Tax Loss Harvesting: what is it and how large is the expected benefit?

Tax Loss Harvesting is the rage now. Robo-advisers do it for you, and every DIY saver should seriously consider the benefits. Let’s look at what Tax Loss Harvesting is, how and why it works and how large (or small) the expected benefits can be.

What is Tax Loss Harvesting?

Tax loss harvesting involves these four simple steps:

1. Continuously monitor the taxable portfolio for tax lots that are underwater, i.e., have a cost basis above today’s value. Otherwise known as a capital loss. This can be a short-term or long-term loss.
2. Sell those tax lots and immediately buy a similar asset (though not identical to avoid the IRS wash sale rule). One might also do the regular portfolio rebalancing at the same time.
3. Up to $3,000 p.a. in such losses can be used to offset ordinary income and lower the tax burden. Any excess over the $3,000 can be carried forward and written off in future years.
4. Reinvest the tax savings from step 3 and watch your portfolio grow even faster! Note: the benefit is not $3,000, but $3,000 times the marginal tax on ordinary income!

For more details on the implementation please refer to our DIY guide on how to be your own DIY Robo-adviser.

How and why does Tax Loss Harvesting work?

The TLH excess return consists of three components and we can later see how much each one of them is really worth:

1. Time value of money: you get a tax benefit today but may pay higher taxes later when you eventually liquidate the position in retirement. Even if the tax rate were to be the same in retirement (or even slightly higher) you stand to gain from TLH because the government gives you an interesting-free loan to invest the TLH proceeds over potentially many years. You increase your future tax liability by exactly what you harvest in benefit today. See the red bar in middle column vs. blue bar on the left in the figure above. But getting the benefit today is still valuable! The green bar in the middle column is the gain from TLH in that case, which is equal to the capital gains from investing the date 0 tax savings net of the capital gains tax on those additional capital gains (orange bar).
2. Tax rate arbitrage 1: between ordinary income and long-term gains. If you do the TLH right, your future capital gains are taxed at the lower long-term capital gains rate, while the benefit from the tax write off today comes from the higher ordinary income marginal rate. You squeeze the orange and red tax liabilities and get to keep more for yourself, see right column in the chart above.
3. Tax rate arbitrage 2: between currently high marginal rates and (hopefully) lower rates in retirement. Chances are that during your working years you have higher taxable income putting you into a higher tax bracket both on your federal and state tax return. Moreover, you might move from a high tax state (NY, CA) to a low or no-income-tax state (FL, NV). In the chart above, tax rate arbitrage squeezes the red and orange tax liability boxes even more and leaves more pure after-tax gain (green bar, right column) for you to keep. In the best possible case the future taxes are zero and you get to keep the entire loot! Woo-hoo!

Tax Loss Harvesting: Numerical Examples

Let’s look at some numerical examples to see how much we can expect to make from harvesting tax losses! The assumptions used in the calculations are:

• The investor starts out on December 31 of year 0 with a given portfolio value and a given harvestable loss.
• The all-equity portfolio generates a certain capital gains return and dividend yield each year. The dividends net of taxes are reinvested in the portfolio every year.
• Without TLH the investor simply keeps the money invested for a number of years then retires on December 31 of year N and liquidates the portfolio on January 1 of year N+1.
• With TLH we assume the investor realizes the entire loss on December 31 of year 0 and invests the proceeds as well as the TLH tax savings in a very similar (but not identical) equity fund to avoid the wash sale rule. Any tax losses beyond the $3,000 maximum are carried forward and invested in future years. Upon retirement, any leftover losses will be used to offset long-term capital gains during retirement.
• In both cases we assume that the investor realizes no other capital gains along the way. He/she will of course earn capital gains, but just not realize them in a taxable account until after retirement (using tax rate arbitrage 2, see above).

What’s the most you can expect in extra annualized % return?

Let’s pick assumptions that would generate a large TLH benefit, so as to give TLH the best possible chance to succeed:

We call this person Mr. A:

• Portfolio value of $10,000 and a harvestable loss of $3,000 (thus, initial cost basis $13,000)
• Very high current marginal taxes: 49.6% for ordinary income (39.6% federal plus 10% state), 33.8% for dividends (20% federal, 10% state, 3.8% ACA). The tax rate on long-term capital gains in retirement is 0% for the maximum tax rate arbitrage.
• Time horizon is 10 years.
• The expected equity return is 7% nominal, 5% from capital gains and 2% from the dividend yield.
• Results, see table below: Almost 1.5% extra return annualized, and over $2,700 of extra after-tax portfolio value in ten years. That’s a lot of dough for two trades, taking probably no more than a few minutes of your time!
• The 1.49% annual tax benefit is significantly better than the 0.77% Betterment calculated.
• Of course, we could have stretched the assumptions much more to make the TLH even more attractive, say, by lowering the current portfolio value to $3,000 while keeping the tax loss at $3,000 as well. The annualized TLH benefit would have been more than 4%. That would have been a bit unrealistic, though; someone earning a high six figure income, who owns only a few thousand dollars and generated tax losses of 50%.

Most of the gain came from the tax rate arbitrage, especially the zero future tax rate (Tax Arbitrage 2), see chart below. But nevertheless, even in the absence of any tax arbitrage, there is money to be made from TLH. Almost $700 are simply due to the time value of investing the TLH proceeds over ten years. Thus, TLH is worthwhile even if the tax in the future were to jump up all the way to an insane 49.6%, equal to today’s high marginal rate. If in addition the future LT Capital gains tax were to stay at today’s Dividend/LT Cap Gains tax rate of 33.8% we gain another $700, with the remaining $1,350 coming from reducing the LT Capital Gains tax rate all the way to 0%.

How much do we diminish the Tax Loss Harvesting benefit when we use consecutively less extreme assumptions?

Mr. A has a pretty impressive excess return from Tax Loss Harvesting. But his case is not typical at all. Very few people are in such a high tax bracket, and in addition Mr. A also had a very tiny investment portfolio and a large pile of tax losses relative to the portfolio value. This means the benefits of a $3,000 write-off are applied to a relatively small base, hence, the impressive increase in the % return.

Let’s see how that annual percentage tax alpha diminishes as we make the assumptions less and less beneficial for TLH.

Mrs. B: same as Mr. A but twice the portfolio value and twice the harvestable loss.

• $20,000 initial portfolio value, $6,000 harvestable loss
• Why should this make a difference? Mrs. B can only write off $3,000 from her taxable income in a year and has to wait until next year for the remaining $3,000. The dollar benefit increases, though by less than a factor of two. The tax alpha suffers a tiny bit due to this but is still very, very impressive! 1.44% p.a.!

Mr. C: Five times the portfolio value and five times the harvestable loss as Mrs. B

• $100,000 initial portfolio value, $30,000 in harvestable loss
• The dollar value of TLH goes up, but by a little bit less than a factor of five and we experience another deterioration in the percent alpha because it take 10 years to work off the pile of tax losses, $3,000 per year. This diminishes the time value gain of TLH, but the total alpha is still north of 1%! 1.16%!

Mrs. D: double the portfolio size and tax loss of Mr. C

• $200,000 initial portfolio value, $60,000 in harvestable losses
• Now there are more tax losses than we can write off in the remaining 11 years of high marginal tax rates (it’s 11 years, because we assume that we use the tax write-off in calendar years zero through ten). The remaining tax losses will now offset the capital gains during retirement. Then the TLH will have zero benefit because LT gains are already taxed at a zero marginal rate.
• The result is a significant deterioration in the tax alpha down to 0.64%, but still a good boost in returns and an impressive $22,000 in extra final portfolio value.

Mr. E: Lower Tax loss available

• Same as Mrs. D, but instead of $60K in tax losses, he is only able to scoop together $20K. That doesn’t sound like very much. True, the losses would only be about 10% of today’s portfolio value, but keep in mind that the initial portfolio value could the product of many years (even decades) of saving and reinvesting dividends. A large share of the tax lots in your account may have a low tax basis so that even a significant drop of 20 or 30% will not produce much of a harvestable loss.
• The tax benefit drops to now 0.44% p.a. and a total of about $15,500. Sounds attractive, but the Betterment and Wealthfront fees of 0.15-0.25% already eat up a big chunk of it.

Mrs. F: Lower current tax rates

• Same assumptions as Mr. E, but more middle-class tax rates: 30% for ordinary income (e.g. 25% federal, 5% state), lower taxes on dividends (15% federal, 5% state), but keep the zero tax rate in retirement.
• Another deterioration in the tax benefit. Only about $9,500 in extra portfolio value after ten years, or about. 0.27% in extra return. That’s still worth the “hassle” of a few trades today but may not justify hiring a Rob-adviser who charges 0.15%-0.25%.

Mr. G: Half-million dollar taxable portfolio with $25K in tax losses

• Now we get into FI (Financial Independence) territory. Again, this kind of portfolio value was not invested as one large lump-sum just a few weeks ago, but is the product of many years worth of saving. Hence, a sizable tax loss is not that easy to generate given that many tax lots are going to be from the early to mid 2000s, not to mention early 2009.
• Tas Loss Harvesting is still going to generate a lot of extra after-tax return, but as a percentage of the principal it’s only 0.11% p.a. If you hire a Robo-adviser the 0.15%-0.25% annual fees will eat up the entire tax advantage and more.

Mrs. H: Same as Mr. G, but low current tax rate

• Now the current ordinary income tax is only 15% (no state tax, 15% federal) and the existing dividend and LT capital gains tax is already 0%. It will also stay 0% in retirement.
• The Tax Loss Harvesting gain is now under $5,000, and only worth 5 basis points per year. Hiring a Robo-adviser in this situation would be a major waste of money.

If anybody has ideas for other robustness checks (Mr./Mrs. I,J,K,L,M,…) for us to check out please let me know. We might be able to reach Mr. T (see below, right) before Physician on FIRE reaches Dr. J (see below, left) in his Four (or Five?) Physician study.

Caveats:

The TLH benefits can be larger or smaller depending on several other factors, especially unrelated capital gains. If you have large additional short-term capital gains from other unrelated investments early on, you might do significantly better than we calculated, because you don’t have to wait to write off $3,000 every calendar year, but instead get the entire benefit upfront and thus a higher time value of money effect (higher Time Value of Money effect). But nobody in their right mind would voluntarily realize short-term gains when they face high marginal taxes, so unless there are circumstances where someone faces forced short-term gains this situation can’t be too common.

If you have large additional long-term gains from unrelated investments, you might do worse than we calculated. That’s because first you have to net your harvested capital losses against capital gains. Even the short-term losses are netted against long-term gains before they can be used to lower your taxable income! In the extreme case where long-term gains are already taxed at zero percent (Mrs. H in the example above) your entire tax losses might be wiped out for no gain at all. Thus, unrelated capital gains will inadvertently eliminate the Tax Rate Arbitrage 1 Effect between ordinary income and long-term capital gains.

Moreover, caution with average vs. marginal benefits. The benefit from TLH of the marginal dollar invested is most likely below the average benefit. Let’s take the comparison between a Roth IRA vs. a taxable account (see our comparison calculations here): normally a Roth IRA would easily beat a taxable account without TLH because dividends and capital gains are taxed in the taxable account. But the difference in returns could be small if capital gains are realized in retirement at 0%. TLH could make the taxable account more attractive or the marginal benefit could be so small to keep the Roth IRA ahead. Whether to put those additional $5,500 into a Roth IRA or taxable account crucially depends on the individual circumstances and the marginal benefit, not the average over the whole portfolio!

Conclusions:

• Tax loss harvesting is a useful tool when saving in a taxable account. Unfortunately, TLH cannot be scaled arbitrarily. Hence, the tax alpha estimates quoted by the Robo-advisers are not that easy to generalize.
• Depending on the tax rate assumptions and the size of the harvestable loss in relation to both the $3,000 annual tax write off and the initial portfolio value, the tax alpha can range from several basis points (hundredth of a percent) to north of one percent per year. For most investors, however, the promise of 0.77% extra return p.a. from TLH (Betterment estimate) is unrealistic. The Wealthfront estimate of upto 2% p.a. is borderline preposterous. We have written earlier on how the two Robo-advisers exaggerate their estimates and depending on where you fall in the spectrum of Mr. A through Mrs. H, you may have a very different experience from what the Robo-advisers advertise.
• For current early retirees in low tax brackets and low (or no) state taxes, the expected extra return from newly generated tax losses is minimal, likely below 0.10%. That’s still a lot of money, considering the large sums involved. But it would likely not justify hiring a Robo-adviser just for the Tax Loss Harvesting.