directions: the plan side got more generous in dollars, while the credit side got more generous in rate. Which one wins now depends on a comparison nobody was making a year ago.

What Each One Actually Is

The §21 credit: a percentage of a capped expense base

IRC §21 gives a nonrefundable credit equal to an "applicable percentage" of employment-related care expenses. As rewritten by Pub. L. 119-21 §70405(a), the statute now reads:

"For purposes of paragraph (1), the term 'applicable percentage' means 50 percent— (A) reduced (but not below 35 percent) by 1 percentage point for each $2,000 or fraction thereof by which the taxpayer's adjusted gross income for the taxable year exceeds $15,000, and (B) further reduced (but not below 20 percent) by 1 percentage point for each $2,000 ($4,000 in the case of a joint return) or fraction thereof by which the taxpayer's adjusted gross income for the taxable year exceeds $75,000 ($150,000 in the case of a joint return)." — 26 U.S.C. §21(a)(2)

The expense base it applies to is capped by §21(c) at $3,000 for one qualifying individual and $6,000 for two or more, and further capped by §21(d) at the lower of the two spouses' earned income. Married taxpayers must file jointly (§21(e)(2)), and the provider's TIN must be on the return (§21(e)(9)).

One point that trips up freelancers with a thin year: for §21 purposes, Publication 503 defines your earned income this way —

"For purposes of the child and dependent care credit, net earnings from self-employment generally means the amount from Schedule SE (Form 1040), line 3, minus any deduction for self-employment tax on Schedule 1 (Form 1040), line 15."

— and adds, flatly, "A net loss from self-employment reduces earned income."

The §129 program: an exclusion, or for the self-employed, a deduction

IRC §129 excludes employer-paid dependent care assistance from the employee's gross income, capped as amended for 2026 at:

"The amount which may be excluded under paragraph (1) for dependent care assistance with respect to dependent care services provided during a taxable year shall not exceed $7,500 ($3,750 in the case of a separate return by a married individual)." — 26 U.S.C. §129(a)(2)(A)

That exclusion is worth more than an equivalent income-tax deduction, because §3121(a)(18) also keeps the amount out of FICA wages — it excludes "any payment made, or benefit furnished, to or for the benefit of an employee if at the time of such payment or such furnishing it is reasonable to believe that the employee will be able to exclude such payment or benefit from income under section 127, 129, 134(b)(4), or 134(b)(5)."

For a self-employed participant the mechanism is different, and Publication 503 spells it out:

"If you are self-employed and receive benefits from a qualified dependent care benefit plan, you are treated as both employer and employee. Therefore, you wouldn't get an exclusion from wages. Instead, you would get a deduction on Schedule C (Form 1040), line 14; Schedule E (Form 1040), line 19 or 28; or Schedule F (Form 1040), line 15. To claim the deduction, you must use Form 2441."

A Schedule C line 14 deduction reduces net profit, so it reaches self-employment tax as well as income tax — a genuinely better shape than the W-2 exclusion, if you can get to it. Whether you can is the next section. (For what else belongs on that line, see Schedule C Line 14 — employee benefit programs.)

The Label Is Not the Test

"Dependent Care FSA" is a benefits-industry product name. It describes a §125 cafeteria-plan salary reduction feeding a §129 dependent care assistance program. Neither half of that is the legal test, and confusing the label for the test sends freelancers down two different wrong paths.

Wrong path one: "I'm self-employed, so I'm not an employee, so §129 is closed to me." That is the correct answer for a health FSA, and people generalize it. It is wrong here. §129 says the opposite in as many words:

"The term 'employee' includes, for any year, an individual who is an employee within the meaning of section 401(c)(1) (relating to self-employed individuals)." — 26 U.S.C. §129(e)(3)

"An individual who owns the entire interest in an unincorporated trade or business shall be treated as his own employer." — 26 U.S.C. §129(e)(4)

Treasury itself leaned on exactly this distinction in the August 2026 proposed regulations, when explaining why the new §128 Trump-account rules define "employee" more narrowly than §129 does:

"In contrast to section 128, however, section 129(e)(3) states the term employee includes 'an individual who is an employee within the meaning of section 401(c)(1) (relating to self-employed individuals).' Section 129(e)(3) is not among the paragraphs of section 129 that are incorporated into section 128." — Preamble to proposed regulations, 91 FR 51611 (Aug. 11, 2026)

The same preamble then says why §128 could safely drop one of §129's four nondiscrimination tests: §128 "does incorporate three of the four nondiscrimination provisions of section 129(d), omitting only section 129(d)(4), which tests owner concentration, a rule that is generally unnecessary when self-employed individuals, including owners in that capacity, are not eligible for the benefit." Read that backwards and it tells you what §129(d)(4) is for: it exists precisely because self-employed owners are eligible for §129 — and it is the rule that stops them.

Wrong path two: "So I'll set up a plan for my LLC and put in $7,500." This is where the label costs real money, because a payroll provider will happily sell a one-person business a plan document. The binding test is:

"Not more than 25 percent of the amounts paid or incurred by the employer for dependent care assistance during the year may be provided for the class of individuals who are shareholders or owners (or their spouses or dependents), each of whom (on any day of the year) owns more than 5 percent of the stock or of the capital or profits interest in the employer." — 26 U.S.C. §129(d)(4)

A sole proprietor owns the entire capital and profits interest, which is "more than 5 percent" under the §416(i)(1)(B)(i)(II) definition §414(q)(2) points to. In a plan with one participant who is also the owner, 100% of the benefits go to that class. The test fails on arithmetic, not on judgment.

And the escape hatch in §129(d)(1) does not reach the owner. Its flush text preserves a failed plan "in the case of employees who are not highly compensated employees" — and §414(q)(1)(A) makes anyone who "was a 5-percent owner at any time during the year or the preceding year" a highly compensated employee. The rescue is built for the staff, not the founder.

What the 25% rule costs, in dollars

The proposed regulations published August 11, 2026 add a correction mechanism: rather than disqualifying the plan outright, the excess can be forced back into income. Proposed Treas. Reg. §1.129-2(j)(3)(ii)(A) defines the arithmetic — "the permitted concentration amount is 25 percent of the total dependent care benefits provided by the employer to all participants during the year divided by the number of such individuals to whom benefits were provided."

node -e "
const maxExclusion = 7500;  // IRC 129(a)(2)(A), as amended by Pub. L. 119-21 sec. 70404(a)
// Solo freelancer, no employees: she is the only participant and a more-than-5-percent owner.
const totalBenefits = maxExclusion, principalOwners = 1;
const permitted = 0.25 * totalBenefits / principalOwners;  // Prop. Reg. 1.129-2(j)(3)(ii)(A)
console.log('Total plan benefits:', totalBenefits.toFixed(2));
console.log('Permitted concentration amount:', permitted.toFixed(2));
console.log('Excess ownership concentration forced back into income:', (totalBenefits - permitted).toFixed(2));
console.log('Share of the 7500 that survives:', ((permitted / maxExclusion) * 100).toFixed(2) + ' percent');
const requiredTotal = maxExclusion / 0.25;
console.log('Total plan benefits needed to keep the whole 7500:', requiredTotal.toFixed(2));
console.log('Of which must go to non-owner employees:', (requiredTotal - maxExclusion).toFixed(2));
console.log('Non-owner employees at a full 7500 each:', ((requiredTotal - maxExclusion) / maxExclusion).toFixed(2));
"

Total plan benefits: 7500.00
Permitted concentration amount: 1875.00
Excess ownership concentration forced back into income: 5625.00
Share of the 7500 that survives: 25.00 percent
Total plan benefits needed to keep the whole 7500: 30000.00
Of which must go to non-owner employees: 22500.00
Non-owner employees at a full 7500 each: 3.00

A one-person plan keeps $1,875 and reports $5,625 as income. To keep the full $7,500, the business has to be paying $30,000 of dependent care benefits in total, $22,500 of it to non-owner employees — the equivalent of three staff members maxing out the benefit alongside you. That is a real structure for an agency with a payroll; it is not a structure a solo Schedule C filer can conjure.

Two caveats on that correction, both load-bearing. These regulations are proposed, not final: their stated applicability date is "plan years beginning on or after the date final regulations are published in the Federal Register," with the preamble adding that "Taxpayers may rely on these proposed regulations for plan years beginning before the date final regulations are published." Without relying on them, the statutory consequence of failing §129(d)(4) is harsher, not gentler — the plan simply is not a dependent care assistance program as to the owner, and the whole $7,500 is income.

Side by Side

The asymmetry at the bottom of that table is the whole game. §21(c)'s closing sentence is one-directional: "The amount determined under paragraph (1) or (2) (whichever is applicable) shall be reduced by the aggregate amount excludable from gross income under section 129 for the taxable year." Nothing reduces the §129 cap by the credit. So the plan election is made first, in October, by a household that usually has not run the §21 math at all.

The Worked Example

Priya Raman is a freelance UX designer filing Schedule C as a sole proprietor with no employees. Her husband Danny is a W-2 employee earning $62,000 whose employer offers a dependent care assistance program. They file jointly for 2026, have two children in licensed daycare and after-school care costing $12,000 for the year, take the standard deduction, and have no other income. Priya's Schedule C net profit is $78,000.

Because Priya has no employees, her own business cannot run a plan — the 25% rule above. The live question is whether Danny should elect the full $7,500 through his employer's plan.

Step 1 — the self-employment layer

node -e "
const netProfit = 78000;         // Priya, Schedule C net profit
const spouseWages = 62000;       // Danny, W-2 salary
const wageBase2026 = 184500;     // SSA, 2026 OASDI contribution and benefit base

const netSeEarnings = netProfit * 0.9235;
const seTax = netSeEarnings * 0.153;               // both halves, under the wage base
const seTaxDeduction = seTax / 2;                  // Schedule 1, line 15
const priyaEarnedIncome = netProfit - seTaxDeduction; // Pub. 503: Sch SE line 3 minus Sch 1 line 15

console.log('Net earnings from self-employment (92.35%):', netSeEarnings.toFixed(2));
console.log('Combined OASDI-covered earnings (SE + spouse wages):', (netSeEarnings + spouseWages).toFixed(2), '< wage base', wageBase2026);
console.log('Self-employment tax (15.3%):', seTax.toFixed(2));
console.log('Deductible half of SE tax:', seTaxDeduction.toFixed(2));
console.log('Priya earned income for section 21:', priyaEarnedIncome.toFixed(2));
"

Net earnings from self-employment (92.35%): 72033.00
Combined OASDI-covered earnings (SE + spouse wages): 134033.00 < wage base 184500
Self-employment tax (15.3%): 11021.05
Deductible half of SE tax: 5510.52
Priya earned income for section 21: 72489.48

Both spouses clear the §21(d) and §129(b) earned income limits comfortably — $72,489.48 and $62,000 against a $6,000 credit base and a $7,500 exclusion cap — so neither limitation binds in either path below.

Step 2 — both paths, end to end

node -e "
// Shared 2026 constants, each verified from a raw primary source
const stdDeductionMFJ = 32200;   // Rev. Proc. 2025-32 sec. 4.14
const qbiThresholdMFJ = 403500;  // Rev. Proc. 2025-32 sec. 4.26
const ctcPerChild = 2200;        // Rev. Proc. 2025-32 sec. 4.05
const careExpenses = 12000;      // qualifying expenses actually paid
const kids = 2;
const dollarLimit = 6000;        // IRC 21(c)(2), two qualifying individuals

const netProfit = 78000, spouseWages = 62000;
const seTax = netProfit * 0.9235 * 0.153;
const seTaxDeduction = seTax / 2;
const qbi = netProfit - seTaxDeduction;

function tax2026MFJ(ti) {
  const b = [[0,.10],[24800,.12],[100800,.22],[211400,.24],[403550,.32],[512450,.35],[768700,.37]];
  let t = 0;
  for (let i = 0; i < b.length; i++) {
    const [start, rate] = b[i], end = i + 1 < b.length ? b[i+1][0] : Infinity;
    if (ti > start) t += (Math.min(ti, end) - start) * rate;
  }
  return t;
}
// IRC 21(a)(2), as amended by Pub. L. 119-21 sec. 70405(a)
function applicablePercentage(agi, joint) {
  const p = Math.max(50 - Math.ceil(Math.max(0, agi - 15000) / 2000), 35);
  const step2 = joint ? 4000 : 2000, start2 = joint ? 150000 : 75000;
  return Math.max(p - Math.ceil(Math.max(0, agi - start2) / step2), 20);
}

function household(dcapElection) {
  const w2BoxOne = spouseWages - dcapElection;                   // IRC 129(a)(1)
  const spouseFica = w2BoxOne * 0.0765;                          // IRC 3121(a)(18)
  const agi = netProfit + w2BoxOne - seTaxDeduction;
  const pct = applicablePercentage(agi, true);
  const reducedLimit = Math.max(0, dollarLimit - dcapElection);   // IRC 21(c), flush text
  const creditable = Math.min(careExpenses, reducedLimit, netProfit - seTaxDeduction, w2BoxOne);
  const credit = creditable * pct / 100;
  const tiBeforeQbi = agi - stdDeductionMFJ;
  const qbiDeduction = Math.min(qbi * 0.20, 0.20 * tiBeforeQbi);  // no wage/SSTB limit below threshold
  const ti = tiBeforeQbi - qbiDeduction;
  const taxBeforeCredits = tax2026MFJ(ti);
  const ctc = Math.min(kids * ctcPerChild, Math.max(0, taxBeforeCredits - credit));
  const incomeTax = taxBeforeCredits - credit - ctc;
  return { w2BoxOne, spouseFica, agi, pct, reducedLimit, creditable, credit,
           tiBeforeQbi, qbiDeduction, ti, taxBeforeCredits, ctc, incomeTax,
           total: incomeTax + seTax + spouseFica };
}

for (const e of [0, 7500]) {
  const h = household(e);
  console.log('--- Dependent care assistance program election: ' + e.toFixed(2) + ' ---');
  console.log('  Spouse W-2 box 1 wages:', h.w2BoxOne.toFixed(2));
  console.log('  Spouse employee FICA at 7.65 percent:', h.spouseFica.toFixed(2));
  console.log('  AGI:', h.agi.toFixed(2));
  console.log('  Section 21 applicable percentage:', h.pct + ' percent');
  console.log('  Section 21(c) dollar limit after the section 129 reduction:', h.reducedLimit.toFixed(2));
  console.log('  Creditable expenses:', h.creditable.toFixed(2));
  console.log('  Child and dependent care credit:', h.credit.toFixed(2));
  console.log('  Taxable income before QBI (threshold ' + qbiThresholdMFJ + '):', h.tiBeforeQbi.toFixed(2));
  console.log('  QBI deduction:', h.qbiDeduction.toFixed(2));
  console.log('  Taxable income:', h.ti.toFixed(2));
  console.log('  Income tax before credits:', h.taxBeforeCredits.toFixed(2));
  console.log('  Child tax credit applied:', h.ctc.toFixed(2));
  console.log('  Income tax after credits:', h.incomeTax.toFixed(2));
  console.log('  Self-employment tax:', seTax.toFixed(2));
  console.log('  TOTAL FEDERAL TAX:', h.total.toFixed(2));
}
console.log('Electing the full 7500 costs this household:', (household(7500).total - household(0).total).toFixed(2));
"

--- Dependent care assistance program election: 0.00 ---
  Spouse W-2 box 1 wages: 62000.00
  Spouse employee FICA at 7.65 percent: 4743.00
  AGI: 134489.48
  Section 21 applicable percentage: 35 percent
  Section 21(c) dollar limit after the section 129 reduction: 6000.00
  Creditable expenses: 6000.00
  Child and dependent care credit: 2100.00
  Taxable income before QBI (threshold 403500): 102289.48
  QBI deduction: 14497.90
  Taxable income: 87791.58
  Income tax before credits: 10038.99
  Child tax credit applied: 4400.00
  Income tax after credits: 3538.99
  Self-employment tax: 11021.05
  TOTAL FEDERAL TAX: 19303.04
--- Dependent care assistance program election: 7500.00 ---
  Spouse W-2 box 1 wages: 54500.00
  Spouse employee FICA at 7.65 percent: 4169.25
  AGI: 126989.48
  Section 21 applicable percentage: 35 percent
  Section 21(c) dollar limit after the section 129 reduction: 0.00
  Creditable expenses: 0.00
  Child and dependent care credit: 0.00
  Taxable income before QBI (threshold 403500): 94789.48
  QBI deduction: 14497.90
  Taxable income: 80291.58
  Income tax before credits: 9138.99
  Child tax credit applied: 4400.00
  Income tax after credits: 4738.99
  Self-employment tax: 11021.05
  TOTAL FEDERAL TAX: 19929.29
Electing the full 7500 costs this household: 626.25

Two things worth checking rather than assuming, both of which this example was constructed to make checkable. The election drops AGI by $7,500 but the §21 applicable percentage stays at 35% in both columns — it is already pinned to the statutory floor of the first phase-down and still short of the $150,000 joint AGI where the second one starts. And the $7,500 of taxable income removed sits entirely inside the 12% bracket in both columns, since taxable income moves from $87,791.58 to $80,291.58 and the 12% band runs from $24,800 to $100,800. No bracket is straddled, so a single marginal rate is honest here. The QBI deduction is identical in both columns at $14,497.90 — the plan election runs through Danny's wages, not Priya's business, so qualified business income never moves; it is also far below the 2026 §199A threshold of $403,500 for a joint return, so neither the SSTB phase-out nor the W-2-wage cap can bind, and the 20%-of-taxable-income cap ($20,457.90 in the first column) is not the lesser figure. The full three-limit mechanics are in the QBI deduction guide.

The Decision Rule

Because §21(c) knocks out creditable expenses dollar for dollar until the base hits zero at $6,000, while the plan cap runs to $7,500, the household is choosing between two corners, not sliding along a line. Write c for the §21 applicable percentage and r for the combined marginal rate the excluded wages would otherwise have carried. Electing the full $7,500 beats electing nothing exactly when 7500 × r > 6000 × c, which is r greater than 0.8c.

node -e "
// IRC 21(a)(2), as amended by Pub. L. 119-21 sec. 70405(a)
function applicablePercentage(agi, joint) {
  const p = Math.max(50 - Math.ceil(Math.max(0, agi - 15000) / 2000), 35);
  const step2 = joint ? 4000 : 2000, start2 = joint ? 150000 : 75000;
  return Math.max(p - Math.ceil(Math.max(0, agi - start2) / step2), 20);
}
// Electing the full 7500 beats electing nothing when 7500*r exceeds 6000*c, i.e. r exceeds 0.8c
for (const agi of [15000, 44000, 100000, 134489.48, 150000, 180000, 210000, 260000]) {
  const c = applicablePercentage(agi, true);
  console.log('Joint AGI ' + agi.toFixed(2).padStart(10) +
    '  credit rate ' + String(c).padStart(2) + ' percent' +
    '  DCAP wins only above a combined marginal rate of ' + (0.8 * c).toFixed(1) + ' percent');
}
console.log('');
console.log('Example household: 12 percent bracket plus 7.65 percent employee FICA =', (12 + 7.65).toFixed(2));
console.log('Value of electing nothing: 6000 x 0.35 =', (6000 * 0.35).toFixed(2));
console.log('Value of electing 7500:    7500 x 0.1965 =', (7500 * 0.1965).toFixed(2));
console.log('Difference:', (6000 * 0.35 - 7500 * 0.1965).toFixed(2));
"

Joint AGI   15000.00  credit rate 50 percent  DCAP wins only above a combined marginal rate of 40.0 percent
Joint AGI   44000.00  credit rate 35 percent  DCAP wins only above a combined marginal rate of 28.0 percent
Joint AGI  100000.00  credit rate 35 percent  DCAP wins only above a combined marginal rate of 28.0 percent
Joint AGI  134489.48  credit rate 35 percent  DCAP wins only above a combined marginal rate of 28.0 percent
Joint AGI  150000.00  credit rate 35 percent  DCAP wins only above a combined marginal rate of 28.0 percent
Joint AGI  180000.00  credit rate 27 percent  DCAP wins only above a combined marginal rate of 21.6 percent
Joint AGI  210000.00  credit rate 20 percent  DCAP wins only above a combined marginal rate of 16.0 percent
Joint AGI  260000.00  credit rate 20 percent  DCAP wins only above a combined marginal rate of 16.0 percent

Example household: 12 percent bracket plus 7.65 percent employee FICA = 19.65
Value of electing nothing: 6000 x 0.35 = 2100.00
Value of electing 7500:    7500 x 0.1965 = 1473.75
Difference: 626.25

The $626.25 the shortcut produces is the same $626.25 the full return computation produced above — which is the point of running both. The pattern in the table is the practical takeaway: across the entire $44,000–$150,000 joint AGI band, the credit rate is pinned at 35%, so the plan only pays if the household's combined marginal rate exceeds 28%. A 12% or 22% bracket plus 7.65% of FICA does not get there. It is only once joint AGI climbs past $150,000 and the second phase-down starts eating the credit rate that the plan pulls ahead — and by $210,000 of joint AGI the credit is down to its 20% floor, where a 24% bracket plus payroll tax wins easily.

Where Each One Wins

The §21 credit wins when:

• You are a solo freelancer with no employees. This is not a preference; §129(d)(4) settles it.
• Joint AGI sits in the broad 35%-credit band and neither spouse is in a high bracket. This covers most freelance households with one moderate W-2 income.
• Your care expenses are modest. If you pay $4,000 rather than $12,000, the $6,000 credit base is never the binding constraint, and giving any of it up for exclusion costs more than it buys.
• One spouse's earned income is low but positive. The credit base caps at $6,000 anyway, so the §21(d) limit bites later than the §129(b) limit does on a $7,500 election.

The §129 program wins when:

• Joint AGI is high enough that the applicable percentage has fallen toward 20%, and the household's combined marginal rate clears 0.8c.
• Your own business has real non-owner payroll. With at least $22,500 of dependent care benefits going to non-owner employees, the 25% cap stops binding, and the benefit arrives as a Schedule C line 14 deduction that reduces self-employment tax as well as income tax — a shape no credit can match. Budget for the §199A drag, though: lowering net profit lowers qualified business income too, so roughly 80 cents of each deducted dollar actually leaves taxable income.
• Care expenses comfortably exceed $7,500, so the full cap is usable.
• Hiring a spouse into the business is already on the table for other reasons — see hiring your spouse — since a bona fide non-owner employee changes the concentration arithmetic. A spouse who is also an owner does not: §129(d)(4) reaches "shareholders or owners (or their spouses or dependents)" by its own terms.

Common Mistakes

1. Treating "Dependent Care FSA" as the thing the law regulates. It is a product name for a §125 salary reduction feeding a §129 program. The test that decides whether a freelancer can use it is §129(d)(4)'s owner concentration cap, which no amount of plan-document shopping changes. Note also that §125(d)(1)(A) defines a cafeteria plan as one "under which all participants are employees" and contains nothing corresponding to §129(e)(3)'s express inclusion of self-employed individuals — so the "FSA" half of the label is the half that fits a freelancer worst, while the §129 half is the half that at least opens the door.
2. Assuming self-employed people are categorically shut out, the way they are from a health FSA. §129(e)(3) and §129(e)(4) say otherwise, and a freelancer with genuine non-owner payroll can use a program. Generalizing the health-FSA answer costs a real deduction for a business with employees. (The health-side rule, which genuinely is a flat no, is covered in HSA vs. FSA for the self-employed.)
3. Electing a partial amount "to get a bit of both." §21(c) reduces the creditable base from the first dollar excluded. Any election between $1 and $6,000 trades credit at rate c for exclusion at rate r on the same dollars, which is dominated by whichever corner the 0.8 rule picks.
4. Using a 2025 publication for a 2026 return. The 2025 edition of Publication 503 prints $5,000 and a 35% top rate throughout — correct for 2025, wrong for 2026. Both figures were changed by Pub. L. 119-21 effective for taxable years beginning after December 31, 2025. And because §129's cap is a fixed statutory amount with no cost-of-living adjustment, it does not appear among Rev. Proc. 2025-32's inflation-adjusted items at all, so there is no annual notice to watch for.
5. Forgetting the credit is nonrefundable again. §21(g)'s refundability applied only to taxable years "beginning after December 31, 2020, and before January 1, 2022." A thin freelance year can waste the credit entirely, and the §21(d) earned income limit can zero it out before the nonrefundability even matters.
6. Skipping the provider's TIN. §21(e)(9) denies the credit unless the care provider's identifying information is on the return. Get a Form W-10 signed while the child is still enrolled, not after you have switched providers.

How CentSense Helps

The §21-versus-§129 comparison is only as good as the expense records behind it, and both paths are reconciled on the same Form 2441:

• Scan daycare, after-school, and sitter invoices with AI receipt capture as they arrive, so the year's qualifying-expense total is a lookup in October when the plan election is due — not a shoebox reconstruction in April
• Keep a dedicated category for dependent care so the total is never tangled with household spending the credit cannot reach
• Store provider invoices with the provider's name and address attached to each receipt image, which is exactly the trail a Form W-10 request and a §21(e)(9) TIN entry depend on
• Track Schedule C net profit continuously, since both the §21(d) earned income limit and your projected AGI — the input that sets the applicable percentage — move with it all year
• Export a clean CSV of the year's care expenses for your CPA or EA alongside the rest of the Schedule C file

For the broader question of when a credit beats a deduction, see tax deductions vs. tax credits, and for how the child-related credits interact with self-employment income, see Schedule C earned income, EITC, and the child tax credit.

Start free →

Authoritative References

• 26 U.S.C. §21 — Expenses for household and dependent care services necessary for gainful employment
• 26 U.S.C. §129 — Dependent care assistance programs
• 26 U.S.C. §125 — Cafeteria plans
• 26 U.S.C. §414 — Definitions and special rules (highly compensated employee, §414(q))
• 26 U.S.C. §416 — Special rules for top-heavy plans (5-percent owner, §416(i)(1)(B))
• 26 U.S.C. §3121 — Definitions (FICA wage exclusion, §3121(a)(18))
• IRS — Publication 503, Child and Dependent Care Expenses
• IRS — About Form 2441, Child and Dependent Care Expenses
• IRS — Internal Revenue Bulletin 2025-45 (Rev. Proc. 2025-32, 2026 inflation adjustments)
• Federal Register — Employer Contributions to Trump Accounts and Nondiscrimination Rules for Dependent Care Assistance Programs (proposed, Aug. 11, 2026)
• GovInfo — Social Security Administration, Cost-of-Living Increase and Other Determinations for 2026 (2026 OASDI contribution and benefit base)

This guide is general education for U.S. freelancers and self-employed parents covering the 2026 tax year, not personalized tax or financial advice. The worked example assumes a calendar-year joint filer with no other income, no itemized deductions, no state tax effects, and no employer-side payroll considerations, and the proposed regulations cited here were still proposed as of publication. Whether a dependent care assistance program can work in your specific business — and what an election costs you against the credit — depends on your payroll, your ownership, and your projected AGI. Run the numbers with a CPA or EA before your open-enrollment deadline, because a plan election generally cannot be undone mid-year.