• KEY TAKEAWAYS
  • Tax loss harvesting can reduce your taxes.

  • TLH+ is twice as effective as available automated loss harvesting strategies.

  • Betterment’s best-in-class system avoids pitfalls found in existing tax loss harvesting strategies in order to maximize your total tax benefit.

Introduction

Summary: Tax loss harvesting is a sophisticated technique to get more value from your investments—but doing it well requires expertise.

There are many ways to get your investments to work harder for you—better diversification, downside risk management, and the right mix of asset classes for your risk level. Betterment does all of this automatically via its low-cost index fund ETF portfolio.

But there is another way to get even more out of your portfolio—using investment losses to improve your after-tax returns with a method called tax loss harvesting. In this white paper, we introduce Betterment’s new Tax Loss Harvesting+™ (TLH+™): a sophisticated, fully automated service for Betterment customers.

Betterment’s new TLH+ service scans portfolios every day for opportunities (temporary dips that result from market volatility) to realize losses which can be valuable come tax time. While the concept of tax loss harvesting is not new for wealthy investors, TLH+ utilizes a number of innovations that typical implementations lack. It takes a holistic approach to tax-efficiency, optimizing every user-initiated transaction in addition to adding value through automated activity, such as rebalances.

As a best-in-class service, TLH+ not only improves on this powerful tax-saving strategy—over the last 13 years, it would have provided an estimated 0.77% to a typical customer’s after-tax returns, annually—but also makes tax loss harvesting available to more investors than ever before.

Backtest of TLH+

What is tax loss harvesting?

Capital losses can lower your tax bill by offsetting gains, but the only way to realize a loss is to sell the depreciated asset. However, in a well-allocated portfolio, each asset plays an essential role in providing a piece of total market exposure. For that reason, an investor should not want to give up the expected returns associated with each asset just to realize a loss.

At its most basic level, tax loss harvesting is selling a security that has experienced a loss—and then buying a correlated asset (i.e. one that provides similar exposure) to replace it. The strategy has two benefits: it allows the investor to “harvest” a valuable loss, and it keeps the portfolio balanced at the desired allocation.

How does it lower your tax bill?

Capital losses can be used to offset capital gains you’ve realized in other transactions over the course of a year—gains on which you would otherwise owe tax. Then, if there are losses left over (or if there were no gains to offset), you can offset up to $3,000 of ordinary income for the year. If any losses still remain, they can be carried forward indefinitely (see here for more on how these rules apply).

Tax loss harvesting is primarily a tax deferral strategy, and its benefit depends entirely on individual circumstances. Over the long run, it can add value through some combination of these distinct benefits:

  • Tax deferral: Losses harvested can be used to offset unavoidable gains in the portfolio, or capital gains elsewhere (e.g., from selling real estate), deferring the tax owed. Savings should be invested— assuming a conservative growth rate of 5% over a 10-year period, a dollar of tax deferred would be worth $1.63. Even after belatedly parting with the dollar, and paying tax on the $0.63 of growth, you’re ahead.
  • Pushing capital gains into a lower tax rate: If you’ve realized short-term capital gains (STCG) this year, they’ll generally be taxed at your highest rate. However, if you’ve harvested losses to offset them, the corresponding gain you owe in the future could be long-term capital gain (LTCG). You’ve effectively turned a gain that would have been taxed up to 50% today into a gain that will be taxed more lightly in the future (up to 30%).
  • Converting ordinary income into long-term capital gains: A variation on the above: offsetting up to $3,000 from your ordinary income shields that amount from your top marginal rate, but the offsetting future gain will likely be taxed at the LTCG rate.
  • Permanent tax avoidance: Tax loss harvesting provides benefits now in exchange for increasing built-in gains, subject to tax later. However, under certain circumstances (charitable donation, bequest to heirs), these gains can avoid taxation entirely.

Navigating the Wash Sale Rule

Summary: Wash sale rule management is at the core of any tax loss harvesting strategy. Unsophisticated approaches can detract from the value of the harvest or place constraints on customer cash flows in order to function.

If all it takes to realize a loss is to sell a security, it would seem that maintaining your asset allocation is as simple as immediately repurchasing it. However, the IRS limits a taxpayer’s ability to deduct a loss when it deems the transaction to have been without substance.

At a high level, the so-called “wash sale rule” disallows a loss from selling a security if a “substantially identical” security is purchased 30 days after or before the sale. The rationale is that a taxpayer should not enjoy the benefit of deducting a loss if he did not truly dispose of the security.

The wash sale rule applies not just to situations when a “substantially identical” purchase is made in the same account, but also when the purchase is made in the individual’s IRA account, or even in a spouse’s account. This broad application of the wash sale rule seeks to ensure that investors cannot utilize nominally different accounts to maintain their ownership, and still benefit from the loss.

A wash sale involving an IRA account is particularly unfavorable. Generally, a “washed” loss is postponed until the replacement is sold, but if the replacement is purchased in an IRA account, the loss is permanently disallowed.

If not managed correctly, wash sales can undermine tax loss harvesting. Handling proceeds from the harvest is not the sole concern—any deposits made in the following 30 days (whether into the same account, or into the individual’s IRA) also need to be allocated with care.

Avoiding the wash

The simplest way to avoid triggering a wash sale is to avoid purchasing any security at all for the 30 days following the harvest, keeping the proceeds (and any inflows during that period) in cash. This approach, however, would systematically keep a portion of the portfolio out of the market. Over the long term, this “cash drag” would hurt the portfolio’s performance.

More advanced strategies repurchase an asset with similar exposure to the harvested security that is not “substantially identical” for purposes of the wash sale rule. In the case of an individual stock, it is clear that repurchasing stock of that same company would violate the rule. Less clear is the treatment of two index funds from different issuers (e.g., Vanguard and Schwab) that track the same index. While the IRS has not issued any guidance to suggest that such two funds are “substantially identical,” a more conservative approach when dealing with an index fund portfolio would be to repurchase a fund whose performance correlates closely with that of the harvested fund, but tracks a different index.

Selecting a viable replacement security, however, is just one piece of the accounting and optimization puzzle. Manually implementing a tax loss harvesting strategy is feasible with a handful of securities, little to no cash flows, and infrequent harvests (once a year is common for DIY practitioners). However, assets will often dip in value but recover by the end of the year, so annual strategies leave many losses on the table. The wash sale management and tax lot accounting necessary to support more frequent (and thus more effective) harvesting quickly become overwhelming in a multi-asset portfolio—especially with regular deposits, dividends, and rebalancing.

Software is ideally suited for this complex task. But automation, while necessary, is not sufficient. The problem can get so complex that basic tax loss harvesting algorithms may choose to keep new deposits and dividends in cash for the 30 days following a harvest, rather than tackle the challenge of always maintaining full exposure at the desired allocation.

A best-in-class tax loss harvesting algorithm should be able maximize harvesting opportunities across a full range of volatility scenarios, without sacrificing the investor’s precisely tuned global asset allocation. It should reinvest harvest proceeds into closely correlated alternate assets, all while handling unforeseen cash inflows from the investor without ever resorting to cash positions. And most of all, it should do everything to avoid leaving a taxpayer worse off. TLH+ was created because no available implementations solved all of these problems.

Existing strategies and their limitations

Every tax loss harvesting strategy shares the same basic goal: to maximize a portfolio’s after-tax returns by realizing built-in losses while minimizing the negative impact of wash sales.

Approaches to tax loss harvesting differ primarily in how they handle the proceeds of the harvest to avoid a wash sale. Below are the three strategies commonly employed by manual and algorithmic implementations.

After selling a security that has experienced a loss, existing strategies would have you…

Existing strategy Problem
Delay reinvesting the proceeds of a harvest for 30 days, thereby ensuring that the repurchase will not trigger a wash sale. While it’s the easiest method to implement, it has a major drawback: no market exposure—also called cash drag. Cash drag hurts portfolio returns over the long term, and could offset any potential benefit from tax loss harvesting.
Reallocate the cash into one or more entirely different asset classes in the portfolio. This method throws off an investor’s desired asset allocation. Additionally, such purchases may block other harvests over the next 30 days by setting up potential wash sales in those other asset classes.
Switch back to original security after 30 days from the replacement security. Common manual approach, also used by some automated investing services. A switchback can trigger short-term capital gains when selling the replacement security, reducing the tax benefit of the harvest. Even worse, this strategy can leave an investor owing more tax than if it did nothing.
The hazards of switchbacks

In the 30 days leading up to the switchback, two things can happen: the replacement security can drop further, or go up. If it goes down, the switchback will realize an additional loss. However, if it goes up, which is what any asset with a positive expected return is expected to do over any given period, the switchback will realize short-term capital gains (STCG)—kryptonite to a tax-efficient portfolio management strategy.

To be sure, the harvested loss should offset all (or at least some) of this subsequent gain, but it is easy to see that the result is a lower-yielding harvest. Like a faulty valve, this mechanism pumps out tax benefit, only to let some of it leak back. An algorithm that expects to switch back unconditionally must deal with the possibility that the resulting gain could exceed the harvested loss, leaving the taxpayer worse off.

An attempt to mitigate this risk could be setting a higher threshold based on volatility of the asset class—only harvesting when the loss is so deep, that the asset is unlikely to entirely recover in 30 days. Of course, there is still no guarantee that it will not, and the price paid for this buffer is that your lower-yielding harvests will also be less frequent than they could be with a more sophisticated strategy.

Examples of negative tax arbitrage

Let’s look at how an automatic 30-day switchback can destroy the value of the harvested loss, and even increase tax owed, rather than reduce it.

Below is actual performance for Emerging Markets—a relatively volatile asset class that’s expected to offer some of the biggest harvesting opportunities in a global portfolio. We assume a position in VWO, purchased prior to January 1, 2013. With no harvesting, it would have looked like this:

No Tax Loss Harvesting

Emerging Markets, 1/2/2014 – 5/21/2014

A substantial drop in February presented an excellent opportunity to sell the entire position and harvest a $331 long-term loss. The proceeds were re-invested into IEMG, a highly correlated replacement (tracking a different index). By March, however, the dip proved temporary, and 30 days after the sale, the asset class more than recovered. The switchback sale results in STCG in excess of the loss that was harvested, and actually leaves the investor owing tax, whereas without the harvest, he would have owed nothing.

TLH with 30-day Switchbacks

Emerging Markets, 1/2/2014 – 5/21/2014

Under certain circumstances, it can get even worse. Due to a technical nuance in the way gains and losses are netted, the 30-day switchback can result in negative tax arbitrage, by effectively pushing existing gains into a higher tax rate.

When adding up gains and losses for the year, the rules require netting of like against like first. If any long-term capital gain (LTCG) is present for the year, you must net a long-term capital loss (LTCL) against that first, and only then against any STCG. In the scenario above, the harvested $331 LTCL was used to offset the $360 STCG from the switchback; long can be used to offset short, if we assume no LTCG for the year.

Negative tax arbitrage when unrelated long-term gains are present

Now let’s assume that in addition to the transactions above, the taxpayer also realized a LTCG of exactly $331 (from selling some other, unrelated asset). If no harvest takes place, the investor will owe tax on $331 at the lower LTCG rate. However, if you add the harvest and switchback trades, the rules now require that the harvested $331 LTCL is applied first against the unrelated $331 LTCG. The harvested LTCL gets used up entirely, exposing the entire $360 STCG from the switchback as taxable. Instead of sheltering the highly taxed gain on the switchback, the harvested loss got used up sheltering a lower-taxed gain, creating far greater tax liability than if no harvest had taken place.

Tax Strategy STCG Realized LTCG Realized Taxes Owed
No TLH $0 $331 $109
TLH with 30-day switchbacks $360

($331-$331)

$0 net

$187

In the presence of unrelated transactions, unsophisticated harvesting can effectively convert existing LTCG into STCG. Some investors regularly generate significant LTCG (for instance, by gradually diversifying out of a highly appreciated position in a single stock). It’s these investors, in fact, that would benefit the most from effective tax loss harvesting. However, if their portfolios are harvested with unconditional 30-day switchbacks over the years, it’s not a question of “if” the switchbacks will convert some LTCG into STCG, but “when” and “how much.”

Negative tax arbitrage with dividends

Yet another instance of negative tax arbitrage can result in connection with dividend payments. If certain conditions are met, some ETF distributions are treated as “qualified dividends”, taxed at lower rates. One condition is holding the security for more than 60 days. If the dividend is paid while the position is in the replacement security, it will not get this favorable treatment: under a rigid 30-day switchback, the condition can never be met. As a result, up to 20% of the dividend is lost to tax (the difference between the higher and lower rate).


The Betterment Solution

Summary: TLH+ substantially improves on existing strategies by managing parallel positions within each asset class indefinitely, as tax considerations dictate. It approaches tax-efficiency holistically, optimizing every transaction, including customer activity.

The fundamental advance of Betterment’s TLH+ is that tax-optimal decision making should not be limited to the harvest itself—the algorithm must remain vigilant with respect to every transaction.

An unconditional 30-day switchback, whatever the cost, is plainly suboptimal, and could even leave the investor owing more tax that year—unacceptable for an automated strategy that seeks to lower tax liability. Intelligently managing a bifurcated asset class following the harvest is every bit as crucial to maximizing tax alpha as the harvest itself.

The innovations of TLH+ include:

  • No exposure to short-term capital gains in an attempt to harvest losses. Through our proprietary Parallel Position Management system, a dual-security asset class approach enforces preference for one security without needlessly triggering capital gains in an attempt to harvest losses, all without putting constraints on customer cash flows.
  • No negative tax arbitrage traps associated with less sophisticated harvesting strategies (e.g., 30-day switchback), making TLH+ especially suited for those generating large long-term capital gains on an ongoing basis.
  • Zero cash drag at all times. With fractional shares and seamless handling of all inflows during wash sale windows, every dollar is invested at the desired allocation risk level.
  • Tax loss preservation logic extended to user-realized losses, not just harvested losses, automatically protecting both from the wash sale rule. In short, user withdrawals always sell any losses first.
  • No disallowed losses through overlap with a Betterment IRA. We use a tertiary ticker system to eliminate the possibility of permanently disallowed losses triggered by subsequent IRA activity. This makes TLH+ ideal for those who invest in both taxable and tax-advantaged accounts.
  • Harvests also take the opportunity to rebalance across all asset classes, rather than re-invest solely within the same asset class. This further reduces the need to rebalance during volatile stretches, which means fewer realized gains, and higher tax alpha.

Through these innovations, TLH+ creates significant value over manually-serviced or less sophisticated algorithmic implementations. TLH+ is accessible to more investors than any other harvesting service—fully automated, more effective, and at no additional cost.

Parallel securities

To ensure that each asset class is supported by optimal securities in both primary and alternate positions, we screened by expense ratio, liquidity (bid-ask spread), tracking error vs. benchmark, and most importantly, covariance of the alternate with the primary.¹

With those four measures in place, we assumed annual portfolio turnover of 10% and aggregated them into a single number representing the total annual cost of ownership.

expense-ratios-03

While there are small cost differences between the primary and alternate securities, the cost of negative tax arbitrage from tax-agnostic switching vastly outweighs the cost of maintaining a dual position within an asset class.

For a 70% stock portfolio composed only of primary securities, the average underlying expense ratio is 0.136%. If each asset class consisted of a 50/50 split between primary and alternate, that cost would be 0.152%—a difference of less than two basis points.

Of the 12 asset classes in Betterment’s core taxable portfolio, nine were assigned alternate tickers. Short-term Treasuries (SHV) and Inflation-protected Bonds (VTIP) are insufficiently volatile to be viable harvesting candidates (and disappear entirely above a 57% stock allocation). There is currently no suitable alternative available for International Bonds (BNDX).

Below is a look at the primary and alternate securities in the Betterment portfolio.

Additionally, TLH+ features a special mechanism for coordination with IRAs that required us to pick a third security in each harvestable asset class (except in municipal bonds, which are not in the IRA portfolio). While these have a higher cost than the primary and alternate, they are not expected to be utilized often, and even then, for short durations (more below in IRA protection).

Parallel Position Management

As demonstrated, the unconditional 30-day switchback to the primary security is problematic for a number of reasons. To fix those problems, we engineered a platform to support TLH+, which tax-optimizes every user and system-initiated transaction: the Parallel Position Management (PPM) system.

PPM allows each asset class to be comprised of two closely correlated securities indefinitely, should that result in a better after-tax outcome. Here’s how a portfolio with PPM looks to a Betterment customer.

portfolio-tab-dual-tickers-01

PPM provides several improvements over the switchback strategy. First, unnecessary gains are avoided. Second, the parallel security (could be primary or alternate) serves as a safe harbor to minimize wash sales—not just from harvest proceeds, but any cash inflows. Third, the mechanism serves to protect not just harvested losses, but losses realized by the customer as well.

PPM not only facilitates effective, daily tax loss harvesting, but also extends maximum tax-efficiency to customer-initiated transactions. Every customer withdrawal is a potential harvest (losses are sold first). And every customer deposit and dividend is routed to the parallel position that would minimize wash sales, while shoring up the target allocation.

PPM has a preference for the primary security when rebalancing and for all cash flow events—but always subject to tax considerations. This is how PPM behaves under various conditions:

Transaction PPM behavior
Withdrawals and sales from rebalancing Sales default out of the alternate position (if such a position exists), but not at the expense of triggering STCG—in that case, PPM will sell lots of the primary security first. Rebalancing will always stop short of realizing STCG. Taxable gains are minimized at every decision point—STCG tax lots are the last to be sold on a user withdrawal.
Deposits, buys from rebalancing, and dividend reinvestments PPM directs inflows to underweight asset classes, and within each asset class, into the primary, unless doing so incurs greater wash sale costs than buying the alternate.
Harvest events TLH+ harvests can come out of the primary into the alternate, or vice versa, depending on which harvest has a greater expected value. After an initial harvest, it could make sense at some point to harvest back into the primary, to harvest more of the remaining primary into the alternate, or to do nothing. Harvests that would cause more washed losses than gained losses are avoided.

PPM eliminates the negative tax arbitrage issues previously discussed, while leaving open significantly more flexibility to engage in harvesting opportunities. TLH+ is able to harvest more frequently, and can safely realize smaller losses, all without putting any constraints on user cash flows.

Let’s return to the Emerging Markets example from above, demonstrating how TLH+ harvests the loss, but remains in the appreciated alternate position (IEMG), thereby avoiding STCG.

Smarter Harvesting – Avoid the Switchback

Emerging Markets, 1/2/2014 – 5/21/2014

Better wash sale management

Managing cash flows across both taxable and IRA accounts without needlessly washing realized losses is a complex problem. Some automated tax loss harvesting services recommend that their customers make infrequent deposits (e.g., quarterly), since more frequent schedules would render their algorithms ineffective.

TLH+ operates without constraining the way that customers prefer contributing to their portfolios, and without resorting to cash positions. With the benefit of parallel positions, it weighs wash sale implications of every deposit and withdrawal and dividend reinvestment, and systematically chooses the best investment strategy. This system protects not just harvested losses, but also losses realized through withdrawals. When appropriate, TLH+ even optimizes deposits prior to a harvest, potentially allocating to alternate assets in anticipation of harvesting.

Betterment customers need never worry how their transactions might impact activity generated by the automated daily harvest run. A customer can be assured that each bi-weekly auto-deposit (timed with a pay period, for example) will be diversified across 12 asset classes, and will not preclude harvesting.

IRA protection

The wash sale rule applies when a “substantially identical” replacement is purchased in an IRA account. Taxpayers must calculate such wash sales, but brokers are not required to report them. Even the largest ones leave this task to their customers.² This is administratively complicated for taxpayers and lead to tax issues.

At Betterment, we felt we could do more than the bare minimum. Being far better equipped to perform this calculation, we do it so that our customers don’t have to. Since IRA wash sales are particularly unfavorable—the loss is disallowed permanently—TLH+ goes another step further, and ensures that no loss realized in the taxable account will ever be washed by a subsequent deposit into a Betterment IRA. In doing so, TLH+ always maintains target allocation in the IRA, without cash drag.

No harvesting is done in an IRA, so if it weren’t for the potential interaction with taxable transactions, there would be no need for IRA alternate securities. However, on the day of an IRA inflow, both the primary and the alternate security in the taxable account could have realized losses in the prior 30 days. Accordingly, each asset class supports a third correlated security (tracking a third index). The tertiary security is only utilized to hold IRA deposits within the wash window temporarily.

Let’s look at an example of how TLH+ handles a potentially disruptive IRA inflow when there are realized losses to protect, using real market data for the Developed Markets asset class.

The customer starts with a position in VEA, the primary security, in both the taxable and IRA accounts. We then harvest a loss by selling the entire taxable position, and repurchase the alternate security, SCHF.

Loss Harvested in VEA

ira-alternates-slide1-02

Two weeks pass, and the customer makes a withdrawal from the taxable account (the entire position, for simplicity), intending to fund the IRA. In those two weeks, the asset class dropped more, so the sale of SCHF also realizes a loss. The VEA position in the IRA remains unchanged.

Customer Withdrawal Sells SCHF at a Loss

ira-alternates-slide2-02

A few days later, the customer contributes to his IRA, and $1,000 is allocated to the Developed Markets asset class, which already contains some VEA. Despite the fact that the customer no longer holds any VEA or SCHF in his taxable account, buying either one in the IRA would permanently wash a valuable realized loss. The Tertiary Ticker System automatically allocates the inflow into the third option for developed markets, IEFA.

IRA Deposit into Tertiary Ticker

ira-alternates-slide3-03

Both losses have been preserved, and the customer now holds VEA and IEFA in his IRA, maintaining desired allocation at all times. Because no capital gains are realized in an IRA, there is no harm in switching out of the IEFA position and consolidating the entire asset class in VEA when there is no danger of a wash sale.

The result: Customers using TLH+ who also have their IRA assets with Betterment can rest assured that no deposit pattern into their IRA (lump rollover, auto-deposits or even dividend reinvestments) will ever endanger valuable realized losses—they are automatically protected.

Smart rebalancing

Lastly, TLH+ directs the proceeds of every harvest to rebalance the entire portfolio, the same way that a Betterment account handles any incoming cash flow (deposit, dividend). Most of the cash is expected to stay in that asset class and be reinvested into the parallel asset, but some of it may not. Recognizing every harvest as a rebalancing opportunity further reduces the need for additional selling in times of volatility, further reducing tax liability. As always, fractional shares allow the inflows to be allocated with perfect precision.


TLH+ Model Calibration

Summary: To make harvesting decisions, TLH+ optimizes around multiple inputs, derived from rigorous Monte Carlo simulations.

The decision to harvest is made when the benefit, net of cost, exceeds a certain threshold. The potential benefit of a harvest is discussed in detail below (“Results”). Unlike a 30-day switchback strategy, TLH+ does not incur the expected STCG cost of the switchback trade. Therefore, “cost” consists of three components: trading expense, execution expense, and increased cost of ownership for the replacement asset (if any).

All trades are free for Betterment customers. TLH+ is engineered to factor in the other two components, configurable at the asset level, and the resulting cost approaches negligible. Bid-ask spreads for the bulk of harvestable assets are extremely narrow, and Betterment’s vertically integrated trading platform is able to net out a substantial amount of customer trades, often avoiding the bid-ask spread entirely. Expense ratios for the major primary/alternate ETF pairs are extremely close, and in the case where a harvest back to the primary ticker is being evaluated, that difference is actually a benefit, not a cost.

A harder cost to quantify could result from what can be thought of as an “overly itchy TLH trigger finger.” Without the STCG switchback limitation, even very small losses appear to be worth harvesting, but only in a vacuum. Harvesting a loss “too early” could mean passing up a bigger (temporary) loss—made unavailable due to wash sale constraints stemming from the first harvest. Optimizing the thresholds to maximize loss yield becomes a statistical problem, ripe for an exhaustive simulation.

There are two general approaches to testing a model’s performance: historical backtesting and forward-looking simulation. Optimizing a system to deliver the best results for only past historical periods is relatively trivial, but doing so would be a classic instance of data snooping bias. The maturation of the global ETF market is a relatively recent phenomenon. Relying solely on a historical backtest of a portfolio composed of ETFs that allow for 10 to 20 years of reliable data when designing a system intended to provide 40 to 50 years of benefit would mean making a number of indefensible assumptions about general market behavior.

The superset of decision variables driving TLH+ is beyond the scope of this paper—optimizing around these variables required exhaustive analysis. TLH+ was calibrated via Betterment’s rigorous Monte Carlo simulation framework, spinning up thousands of server instances in the cloud to run through tens of thousands of forward-looking scenarios testing model performance.


TLH+ Results

Summary: In backtesting performance between 2000 and 2013, we found Betterment’s TLH+ produced twice as many tax offsets as a commonly used 30-day switchback strategy.

While forward-looking simulations are a far better approach to design and calibration, pure backtesting allows us to observe model performance during familiar market conditions.

The precise value of harvesting depends on many variables that are unknown at the time of the trade (and could remain unknown for decades)—in particular, assumptions about liquidation or other disposition of assets in the portfolio. Other key assumptions include when the loss will be used, and which type of gain the loss will offset, since in the absence of matching gains, a short-term loss could offset a long-term gain, and vice versa.

Tax alpha and IRR

It is relatively simple to evaluate the immediate benefit of tax loss harvesting for a single year, ignoring any offsetting future costs. The task becomes more difficult when a number of these years are strung together, and when subsequent events (i.e. taxable liquidation) significantly impact overall performance. A measure called the Internal Rate of Return (IRR) is well-suited for the task, in particular when assuming ongoing deposits into the portfolio. To properly measure tax alpha, we used IRR to calculate the excess return that a TLH+ portfolio would have generated relative to the baseline Betterment portfolio over the last 13 years.

For both portfolios, we assumed a constant 70% stock allocation, an initial $50,000 investment on 1/1/2000, and twice monthly auto-deposits of $750, escalated 5% annually (to account for inflation and salary growth). All dividends were reinvested to rebalance the portfolio (buying underweight assets rather than all assets pro-rata) and taxed under the current rules over the entire period. For tax rates, we assume a single California resident (where Betterment has the most customers) making $100,000/year (federal: 28% on income, 15% on LTCG; state: 9.3%).

Savvy investors avoid STCG, and Betterment will never trigger it on behalf of customers, so we assumed that there is never any STCG to offset. Instead, we assume that harvested losses are offset against ordinary income up to $3,000, and the excess, if any, against LTCG outside the portfolio. This mimics a typical scenario where the investor waits until the end of the year, when he knows exactly how much LTCG he can realize tax-free, and still leave enough losses to fully utilize the more valuable ordinary income offset (for example, if harvested losses add up to $8,000 by year-end, he would trigger $5,000 in LTCG). All tax savings were reinvested into the portfolio the following year.

Finally, to factor in liquidation of embedded gains, we calculated IRR for three scenarios: (1) full liquidation on 1/1/2014, (2) liquidation of 50% of the portfolio on 1/1/2014, and (3) no liquidation. Note that the comparison is like-for-like: the non-TLH+ portfolio is also liquidated, and taxes paid.

Even with full liquidation, TLH+ would have delivered tax alpha of 0.62%.

Tax alpha for investor earning $100,000 (2000-13)

Tax loss harvesting

A few words on top of standard disclosure are in order. No reliable data was available prior to 2000 that could adequately represent the performance of the full Betterment portfolio, constraining our ability to backtest over a longer period. The annualized results benefit from the fact that much of the value of harvesting came in the first few years of the period (the bear market of 2000-02). That allowed maximum time for the reinvested savings to compound, thus maximizing the benefit of tax deferral. The next 13 years may look nothing like the last 13.

On the other hand, 13 years is brief compared to a lifetime of tax deferral—an investor in his thirties can reasonably expect his tax savings to compound for 30 years before liquidating. To strike a balance, we believe that a tax alpha of 0.77% (50% liquidation after 13 years) is a reasonable estimate of what TLH+ can deliver to a typical Betterment customer.

These assumptions apply to many, but not all Betterment customers. Higher earners will see more benefit. To further demonstrate how tax loss harvesting adds value, we changed just one assumption: using the maximum California tax rates (federal: 39.6% on income, 23.8% on LTCG; state: 12.3%). As expected, the tax alpha under all liquidation scenarios jumps significantly.

Tax alpha for investor earning $500,000+ (2000-13)

Tax loss harvesting

While passive investors should be able to avoid STCG entirely, deferring these highly taxed gains with harvested losses is especially valuable. To layer on this benefit, we ran the backtest again, this time assuming that no matter how much STCL is harvested in a given year, there is always sufficient STCG available for offset. Tax alpha climbs higher, reaching nearly 1% with full liquidation.

Tax alpha for investor earning $500,000+, with STCG available (2000-13)

Tax loss harvesting

Annual Tax Offsets

While IRR is the correct way to measure after-tax returns over a multi-year period, it can be useful to examine how a tax loss harvesting strategy behaved in a given year, ignoring the uncertainty of liquidation.

The expected value of a loss is calculated by multiplying the amount of the loss by either the short-term or long-term rate, whichever is applicable. This approximates the direct tax savings from the loss in the year it is harvested, without accounting for the future liability that the harvest embeds into the portfolio. The total tax offset for the year’s harvesting activity is calculated via this formula:

tax_alpha_calculation-642

Assuming the maximum California tax rates, and the same initial deposit and auto-deposit schedule as above, we calculated the annual tax offsets that a Betterment portfolio with TLH+ would have generated over the last 13 years.

Annual Tax Offsets

TLH+ was able to provide positive tax offsets in 11 out of 13 years, though as one would expect, most of the value came during the years of the 2000-03 bear market, and the 2008 financial crisis. Because TLH+ avoids tax-indifferent switchbacks, it never caused negative tax offsets over the period, even though the portfolio was regularly rebalanced. The mean annual tax offset was 1.94%. 

It is very important to emphasize that an annual tax offset is not a true “after-tax” measure, because it does not take into account that some of the benefit will be clawed back. It assumes that tax deferral is synonymous with permanent tax avoidance, which is misleading. By definition, it will be a higher number than any measure that attempts to factor in liquidation. However, because it is sometimes erroneously referred to as “tax alpha”, it can be useful for apples-to-apples comparisons between various strategies.

We also ran the same 70% Betterment portfolio, with all the same assumptions, but with a strategy utilizing the 30-day switchback. It is not surprising that TLH+ would outperform the 30-day switchback approach, given the discussion of the latter’s limitations. The extent of the advantage, however, is striking—the mean annual tax offset is double.

Tax loss harvesting


Best Practices for TLH+

Summary: Tax loss harvesting can add some value for most investors, but high earners with a combination of long time horizons, ongoing realized gains, and plans for some charitable disposition will reap the largest benefits.

This is a good point to reiterate that tax loss harvesting delivers value primarily due to tax deferral, not tax avoidance. A harvested loss can be beneficial in the current tax year to varying degrees, but harvesting that loss generally means creating an offsetting gain at some point in the future. If and when the portfolio is liquidated, the gain realized will be higher than if the harvest never took place.

Let’s look at an example:

Year 1: Buy asset A for $100.

Year 2: Asset A drops to $90. Harvest $10 loss, repurchase similar Asset B for $90.

Year 20: Asset B is worth $500 and is liquidated. Gains of $410 realized (sale price minus cost basis of $90)

Had the harvest never happened, we’d be selling A with a basis of $100, and gains realized would only be $400 (assuming similar performance from the two correlated assets.) Harvesting the $10 loss allows us to offset some unrelated $10 gain today, but at a price of an offsetting $10 gain at some point in the future.

The value of a harvest largely depends on two things. First, what income, if any, is available for offset? Second, how much time will elapse before the portfolio is liquidated? As the deferral period grows, so does the benefit—the reinvested savings from the tax deferral have more time to grow.

While nothing herein should be interpreted as tax advice, examining some sample investor profiles is a good way to appreciate the nature of the benefit of TLH+.

Who benefits most?

The Bottomless Gains Investor A capital loss is only as valuable as the tax saved on the gain it offsets. Some investors may incur substantial capital gains every year from selling highly appreciated assets—other securities, or perhaps real estate. These investors can immediately use all the harvested losses, offsetting gains and generating substantial tax savings.

The High Income Earner Harvesting can have real benefit even in the absence of gains. Each year, up to $3,000 of capital losses can be deducted from ordinary income. Earners in high income tax states (such as New York or California) could be subject to a combined marginal tax bracket of up to 50%. Taking the full deduction, these investors could save $1,500 on their tax bill that year.

What’s more, this deduction could benefit from positive rate arbitrage. The offsetting gain is likely to be LTCG, taxed at around 30% for the high earner—less than $1,000—a real tax savings of over $500, on top of any deferral value.

The Steady Saver – An initial investment may present some harvesting opportunities in the first few years, but over the long term, it’s increasingly unlikely that the value of an asset drops below the initial purchase price, even in down years. Regular deposits create multiple price points, which offer continuous harvesting opportunities. (This is not a rationale for keeping money out of the market and dripping it in over time—tax loss harvesting is an optimization around returns, not a substitute for market exposure.)

The Philanthropist – In each scenario above, any benefit is amplified by the length of the deferral period before the offsetting gains are eventually realized. However, if the appreciated securities are donated to charity or passed down to heirs, the tax can be avoided entirely. When coupled with this outcome, the scenarios above deliver the maximum benefit of TLH+. Wealthy investors have long used the dual strategy of loss harvesting and charitable giving.

Even if an investor expects to mostly liquidate, any gifting will unlock some of this benefit. Using losses today, in exchange for built-in gains, offers the partial philanthropist a number of tax-efficient options later in life.

Who benefits least?

The Aspiring Tax Bracket Climber – Tax deferral is undesirable if your future tax bracket will be higher than your current. If you expect to achieve (or return to) substantially higher income in the future, tax loss harvesting may be exactly the wrong strategy—it may, in fact, make sense to harvest gains, not losses. Below a certain income cutoff, taxpayers can actually realize some capital gains at a 0% rate—an opportunity that should not be missed.

Graduate students, those taking parental leave, or just starting out in their careers should ask “What tax rate am I offsetting today” versus “What rate can I reasonably expect to pay in the future?”

The Scattered Portfolio – TLH+ is carefully calibrated to manage wash sales across all assets managed by Betterment, including IRA assets. However, the algorithms cannot take into account information that is not available. To the extent that a Betterment customer’s holdings (or a spouse’s holdings) in another account overlap with the Betterment portfolio, there can be no guarantee that TLH+ activity will not conflict with sales and purchases in those other accounts (including dividend reinvestments), and result in unforeseen wash sales that reverse some or all of the benefits of TLH+. We do not recommend TLH+ to a customer who holds (or whose spouse holds) any of the ETFs in the Betterment portfolio in other accounts, and we currently do not offer TLH+ to customers whose spouses also have Betterment accounts.

The Rapid Liquidator – What happens if all of the additional gains due to harvesting are realized over the course of a single year? The federal LTCG rate for the majority of taxpayers is currently 15%, but goes as high as 23.8%. In a full liquidation of a long-standing portfolio, the additional gains due to harvesting could push the taxpayer into a higher bracket, potentially reversing the benefit of TLH+. For those who expect to draw down with more flexibility, smart automation will be there to optimize the tax consequences.


Conclusion

Summary: Tax loss harvesting is a highly effective way to improve your investor returns without taking additional downside risk.

Tax loss harvesting may get the spotlight, but under the hood, our algorithms labor to minimize taxes on every transaction, in every conceivable way. Historically, tax loss harvesting has only been available to extremely high net worth clients. Betterment is able to take advantage of economies of scale with technology and provide this service to all qualified customers.

  • We do no harm: we never trigger short-term capital gains (others often do, through unsophisticated automation).
  • We do it holistically: we don’t just harvest daily—we make every transaction tax efficient—withdrawals, deposits, rebalancing, and more.
  • We coordinate wash sale management across both taxable and IRA accounts seamlessly.

¹Covariance is a measure of co-movement, which takes into account not only correlation, but also whether or not the two securities have similar absolute levels of risk. A covariance of 1 indicates that the two securities not only move together, but also have a similar range of returns.

²“Vanguard doesn’t track wash sales across accounts with different registrations—for example, between your individual account and your joint account or between your individual account and your IRA. However, you’re still obligated to correctly account for all wash sales when filing your income tax return.” (See “What are wash sales?” )

³ See here for more on how we backtest performance.


Disclosure

The potential after-tax benefit of TLH+ was calculated through historical backtesting of the strategy as applied to the Betterment model portfolio and is not based on actual client trading history, with all relevant assumptions stated within the text.  Actual Betterment clients may experience different results. Factors which will determine the actual benefit of TLH+ include, but are not limited to, market performance, the size of the portfolio, the stock exposure of the portfolio, the frequency and size of deposits into the portfolio, the availability of capital gains and income which can be offset by losses harvested, the tax rates applicable to the investor in a given tax year and in future years, the extent to which relevant assets in the portfolio are donated to charity or bequeathed to heirs, and the time elapsed before liquidation of any assets that are not disposed of in this manner.

The backtesting analysis that produced this estimate was done over the 13-year period from 2000 to 2013. No reliable data was available prior to 2000 that could adequately represent the performance of the full Betterment portfolio. The years in question featured a number of exceptional periods of market activity, and past performance is no guarantee of future results. The information used as the foundation for historical backtesting was compiled from third-party sources, and while we believe the information provided here is reliable, we do not warrant its accuracy or completeness. For more on how we backtest against historical data, see here.

Tax loss harvesting is not suitable for all investors. Nothing herein should be interpreted as tax advice, and Betterment does not represent in any manner that the tax consequences described herein will be obtained, or that any Betterment product will result in any particular tax consequence. Please consult your personal tax advisor as to whether TLH+ is a suitable strategy for you, given your particular circumstances. The tax consequences of tax loss harvesting are complex and uncertain and may be challenged by the IRS. You and your tax advisor are responsible for how transactions conducted in your account are reported to the IRS on your personal tax return. Betterment assumes no responsibility for the tax consequences to any client of any transaction.