Most tax-loss harvesting writing assumes the loss is idiosyncratic: one stock fell, the market did not, and the harvest captures the difference. Fixed income inverts that assumption. When prevailing interest rates rise, nearly every bond and bond fund of comparable duration reprices downward at once — the losses are correlated, broad, and arrive on a schedule set by the rate environment rather than by any single issuer's fortunes. That changes which positions carry a harvestable loss, how a replacement is chosen, and which sections of the Internal Revenue Code govern the result. Two of those sections — IRC §852(b)(4) and IRC §1276 — have no equivalent in an equity harvest at all.
Why do bond funds produce harvest opportunities that equity positions often do not?
Why does a rising interest-rate environment create tax-loss harvesting opportunities across an entire bond sleeve at once, when equity losses are usually position-specific?
A bond's price moves inversely to prevailing yields, so a general rate increase pushes the market value of most existing fixed-rate bonds and bond funds below their purchase price simultaneously — meaning a fixed-income sleeve may present many harvestable positions at the same moment, whereas an equity sleeve typically presents scattered individual losers against a rising index.
The mechanism is arithmetic rather than sentiment. A bond promises a fixed schedule of coupon payments and a return of principal at maturity. When newly issued bonds of comparable credit quality and maturity begin paying a higher coupon, the older, lower-coupon bond becomes less attractive, and its market price falls until its yield to maturity is competitive with the new issue. Duration measures the sensitivity of that price to a change in yield: a fund with a longer average duration may reprice more sharply for the same move in rates than a short-duration fund holding otherwise comparable credits.
Two consequences follow for a harvesting program. First, the unrealized loss in a bond fund is often not a signal that the underlying credits deteriorated — the bonds may still be expected to pay in full, and the fund's share price may recover as its holdings roll toward maturity. Second, because the decline is broad, the same rate move that puts one bond position underwater tends to put comparable positions underwater too, which makes finding a replacement that preserves exposure straightforward in one sense and legally delicate in another. The asset location article covers the prior question of whether taxable bond exposure belongs in the taxable account at all, since ordinary interest income is a recurring annual tax event there.
It is worth separating a fund from an individual bond. An investor holding an individual bond to maturity receives par regardless of interim price moves, so the interim unrealized loss never has to be recognized — selling to harvest it is a deliberate choice to abandon that hold-to-maturity path. A bond fund has no maturity date; its share price is a continuously marked pool, so the position is marked whether or not the investor acts.
Does the wash-sale rule apply the same way to bond funds?
Does IRC §1091's wash-sale rule cover bonds and bond funds, or is it limited to stocks?
IRC §1091 applies to losses on the sale of "stock or securities," which includes bonds, bond mutual funds, and bond ETFs — so a loss sale followed by acquiring a substantially identical security within the 61-day window running from 30 days before the sale to 30 days after it is disallowed in the same way an equity wash sale would be.
The 61-day window and the basis-adjustment mechanics carry over without modification. A disallowed loss is not lost outright in a taxable account: under IRC §1091(d), the disallowed amount is added to the basis of the replacement security, and the replacement inherits the original position's holding period under IRC §1223(3). The economic effect is deferral rather than forfeiture — with the well-documented exception that a replacement purchase inside an IRA produces a permanent loss under Rev. Rul. 2008-5, because the IRA receives no basis increase. The wash-sale rule explainer walks the general mechanics; the DRIP article covers the reinvestment path that most often triggers an accidental one.
Bond funds make the reinvestment path unusually easy to trip. Bond funds commonly distribute interest income monthly rather than quarterly, and automatic reinvestment of a monthly distribution is a purchase of fund shares. An investor who harvests a loss in a bond fund on the 10th of a month and leaves automatic reinvestment switched on may acquire replacement shares of the identical fund before the month ends — squarely inside the 30-day forward window, and for a security whose identity with the sold position is not arguable. Because the reinvested amount is usually small relative to the position, the disallowance is typically partial: only the loss attributable to the number of shares repurchased is deferred.
HarvestEngine quantifies that partial split rather than treating any overlap as an all-or-nothing
block. compute_wash_impact() in src/tlh/wash_engine/impact.py returns the
allowed and disallowed portions of a proposed loss sale together with the per-replacement-lot
basis carryover, using same-symbol lots acquired within the surrounding window in the same account.
After a loss sale executes, auto_create_wash_cooldowns records a 30-day cooldown on
that symbol — created only for actual loss sales, since §1091 does not reach gains.
What makes two bond funds substantially identical, and what makes them merely similar?
When are two different bond funds treated as substantially identical for wash-sale purposes, and what distinguishes a defensible replacement from a risky one?
The IRS has issued no bond-fund-specific safe harbor, so the question is answered on facts and circumstances: two funds tracking the same index with the same issuer, maturity, and credit composition are the strongest candidates for identity, while funds tracking different indexes with materially different portfolios have generally been treated by practitioners as not substantially identical — with the analysis growing weaker as the two portfolios converge.
The governing authority is thin, and honestly stated it is thinner for bonds than for equities. Rev. Rul. 59-44 addressed bonds of the same issuer differing in maturity and interest rate and found them not substantially identical, which is the most commonly cited starting point. But that ruling concerned individual bonds of one issuer, not diversified funds, and the Service has never extended it to fund shares by published guidance. Anyone reading a bright line into it is reading in more than the ruling says.
What practitioners generally examine instead is the degree of portfolio overlap. Two funds that both track the identical benchmark — same index provider, same index name, same rebalancing rules — hold nearly the same bonds in nearly the same weights, and the case that they are economically distinct is weak regardless of which sponsor issues them. Two funds tracking different benchmarks may differ in average duration, credit-quality distribution, sector mix (Treasury versus corporate versus securitized), and issuer concentration, and each of those differences is a fact supporting distinctness. The substantially identical deep dive works through the same analysis for equity index funds, where the fact pattern is better developed and the reasoning transfers.
The practical tension is that the differences which make a replacement legally defensible are the same differences that make it an imperfect economic substitute. A Treasury fund swapped for a corporate fund is plainly a different portfolio — and plainly a different credit exposure. A short-duration fund swapped for a long-duration fund is plainly distinct — and plainly a different rate sensitivity. Choosing a replacement is therefore a two-sided decision, not a search for the closest available match.
How do municipal bond funds change the harvest calculation?
If municipal bond interest is tax-exempt, can a loss on a municipal bond fund still be harvested — and are there rules that apply only to muni funds?
Yes: tax-exempt interest and capital loss are separate items, so a loss on the sale of municipal bond fund shares is generally a deductible capital loss even though the fund's income was exempt — but IRC §852(b)(4)(B) disallows that loss to the extent of exempt-interest dividends received when the shares were held for six months or less, a holding-period rule with no equivalent in an equity harvest.
The first point is often misunderstood. The exemption in IRC §103 covers interest on qualifying state and local obligations; it does not exempt gain or loss on the disposition of the obligation or of fund shares holding it. A municipal bond fund purchased at one price and sold at a lower price generates an ordinary capital loss subject to the same netting and carryforward rules as any other capital loss. The capital loss carryforward article covers how that loss applies against gains and, subject to the annual ordinary-income limit, against other income.
The second point is the trap. IRC §852(b)(4)(B) provides that where a shareholder of a regulated investment company has held shares for six months or less and has received exempt-interest dividends on those shares, any loss on their sale or exchange is disallowed to the extent of the exempt-interest dividends received. A muni fund position bought recently, held through several monthly distributions, and sold at a loss inside the six-month window may therefore find part of that loss eliminated — not deferred into a replacement's basis, as a wash sale would be, but disallowed under a separate provision. Its companion, IRC §852(b)(4)(A), recharacterizes a loss on RIC shares held six months or less as long-term to the extent of any long-term capital gain distributions received on them, which can matter for loss character even outside the muni context.
Holders of individual municipal bonds face a further wrinkle from the market discount rules. Under IRC §1276, accrued market discount on a bond acquired below its revised issue price is generally treated as ordinary income on disposition rather than capital gain, and IRC §1278(a)(2)(C) supplies a de minimis threshold below which discount is treated as zero. A position bought at a discount after rates rose can therefore carry an embedded ordinary-income component that a purely capital-gain framing would miss. This is a bond-specific interaction; the lot selection article covers the parallel question of which lots a sale draws from, which determines the basis the loss is measured against.
What does a bond-fund harvest cost in tracking and portfolio terms?
What are the portfolio-level costs of harvesting a bond fund loss, and how are they weighed against the tax benefit?
The costs are a change in duration or credit exposure for as long as the replacement is held, a bid-ask and possibly a premium-or-discount cost on the round trip, a reset of the holding period toward short-term unless the replacement inherits it under a wash sale, and a 30-day constraint on returning to the original fund — each of which is measurable and none of which is automatically smaller than the tax benefit.
Duration drift is the dominant one. If the replacement fund's average duration differs from the original's, the sleeve's sensitivity to further rate moves changes for the holding period of the replacement. An investor who harvests a long-duration fund into a short-duration fund has reduced the sleeve's rate exposure — which may look prescient if rates continue rising and costly if they fall back. Credit drift is the same argument in a different dimension: a Treasury-for-corporate swap alters the sleeve's exposure to credit spreads, not just to the level of rates. Neither drift is inherently wrong; both are positions being taken, and describing them as neutral would be inaccurate.
Transaction cost is more visible in fixed income than in large-cap equity. Bond ETFs can trade at a premium or discount to net asset value, particularly in less liquid segments, and the underlying bonds themselves carry wider dealer spreads than listed equities. A round trip out of one fund and into another pays that cost twice. For a small unrealized loss, the round-trip cost and the deferred nature of the benefit can leave little margin — which is a reason many programs apply a minimum-loss threshold rather than harvesting every position that shows red.
In HarvestEngine's portfolio decomposition, broad bond ETFs are classified into the passive
beta sleeve rather than the direct-index long sleeve: is_index_etf()
in src/tlh/portfolio/decomposition.py matches a hardcoded broad-market list that
includes core bond funds such as BND, AGG, BIV, BSV and VCIT. That classification is a statement
about where the harvest surface lives — the granular, per-security loss surface described in the
TLH vs ETF rebalancing article belongs to the
individual-equity sleeve, while a bond ETF presents one position and therefore one loss
opportunity at a time.
Read this next with the wash-sale rule explainer, substantially identical: a deep dive, asset location, capital loss carryforwards, and DRIP wash-sale traps.