Eight tax levers, ranked by what they are worth

The tax levers on a stock portfolio rank in almost the reverse order of how loudly they are sold. The four worth the most are decisions with no product attached. The one with the marketing budget is worth the least once an account is a few years old. Here is the ranking for an investor at the top California rates, biggest first, then what each lever does and what it costs.

1. Never realize the gain Hold until the basis steps up at death, or give the shares away. The whole tax disappears. 35¢ per $ of gain
2. Cross the one-year line Long-term instead of short-term. Free, apart from patience; known by everyone and still crossed early. 17¢ per $ of gain
3. Hold an ETF instead of a mutual fund Average tax drag 0.7% against 1.9% a year over ten years. Costs nothing extra. ≈ 120 bps a year
4. Put each return type in the right account High-turnover return belongs where turnover is not taxed. Free. 52% vs 35% on what it throws off
5. Harvest losses Big in year one, small after year five unless new cash or recurring gains keep it alive. Fees run from 9 to 115 bps a year. 155 → 0 to 30 bps a year, year one to year five
6. Move an appreciated portfolio without selling Section 351 seeding into an ETF, or an exchange fund for one big stock. Niche, and only for low-basis positions. Defers the whole gain
7. Tax-aware long/short Multiplies the harvest three to ten times, but the losses still need somewhere to go. Not a general strategy below about $5 million. More losses, same limit
8. Option overlays Covered calls and income ETFs convert the kind of income you receive. They do not reduce it. Changes the label, defers nothing

The four that cost nothing

Each of these is a decision, not a purchase.

1. Never realize the gain. A $100,000 position bought for $60,000 costs $14,040 of tax to sell, 35.1% of the gain. Held until death, its basis resets to market value and the $40,000 gain is never taxed by anyone; California follows the federal rule, and with a $15,000,000 estate exemption per person the estate tax does not enter for most households. Given to a donor-advised fund, the same shares deduct at full market value and the charity sells them untaxed. Every other lever on this list reduces or defers the tax on a gain. This one removes it, which is why a harvesting program that eventually sells its winners can never catch up with an investor who did not.

2. Cross the one-year line. A gain on shares held more than a year is taxed at the 20% federal rate instead of 37%. The 3.8% investment tax and California’s rate apply on both sides of the line, so the whole 17-point gap is federal, and the same gap exists in every bracket: 9 points for a 24%-bracket investor. It costs nothing except patience, and it gets missed anyway: in 2015, the last year the IRS published sales by holding period, individuals realized $3.3 billion of stock gains on shares held nine to eleven months.

3. Hold an ETF instead of a mutual fund. Morningstar’s tax cost ratio over the ten years to 2024 puts the average equity mutual fund’s tax drag at 1.9% a year and the average ETF’s at 0.7%. The gap is the in-kind redemption: an ETF hands its lowest-basis shares to a departing market maker and the gain leaves the fund untaxed, while a mutual fund sells and distributes. What remains inside the ETF is the tax on dividends, which no wrapper removes.

4. Put each return type in the right account. Whatever a traditional IRA or 401(k) earned, its distributions come out as ordinary income; a Roth or an HSA pays nothing on the way out. So a holding whose return arrives as long-term gains and qualified dividends belongs in a taxable or Roth account, where those rates apply, and a holding whose return arrives as ordinary income or short-term gains belongs in a deferred account, where its turnover is not taxed. The lever is worth the difference between 52% and 35% on whatever the holding throws off.

The one with a price tag

The lever the industry sells, and the only one on the list with a fee attached to the lever itself.

5. Harvest losses. Sell a position that is down, buy a similar one the same day, and the loss offsets gains elsewhere. AQR’s simulation puts the benefit at 155 basis points a year in year one and 0 to 30 by year five, because a portfolio that has risen runs out of lots below cost unless new cash or recurring gains keep feeding it. The product that does this for you costs from 9 basis points a year direct to a retail investor to about 115 through an adviser and a platform. At the adviser-and-platform price of about 115 basis points the fee outruns even the high end of that year-five benefit; the arithmetic only works in year one, or on the 9-basis-point retail product. Harvesting is not useless; it is the fifth thing to do, priced right, rather than the first thing to buy.

Three for special situations

Each solves one problem for one kind of holder.

6. Move an appreciated portfolio without selling. A diversified portfolio can be contributed into a new ETF under Section 351 without recognizing its gains, and one big stock can go into an exchange fund that pools it with other investors’ stock. Both defer the whole gain rather than paying 35 cents of it up front, and both only make sense for positions with a low cost basis.

7. Tax-aware long/short. Adding a short book multiplies the losses available to harvest three to ten times. The losses still need gains to offset, and the strategy carries adviser, financing and borrow costs, so it is not a general strategy below about $5 million of taxable assets.

8. Option overlays. Covered calls and the income ETFs built on them convert the kind of income you receive: option premium is short-term, or 60% long-term and 40% short-term on a broad index. They change the label on the income. They do not reduce it or defer it.

Why the order is upside down from the ads

Nobody earns a fee when an investor waits a year, holds an ETF, or leaves a winner in the estate. Those are the levers worth the most. Loss harvesting is the lever with a product attached, and the product carries the fee stack. The three at the bottom are real tools for the situations they were built for, sold to many people who are not in them.