Prepay or Invest? Compare the Right Two Things
Prepaying a mortgage is not an investment decision against the stock market. It is the retirement of a bond you issued (section “The Mortgage as a Negative Bond”), and its return is exactly the after-tax mortgage rate, earned risk-free. So compare it against a risk-matched alternative, not against equities:
where is the yield on a bond or cash instrument of comparable duration and its tax rate. Two consequences follow immediately, both cutting against conventional wisdom.
First, is probably smaller than you think, which makes prepayment more attractive than the traditional advice assumes. If you take the standard deduction, the mortgage interest deduction is worth zero and the after-tax cost of the loan is the full note rate. If you itemize, the 2/37 rule caps the benefit near 35 cents on the dollar (section “Tax Deductions of The Mortgage Interest”), and interest on acquisition debt above $750,000 generates no deduction at all — so on a $1.5 million loan, half your interest is non-deductible regardless. A 6.5% mortgage at costs 5.36% after tax, risk-free. Very few fixed-income alternatives clear that.
Second, the comparison is not against your whole portfolio’s expected return, because prepayment is certain and equity returns are not. Holding a bond allocation while carrying a higher-rate mortgage is a guaranteed negative spread; that is the negative-bond argument, and prepayment is how you close it. But this only reaches the money left after the waterfall above, because the tax-advantaged rungs beat both sides of this comparison — which is precisely the finding of the Amromin, Huang, and Sialm work quoted in section “Surviving Job Loss: How Strategic Financial Planning Can Save Your Home”: a large minority of households accelerating mortgage payments instead of funding tax-deferred accounts are simply making the wrong trade, at a cost of 11 to 17 cents on the dollar.
If you do decide to prepay, decide first what you want from it. Prepaying alone shortens the term and leaves the payment unchanged; prepaying and then recasting lowers the payment and leaves the term (section “Recasting: The Refinance Nobody Offers You”). If your motive is resilience against an income shock, neither one helps as much as the same money left liquid — the servicer wants November’s payment in November regardless of how far ahead you are.
Paying it off in retirement, and where the money comes from. Once the paycheck stops, the funding source carries its own tax and the question changes shape. The framing that gets this wrong compares the tax on the withdrawal against the interest avoided: pull $141,000 from a pre-tax IRA to net the $111,000 you still owe, incur $30,000 of tax, observe that only $20,000 of interest remains on a loan in its final years, and conclude the trade is absurd. That is the wrong comparison. The $30,000 is not a cost of retiring the mortgage — it is the cost of withdrawing, and those dollars are taxable whenever they come out. The genuine incremental cost is the acceleration: the spread between your marginal rate this year and the rate the same dollars would have met spread over later years, plus any cliff the lump crosses — an IRMAA tier on its two-year lookback (section “Cliff Choreography”), a larger share of Social Security dragged into the tax base (section “The Social Security Tax Torpedo”), the NIIT threshold — plus the sheltered compounding you surrendered. Often still a poor trade, but for the right reason and by a far smaller margin than the arithmetic above suggests.
Which makes the funding order the real decision. Taxable dollars first, choosing high-basis lots so the realized gain is small. Pre-tax only once taxable is exhausted — and never in a single year if the balance is large, since splitting a payoff across two or three calendar years keeps each slice inside a bracket and under the cliffs. Roth never, for this purpose: you would be spending the best-taxed asset you will ever own to retire the cheapest debt you will ever hold.
Two retirement-specific effects the accumulation-phase formula does not capture. The first argues against payoff: the low-bracket years between retirement and RMDs are finite and valuable, and a pre-tax withdrawal to retire a mortgage consumes exactly the same cheap bracket space as a Roth conversion (section “The Conversion Window”). The conversion buys permanently tax-free growth; the payoff buys a bond-equivalent yield equal to your after-tax mortgage rate. At ordinary mortgage rates the conversion is the better use of scarce bracket room, so retire the loan with taxable money or leave it standing. The second argues for payoff, and it is the one spreadsheets miss: killing the payment lowers your required withdrawal in every remaining year, and therefore your MAGI in every remaining year. A household that drops permanently below an IRMAA tier, or out of the torpedo zone, has bought something no interest calculation shows. The same elimination removes a fixed obligation that would otherwise force selling into a down market, which is the sequence-risk version of the identical point.