Rebalancing Cost: The Tax and Transaction Drag of Staying on Target

You rebalance on a Tuesday in March. Your 60/40 portfolio has drifted to 64/36 after a good year in stocks, so you sell $20,000 of a broad US index fund inside a $500,000 taxable brokerage account and buy bonds with the proceeds. The confirmation shows a commission of $0.00, and the whole thing takes ninety seconds.

That trade cost you roughly $863. The commission really was zero, and the bid-ask spread took about $6. The other $857 is capital gains tax, and it doesn’t appear on the confirmation at all — it shows up in April of the following year, by which point almost nobody connects it back to the Tuesday.

That $863 is 17.3 basis points of the whole portfolio. Here is the arithmetic. The shares you sold were up 40% from what you paid, so your cost basis was $20,000 ÷ 1.40 = $14,286 and the realised gain was $5,714. At a 15% long-term rate, tax is $857. Spread on a liquid ETF ran about 3 basis points of the amount traded, which is $6. Total $863, divided by $500,000, is 0.173%.

Nearly everything written about rebalancing cost still leads with commissions, which US brokers cut to zero years ago. The number that decides your actual cost is different: your embedded gain, times your capital gains rate, times how much you turn over. Trading friction is a rounding error sitting next to it.

Where the money actually goes

Four things cost you money, and they are nowhere near equal.

Capital gains tax dominates. You only pay it in a taxable account, and only on positions sold at a gain. IRS Topic 409 sets out the structure: hold more than a year and the gain is taxed at 0%, 15%, or 20% depending on your taxable income; hold a year or less and it’s taxed at your ordinary rate. Nothing you do at the broker changes that.

The Net Investment Income Tax adds 3.8% on top. It applies to the lesser of your net investment income or the amount your modified AGI exceeds a threshold — $250,000 married filing jointly, $200,000 single or head of household, $125,000 married filing separately, per IRS Topic 559. Those thresholds were never indexed to inflation, so more households cross them every year without their real income changing. If it applies to you, 15% becomes 18.8% and 20% becomes 23.8%.

The bid-ask spread is the cost of crossing the market. You sell at the bid and buy at the ask, and on a large-cap US equity ETF that gap is often a penny on a share priced in the hundreds. Half a basis point to two basis points each way. Mutual funds transact at NAV with no spread you can see, but the fund’s own trading still comes out of performance.

Commissions and fund-level fees are mostly gone for stocks and ETFs, and mostly not gone for mutual funds. Redemption fees, exchange fees, and purchase fees sit outside the expense ratio entirely, which is why they surprise people; the SEC’s Investor.gov page on mutual fund fees and expenses lists each of them separately from the annual operating cost. A frequent rebalancer in a fund with a 2% redemption fee on shares held under 90 days is playing a different game from the ETF holder above.

The assumptions behind every number here

Change any one of these and the answer moves, so they’re worth stating flatly. The account is taxable — inside an IRA or 401(k) the tax column is zero and the page is nearly pointless. The holding period is over twelve months, so long-term rates apply. Round-trip trading cost is 3 basis points of the amount traded, which suits a liquid broad-market ETF and badly understates a thinly traded one. Commissions are zero. State income tax is excluded, and many states tax gains as ordinary income. Cost basis is averaged across the lot sold; picking specific lots would lower the bill. Wash sale rules never come up, because every sale here is at a gain.

Every dollar figure below is an illustration I calculated. The rate structure comes from the IRS pages linked above. The arithmetic is mine, and you should redo it with your own numbers.

One more thing worth knowing before the tables: the cost is proportional, not fixed. Run the same $500,000 example at $50,000 or at $5,000,000 and the basis-point figure stays at 17.3 in every case, because both the tax and the spread scale with the amount traded. A $50,000 portfolio and a $5,000,000 portfolio pay the same percentage for the same behaviour. The exception is at the very bottom. On $1,000 traded, a $10 flat fee is 100 basis points of the trade, and odd-lot ETF fills carry wider effective spreads, so below roughly $25,000 the trading line stops being noise.

What changes the number: the gain and the rate

Two variables move the answer, and you have partial control over both. Every row below assumes you sell 4% of the portfolio, the gain is long-term, and trading costs 3 basis points of the amount traded. The figure shown is total cost in basis points of the whole portfolio.

Check the 100%-gain row at 23.8% yourself, because the logic generalises. Shares that doubled mean half the sale proceeds are gain. Sell 4% of the portfolio and you realise 2% of the portfolio as gain. Tax at 23.8% is 0.476% of the portfolio — 47.6 basis points — plus 0.1 for trading. That’s the 47.7.

Now compare the corners. Someone in the 0% long-term bracket rebalancing a recently funded account pays 0.1 basis points. Someone at 23.8% selling shares that tripled pays 63.5. Identical action, 635 times the cost, and the difference is entirely tax posture.

For scale, 63.5 basis points a year exceeds the expense ratio of most retail index funds by a wide margin. The SEC page linked above illustrates over ten years how a small annual fee difference compounds into thousands of dollars of foregone value. Tax drag compounds through the same mechanism, because a dollar handed to the Treasury in year one is not there to grow in years two through thirty.

The twelve-month line

This is the largest avoidable multiplier on the page, and it turns on a calendar date rather than anything about markets.

Same trade as before: $500,000 portfolio, $20,000 sold, $5,714 of realised gain. Sell at month eleven and the gain is ordinary income. Wait until month thirteen and it’s long-term.

The bracket pairings follow the general structure in IRS Topic 409, but the income thresholds separating the 0%, 15%, and 20% long-term rates are adjusted for inflation every year. Look up the current figures rather than assuming these pairings hold at your income.

Read that table practically. If drift triggers a rebalance at month ten, waiting two months saves somewhere between 8 and 23 basis points of the whole portfolio on this single trade. Whether two more months of being off-target is worth 8 to 23 basis points depends on how far off you actually are — a portfolio at 64/36 and one at 78/22 are not facing the same question.

Working backwards from a tax budget

Sometimes the question runs the other way. You’ve decided you can absorb $1,000 of tax this year and want to know how much rebalancing that buys.

Divide the budget by your effective rate to get the maximum realised gain. Then divide that by the gain’s share of sale proceeds, which is the embedded gain divided by one plus the embedded gain.

At $1,000, an 18.8% effective rate, and shares up 50%: maximum realised gain is $1,000 ÷ 0.188 = $5,319. A 50% embedded gain means gain is 33.33% of proceeds. So the maximum sale is $5,319 ÷ 0.3333 = $15,957.

That’s a useful, deflating number. If getting fully back on target required moving $40,000, you now know your budget covers 40% of the job. The remaining choice is whether to go partway this year and finish in January, or to find the untaxed routes below.

Rebalancing without a tax bill

Two ideas about this cost are wrong often enough to be worth naming.

The first is that commissions and spreads are the main event. They were, in 1998. In the trade at the top of this page, friction cost $6 and tax cost $857 — trading is 0.7% of the total. Grinding away at execution quality while ignoring which lots you sell is optimising the wrong 0.7%. Trading cost does still matter in specific places: thinly traded closed-end funds, individual small caps, some international and emerging-market ETFs where spreads run 20 to 50 basis points, and any fund with a redemption fee. Read the fee table before you assume 3 basis points.

The second is that the tax is the price of staying on target. It isn’t, or not entirely. Selling inside an IRA or 401(k) costs nothing in tax, and works whenever the drifted asset is held there in enough size. Directing new contributions to the underweight asset costs nothing either — if you’re putting $2,000 a month into that $500,000 portfolio, that’s $24,000 a year, 4.8% of the portfolio, which is more rebalancing capacity than the 4% sale modelled here. The catch is timing, since contributions arrive across twelve months while drift happens whenever it happens. Redirecting dividends instead of reinvesting them adds another 1 to 2% a year, limited by yield. And when you do have to sell in taxable, choosing specific high-basis lots rather than accepting the broker’s default can cut the realised gain to a fraction of the average.

When this arithmetic breaks

Several situations move the answer, some sharply in your favour.

Carryforward losses. Capital losses offset capital gains dollar for dollar. Carrying $30,000 of losses from a prior year means the first $30,000 of gains costs nothing, and the 17.3 basis points collapses to 0.1 until the carryforward runs out. The IRS pages linked above describe the offset and the $3,000 annual cap on deducting excess losses against ordinary income.

A 0% long-term year. Taxable income low enough to qualify zeroes the tax column entirely. That happens in a gap year between jobs, in early retirement before Social Security starts, or in a year with unusually large deductions. Some people deliberately realise gains in those years to reset basis higher for later.

State tax. California, among others, has no preferential rate for capital gains. Add your state marginal rate to every figure above. At 9.3% state on top of 18.8% federal, the 40%-embedded-gain row goes from 21.6 to about 32 basis points.

Positions held at a loss. Then rebalancing produces a deduction rather than a bill — but buying something substantially identical within 30 days on either side triggers the wash sale rule and defers the loss.

Fund distributions you didn’t ask for. An actively managed fund passes realised gains through whether or not you trade. A fund distributing 5% of NAV as long-term gain hands you a tax bill with no rebalancing involved. Reinvesting that distribution raises your basis, which quietly lowers the cost of the next rebalance.

Bonds rather than stocks. Bond funds bought before the 2022 rate move may carry embedded losses instead of gains, which inverts the whole calculation.

Step-up and charity. Inherited assets generally receive a basis step-up, and appreciated shares donated to a qualified charity can avoid realising the gain. Both change the long-run answer for exactly the largest embedded gains that the table above marks as most expensive to sell.

What this does not tell you

Everything above is the cost side. None of it is the benefit side, and rebalancing does have one: it holds your risk exposure where you set it. A 60/40 portfolio left alone through a long equity run drifts to 75/25 without announcing itself, and that’s a materially different portfolio with a different drawdown in the next bad quarter.

I haven’t put a number on that benefit, because it isn’t a fixed number. It depends on the return and correlation path that actually happens, which nobody knows ahead of time. Anyone quoting a precise “rebalancing bonus” is quoting a backtest over one particular history.

Also outside these figures: the behavioural cost of a rule you won’t follow, since rebalancing means selling what went up and buying what went down, and plenty of people can’t. Cash drag while trades settle. Your actual effective rate, which depends on your whole return — including whether this gain itself pushes you into a higher bracket or across the NIIT threshold. And the question of how often or at what threshold to rebalance at all, which is an allocation decision, not a cost calculation.

FAQ

How much does rebalancing cost per year?

For a taxable account turning over 4% a year with 40% embedded gains at a 15% long-term rate, about 17 basis points — $173 per $100,000. At a 23.8% rate on positions that doubled, it reaches 48 basis points. Inside an IRA or 401(k) it’s roughly 0.1 basis points, which is nothing.

Is it better to rebalance in my IRA or my taxable account?

On cost alone the IRA wins every time, because trades inside it don’t realise taxable gains. On the $500,000 example above, that’s $857 saved on a single rebalance. The real constraint is whether the IRA holds enough of the drifted asset to do the whole job.

Does rebalancing trigger capital gains tax?

Yes, in a taxable account, on any position sold at a gain. Selling $20,000 of a fund that’s up 40% realises $5,714 of gain, which costs $857 at a 15% rate. Buying triggers nothing, and selling at a loss produces a deduction instead.

What is the 3.8% net investment income tax and will I pay it when I rebalance?

It’s a surtax on investment income that applies once modified AGI exceeds $200,000 single or $250,000 married filing jointly, per IRS Topic 559, and it hits the lesser of your net investment income or the excess over the threshold. It converts a 15% rate to 18.8% and a 20% rate to 23.8%. On a 4% rebalance that’s roughly 4 to 8 extra basis points of portfolio cost.

How long do I have to hold before selling to get the lower rate?

More than one year. At a 24% ordinary bracket, $5,714 of gain costs $1,371 short-term against $857 long-term — a $514 difference, or 10 basis points of a $500,000 portfolio. At 35% the gap widens to $1,143, about 23 basis points.

Do zero-commission brokers mean rebalancing is free now?

No. Commissions were always the small part. In the worked example, $0 of commission and roughly $6 of spread sit beside an $857 tax bill, so free trading removed about 0.7% of the total cost of rebalancing a taxable account.

Can I rebalance without selling anything?

Yes, by pointing new money at the underweight asset. Contributing $2,000 a month to a $500,000 portfolio adds 4.8% of portfolio value a year, more capacity than the 4% sale modelled here, at zero tax cost. Redirecting dividends rather than reinvesting them adds another 1 to 2%.

Before you apply any of this

Pull three of your own numbers first. The unrealised gain percentage on each taxable holding, which your broker shows per lot. Your long-term capital gains rate for the current tax year. And whether your modified AGI sits near the NIIT thresholds. With those three you can find your own row in the table above instead of borrowing mine.

Then check one setting: whether your broker lets you choose specific tax lots at the moment of sale, or silently defaults to first-in-first-out. That single toggle controls the embedded gain figure that drives every number on this page. And if you hold mutual funds rather than ETFs, read the fee table for redemption and exchange charges using the SEC’s guide to mutual fund fees and expenses, because those don’t show up in the expense ratio and won’t show up on your trade confirmation either.

This article is general information, not financial advice. See our disclaimer.

Also worth reading: Qualified vs Ordinary Dividends: The Tax Rate Gap on the Same Payout

More in Costs and Fees · all topics · calculators · how this was checked