Tax Loss Harvesting is Overrated

A friend of mine who reads this blog emailed me recently for advice about about tax loss harvesting (abbreviated TLH in this article). The allure and promise of TLH, to pay less in taxes, is indeed sexy. Who doesn’t like this promise?

TLH is one of the major supposed value-adds of robo-advisors like Wealthfront and Betterment (these links describe in great detail their efforts to TLH).

The purpose of this article is to convince you that the above claims are hogwash. Or perhaps I’m wrong and the purpose of this article is to expose how ignorant I am on the topic.


Wealthfront’s TLH claims right on its homepage.


The basic premise of TLH:

The purposes of TLH is to realize losses (i.e. sell shares that have gone down in price) to get tax relief today. My major qualms with TLH are the following:

  • By TLH today, you’re simply lowering your cost basis. However, this lower cost basis means higher taxes down the road. None of the proponents of TLH fully discuss the ramifications of higher taxes down the road.
  • TLH oftentimes presumes you have capital gains to offset. However, if you’re a tax-conscious long-term investor you should optimally not have gains to offset.
    • In the absence of capital gains, one may use capital losses can offset up to $3k of ordinary income (e.g. W2 wages). Any “unused” tax losses carried forward to future years.


Let’s quantify the benefits of TLH then:


  • You purchased $100k of shares and they went down to $70k during a recent market downturn. You sell and generate a $30k loss.
  • You immediately purchase a “not-substantially identical” tax loss harvest partner (e.g. sell VTSAX and buy VFIAX). Since you’re a prudent long-term investor, you don’t sell the $70k of VFIAX for decades.
    • This is an minor nuisance with TLH; you end up in an investment that you may not have wanted (e.g. VFIAX rather than VTSAX).
  • Since you have no investment gains, the $30k of tax losses cannot be used to offset any investment gains. Instead, you use $3k/year of losses to offset ordinary income with any used losses “carried forward” to future years.
  • Your marginal (federal + state) tax rate on labor income is 33.333%.


  • You’ll benefit $1k (=$3k*33.33%) in reduced federal + state taxes each year for 10 years.
  • Assume you sell the investments in your brokerage account in 10 years. Because your cost basis is $30k lower than it would have been had you not TLH’d, you’re going to pay more taxes at that point. Assume that tax law doesn’t change significantly over the next 10 years (a big assumption) and that long-term capital gains are still taxed less than labor income. Let’s assume that your marginal tax rate on investments (federal + state) is 25%. Therefore, you’ll face a $7,500 (=$30k*0.25) higher tax burden in 10 years as a result than had you not TLH’d today.
  • Let’s compute the present value of this entire transaction, appropriately accounting for the fact that the TLH has increased our long-term investment taxes due to the lower cost basis. Assume a 5% discount rate. Assume first tax benefit is claimed immediately and that the eventual sale of the securities is 10 years down the road.
    • PV = $1k + $1k/(1.05) + $1k/(1.05^2) + $1k/(1.05^3) + $1k/(1.05^4) + $1k/(1.05^5) + $1k/(1.05^6) + $1k/(1.05^7) + $1k/(1.05^8) + $1k/(1.05^9) – $7.5k/(1.05^10)
    • PV = $3,503.47

Discussion of Results:

  • A present value of $3.5k of TLH benefits is not nothing, but it’s certainly not the panacea that is worth the disproportionate amount of attention that investors give to it.
  • I think the disproportionate emphasis of TLH comes from marketing from robo-advisors and the fact that uniformed investors don’t appropriately account for the “costs” of TLH (the higher tax burden down the road).

How we could improve upon our benchmark results:

  • Rather than selling the TLH’d shares in 10Y, we could have completely avoided the corresponding increase in capital gains taxes with the following:
    • Donating the appreciated shares (now with lower cost basis) to charity.
    • Dying, so that our heirs benefit from the “step-up in cost basis” at death.
    • Doing something similar to GoCurryCracker, and exploit the 0% region for federal taxes on LTCG (e.g.

With the above framing, TLH with the intent to donate low-cost-basis shares to charity/heirs is indeed pure arbitrage. But this isn’t what I hear promoted on the internet.


Is it worth paying a robo-advisor to TLH for you?

Hell no.

Let’s assume a 0.5% AUM fee. Assume you have $1M with an advisor. Each year you’re paying $5,000 (=$1M*0.5%) in fees to achieve $1,000 in TLH benefits (and recall that these benefits are diminished once we appropriately account for the higher tax burden downstream).

Pushing TLH strategies to the forefront of their marketing materials is a baseless attempt to convince you that their AUM is fairly earned. It is not.


Despite all of the above, let’s say you still want to TLH. Here’s how I’d do it:

  • Have all tax-sheltered accounts at one brokerage (e.g. I have all of my 401k, IRA, HSA at Fidelity).
  • Have all taxable accounts at another brokerage (e.g. I have all of my taxable stuff at Vanguard).
  • Turn dividend reinvestment off in your taxable brokerage account.
  • Why is the above important?
    • A “wash sale” occurs when you sell a security and buy one that is “substantially identical” within 30 days. Since VG and Fidelity track slightly different indices (for their total US stock market indices, for example), you can have dividends reinvested automatically in your tax-sheltered accounts without fear that these auto-reinvestments of dividends will invalidate your TLH efforts in your taxable brokerage account.
  • When you go to sell, be sure you have an appropriate “tax loss harvesting partner” (e.g. VTSAX and VFIAX). Google “tax loss harvesting partner” for more examples. Ensure that you’re not selling a security that you bought less than 30 days ago in your taxable brokerage account.
    • If you don’t want to stay in the “partner” security after 30 days, then go back to your original investment.
  • I’d look for large investment losses so that you only have to bother with this once a decade or so. Anything less than $10k in losses does not strike me as being worth your time.
  • Importantly, since you’re a prudent long-term investor, you should avoid selling appreciated shares to generate taxable gains in the first place (unless you’re in the GCC sweet spot of 0% MTR).



I realize that TLH may be useful in the following scenarios:

  • Allowing an investor to avoid LTCG when there is a big liquidity event (e.g. paying for kid’s college).
  • Helping an active trader to offset big gains.

I just don’t think the above are sufficiently important to warrant the disproportionate amount of attention I see to TLH on the internet. The much more impactful way of avoiding taxes in a taxable brokerage account is the following:

  • Buy index funds and hold them until you die.
    • If you can’t wait that long, then:
      • Hold them for as long as you can
      • Or dump them to charity
      • Or wait to sell until you get to the GCC 0% region

I realize the above isn’t sexy advice, but it’s the best advice one could give for minimizing one’s tax burden while investing in taxable brokerage accounts.


Reader Poll

Do you find the assertion that TLH is overrated convincing? Or is the above analysis flawed/unconvincing? If so, what am I missing?

30 thoughts on “Tax Loss Harvesting is Overrated”

    • Thanks for sharing. A colleague sent me the article a few days ago and I didn’t see any mention of the $3k/year deduction limit for W2 income. Given the constraint, I’m unsure how the authors arrive at their conclusions, but I’ll take another look at the paper later to see if I can figure it out.

      • To me, I think dividend is a big drag when investing in taxable account long term. If investing in things that generate minimum or no dividend, then I do not see much difference between taxable and Roth. Of course, tax code could change in the future…

        • I agree entirely with your assessment on the importance of dividend tax drag over a multi-decade horizon. If you could avoid that and arrive to the 0% MTR on investments in retirement, it’d effectively be a Roth account. In fact, it’d be even better than a Roth because you’d still have the free option to TLH along the way without paying the “cost” of higher capital gains taxes down the road.

          BTW, I emailed an author on that SSRN paper about their lack of consideration of any $3k/year deduction limits. I’ll report back if they respond. If they ignored the constraint, it seems like a pointless paper to me.

  1. There’s a pretty good Mad Fientist post on it, but I don’t think anything disagrees with your analysis. If I recall correctly, he assesses the value assuming a 0% long-term capital gains tax due to low income in retirement.

    Personally, I take advantage of it because it’s not that much work. I’d also rather have tax savings now than later as I anticipate being in a lower tax bracket in retirement.

    • If this is the article you’re referring to, I agree that some of the aspects have been covered already: However, the fact that there is a big cost — the bigger tax burden down the road — is not something I’ve seen appropriately covered before. I’m unsure why not because it is so obvious.

      I’m glad that you’re TLH’ing successfully. Are you doing it annually or just periodically during huge market declines? Do you have capital gains that you are offsetting with the losses or are you just offsetting the $3k/year of ordinary income?

      (Un)fortunately for me, I don’t have a penny of unrealized losses to realize. Unfortunately, I dragged my feet during the >30% covid drop and didn’t TLH when I should have. In hindsight, this was an obvious mistake. Luckily, the size of the mistake is muted when I appropriately account for the future increased tax liability.

      • Yep, that’s the article.

        I periodically TLH. In fact, the covid drop was actually the first time that I was able to do it in any significant way. It took me about 10-15 minutes to execute and I now have enough losses to carry over for several years. I also don’t typically have any long-term gains to offset since I’m not living off of investments yet. The 3K ordinary income offset is more valuable anyway given my marginal tax rate now vs where I think it will be when I retire.

        Do you not agree with the Mad Fientist’s suggestion that you can withdraw your money paying 0% LTCG by being in a low tax bracket? Or just think it’s unlikely? Even if you do pay taxes on the LTCG, do you not think it’s likely that the tax rate will be lower than your current marginal rate while in your prime earning years?

        • Glad you TLH’d during covid!

          I think the point of my article is that TLH’ing is generally sold as a free lunch without a downstream LTCG “cost”, but I think that’s mostly misguided. If you are able to pull off the 0% LTCG downstream, then it is indeed a free lunch. I’m unsure how likely this will remain an option. I don’t think the general public will want to retire near the poverty level to fully exploit the 0% LTCG region. I’m pretty comfortable living around there so I think it’ll be in play for us.

          In conclusion, I regret not TLH’ing when I had the chance. If another chance presents itself I’ll do it. But in the grand scheme of things it’s a tiny “mistake” I made given the pretty limited benefits. I generally think that TLH is overblown, but I suppose you could make the same arguments about irrationally strategizing around CC rewards, CC bonuses, grocery coupons, etc (some of which I’m guilty of).

  2. Many of the proponents obsessed with TLH are also the same ones obsessed with Roth over Tradβ€” even though the former obsession assumes tax rates will be lower in retirement, and the latter obsession assumes tax rates will be higher.

    • Not sure if you are implying that I’m in that group, but I think the strategy around Roth vs Trad is much higher stakes than the TLH silliness. Admittedly the Roth vs Trad debate hinges entirely on unknown tax rates 40 years down the road.

      That said, I’m reasonably confident that a GCC 0% Roth conversion strategy will exist in a few decades, particularly for those living frugally near/slightly above the poverty line.

      • Not talking about you at all (I don’t think??). Merely pointing out that many pushing TLH ignore future taxes, as you point out, but are strong supporters of Roths instead of Trads…because they believe future taxes may be higher. This is contradictory. If one truly believes taxes will be higher in the future, they should harvest gainsβ€” not losses. Upfront, I think choose either Trad+TLH or Roth+TGH. What are your thoughts? I am biased towards the first since I will be ER and easily backdoor Trads empty with $0 taxes by normal retirement age.

        • I’m in complete agreement with what you are saying, with one caveat. Mainly, your framing ignores the difference in taxation of labor income vs investment income (e.g. the relative large 0% region of LTCG now). Since TLH today reduces my labor income marginal rate at the cost of a future capital gains marginal rate, is makes the problem slightly more difficult.

          But I think I’m 99% on board with what you’re saying. Like you, I’m 100% in the camp of Trad + TLH now and tax-free Roth conversions and capital gains harvesting in a the future while in the 0% region (a la GCC + many others).

  3. “sell VTSAX and buy VFIAX” to TLH scenario.
    I was advised that that you are likely violating IRS rules so recognize you’re taking a risk. If your positions are in separate brokerage accounts, I was told it is very unlikely the IRS will catch it but it is still a risk.

    • Thanks for the feedback!

      Check out this Bogleheads link:

      The link, which is admittedly not IRS doctrine, is way more aggressive than I am with proposed TLH partners. The article mentions VTI <=> ITOT, while both are total US stock market ETFs. This supposedly avoids wash sale rules because they track different total stock market indices.

      In contrast, VTSAX holds 3600 firms while VFIAX holds 500 firms. This seems substantially more defensible (and thus, less risky) than the above. Given the 3100 firm differential between VTSAX and VFIAX, I’d sleep well at night with that particular strategy.

      I agree that there is a ton of grey area here with very little official guidance from the IRS on the topic. Nonetheless, the internet is full of opinions of random internet strangers like you and me. I certainly don’t claim to be an expert, but again, I personally think a VTSAX <=> VFIAX TLH pair is quite defensible.

      That said, I’ve never TLH’d in my life. I’m open to the possibility, but I need to not drop the ball like I did during Covid. But I’m realistic about the muted benefits of doing so.

      • Anything may be debatable, but saying that VTSAX and VFIAX are not substantially identical seems an extremely safe position. Phillip, it could be interesting to understand the rationale of whoever told you that would violate IRS rules.

        • Glad to hear that the VTSAX <=> VFIAX TLH partner passes the MDM sniff test. When I wrote that in the article, I felt like I was erring on the side of extreme caution.

        • The advisor was my “free” Fidelity advisor that you get when you have a certain level of assets. With Fidelity’s system, most trades automatically test against wash sales rules. This guy didn’t explicitly refer to VTSAX VFIAX but mentioned that this practice is a grey area in general which isn’t exactly black and white as to which securities are and aren’t sufficiently similar. He went on to suggest that if you do want to TLH with potentially similar securities, doing it via separate brokerage accounts is a “safer” way to do it.

  4. Great analysis. I agree that TLH is oversold, and even though Wealthfront charges only .25%, I personally don’t think it’s worth it for me. However, for the casual investor who is too lazy/ignorant to buy-and-hold a broad market index fund, it beats paying 1% to a more expensive, active advisor that churns your account.

    A couple things to note. For simplicity, in my TLH scheme, I share the same pairings in my taxable and tax advantaged accounts. But this means I have to turn off auto-reinvestment in my tax advantaged accounts. Otherwise, I could end up with wash sales between the taxable and non-taxable and a worthless/disallowed adjusted cost basis in my tax-advantaged accounts. I should probably change this to make life easier for myself, but it also makes rebalancing easier as the dividend flows towards the imbalanced investment. I’m sure there’s a small efficiency loss.

    On the sell side, I either keep my appreciated shares (i.e. don’t sell) or donate them to charity (after holding them at least one year) in place of cash to get the most advantageous tax arbitrage.

    An additional note: if you’re doing TLH frequently and not just targeting major dips in the market, you need a constant inflow of funds and be constantly buying shares. As time marches on and the market moves ever upward, lots become less likely to have a lower cost basis. In your example, your $70k shares can only be used to score another loss if the price drops again below that low water mark. To unlock the money for more gains, you’ll either need to sell it and realize the gain. Otherwise it will be “locked up” with the lower cost basis. Hope that makes sense.

    • Thanks for stopping by!

      If you’re donating appreciated shares to charity, then TLH is indeed pure arbitrage because there is only benefits and no costs. Thanks again for teaching me about the 1Y holding period on gains.

      I agree that TLH’ing less frequently makes it less likely to find good opportunities since assets tend to rise. If not for the $3k/year limitation, I’d be actively pursuing this every single month. As is, however, I just can’t get that excited about the hassle unless it’s a huge loss. I like your “water mark” analogy.

  5. Thanks for another good post. I’ve always felt the hype of TLH is overblown, but mostly because the savings just don’t seem worth it for anyone in the lower tax brackets. For me, prioritizing tax-advantaged accounts gets me into the 12% zone. Best case scenario, I “save” $360/year. In order to pull off successful TLH, I would need to 1) remember to log into my brokerage more than once a year, 2) turn off dividend reinvestment, 3) adjust funds in retirement accounts (410k x2, IRA x2, HSA) to avoid wash sales, 4) likely sell off losses from many small lots (the downside of biweekly automatic investing?) after a big downturn. My fear is that taking things off of automatic, even for a short time, would cost me more than $360.

  6. Hey Professor,

    My position has been mentioned by others in comments here and on your Trad vs Roth post. It can be summarized as this principle: you make contributions at your *marginal* tax rate while working, and you make withdrawals at your *average* tax rate while retired.

    The first part is clear enough: when you contribute to a Traditional account, or do TLH, you realize an immediate benefit in the amount of your current marginal rate. This is up to certain limits: $6000 per year for your IRA; $3000 per year for TLH.

    The second part is more subtle. It comes from the fact that you fill up your entire “retirement savings bucket”, from empty to full, over the course of your working years. By definition, it will be withdrawn at the *average* rate in retirement: the first $XX at 0%, the next $YY at 12%, and yes, the last $1 at the marginal retirement rate.

    It is correct, strictly, to do a marginal vs. marginal analysis, for the *last* infinitesimal dollar you contribute to your retirement accounts, ever. But the holistic analysis — such as when giving advice to a young person just starting out — should be based on marginal vs. average.

    It took me a long of time to grasp this after some colleagues introduced me to the idea, but I think it’s a powerful way to talk about this. Hope it can be useful to you: maybe you’ll explain it better than I have to your readers! πŸ™‚

    • IP,

      Thanks for stopping by.

      I’m very sympathetic to the logic of marginal now vs average in retirement. In fact, my draft “book” illustrates this point ad nauseam (download available in menu at top of site).

      A few years after posting it, the legend MDM corrected me. His rebuke spawned this post:

      Although it took a lot of convincing, I’m now a believer in the arguments put forth by MDM. I like the assumption that social security income, pension income, & inheritance income, for example, fill up the lowest tax brackets first (0%, then 10%, then 12%, etc). Then, any income produced by trad 401k/IRA withdrawals is taxed at the marginal rate.

      I think the above framing is the most economically accurate way of thinking about things. It’s a subtle shift in framing to get from marginal now vs average later (the IP and former frugal professor belief) to the marginal now vs marginal (after social security & pension income) later, but I think it’s the most correct.

      Are you convinced of the MDM logic? Or do you think he (and now I) is wrong?

      • MDM is absolutely right. The refinement provided by the first commenter, Decius, and recognized in passing by MDM, is valuable too: the examples use a “greedy” algorithm in time, which prevents taking advantage of the variation in marginal rates during one’s working life. To make the solution truly optimal, I suppose you should sort the contribution years not chronologically, but in order of _decreasing_ marginal rates; and _then_ run the greedy algorithm, contributing to Trad in your highest-income working years.

        In any case, you’re totally right in general. I apologize for preaching about marginal-marginal/marginal-average, without checking if you’d already addressed the question directly. πŸ™‚

        Back to TLH, and to marginal rates. I sort of optimize for the worst case (not enough in retirement); so I tend to significantly discount my estimates of future income, and therefore tax. This is a matter of psychology, and probably one I should really work on. 😊 But the nice thing about gigantic unrealized capital gains is that it helps you both if you undershoot and if you overshoot. If you are “poor” in retirement, your LTCG taxes will be nil. If you are “rich” in retirement, you will end up bequeathing the shares to heirs, and they will benefit from the step-up in basis on death, and LTCG taxes will again be nil.

        As usual on this blog, it’s a safe assumption that just about everyone is in violent agreement. In particular, I want to affirm that I fully support the statement “Hell no” in regards to robo-advisors. πŸ™‚

        Happy Thanksgiving!

        • Happy belated Thanksgiving to you as well!

          It’s indeed interesting how complicated the Roth vs Trad debate really is, once you start peeling back the layers of the onion to consider the true economics. I’m glad there is another advocate of the MDM argument.

          I understand your approach to optimize for the worst cast scenario (not enough in retirement), but this will necessarily lead to sub-optimized decision making in the event that those scenarios don’t come to fruition. However, given the dire consequences of the worst-case coming to pass while not having optimized for it, I think your approach is understandable.

          In many respects, preparing for the worst case in the future is kind of how I budget. It also helps that I’m naturally frugal and can’t really think of any expenditure that I could make to make the family significantly more happy.

          Regarding the apparent consensus among what I write, I’m unsure what to make of it. It seems that readers self-select into blogs like this and tend to agree with what is written. Hopefully the echo-chamber isn’t harmful, but I truly love to have my ideas challenged as MDM did with my incorrect simplification of the Roth vs Trad decision.

  7. TLH in mid career changes the discount rate calculous. Use 3 or 4% (dont need the cash now and PV less of a concern) and the stakes change, no?

    • That’s an interesting idea.

      TO me, the discount rate should be a reflection of the opportunity cost of capital, which I interpret to be the expected return on investments. Given this framing, I’m unsure why a discount rate would change through the lifecycle of one’s working career.

      But I suppose one could also rationalize a behavioral reason to lower the discount rate. I think it’s more rational to not do so.


Leave a Reply