401(k) Loan Calculator: The Ceiling, the Payment, and the Real Cost

The ceiling has three regimes

Section 72(p) of the tax code sets the most a plan can lend. It is the lesser of two quantities, measured against every loan you hold from the plan at once:

$$L_{\max} = \min\!\Big(50{,}000 - R,\;\; \max\big(\tfrac{1}{2}V,\;10{,}000\big)\Big)$$

where:

• $L_{\max}$ is the statutory ceiling on your total plan loans, in dollars.
• $V$ is your vested balance, which the statute calls the nonforfeitable accrued benefit.
• $R$ is the reduction applied to the dollar prong, defined in the next block.

Most calculators stop at "half your balance, capped at fifty thousand". That description is wrong at both ends of the range. The $\max$ term and the $\min$ term bind in opposite directions, so the ceiling moves through three distinct regimes as $V$ rises.

Below $20,000 of vested balance, half the balance comes to less than $10,000 and the floor clause takes over, which puts the statutory figure at $10,000. That figure is not always reachable. The statute can permit more than the account actually holds, so the module takes the lesser of the statutory ceiling and the vested balance itself. Below $10,000 that second clamp is the binding one and the ceiling is the balance, which is why the low end of the signature chart runs at 45 degrees rather than flat. Only between $10,000 and $20,000 does the ceiling sit flat at $10,000 and stop moving with the account.

Between $20,000 and $100,000, neither clamp binds. The ceiling is exactly half the vested balance, and every extra dollar saved buys fifty cents of borrowing capacity.

Above $100,000, the dollar cap binds. The ceiling stops at $50,000 and stays there. A $400,000 vested balance and a $100,000 vested balance carry the same ceiling.

The signature chart draws that function and marks where your own vested balance sits. The two kinks are the whole story.

The floor clause is permissive, not required

The statute lets a plan lend up to $10,000 where half the vested balance falls short. It does not force the plan to. Many plan documents skip the clause and cap every loan at half the vested balance regardless of size. The module models the statutory maximum, which is a ceiling on what a plan may offer rather than a promise of what yours does.

The base is vested, and that gap is invisible on a statement

Your own salary deferrals vest the moment they land. Employer match money often does not, and unvested dollars stay out of the base until the schedule clears.

Inside the middle band the arithmetic is direct: each unvested dollar removes fifty cents of capacity. A $70,000 statement carrying $12,000 of unvested match is a $58,000 base, so the ceiling is $29,000 rather than $35,000. The module takes both figures and shows the gap, because the number on the statement is the one people plan around.

The dollar prong remembers the last twelve months

The $50,000 side of the test is reduced before the comparison runs:

$$R = \max\!\big(0,\; H - O\big)$$

where:

• $H$ is the highest total balance you owed the plan on any day in the prior twelve months.
• $O$ is what you owe the plan on the day the new loan is made.

A loan repaid in November, followed by a new one in December, still carries the November peak against the cap. With the earlier loan fully repaid, $O$ is zero and the whole peak comes off the dollar prong.

Two effects get conflated here. Any balance you still carry counts against both prongs, because the ceiling applies to your loans in aggregate. The twelve-month lookback is a separate penalty, and it lands only on the dollar prong. Raising $R$ moves the second kink in the chart, since the cap regime now begins at $2(50{,}000 - R)$ instead of at $100,000.

The repayment schedule

The statute requires substantially level amortization, payments no less frequently than quarterly, and repayment within five years. A loan used to acquire your principal residence is exempt from the five-year limit, and the plan document sets the term instead.

Level amortization is the standard annuity payment:

$$\text{payment} = \frac{P\,i}{1 - (1+i)^{-N}}$$

where:

• $P$ is the amount borrowed, in dollars.
• $i$ is the periodic interest rate, taken as the annual loan rate divided by the number of payments per year.
• $N$ is the number of payments, equal to the term in years times the payment frequency.

Total interest is $N \times \text{payment} - P$. On $25,000 at 8.5% over five years, paid monthly, the payment is $512.91, sixty of them come to $30,774.80, and the interest totals $5,774.80.

What bi-weekly actually changes

The module runs bi-weekly as twenty-six payments a year at a periodic rate of $i = r/26$, over the same term. Two things follow. The payment falls to $236.37 on the loan above, and total interest falls to $5,727.91.

That saving is $47 over five years, which is roughly nothing. Bi-weekly reduces principal more often, so slightly less balance sits outstanding earning interest, and that is the entire effect.

This is not the accelerated schedule sold under the same name for mortgages. That version pays half the monthly figure every two weeks, which delivers thirteen monthly payments a year and retires the loan early. A plan loan on payroll deduction is the first kind, not the second.

The interest is not the cost

Every competing calculator reports the payment and the total interest, then treats the interest as the price of the loan. It is not. The interest goes into your own account. A bank charges you interest and keeps it; a plan charges you interest and credits it back to you.

The real cost is that the borrowed principal leaves the market while the loan runs. To size it, the module follows two paths over the term and compares where they land.

Do not borrow. The amount stays invested at the plan return throughout:

$$V_{\text{keep}} = P\,(1+\rho)^{N}$$

where:

• $\rho$ is the periodic plan return, converted geometrically from the annual figure: $\rho = (1 + r)^{1/f} - 1$, where $f$ is the number of payments a year.
• $P$ and $N$ carry the meanings given above.

The two rates are converted differently, and on purpose. A loan APR is a nominal quote, and the recordkeeper amortises it at $i = r_{\text{loan}}/f$, so that side is a plain division. An expected market return is an effective annual figure, and dividing it by twelve would quietly claim more than a year's growth over a year. So $\rho$ takes the twelfth root instead. At an annual plan return of 7% with monthly payments, $\rho$ is 0.565415% a period rather than 0.583333%. The gap looks trivial and is not: it is what moves the crossover in the next section.

Borrow. The plan swaps a market investment for a loan to you. It holds a declining note that credits the loan rate, and it receives each repayment as it arrives, which is then invested at the plan return. The note reaches zero at the end of the term, so only the repayment stream survives:

$$V_{\text{borrow}} = \text{payment} \times \frac{(1+\rho)^{N} - 1}{\rho}$$

where:

• the fraction is the future value of a one-dollar payment made every period, accumulated at $\rho$.

The cost is the difference, $K = V_{\text{keep}} - V_{\text{borrow}}$. The rest of the account is invested identically on both paths, so it cancels exactly and never enters the arithmetic.

The second chart draws both paths across the term. The gap at the right-hand edge is $K$.

The crossover, which is exact

Write the payment as $P / a_N$, where $a_N = \big(1 - (1+i)^{-N}\big)/i$ is the present value of a one-dollar annuity at the loan rate. Write the accumulation factor as $s_N = \big((1+\rho)^{N} - 1\big)/\rho$. Then:

$$K = P\left[(1+\rho)^{N} - \frac{s_N}{a_N}\right]$$

where:

• $K$ is the cost of borrowing, in dollars at the end of the term, positive when you end behind.

Now set the two periodic rates equal, $i = \rho$. A standard annuity identity gives $s_N = a_N (1+\rho)^{N}$ when they coincide, the bracket collapses, and $K$ is exactly zero. Not approximately, and not for reasonable inputs: zero, at every principal, every term and every level of rates.

The condition is on the periodic rates, and that is not the same as the quoted ones being equal. The loan runs at $i = r_{\text{loan}}/f$ and the plan at $\rho = (1 + r_{\text{plan}})^{1/f} - 1$, so solving $i = \rho$ puts the crossing at a loan APR of:

$$r_{\text{loan}}^{*} = f\left[(1 + r_{\text{plan}})^{1/f} - 1\right]$$

which sits a little below the quoted plan return rather than level with it. Against an 8.5% plan return with monthly payments the crossing is at 8.185792%, about 31 basis points low. Inside that band the loan rate is beneath the quoted plan return and the account still ends ahead. Read from the other side the crossing is the same point, $r_{\text{plan}}^{*} = (1 + r_{\text{loan}}/f)^{f} - 1$, and the module prints both figures so the two can be checked against each other.

That crossing is the honest result almost nobody publishes. The sign of the cost is the sign of the spread between the two periodic rates, and nothing else.

The mechanism is clearer in a second exact form. Roll the two paths forward one period at a time and the difference telescopes:

$$K = \sum_{k=1}^{N} (\rho - i)\,L_{k-1}\,(1+\rho)^{\,N-k}$$

where:

• $L_{k-1}$ is the loan balance outstanding at the start of period $k$.
• $(\rho - i)$ is the periodic spread between the plan return and the loan rate.

Each period you give up the spread on the balance still outstanding, never the whole return on it. Every $L_{k-1}$ is positive, so the sign of $K$ follows the sign of $\rho - i$ with no exceptions. On the $25,000 loan above, the average outstanding balance across the five years is 54% of the principal, which is the base the spread applies to.

Numbers make the size of it plain. That loan at 8.5% costs $993.48 if the money would have compounded at 10%. At a 7% plan return it leaves the account $1,453.51 ahead, because the note credited 8.5% while the market delivered 7%. And at a plan return of exactly 8.5% the two paths do not land on the same dollar: the account ends $278.91 ahead, because an 8.5% APR clears the 8.185792% crossing that an 8.5% plan return implies.

The chart annotates the crossover so you can read which side of it your assumptions fall on. Whether a plan return above the loan rate is the right assumption is a separate question, and this module does not answer it.

The interest is taxed a second time

The most repeated claim about plan loans is that you get taxed twice. It is half right, and the half matters.

You received $P$ untaxed, because a loan meeting the 72(p) conditions is not a distribution. You return $P$ out of pay that has already been taxed. Those two facts cancel. The principal is taxed once, at withdrawal, exactly as it would have been had you never borrowed.

Interest is different. It is new money, not principal coming home. You earn it, pay income tax on it, deposit it into a pre-tax account, and pay income tax on it again when it comes out:

$$T_{\text{interest}} = t_w \times I, \qquad I = N \times \text{payment} - P$$

where:

• $T_{\text{interest}}$ is the second layer of tax, in dollars measured at the time of repayment.
• $t_w$ is your marginal tax rate at withdrawal, written as a decimal.
• $I$ is total loan interest, which is what the second layer applies to.

Those dollars compound before they are withdrawn, so the tax charged later is $t_w I (1+\rho)^{k}$. Discounted back at the same plan return, it is $t_w I$ regardless of how long the wait runs. On $5,775 of interest at a 22% withdrawal rate, the second layer is about $1,270.

Note the scale. It is a fraction of the interest, not a fraction of the loan. Articles that stretch this into "you are taxed twice on the borrowed money" overstate it by roughly the ratio of principal to interest, which on a five-year loan is around four to one.

One thing the model leaves out of this figure. Those interest dollars now grow tax-deferred rather than in a taxable account, which carries its own tax and is not netted against the second layer here.

The separation trap

Payroll deduction needs a payroll. When employment ends, plans commonly call the balance due, and an unpaid balance is subtracted from your account as a plan loan offset.

The IRS treats an offset as an actual distribution rather than a bookkeeping entry. It is ordinary income in the year it happens, reported on Form 1099-R, plus the 10% additional tax if you are under 59½ and no exception applies:

$$T_{\text{offset}} = O_{\text{sep}} \times \big(t_w + p\big)$$

where:

• $O_{\text{sep}}$ is the balance outstanding on the day of separation, in dollars.
• $t_w$ is the marginal rate that applies to the offset as ordinary income.
• $p$ is 0.10 before age 59½ and zero otherwise.

Separating from service during or after the year you turn 55 is one of the exceptions, so $p$ drops to zero there. The module reads your age against that threshold rather than assuming the penalty.

A separation two years into the $25,000 loan leaves $16,248 still outstanding. At a 22% rate with the additional tax that offset costs $5,199, and the $16,248 leaves the account for good.

The tax is the smaller half of the damage. The balance itself is gone from the shelter, and its future compounding with it.

The offset can be undone, and the clock is specific

An offset caused by severance from employment, or by the plan terminating, is a qualified plan loan offset. It gets a longer rollover window than the usual sixty days. An equivalent amount can go into an IRA or another employer plan up to the due date of that year's tax return, including extensions, which reaches October 15 for a timely extended filing.

Two conditions attach. The offset has to occur within twelve months of the severance date, and the loan has to have satisfied 72(p)(2) immediately before the triggering event. A loan that already defaulted while you were employed is a deemed distribution instead, and it cannot be rolled over at all.

The practical catch is arithmetic rather than legal. The rollover money has to come from outside the plan, at the moment the paycheck stopped.

What this model does not do

Being explicit about this matters more than the cost figure itself.

The forgone employer match is absent, and it is usually the largest omission. Loan repayments are not plan contributions. They earn no match and they do not count toward the elective deferral limit. Where the payment crowds out a deferral that was being matched, the lost match commonly runs several times the market spread computed here. The module assumes your contribution rate is unchanged while the loan runs.

Nothing about what the money bought is modelled. The comparison covers the plan side only. If the loan retired credit card debt at 22%, that saving sits entirely outside this arithmetic and is not small. The cost figure answers what the account gave up, not whether the borrowing achieved anything.

A constant plan return is not reality. Markets deliver a sequence, not an average, and the sequence matters here because the amount exposed to it changes every period. The crossover result survives this better than the dollar figures do, since it depends on the sign of a spread rather than on its level.

Plan rules are approximated by statutory maximums. Origination and annual maintenance fees, minimum loan sizes, limits on the number of loans outstanding, spousal consent requirements, and rules suspending contributions during repayment all vary by plan and none are modelled. The loan rate is set by the plan, often at prime plus one point, and is an input here rather than a derived figure.

The cure period is not modelled. A missed payment is not instantly fatal. A plan may allow a cure running to the end of the calendar quarter following the quarter of the missed payment. Past that, the outstanding balance becomes a deemed distribution.

Timing inside the period is simplified. Each repayment is credited at the end of its period and invested immediately at the plan return. Real payroll timing and settlement lags move the answer by a few dollars.

Inflation is absent. Every figure is nominal. A cost of $993 five years out is not $993 of today's purchasing power.

Frequently asked

Why does my plan offer less than this ceiling?

Because the ceiling is the maximum a plan can permit, not the amount it has to. Plan documents routinely stop short of the statutory line, decline the $10,000 floor clause, or allow only one loan at a time. The loan agreement is where the operative number lives.

Do bi-weekly payments save money?

Marginally, and much less than the query volume suggests. Twenty-six payments a year at the same nominal rate and the same term reduce principal more often, which on a five-year $25,000 loan at 8.5% saves about $47 of interest in total. The large savings people associate with bi-weekly mortgages come from making thirteen monthly payments a year and finishing early, which is a different schedule.

Why is the cost sometimes negative?

Because the plan is holding your loan as an asset. When the loan rate exceeds what the market delivered over the same window, the account earned more on the note than it would have earned in the market. Over a five-year stretch where returns are poor, the borrower ends ahead, which is the opposite of what the folklore predicts. The chart shows the crossover explicitly rather than assuming the loan always costs something.

Is a 401(k) loan really taxed twice?

Only the interest is, and the module sizes that separately. The principal is untaxed on the way out and repaid with taxed money, so the two cancel and it is taxed once at withdrawal. Stretching the claim to cover the whole loan overstates the effect several times over.

What loan rate should I enter?

That is a plan fact rather than a modelling choice. Plans commonly set it at the prime rate plus one percentage point, fixed for the life of the loan, and your loan agreement states the figure. It is an input here because the statute does not set it.

Should I take the loan?

That is not a question this tool answers, and nothing here is advice. It computes the ceiling, the schedule, the plan-side cost against a stated return assumption, the second layer of tax on the interest, and the offset exposure at your age. What weight to place on those figures, and on the rules the model omits, is a matter for you and for a licensed professional.

Sources and further reading

• Internal Revenue Service, Retirement plans FAQs regarding loans. The plain-language statement of the ceiling, the twelve-month lookback, and the repayment requirements, with worked examples.
• U.S. Government Publishing Office, 26 U.S.C. § 72(p), loans treated as distributions. The statute itself, including the dollar limitation, the five-year term, and the level amortization requirement.
• Internal Revenue Service, Plan loan offsets. What separates an offset from a deemed distribution, and the extended rollover window for a qualified plan loan offset.
• Internal Revenue Service, Exceptions to tax on early distributions. The list the 10% additional tax is tested against, including separation from service at 55.
• The companion article, how much can you borrow from your 401(k), covers the rules narratively and compares a loan against a hardship distribution.

Educational research only. Not investment advice, not tax advice, and not a projection of your individual outcome.

Methodological sources

• Internal Revenue Service (2026). Retirement plans FAQs regarding loans.
• Office of the Law Revision Counsel, U.S. House of Representatives (2023). 26 U.S.C. § 72(p) — Loans treated as distributions.
• Office of the Law Revision Counsel, U.S. House of Representatives (2023). 26 U.S.C. § 402(c)(3)(C) — Rollover of certain plan loan offset amounts.
• Employee Benefits Security Administration, U.S. Department of Labor (2024). 29 CFR § 2550.408b-1 — General statutory exemption for loans to plan participants and beneficiaries who are parties in interest with respect to the plan.
• Internal Revenue Service (2026). Topic no. 558, Additional tax on early distributions from retirement plans other than IRAs.
• Internal Revenue Service (2025). Publication 575, Pension and Annuity Income.
• Timothy (Jun) Lu, Olivia S. Mitchell, Stephen P. Utkus, Jean A. Young (2017). Borrowing from the Future? 401(k) Plan Loans and Loan Defaults.

Read the note behind this module

• How Much Can You Borrow From Your 401(k)? The Limit and the Repayment Trap
• Can I Use My 401(k) to Buy a House? The Loan, the Withdrawal, and What Each Costs
• Hardship Withdrawal From a 401(k): What Qualifies, and What It Permanently Costs

Educational purposes only. Not investment advice.