White Paper: Tax Loss Harvesting – ETFs v.s. Direct Indexing
Frec Direct Indexing aims to improve on traditional broad-market ETF investing by matching the return of these assets at a comparable fee, while offering additional tax-loss-harvesting (TLH) benefits. For a more detailed background on the direct indexing approach we invite the reader to review the Direct Indexing White Paper . In this report we consider an alternative comparison, specifically we consider the performance of automated TLH applied to a pair of similar ETFs in contrast with the Direct Indexing approach approach employed at Frec.
Summary of Results
In this section we summarize the main findings of this report, more details on these and other findings may be found in the sections below.
This report includes results from simulations over the period 12/17/2003 to 07/25/2023 for Direct Indexing into the S&P 500 and a ETF-based TLH algorithm using the IVV/SPY ETFs. The reasons for these choices and other details provided in following section. We consider Frec’s 0.1% annual fee in contrast with the 0.25% fee charged at popular robo-advisor platforms1.
These simulations indicate that the Direct Indexing approach allows the investor to harvest 1.9x more losses over a 10 year period, and that this ratio climbs to 2.1x when one considers the tax consequences of selling assets to pay fees.
We observe that deposits initiated during market lows can result in an ETF-based TLH approach not having a single opportunity to harvest losses, resulting in no tangible benefit to the investor. Further, when assets are sold to cover management fees the investor can incur a taxable gain in addition to the charged fee. For example, the figure below illustrates the time series of capital losses over a simulation period 07/27/2013 – 07/19/2023 for Direct Indexing and ETF-based TLH with the same (0.1%, 0.25%) fee structures applied. The observed capital gains (negative losses) are the result of the algorithm selling assets in order to pay fees.
Finally, in an attempt to put an actual dollar figure on the benefits of tax loss harvesting (and fees) one can simulate the reinvestment of tax savings into the portfolio. We conducted such simulations where capital losses are reinvested at an assumed tax rate of 42.3% and observed that the average excess return (over the SPY ETF) for Direct Indexing was 45.3% while the ETF-based approach had an average excess return of 17.3% over a 10 year period. Said differently, the average alpha (excess CAGR) over SPY was found to be 1.87% v.s. 0.84% respectively.
ETF-Based Tax Loss Harvesting
At a high-level the approach to harvesting tax losses using ETFs is generally straightforward; the algorithm monitors the cost basis of the held ETF tax lots relative to the market price at each trading opportunity. If any of these tax lots are trading below their cost basis, by a set threshold, then the lot is sold, the loss realized, and the proceeds are invested into a similar ETF. Provided, of course, that no wash sales are triggered by this trading action. Harvesting can occur between pairs of ETFs or larger sets (or even sets-of-sets) and can include other special considerations such as avoiding wash sale interaction with IRA accounts.
Swapping between similar ETFs does raise the question of how one can determine whether two similar ETFs are “substantially identical” for the purposes of satisfying wash sale rules. In this report, however, we simplify this issue by assuming a favorable assumption for the ETF-based harvesting approach: that a perfect substitute ETF may be swapped without “substantially identical” implications. Specifically, as we are interested in the S&P 500 index as a baseline, the ETFs used are SPY and IVV, both of which track this index and have sufficient historical data for the last 20 years.
The ETF-based algorithm implemented for this report is a simple one. Prices and lot cost basis values are monitored weekly, if a tax lot has fallen below a threshold value of 3% it will be sold to harvest a loss (respecting the 30-day wash sale window). Also, if there are no tax implications as a tie-breaker, the algorithm will prefer to hold the lower-fee ETF (IVV) with an expense ratio of 0.03%. Importantly, this algorithm is used for comparison purposes only and it is not claimed to simulate the performance as the algorithms used by popular robo-advisors or any other platform. For detailed information on more advanced tax loss harvesting with ETFs, and the many related considerations, we recommend reviewing the Tax Loss Harvesting+ white paper published by Betterment .
We evaluate the performance of the ETF-based TLH approach in contrast with direct indexing over multiple historical backtesting simulations, following standard experimental methodology (e.g. Moehle et al. ). Specifically simulations were conducted on overlapping 10 year periods, with starting dates every 90 days, using historical data spanning the period 12/17/2003 to 07/25/20232.
In each of the 40 simulations runs algorithms are executed on a weekly basis and the resulting trade actions are simulated using closing prices on the same day. Here we consider the scenario where there is a one-time deposit of $50,000 at the beginning of the simulation period. While TLH does generally does work best when regular deposits are used, the performance implications of each of these independent deposits can really be considered in isolation.
Importantly, two simulations were conducted for each approach, the first simulation does not assume any of the tax losses harvested are reinvested into the portfolio, while the second assumes that any tax losses can be yield tax savings as cash and are reinvested quarterly, an assumed tax rate of 42.3% is used. This tax rate represents an investor in the 95-98th percentile tax bracket as is described in the “Investor A” profile of Khang et al. . These two scenarios (with and without reinvestment) represent somewhat of a best-case and worst-case scenario when it comes to the investor’s ability to utilize tax losses.
The quality of any simulation is only as good as the underlying data used. We are particularly proud of the considerable efforts taken to produce high-quality simulations and continue to refine our simulations going forward. The dataset used for our simulations come from two sources, first from S&P Global we obtained the historical weighting used by the S&P 500 index on a daily basis going back to 2003. Additionally, we license dividend and split-adjusted daily stock prices from our primary data provider Xignite. Even with high quality data sources there were still dozens of data artifacts (e.g. mergers & acquisitions) and inaccuracies that had to be detected and corrected for, which was done manually by our engineering teams.
For each of the 40 simulation periods 4 separate TLH simulations were performed, ETF-based with and without reinvestment, and Direct Indexing with and without reinvestment. All simulations assume only a single deposit, of $50,000, initiated on day 1.
In order to develop an understanding of the experimental setup let’s consider an example, specifically the very first simulation run from 12/17/2003 to 12/14/2013 with no reinvestment of harvested losses.
The above figure illustrates the difference between the portfolio values for the Direct Indexing algorithm as well as the ETF-based algorithm in contrast to the SPY ETF. Considering that none of the tax losses are reinvested into the index it’s not surprising to see that the portfolios track the performance of the baseline (SPY) very closely. Both the TLH approaches slightly out-perform SPY, which is likely due to a mixture of variance and reduced fees. Note that management fees for the loss-harvesting algorithms are not included in the portfolio values above, we discuss fees later in this section.
Naturally, without reinvestment, any benefits of TLH are not reflected int the portfolio values shown above. So, for this setting, the losses that are harvested are assumed to be carried forward over the length of the simulation.
The figure above shows a time series of these carried-forward capital losses for the two approaches on the same simulation run. Here we can see the “all-or-none” loss-harvesting behavior for the ETF-based algorithm in contrast to the Direct Indexing approach in the first 4 years of simulation. Specifically, there are no losses harvested until the ETF drops below the 2003 cost basis during the market downturn in the 2008 financial crisis3. The Direct Indexing approach, on the other hand, is able to harvest losses from some of the constituents throughout the simulation period. Overall on this single simulation we can see that the Direct Indexing approach harvests losses at approximately a 2.4x higher rate.
We next consider how reinvesting the harvested losses quarterly at an assumed tax rate of 42.3% would affect the portfolio value over time. Again, this scenario is helpful in understanding somewhat of a “best-case” scenario with respect to an individual’s tax situation.
The above figure shows the portfolio value for the two TLH approaches for the same time period relative to the performance of a buy-and-hold investment into the SPY ETF. Here we can see the value of both harvesting more losses but also harvesting earlier in the time period leading to an additional $15.5k in portfolio returns by the end of the simulation. Interestingly, this reinvestment also creates more opportunities for loss harvesting as well, as illustrated in the following figure.
Here we can see that both algorithms are able to harvest more losses when reinvestment occurs, though the Direct Indexing approach is able to capture significantly more, $5,063 compared to $320 (relative to the no reinvest chart above). This is of course only a single example, in the following section we consider the results aggregated over all 40 simulation runs to present a more robust view of the differences between these approaches, particularly with respect to loss harvesting efficiency.
Multiple Simulation Results
We begin first with observations on the average amount losses harvested by these two approaches. Recall that these simulations consist of overlapping 10 year periods, with starting dates offset from each other by 90 days, overall data is used from the period from 12/17/2003 to 07/25/2023. Additionally, recall that each simulation includes a one-time $50,000 deposit on day 1, also that the simulations were repeated under 2 scenarios: reinvestment where tax savings from harvested losses are reinvested quarterly at 42.3% assumed tax rate and no-reinvestment where not reinvestment occurs.
The above figure shows average amount of harvested capital losses for each of these algorithms, here we find that in the reinvestment scenario Direct Index algorithm harvests an average of $22,035.50 (44.1% of deposit) in losses while the ETF-based algorithm harvests an average of $10,434.53 (20.9% of deposit). In the no-reinvestment scenario these losses are $19,223.35 (38.4%) for Direct Index and $10,098.77 (20.2%) ETF-based. The increase in harvested losses in the reinvestment case is likely because each $1 of new re-investment, in turn, yields an average of $0.38 in harvested losses down the line.
Said simply, the empirical data suggests that by utilizing a Direct Indexing approach one can expect to harvest 1.9x to 2.1x more capital loss than a comparable ETF-based TLH approach, depending on the level of reinvestment.
Breaking down the losses harvested by year for the no-reinvestment case (see above figure) we observe that the Direct Indexing approach is able to harvest losses for a greater period of time. This occurs even though no additional funds were deposited because each loss harvest opportunity will create new tax lots upon the reinvestment of the proceeds. This effect is further magnified when tax savings are reinvested.
Finally, in order to better understand the potential upside of Direct Indexing and ETF-based tax loss harvesting we can compare the excess returns (portfolio growth – benchmark growth) relative to the SPY ETF benchmark in the reinvestment scenario.
The above figure illustrates how reinvesting the tax savings from harvested losses can lead to increased portfolio returns. Here, the Direct Indexing portfolio has an average excess return of 51.5% (average excess CAGR (a.k.a. tax alpha) of 2.02%)5 while the ETF-based portfolio has an average excess return of 27.2% (1.15% average excess CAGR). While each investor’s tax situation will affect how much of this excess return can be realized in practice, the relative performance between Direct Indexing and and ETF-based strategies should remain relatively consistent.
Tax Implications of Withdrawals and Management Fees
One major difference between an ETF and a Direct Index portfolio relates to the flexibility over what assets may be sold to cover withdrawals. Because there are more stocks to choose from, and the loss harvesting activity is more frequent, simulations indicate that the Direct Index portfolios are able to sell stock for cash with reduced tax implications. This detail becomes important an important point of comparison as popular robo-advisor platforms charge a management fee of at least 0.25% for ETF-based tax loss harvesting1.
The above figure illustrates the impact of fees on the total losses harvested for Direct Indexing with the 0.1% fee charged by Frec in contrast to ETF-based harvesting at a 0.25%, for the same simulations described above. Here we observe that for the Direct Indexing approach withdrawals to cover fees has a negligible effect on the total losses — i.e. incurs minimal capital gain — as net losses drop from 38.5% to 38.3% of the initial deposit ($19,223 in net harvested losses to $19,129). For the ETF-based approach the gains incurred by selling stock result in drop in the loss harvesting rate of 20.2% to 18.0%.
Of course, the tax implications of selling to cover the fee are less pronounced than the cumulative dollar amount of the fee itself. We can consider this dollar impact by looking at average excess return of the portfolio under the aforementioned tax-saving reinvestment scenario using an assumed tax rate of 42.3%.
The above figure illustrates the difference between the excess returns (portfolio growth – benchmark growth) over SPY for the different alternatives. Here we can see that the low-cost ETF IVV, expense ratio of 0.03%, slightly out-performs the SPY ETF having an expense ratio of 0.09%. However, these relatively modest improvements are significantly out-paced by the tax loss harvesting algorithms as tax savings are reinvested. In these simulations the average excess return for Direct Indexing was 45.3%, average excess CAGR (alpha) of 1.87%, while the ETF-based approach had an average excess return of 17.3%, average excess CAGR of 0.84%.
Historical backtesting simulations indicate quite conclusively that Direct Indexing can out-perform ETF-based tax loss harvesting by a factor of 1.9x – 2.1x depending on the investor’s ability to reinvest tax savings. Additionally, simulation results suggests that the Direct Indexing approach can allow more tax-efficient withdrawals from the account, which can be a significant factor when management fees are applied. When contrasting the 0.1% fee offered by Frec Direct Indexing with the higher 0.25% fee charged by many popular robo-advisors these advantages may add up to a significantly higher portfolio value over time.
 Direct Indexing White Paper, Frec Inc. https://frec.com/resources/blog/frec-direct-indexing-algorithm, Accessed 17 Oct. 2023.
 Tax Loss Harvesting+ Methodology, Boris Khentov, Jun. 27, 2023, https://www.betterment.com/resources/tax-loss-harvesting-methodology, Accessed 17 Oct. 2023.
 Khang, Kevin and Cummings, Alan and Paradise, Thomas, and O’Connor, Brennan, Personalized Indexing: A Portfolio Construction Plan, (March 2022), https://corporate.vanguard.com/content/dam/corp/research/pdf/personalized_indexing_a_portfolio_construction_plan.pdf
 Moehle, N., Kochenderfer, M.J., Boyd, S. et al. Tax-Aware Portfolio Construction via Convex Optimization. J Optim Theory Appl 189, 364–383 (2021). https://doi.org/10.1007/s10957-021-01823-0
- At the time of writing the simulations in this report are very similar to those conducted for the Direct Indexing White Paper  but benefit from the inclusion of more data into 2023 as well as slightly longer simulation periods.
- Known as the “Great Recession”
- Past performance is not a guarantee of future results. These results are hypothetical, do not reflect actual investment results, and are not a guarantee of future results.
- Here the average CAGR is computed over the 10-year period, but the reader should be aware that TLH effectiveness is highest in the early years and gradually diminishes over time. As a result, the “tax alpha” metric will change significantly depending on the time period used in the analysis. An important consideration when comparing published results across different providers.
This white paper describes implementation and performance details of a Direct Indexing approach similar to that used on the Frec platform at the time of writing (10/17/2023) details may differ from the implementation used in the product now and in future.
This white paper is for information purposes only and is not intended as tax advice. Frec refers to Frec Markets, Inc. and its wholly owned subsidiaries, Frec Advisers LLC and Frec Securities LLC. Frec does not provide legal or tax advice and does not assume any liability for the tax consequences of any client transaction. Clients should consult with their personal tax advisors regarding the tax consequences of investing with Frec and engaging in these tax strategies, based on their particular circumstances. Clients and their personal tax advisors are responsible for how the transactions conducted in an account are reported to the IRS or any other taxing authority on the investor’s personal tax returns. Frec assumes no responsibility for tax consequences to any investor of any transaction.
The S&P 500© index is a product of S&P Dow Jones Indices LLC or its affiliates (“SPDJI”), and has been licensed for use by Frec Markets, Inc. S&P®, S&P 500®, US 500 and The 500 are trademarks of Standard & Poor’s Financial Services LLC (“S&P”); and these trademarks have been licensed for use by SPDJI and sublicensed for certain purposes by Frec Markets, Inc. Frec’s direct indexing strategy is not sponsored, endorsed, sold or promoted by SPDJI, Dow Jones, S&P, their respective affiliates and none of such parties make any representation regarding the advisability of investing in such product(s) nor do they have any liability for any errors, omissions, or interruptions of the S&P 500© index.
The effectiveness of Frec’s tax-loss harvesting strategy to reduce the tax liability of the client will depend on the client’s entire tax and investment profile, including purchases and dispositions in a client’s (or client’s spouse’s) accounts outside of Frec, the type of investments (e.g., taxable or nontaxable) or holding period (e.g., short-term or long-term. The performance of the new securities purchases through the tax-loss harvesting service may be better or or worse than the performance of the securities that are sold for tax-loss harvesting purposes.