Exchange-Traded Funds Use Section 852(b)(6) for Tax Avoidance, Not Just Tax Deferral: So Why Is This Loophole Still Open?

IV. Advocating for Reform by Rebutting the ETF Industry’s Specious Defense of the Tax Break Under Section 852(b)(6)

A. Because ETFs’ use of section 852(b)(6) results in forgone tax revenue and exacerbates income inequality, reforms are needed, despite the nonrecognition rule’s role in ETFs’ share-price-stability.

By virtue of section 852(b)(6), neither the ETF itself, nor the ETF’s shareholders, nor the redeeming authorized participant will recognize any gain when an ETF satisfies an authorized participant’s redemption demand in kind by distributing appreciated portfolio securities. No rationale rooted in tax theory exists to accord this special break to ETFs and their investors. As we have seen, however, the SEC considers ETFs’ nonrecognition of gain on in-kind redemption distributions to be essential to ETFs’ use of creation and redemption transactions to maintain share-price stability—and the Service apparently concurs. At the same time, the ETF industry contends that ETFs’ use of section 852(b)(6) creates little more than a timing issue, and that the ETFs’ unrecognized gain is ultimately taxed when the ETFs’ shareholders sell their ETF shares. The industry also argues that taxing ETF shareholders on such gains only when they sell their shares is more appropriate than taxing them when the ETF realizes those gains.

In light of how securities regulators view the nonrecognition rule’s role in preserving the creation and redemption process as a price-stability mechanism, curbing ETFs’ use of section 852(b)(6) might seem at first to be untenable. And revising how section 852(b)(6) applies to ETFs might not even appear to be all that necessary—if one accepts the industry’s timing-issue argument. Alas, however, section 852(b)(6)’s nonrecognition rule—as manipulated by ETFs—actually creates much more than a simple timing issue.

As detailed below, ETFs’ exploitation of the rule often causes a significant portion of their investment gains never to be taxed, which results in a substantial loss of revenue to the Treasury. Moreover, even the “mere” deferral of such tax—as the ETF industry posits—creates both revenue loss and significant inequities in the way that gains on investments are taxed. Those inequities disproportionately favor the wealthy and exacerbate income and wealth inequality. In addition, the industry’s argument that ETF shareholders should not be taxed on ETFs’ realized gains at the time of the realization cynically mischaracterizes the relationship between the funds and their investors.

For all of these reasons, policymakers should limit ETFs’ use of section 852(b)(6) as a tax-avoidance tool. Yet, at the same time, any revisions to section 852(b)(6) to reign in ETFs’ tax avoidance or evasion must also preserve ETFs’ ability to maintain share-price parity with NAV-per-share. In any event, the first step toward any such reform is to recognize the speciousness of the ETF industry’s defense of the tax break that ETFs receive under section 852(b)(6).

B. ETFs distribute their redemption-related gains to their authorized participants, but argue that they effectively retain those gains in their investment portfolios.

ETF insiders bristle at any suggestion that their funds receive unwarranted tax breaks. Accordingly, the ETF industry is quick to assert that, while section 852(b)(6) may enable ETFs to be more tax efficient than mutual funds, the rule does not facilitate any tax avoidance by ETFs. The industry contends that, although an ETF does not recognize gain at the time of a redemption-related distribution of appreciated portfolio securities, there is no permanent exclusion of such gain. Instead, according to the industry, there is merely a deferral until the ETF’s shareholders sell their shares—at which time, those shareholders will supposedly recognize their ratable portions of the ETF’s previously unrecognized gain.

How exactly does this purported deferral work? Obviously, an ETF shareholder is taxed on any gain that she recognizes in a taxable disposition of her ETF shares. But through what mechanism, and to what extent, does the ETF’s previously unrecognized gain on an in-kind redemption distribution become part of the gain that the shareholder herself realizes (and recognizes) when she ultimately sells her shares?

At the time of a shareholder’s sale of her ETF shares, her realized gain (if any) will equal the excess (if any) of (1) her amount realized (i.e., the sales proceeds that she receives) over (2) her adjusted basis in the shares. There is, of course, no rule in the Code that reduces the shareholder’s basis in her ETF shares by an amount equal to her portion of the ETF’s unrecognized gains. Ergo, a shareholder can be susceptible to taxation on those gains only if (and only to the extent that) those gains are reflected in her amount realized. Assuming that the ETF’s share price equals its NAV-per-share, the shareholder’s amount realized will equal her ratable portion of the ETF’s NAV at the time she sells her shares. Thus, there is only one way that a shareholder’s amount realized in a subsequent sale of her shares could include her portion of the ETF’s realized-but-unrecognized gain on a previous redemption distribution: The ETF’s gain on the distribution would have to remain part of the ETF’s NAV after the distribution. Yet—as a literal matter, at least—that does not actually happen.

An ETF’s NAV, as of any given time, equals the net value of the assets that it holds at such time. Included in the composition of that net value, of course, are any built-in (i.e., unrealized) gains or losses on such assets as of such time. But, at the risk of stating the obvious, an ETF’s NAV at a given time plainly does not include any part of the value of any assets that the ETF has previously distributed.

When an ETF distributes appreciated portfolio securities to its authorized participant in an in-kind redemption, the ETF distributes the entirety of the value of those securities—including the portion of that value constituting the ETF’s gain on those securities—to the authorized participant. Thus, when an ETF distributes such securities to satisfy a redemption in kind, it applies the gain that it realizes in the distribution to the remittance of the redemption proceeds. It does not have to recognize that gain, of course, by virtue of section 852(b)(6). Accordingly, it does not distribute the gain to its shareholders as a dividend. But it does not retain and reinvest that gain, either. Instead, the ETF distributes the tax-free gain to its authorized participant, in partial disbursement of the redemption proceeds. Once it does so, that gain is no longer included in the ETF’s NAV.

Thus, if an ETF shareholder subsequently sells her shares, her pro rata portion of the ETF’s unrecognized gain on the distributed securities cannot literally be included in her NAV-based amount realized. At any point after the redemption distribution, the only gains included in the composition of the ETF’s NAV are the ETF’s unrealized gains on its remaining assets. Because neither the ETF itself nor the ETF’s shareholders are taxed on the ETF’s realized-but-unrecognized gain at the time of the distribution, and because the ETF’s authorized participant takes a cost basis in the distributed securities, that specific gain is actually never taxed to anyone.

At the same time, however, an ETF’s NAV following an in-kind redemption distribution is greater than would be the case if the ETF were required to recognize its gain on the distribution. If such gain were recognized, the ETF would have to distribute the gain to its shareholders as a dividend. To pay such a dividend, the ETF would have to dispose of portfolio assets with an aggregate fair market value at least equal to that gain—in addition to the assets that it transferred to the authorized participant as redemption proceeds. To put it another way, following an in-kind redemption, an ETF’s NAV is greater—by at least the amount of the ETF’s unrecognized gain—than it would be if that gain were recognized and distributed to the ETF’s shareholders. From this, the ETF industry apparently extrapolates that not distributing the unrecognized gain in a shareholder dividend is tantamount to preserving the gain within the fund’s portfolio. In other words, the industry contends that the unrecognized gain is effectively reinvested in the portfolio.

The obvious counterpoint to that argument, of course, is that an ETF does in fact depart with the gain that it realizes in an in-kind redemption distribution: While it may not pay out that gain to its shareholders, it transfers that gain to its authorized participant when it distributes the appreciated securities. The ETF does not keep the unrecognized gain, but rather uses that gain to satisfy its redemption obligations.

Nonetheless, because the ETF’s post-redemption NAV is comparatively higher, by the amount of the unrecognized gain, the ETF’s share price immediately after the in-kind redemption distribution is also comparatively higher—assuming, again, that such share price equals the fund’s NAV-per-share. In turn, if a shareholder of the ETF were to sell her shares immediately after such distribution, her amount realized would be comparatively greater, by an amount equal to her ratable share of the fund’s unrecognized gain on the distribution. By extension, the shareholder’s recognized gain on the sale of her ETF shares would be comparatively greater (or her recognized loss would be comparatively less) by such amount. On this basis, the ETF industry maintains that, while an ETF investor is not taxed on her share of the ETF’s gain at the time of the redemption, the investor is effectively taxed on such gain when she ultimately sells her ETF shares.

C. An example of an ETF’s literal disposition, or effective retention, of its realized gains on an in-kind redemption distribution.

As an illustration of the ETF redemption process and the disposition of an ETF’s redemption-related gains, consider the following example of a hypothetical stock ETF:

Example 1: As of April 1, 2021, SureTrade Equity ETF (“SureTrade ETF”) has an NAV of $100 million, and it has one million fund shares outstanding. An individual shareholder named Etta owns 970 SureTrade ETF shares, and she holds her shares in a taxable account. SureTrade ETF’s investment portfolio consists of stocks in 80 large, publicly traded domestic corporations. Among those investments are shares of ABC Corp. stock that the fund acquired several years ago at an aggregate price of $2 million. As of April 1, 2021, the aggregate fair market value of those ABC Corp. shares is $3 million—a 50% increase from the price that the fund paid to acquire the shares. Also as of April 1, 2021, due to heavier-than-usual investor selling in the market, the trading price of SureTrade ETF’s shares declines to $97 per share—below the fund’s then-current NAV-per-share of $100 per share.

Seeking to profit from the discount of the ETF’s share price in the market to its NAV, SureTrade ETF’s authorized participant, ACME Investment Bank, purchases 30,000 of the ETF’s shares in the market for $2,910,000. It then bundles those shares into a creation unit and redeems the creation unit to SureTrade ETF. As with any other redeemable shares of an open-end RIC, SureTrade ETF must satisfy its authorized participant’s redemption demand at a price equal to the fund’s then-current NAV-per-share. Thus, the fund must redeem the authorized participant at a redemption price of $3 million. Per its usual policy, SureTrade ETF satisfies its authorized participant in kind—by distributing its $3 million worth of ABC Corp. shares in exchange for the redeemed creation unit.

For the authorized participant, this is a taxable exchange. The authorized participant recognizes $90,000 of gain (on which it owes presumably $18,900 in federal income tax) when it redeems the creation unit, and the authorized participant takes a $3 million initial cost basis in the ABC Corp. shares that it receives as redemption proceeds. Immediately after its receipt, the authorized participant resells the ABC Corp. shares in the market at their then-current aggregate share price of $3 million. However, because its $3 million amount realized upon such resale does not exceed its $3 million basis in the ABC Corp. shares, the authorized participant does not recognize any gain on the resale for tax purposes. This result obtains even though the authorized participant resells the ABC Corp. shares for more than it paid to acquire the creation unit that it delivered in exchange for those shares. Ultimately, the authorized participant makes an economic profit on this transaction of $71,100 (i.e., (1) the $3 million it receives when it resells the ABC Corp. shares minus (2) the $2,910,000 price it paid for the ETF shares comprising the redeemed creation unit minus (3) the $18,900 of tax on its recognized gain on the creation unit redemption).

When it distributes its appreciated ABC Corp. shares to its authorized participant, SureTrade ETF realizes a long-term capital gain of $1 million. Section 852(b)(6), however, provides that the fund does not have to recognize that gain. Accordingly, the fund does not have to distribute that gain to its shareholders to avoid entity-level federal income tax. Thus, neither SureTrade ETF nor its remaining shareholders (like Etta) will be taxed on that gain when the distribution occurs. Because the $1 million of gain is embedded in the value of the ABC Corp. shares at the time of the distribution, the fund transfers that gain—as part of its in-kind distribution of redemption proceeds—to the authorized participant. Yet, the authorized participant will not be taxed on that gain either, because it takes a cost basis in the ABC Corp. shares. In the end, therefore, SureTrade ETF’s realized gain on the distribution of the ABC Corp. shares per se is never taxed to anyone.

After the distribution of the ABC Corp. shares, SureTrade ETF’s NAV is comprised solely of the net value of its remaining assets (including any built-in gain on those assets). The $3 million value of the distributed ABC Corp. shares—including the $1 million portion constituting the fund’s realized gain on those shares—is no longer included in the fund’s NAV. Thus, if Etta were to sell her SureTrade ETF shares immediately after the distribution at a share price equal to the fund’s then-current NAV-per-share, her amount realized would not literally include her portion of the fund’s $1 million of unrecognized gain on the ABC Corp. stock. Instead, Etta would recognize a gain on such a sale only if, and to the extent that, (1) her ratable portion of the then-current net aggregate value of the fund’s 79 remaining investments exceeded (2) her ratable portion of the net aggregate value of the fund’s investments as of the date of purchase of her shares.

Nevertheless, SureTrade ETF would have had to take an additional $1 million worth of assets out of its investment portfolio—or perhaps even more—if it had been required to recognize its gain on the ABC Corp. shares at the time of the distribution. In such a case, to avoid entity-level federal income tax, the fund would have had to pay its shareholders a capital gain dividend of at least $1 million. But because SureTrade ETF did not in fact recognize the gain and pay such a dividend, that $1 million (or more) of assets remains in the fund’s portfolio and included in the fund’s NAV. The ETF industry would presumably assert that this is tantamount to SureTrade ETF retaining its $1 million of unrecognized gain and reinvesting that gain in its portfolio. If one were to accept that premise, then the fund’s NAV effectively would continue to include the $1 million of gain after the distribution. By extension, that gain would continue to be reflected in the fund’s NAV-per-share-based share price. Accordingly, on this theory, a shareholder’s pro rata portion of the $1 million gain would effectively be included in her amount realized when she eventually sells her shares.

In this example, immediately following the authorized participant’s redemption of the creation unit and the fund’s distribution of the ABC Corp. stock, the number of SureTrade ETF’s outstanding shares is reduced from 1,000,000 to 970,000, and SureTrade ETF’s NAV is reduced from $100 million to $97 million. Assume that, as a result of the redemption transaction, SureTrade ETF’s share price in the market is restored to parity with the fund’s NAV-per-share of $100. At that point in time, the market value of investor Etta’s SureTrade ETF shares is $97,000. The ETF industry would doubtlessly claim that $1,000 of that value effectively represents Etta’s ratable share of the fund’s unrecognized gain on its distributions of the ABC Corp. stock—even though the fund previously transferred that gain to its authorized participant. On that view, if Etta were to sell her SureTrade ETF shares at a gain of $1,000 or more at such time, the first $1,000 of her recognized gain would effectively equal her portion of the fund’s gain on the distribution. Under this perspective, Etta would then be taxed on her portion of that gain.

D. Even if ETFs effectively retain their redemption-related realized gains, the ETF industry’s defense of its use of section 852(b)(6) for tax advantages is spurious.

Even if one accepts the initial premise that ETFs effectively retain their redemption-related gains—notwithstanding that they literally distribute those gains to their authorized participants as redemption proceeds—there are at least three reasons to reject the ETF industry’s defense of its tax break under section 852(b)(6). First, an ETF’s disposition of portfolio securities is essentially a disposition of those securities by, or on behalf of, the ETF’s shareholders. Second, a deferred payment of capital gains tax is not equivalent to a payment of the tax at the time of the gain. Third, and perhaps most importantly, not all ETF shareholders ultimately sell their shares at a gain equal to or greater than their pro rata portions of their fund’s redemption-related gains.

1. An ETF’s investors are enriched by, and thus should be taxed on, the ETF’s realized gains on dispositions of portfolio securities.

As part of their defense of the section 852(b)(6) tax break, at least some ETF proponents maintain—incorrectly—that an ETF’s shareholders are not enriched when the ETF itself realizes gain on a portfolio investment. They contend that, instead, the shareholders realize income only if and when they sell their own ETF shares at a gain. For example, Dave Nadig, chief investment officer and director of research of ETF Trends and ETF Database (and former managing director of ETF.com), argues in essence that, when an ETF distributes an appreciated portfolio security in an in-kind redemption, it is only the ETF itself (and not the ETF shareholder) that realizes income. On his view, section 852(b)(6) merely prevents an ETF shareholder from being taxed on gain from a disposition of securities that “someone else” (i.e., the ETF) makes. To Nadig, it is thus entirely appropriate that—by virtue of section 852(b)(6)—ETF investors are taxed only on any gain they may recognize when they ultimately sell their ETF shares. As he puts it, “[t]hey’re paying when they sell, not when other people sell.”

This argument is fundamentally flawed because it distorts the relationship between an ETF and its investors. Specifically, it disingenuously posits that ETFs invest for their own accounts, rather than on behalf of their shareholders. In fact, of course, the opposite is the case. By way of analogy, consider a professional investment manager who actively manages her high-net-worth client’s account. When the manager sells securities out of the client’s portfolio, the manager disposes of the client’s property, on the client’s behalf. Thus, if there is any gain on such a sale, it is obviously the client (and not the manager) who realizes that gain.

Similarly, when an ETF acquires or disposes of portfolio securities, it is managing the investments of its individual shareholders. In other words, the fund acquires or disposes of portfolio assets on behalf of those shareholders. Thus, contrary to Nadig’s assertion, when an ETF’s portfolio securities are disposed of, it is in essence the fund’s shareholders themselves who are disposing of them—not “someone else.” Accordingly, when an ETF disposes of portfolio securities at a gain, that gain is properly viewed as having been realized, ratably, by the ETF’s shareholders.

This, in a nutshell, is the premise underlying the RIC rules in Subchapter M. As a practical matter, only the wealthy are able to invest individually in a professionally managed diversified securities portfolio, so Congress determined that smaller investors should be able to pool their investments to attain the same benefits of asset diversification and professional management. To promote “[t]he overarching policy goal of . . . facilitat[ing] pooled investment[,]” the RIC rules generally seek to “tax[] each member of the pool as she would be taxed if she made investments for her own account.” Of course, aligning the tax treatment of pooled-investment-fund shareholders with that of direct investors in an equivalent securities portfolio requires two things: (1) that any realized gains on the fund’s investments not be taxable to the fund itself and (2) that, instead, each shareholder be taxed on her proportionate share of such gains, in the taxable year of the realization. But for ETFs’ exploitation of section 852(b)(6), that is precisely what the RIC rules accomplish.

“[B]y allowing RICs to eliminate their taxable income by paying it out as a dividend,” Subchapter M essentially treats RICs as mere “conduit[s] for investment by small investors.” And, by requiring RICs to distribute their recognized gains to their shareholders on a current basis in order to avoid entity-level federal tax on such gains, Subchapter M generally works to ensure that each shareholder pays a proportionate amount of tax when the fund disposes of appreciated portfolio securities—just as if the shareholder had sold the same proportionate amount of those securities out of her own individual account.

Yet, as a result of ETFs’ routine use of in-kind redemptions, ETF shareholders typically avoid current-year tax liability on their shares of their funds’ realized investment gains. To the extent it enables ETFs to avoid entity-level tax on such gains without having to distribute those gains to their shareholders, section 852(b)(6) thwarts Subchapter M’s “original goal of establishing tax parity between direct investments in securities and investments in RICs.”

2. Even the “mere” deferral of taxation of an ETF’s realized investment gains is costly to the Treasury.

The ETF industry also argues that, even though section 852(b)(6) permits an ETF to avoid gain recognition at the time of an in-kind redemption, the ETF’s shareholders ultimately pay tax on all of that previously unrecognized gain when they dispose of their shares. For example, Elisabeth Kashner, director of ETF research at FactSet Research Systems, Inc. (and former director of research at ETF.com), contends that an ETF’s shareholders ultimately pay the same amount of tax on their fund’s gains as do the shareholders of a mutual fund with an equivalent investment portfolio. The only distinction, according to Kashner, is that the mutual fund shareholders pay the tax incrementally (as they are taxed on their yearly gain dividends), whereas the “ETF shareholders pay all taxes at the end” when they sell. Similarly, ETF Trends’ Nadig claims that, if ETFs lost the benefit of section 852(b)(6) and had to distribute their redemption-related gains to their shareholders (like mutual funds do), the only difference would be that “the Treasury would collect a little more tax from [the ETFs’ shareholders] today, and just not collect that [tax] later.” And he summarily muses that, in such case, “there’s no guarantee that the money the Treasury would collect now would be more than the money collected at a future point in time when investors sell their ETFs.” Nadig thus dismisses the effects of section 852(b)(6) on the tax liability of ETF shareholders as merely “a timing shift, not a net-tax shift.”

This part of the industry’s tax-break defense is also plainly wrong. As detailed below, even if one accepts that an ETF effectively retains its redemption-related gains, there are many instances in which an investor in that ETF is never subsequently taxed on her portion of those gains. Moreover, even to the extent that an ETF investor is eventually taxed on her effective share of the ETF’s previously realized-but-unrecognized investment gains, the deferral of such taxation until the shareholder’s sale of her shares always results in the Treasury receiving less tax revenue—in real-value terms—than it would if the shareholder were taxed on such gains when the ETF itself realized them.

In any case of the latter sort, “time-value-of-money principles” provide that “a dollar of income today is worth more than a dollar of income a year from now, because the dollar today can be invested and will be worth more than a dollar in a year.” As a result, “[t]he ability to defer the tax on asset appreciation confers a benefit on the taxpayer.” Conversely, it confers a harm on the federal fisc: A dollar of tax revenue received tomorrow will always be worth less than if it were received today. For this reason, in present-value terms, deferring tax liability on asset appreciation generally—and on ETF investment gains, in particular—is tantamount to excluding a portion of the gain from tax altogether. Thus, even to the extent that “ETF investors pay at the end, while mutual fund investors pay as they go[,]” the government still loses out. As even some ETF advocates concede, “[d]ollars not received by the U.S. Treasury today cannot supply classrooms, support public health, promote national security or tend to our veterans. Those same dollars do less tomorrow.”

Because of the time-value of money, federal tax law has long reflected a general policy goal of “try[ing] to prevent [investors] from having easy access to tax deferral vehicles.” But there is at least one other reason why the deferral of an ETF shareholder’s tax on her share of the ETF’s realized investment gains causes the Treasury to lose money: Some portion of those ETF gains will typically consist of short-term capital gains. If the ETF pays a dividend of its short-term capital gain to its shareholders (in the taxable year in which those gains were realized), the dividend is treated as a distribution of ordinary income, and the shareholder is taxed on such income at her ordinary marginal rate. In contrast, if an individual ETF shareholder is not taxed on such gain until a later taxable year in which she sells her ETF shares, the gain is converted to long-term capital gain and is taxed to the shareholder at a lower long-term capital gain rate. Thus, to the extent that the ETF’s realized-but-unrecognized gain on a redemption-related in-kind distribution of appreciated portfolio securities is short-term capital gain, section 852(b)(6) causes the Service to collect the tax on such gain not only later—if at all—but also in a smaller amount.

So, even in those instances when section 852(b)(6) causes the taxation of an ETF’s investment gains “merely” to be deferred—and not avoided for all time, as often happens—there is still a big problem. In short, despite the ETF community’s protestations to the contrary, there is a guarantee that deferring taxation of an ETF’s—or any other RIC’s—realized investment gains causes the Treasury to lose revenue. Moreover, as detailed below, deferring tax on ETF investment gains disproportionately benefits high-net-worth investors. Consequently, any such deferral violates basic precepts of tax fairness and exacerbates income and wealth inequality.

It is precisely to avoid such deferral and its consequences that Subchapter M requires “RICs to make annual distributions of their current income” and recognized gains to avoid entity-level federal tax. Because of the distribution requirement, “RIC shareholders owe taxes on actual distributions every year[,]” rather than in some later year when (and if) they choose to dispose of their shares. Section 852(b)(6), however, lets ETFs avoid distribution of their redemption-related realized investment gains, while still avoiding entity-level tax on those gains. In so doing, and in contravention of longstanding tax policy, section 852(b)(6) enables ETFs to become “spectacular vehicle[s] for tax deferral.”

3. ETF investors often do not ultimately pay tax on their share of an ETF’s redemption-related gains.

As explained in Part V below, there are many cases in which an ETF investor is never taxed at all—or is taxed only partially—on her proportionate share of an ETF’s realized gains on in-kind redemption distributions of appreciated portfolio securities. In those cases, the ETF’s reliance on section 852(b)(6)’s nonrecognition rule produces more than a temporary deferral of tax on those gains. Rather, in such cases, section 852(b)(6) results in the permanent avoidance of tax on those gains.

V. ETFs’ Use of Section 852(b)(6) Often Results in Tax Avoidance Because ETF Investors Often Are Never Fully Taxed on an ETF’s Redemption-Related Realized Gains

At its essence, the ETF industry’s defense of its tax break under section 852(b)(6) centers on the claim that an ETF’s investors are taxed on the ETF’s redemption-related realized gains when they eventually sell their shares. Yet there are many instances in which that central claim is false, even if one accedes to the notion that ETFs effectively retain their realized gains on in-kind redemption distributions. For the reasons detailed below, not all of an ETF’s realized-but-unrecognized gains under section 852(b)(6) are ultimately taxed to the ETF’s shareholders when (and if) they sell. In many cases, the taxation of at least some of those gains is avoided permanently—not merely deferred until later.

A considerable number of ETF shareholders will never sell their shares at all. Instead, they will hold their ETF shares until death and then bequeath the shares to a beneficiary. Because death is not a realization event, the decedent will not be taxed on any gain on her ETF shares—including any portion of the shareholder’s gain that may effectively be comprised of the ETF’s realized-but-unrecognized gains. And, because the beneficiary’s initial basis in the shares will be stepped up to fair market value under section 1014, the beneficiary will never be taxed on any such gain either.

Perhaps more significantly, even if an ETF shareholder does ultimately sell her shares, an ETF’s redemption-related gains often go either partially or fully untaxed. When an ETF investor sells her ETF shares, the investor’s recognized gain at the time of the sale may turn out to be less than her proportionate share of the ETF’s realized-but-unrecognized gains on the in-kind redemption distributions of portfolio securities that occurred while she held her ETF shares. In those cases, the investor is not taxed on her entire share of those redemption-related gains when she sells. To see how this works, let us begin with an example of the ETF industry argument that ETF investors are taxed on the ETF’s redemption-related gains when they sell. Let us then turn to some counterexamples which demonstrate how that argument often fails.

A. An example to illustrate the industry’s argument that ETF investors are eventually taxed on an ETF’s redemption-related gains.

To support the proposition that an ETF shareholder pays tax on her portion of the ETF’s previously unrecognized gains when she sells her shares, FactSet’s Elisabeth Kashner offers the following example of a hypothetical ETF that engages in in-kind redemptions as part of so-called “heartbeat” trades (which are discussed further below):

[Example 2A:] Imagine a mutual fund and ETF have the same portfolios, and are managed identically. Each fund starts with $10.00 per share [NAV] and gains 10% per year. Each portfolio manager closes positions that generate $1.00 per share of profits each year. The mutual fund transacts directly in the securities market, while the ETF uses in-kind creations and redemptions. . . . The mutual fund must distribute its $1.00 of realized capital gains to its shareholders. The ETF makes no capital gains distributions; instead, it [keeps its capital gains invested in portfolio securities]. . . . [Assume that] [t]his process continues for five years.

In her example, Kashner assumes a 20% capital gains tax rate. “At the end of the [first] year, the mutual fund’s [NAV-per-share] is back at $10.00, and its investors have $1.00 [per share] in their pockets, which nets to $0.80 [per share] after tax.” At the same time, the ETF’s NAV-per-share rises to $11.00 at the end of year one, and the ETF’s investors “have $1.00 of unrealized gains.” The process repeats in year two, “taking [the ETF’s NAV-per-share] to $12.10,” and providing the ETF’s investors with another $1.10 per share of unrealized gain. In contrast, at the end of year two, the mutual fund distributes another $1.00 per share to its investors (which, again, results in an after-tax gain of $0.80 per share), and the mutual fund’s NAV-per-share returns to $10. The same thing happens in years three, four, and five. At the end of the fifth year, the ETF’s NAV-per-share is $16.11, and the mutual fund’s NAV-per-share is $10. Over the five-year period, the mutual fund’s investors will have received $5.00 per share of capital gains distributions (netting to $4.00 per share of after-tax gains).

Assume in this example that, at the end of year five, one mutual fund investor redeems her shares to the fund and one ETF investor sells her shares in the market. Assume further that, at all relevant times, the mutual fund’s share price equals the mutual fund’s NAV-per-share and the ETF’s share price equals the ETF’s NAV-per-share. On these assumed facts, the mutual fund investor—who purchased her shares for $10 per share and redeems her shares for $10 per share—recognizes no gain or loss at the time of her redemption. At the same time, of course, the mutual fund investor had already recognized a total of $5 per share of capital gains (and paid $1 of tax on those gains) during her five-year holding period. In contrast, the ETF shareholder—who purchased her shares for $10 per share and sells her shares for $16.11 per share—recognizes $6.11 per share of capital gain (and pays $1.22 per share of tax on that gain) at the time of her sale.

To Kashner, the take-away from this example is that, “[o]ver the full holding period, [the] mutual fund [investor’s] and ETF [investor’s] tax bills will be,” in her words, “identical.” She emphasizes that the “[m]utual fund shareholders will have paid taxes every year, but avoided a large tax on sale, while [the] ETF shareholders pay all taxes at the end.” On this basis, Kashner declares her example to support the contention that section 852(b)(6)’s nonrecognition rule “allows [ETF] investors [merely] to postpone capital gains taxes,” not to avoid those taxes permanently.

B. Counterexamples that demonstrate why ETF investors are not always taxed on their entire share of an ETF’s redemption-related gains.

In Kashner’s hypothetical, the ETF investor’s shares and the mutual fund investor’s shares gain in value every year, and each investor sells or redeems her shares at their all-time-high price. Those assumed facts make Kashner’s fact pattern decidedly improbable. In the real-life experience of the average investor, the performance of most investments—and the timing of most decisions to sell—are not nearly so perfect. Securities values do not simply increase in a straight line; in some periods, they decrease instead. Moreover, it is fairly rare for an investor to a sell a given security exactly when it reaches its highest post-purchase-date price. After all, timing the market is notoriously difficult. Far more frequently, an investor instead continues to hold a security after it hits a peak price, and the investor sells only once the security’s value has declined from the peak. This phenomenon is reflected, for instance, by the fact that many investors sell into bear markets. And perhaps most starkly, it is illustrated by how many investors sold at or near the market lows of the “Great Recession” in 2009.

1. An example of an ETF investor who is only partially taxed on her share of an ETF’s redemption-related realized gains.

With the foregoing in mind, consider a more realistic variation of Kashner’s hypothetical:

Example 2B: As in Example 2A (Kashner’s original example), the mutual fund investor and the ETF investor purchase their respective shares at the beginning of year one for $10 per share. At all relevant times, each fund’s share price equals its then-current NAV-per-share. Also as in Example 2A, the value of each fund’s investment portfolio—and, thus, each fund’s NAV—increases by 10% per year, in each of years one through four. This time, however, each fund’s NAV decreases by 15% in year five because of a general market downturn. The mutual fund investor and the ETF investor redeem or sell their respective shares at the end of year five on the belief that their shares would only lose value from there.

In this scenario—just as in Example 2A—the mutual fund and the ETF each realize $1 per share of capital gains by disposing of appreciated portfolio securities in each of years one through four. The mutual fund does so by selling portfolio securities in the market; the ETF does so through in-kind redemption distributions. The mutual fund recognizes the $1 per share of gain on its sales of portfolio securities in each such year. By virtue of section 852(b)(6), the ETF does not recognize gain on its in-kind redemption distributions. At the end of each of years one through four, the mutual fund distributes a $1 per share capital gain dividend to its shareholders. The mutual fund’s shareholders pay tax (at Kashner’s assumed 20% rate) on those dividends. The ETF, however, makes no such distributions—and its shareholders pay no such tax—in those years. The mutual fund’s NAV-per-share (and, thus, its share price) rises from $10 to $11 during the course of the year, in each of years one through four, and then returns to $10 at year-end as a result of the capital gain distributions. The ETF’s share price rises to $11 at the end of year one, to $12.10 at the end of year two, to $13.31 at the end of year three, and to $14.64 at the end of year four.

In each of years one through four, the mutual fund investor’s share of her fund’s realized (and recognized) capital gain is $1 per share. In each such year, the mutual fund investor pays $0.20 per share of tax on that gain. By contrast, the ETF investor’s share of her fund’s realized (but unrecognized) capital gain is $1 per share for each of years one through four—but the ETF investor does not pay any tax in those years on those gains. In each such year, the ETF realized those gains on its investor’s behalf. In substance, in other words, the ETF investor herself realized those amounts of capital gain in those years. (The gains were realized, of course, irrespective of whether they were reinvested in the fund’s portfolio.) The ETF investor should therefore be taxed on all of those gains. Of course, Kashner would contend that the investor will pay a deferred tax on the entirety of those gains when she sells her ETF shares. But in this modified version of her example, that does not happen.

In this version of the example, in year five, each fund’s NAV decreases by 15%. For simplicity’s sake, assume that neither the mutual fund nor the ETF realizes any gain or loss on dispositions of portfolio securities during year five. Because the mutual fund does not recognize any capital gains for the year, it does not distribute any capital gain dividend to its shareholders in year five. Thus, the fund’s NAV-per-share is not further reduced by any such distribution, and the fund’s shareholders are not taxed on any such distribution. At the close of year five, the mutual fund’s NAV-per-share (and share price) is $8.50 and the ETF’s NAV-per-share (and share price) is $12.44. Also at the close of the year, the mutual fund investor redeems her shares to the fund, and the ETF investor sells her shares in the market. The mutual fund investor recognizes a $1.50 per share loss on her redemption, and the ETF investor recognizes a $2.44 per share gain on her sale.

In Example 2B, the disparities between the overall tax consequences to the mutual fund investor, on one hand, and the ETF investor, on the other, are stark. The mutual fund investor ends up paying a greater amount of tax on her investment gains in years one through four, even though the same amount of capital gain is realized by or on behalf of the ETF investor in those years. The mutual fund investor realizes (and recognizes) $1 per share of capital gain in each of years one through four—for an aggregate gain of $4 per share in those years. The investor pays $0.20 per share of tax on her gain in each of those years (or $0.80 per share of tax in total). In year five, the mutual fund investor recognizes a $1.50 per share loss, but that loss is not carried back to offset the investor’s tax gains in the prior four years. This is, of course, exactly the same tax treatment that a direct investor in the stocks comprising the fund’s investment portfolio would receive, just as Subchapter M intends.

The ETF investor also realizes (or the ETF realizes on its investor’s behalf) capital gain in the amounts of $1.00 per share in each of years one through four—for an aggregate gain of $4.00 per share in the first four years. As noted above, the ETF investor pays no tax in those years on those gains. When she sells her shares in year five, the ETF investor recognizes gain of only $2.44 per share of gain and she pays approximately $0.49 per share of tax on that gain. This is more than a timing issue—more than just a deferral of the tax until year five. In this example, when the ETF investor sells in year five, she is not fully taxed on her realized investment gains for years one through four; indeed, she pays less than 62% of the tax that would have been due if all of those gains had been properly recognized. (In total, the ETF investor would have owed approximately $0.80 per share of tax on her $4.00 per share of year-one-through-four gains, if they had been recognized.)

2. An example of an ETF investor who is not taxed on any part of her share of an ETF’s redemption-related realized gains.

To illustrate further that an ETF’s shareholders often do not pay tax on gains that section 852(b)(6) permits the ETF itself to avoid recognizing, let us consider one more permutation of Kashner’s example:

Example 2C: Assume the same facts as in Example 2B, except that this time, the respective NAVs of the mutual fund and the ETF both decline by 32% (instead of 15%) in year five—perhaps due to a more severe market downturn. The mutual fund investor and the ETF investor each sell their respective fund shares in a panic at the end of year five, at the depths of the downturn (as many investors unfortunately do).

In this version of the example, the mutual fund’s NAV-per-share (and share price) is $6.80 and the ETF’s NAV-per-share (and share price) is $9.96 at the end of year five. On these facts, the mutual fund investor pays all of the tax on her realized (and recognized) capital gains in year one through four, and she recognizes a $3.20 per share loss in year five. The ETF investor, by contrast, pays nothing in years one through four and recognizes a $0.04 per share loss in year five. This time around, when the ETF investor sells in year five, she ends up not paying any tax on any of the $4.00 per share of capital gains that were realized on her behalf in years one through four. She does not just defer the tax on those gains until year five; she avoids all of the tax altogether.

C. ETF investors should be taxed on their share of an ETF’s gains on in-kind redemption distributions in the year in which those gains are realized.

In both Examples 2B and 2C, when the ETF investor sells her shares in year five, she is not fully taxed on her portion of the ETF’s realized (but unrecognized) gains for years one through four. This is because the increase in the value of her shares during the first four years—including the portion of that increase attributable to those realized (and reinvested) gains—is offset by the decline in the value of her shares in year five. In substance, the ETF investor’s loss in year five is carried back to offset her gains in the prior four years.

At first glance, that result might seem appropriate. After all, one might ask, shouldn’t the ETF investor simply recognize whatever the overall gain (or loss) on her investment turns out to be as of the time she sells her shares? But, on further consideration, netting the ETF investor’s year-five loss against her year-one-through-four gains clearly is not appropriate. To the extent that the ETF investor’s gains in the first four years are realized gains, those respective gains are properly includible in her income for those respective years—and they should not be offset by any loss that she incurs in a subsequent year. Netting the later year loss against the earlier year realized gains violates the fundamental tax accounting precept that a taxpayer’s taxable income should be determined discretely for each of her taxable years, and that the taxpayer should be taxed separately on each such year’s income.

Arguably, it is impossible to determine a taxpayer’s “true” income with absolute certainty until the end of her life. Similarly, in the case of a transaction that spans many years, it may not be possible to calculate a taxpayer’s overall income from the transaction definitively until the transaction concludes. Nevertheless, in order for our income taxation system to be workable, taxpayers must be required to report, and pay tax on, their taxable incomes at—and for—regular periodic intervals. Tying such reporting and payment requirements to fixed tax accounting periods is necessary both to manage the administrative burden of calculating one’s taxable income and to ensure that the government receives a steady flow of tax revenue. As everyone understands, “no functioning tax system could afford to wait for its collections until the end of an individual’s lifetime or until the termination of a business enterprise.” Accordingly, the U.S. income tax adopts a convention of annual tax accounting periods. This is of course reflected in section 441, which provides that “[t]axable income shall be computed on the basis of the taxpayer’s taxable year.”

Under the U.S.’s annual tax accounting system, “events in later years [generally do not] affect tax liabilities for earlier years.” In other words, with only a few exceptions, “nothing that happens after the close of a taxable year can alter a taxpayer’s tax liability for that year.” For instance, consider a taxpayer who directly invests in stocks. In year one, she sells shares of ABC Corp. for a $100 gain. Then, she reinvests that gain by applying it to the purchase of shares of XYZ Corp. In year two, the taxpayer sells her XYZ Corp. shares at a $100 loss. Imagine that these are the taxpayer’s only dispositions of capital assets in years one or two. Obviously, in year one, the taxpayer is required to pay tax on $100 of capital gain. Her $100 of capital loss in year two is not carried back to reduce her year-one tax liability.

The principle of horizontal equity and the interest of sound tax administration both dictate that a similarly situated ETF investor be taxed in the same way. With that in mind, consider the following fact pattern:

Example 3: A taxpayer purchases shares in an ETF at the beginning of year one and sells her ETF shares at the end of year two. In year one, the ETF distributes appreciated shares of ABC Corp. stock to a redeeming authorized participant and effectively reinvests its realized (but unrecognized) gain on that distribution by purchasing shares of XYZ Corp. stock. In year two, the ETF distributes its shares of XYZ Corp. stock to a redeeming authorized participant, at which time the fair market value of the XYZ Corp. stock is lower than the ETF’s basis in the stock. Assume that, as a result of the ETF’s year-one gain on the ABC Corp. stock, the aggregate share price of the taxpayer’s ETF shares increases by $100 in year one. Also assume that, as a result of the ETF’s year-two loss on the XYZ Corp. stock, the aggregate share price of the taxpayer’s ETF shares decreases by $100 in year two. When the taxpayer sells her ETF shares at the end of year two, her sale proceeds are thus equal to the amount that she originally paid for the shares.

In this example, the ETF shareholder should pay tax on $100 of capital gain in year one—just as the direct investor in the ABC Corp. shares and XYZ Corp. shares did. After all, her $100 share of the ETF’s realized capital gain on the ABC Corp. shares represents true income to the ETF shareholder for year one—an actual accession to her wealth in that year—regardless of whether her income decreases by a concomitant amount in the following year. This is not reflected in how the ETF shareholder is actually taxed, however. The shareholder does not recognize her portion of the year-one gain in year one, and she does not recognize an amount equivalent to that year-one gain when she sells her ETF shares in year two. Instead, the ETF shareholder recognizes no gain or loss when she sells her shares. Once again, the later year loss is inappropriately netted against the earlier year gain. As a result, the ETF shareholder averts taxation in year one on income that she had garnered in year one; indeed, she escapes taxation on that income for all time. In the end, section 852(b)(6)’s nonrecognition rule enables both the ETF and the ETF shareholder to avoid tax on the year-one gain permanently.

Some readers might object that Example 3 fails to lend adequate weight to the $100 capital loss on the XYZ Corp. shares that the direct investor recognizes in year two. But that objection would be misplaced. Assume, for example, that the investor will be able to apply that entire loss to offset future capital gains on which she would otherwise be taxed in subsequent years. In that case, her overall net tax liability on the direct investments in the ABC Corp. shares and the XYZ Corp. shares will work out—over time—to be the same as the $0 of tax liability incurred by the ETF shareholder. Yet this does not mean that the two taxpayers are treated equally, or that the treatment of the ETF shareholder is merited. First, it is far from certain that the direct investor will actually realize gains in the future in an amount sufficient for her to utilize her entire capital loss. If the direct investor fails to offset all of her year-two capital loss against future gains, then she will incur greater tax liability on her year-one and year-two investments, combined, than does the ETF shareholder. This underscores that the ETF shareholder is undertaxed on her year-one investment gains. Second, even in a case in which the direct investor is able to apply her entire capital loss to offset future gains, that still does not justify permitting the ETF shareholder to carry the equivalent loss back to avoid recognition of a prior year’s realized gain. In such a case, the federal fisc is nonetheless deprived, in that prior year, of tax that it should receive on the ETF shareholder’s income for that prior year. The ETF shareholder, in other words, nonetheless fails to remit the appropriate tax revenue at the appropriate time.

D. ETFs’ exploitation of section 852(b)(6) often results in tax avoidance, and not just in tax deferral.

Simply put, Examples 2B, 2C, and 3 above disprove the ETF industry’s claim that routine in-kind redemptions merely defer capital gains taxes until an investor sells her shares. They demonstrate that, in a number of cases, section 852(b)(6) does far more than delay the taxation of capital gains that an ETF realizes but does not recognize in a redemption-related distribution of appreciated portfolio securities. In particular, those examples refute the industry’s contention that the ETF’s investors themselves always pay tax on those previously unrealized gains when they sell their shares. Instead, as those examples show, an ETF’s exploitation of section 852(b)(6) often results in the permanent avoidance of tax by both the ETF and its investors on a substantial portion—if not all—of those gains.

VI. How ETFs Manipulate Section 852(b)(6) to Avoid Gain Recognition on Non-Redemption-Related Dispositions of Portfolio Securities

Avoiding redemption-related recognized gains is obviously critical to ETFs’ tax efficiency, but ETFs’ efforts to avoid recognition of their investment gains do not stop there. ETFs need to dispose of securities in their investment portfolios for reasons other than to satisfy redemption demands. For example, a passively managed ETF that tracks a particular market index must divest itself of securities that have been removed from the reference index. And an actively managed ETF routinely changes the composition of its portfolio as part of its investment strategy. Normally, such dispositions of portfolio securities would take the form of taxable sales or exchanges. ETF structurers, however, have found ways to manipulate section 852(b)(6) to avoid taxation of any gains that they realize in connection with these non-redemption-related dispositions, as well.

A. “Cherry-picking” high-basis securities for in-kind redemption distributions to avoid gain recognition on taxable sales of portfolio assets.

When choosing which portfolio securities to distribute to a redeeming authorized participant, an ETF typically “cherry-picks” securities in which it has the greatest amount of unrealized gain. Those securities are soon replaced within the ETF’s portfolio by equivalent securities that the ETF receives from its authorized participants in subsequent creation transactions, with respect to which the ETF takes a higher basis. This leaves only securities with the comparatively least amount of unrealized gain to remain in the ETF’s investment portfolio. By continually replacing low-basis positions with high-basis positions in the same securities through successive creation and redemption transactions, an ETF can “wash out” a substantial portion of the inherent gain its investment portfolio without ever recognizing that gain. The cherry-picking technique thus greatly reduces—if not altogether eliminates—the gain that the ETF recognizes if it ever has to dispose of portfolio securities in a taxable sale or exchange.

As an ETF engages in consecutive creation and redemption transactions with its authorized participants, it can employ cherry-picking to replace most—though not always all—of its lower basis portfolio securities with higher basis securities of equivalent value. Because ETFs and their authorized participants routinely engage in large creation and redemption transactions, cherry-picking allows ETFs to expunge much—if not all—of the unrealized gain from their investment portfolios over time, without diminishing the fund’s NAV. Ultimately, this minimizes the ETF’s potential capital gains recognition in any prospective taxable dispositions of its portfolio assets.

Imagine, for instance, the following hypothetical involving an ETF that tracks a particular market index:

Example 4: The ETF in this example tracks a certain equity market index. ABC Corp. stock is included in the index. Accordingly, the ETF must hold shares of ABC Corp. in its investment portfolio. One year ago, the ETF acquired 100,000 shares of ABC Corp. stock from its authorized participant in exchange for a newly issued creation unit. At the time of the exchange, the aggregate fair market value of the ETF shares comprising the creation unit was $10 million. The ETF delivered $100 worth of ETF shares to its authorized participant for each share of ABC Corp. stock that it received; thus, the ETF took a $100 cost basis in each of those ABC Corp. shares. During the year that followed, the market price of ABC Corp. stock increased from $100 per share to $150 per share.

In this example, if the ETF were simply to have retained the 100,000 ABC Corp. shares that it acquired two years ago, it would now have $50 of unrealized gain in each of those shares. Under that scenario, if the ETF were now required to sell those shares for some reason (e.g., if ABC Corp. stock were removed from the index that the fund tracks) then it would recognize $50 of long-term capital gain per share. To avoid having to pay entity-level federal income tax on that gain, the fund would have to distribute the gain to its shareholders—as $5 million of capital gain dividends. The ETF shareholders would then be taxed on that gain at the applicable long-term capital gain rate.

To prevent any such prospective gain recognition, however, the ETF does not simply retain its $100-basis ABC Corp. shares. Instead, when the ETF’s share price dips below the fund’s NAV-per-share and the ETF’s authorized participant redeems a creation unit, the ETF satisfies the redemption demand by delivering some of those ABC Corp. shares to the authorized participant. Imagine, for instance, that when ABC Corp.’s shares are trading at $150 per share, the authorized participant acquires shares of the ETF with an aggregate market value of $1.5 million, bundles those ETF shares into a creation unit, and redeems the creation unit to the ETF. The ETF satisfies the redemption demand by delivering 1,000 of its $100-basis ABC Corp. shares to the authorized participant,

Shortly thereafter, when the ETF’s share price rises above the fund’s NAV-per-share, the ETF issues a creation unit, which it delivers to its authorized participant in exchange for new ABC Corp. shares. Say, for instance, that the ETF issues and delivers a creation unit with a market value of $1.5 million when ABC Corp. stock is trading at $150 per share. In exchange for the creation unit, the authorized participant delivers to the ETF 1,000 ABC Corp. shares (which the authorized participant acquires in the market immediately before the trade).

Because an ETF and its authorized participants are incentivized to enter into such creation or redemption transactions whenever the ETF’s share price deviates from NAV-per-share, these trades occur frequently. In fact, an ETF will often issue new creation units to its authorized participant and redeem creation units from its authorized participant during the same trading day. And the amounts involved in each such trade are substantial. For instance, the minimum size of a creation unit that an ETF will accept in redemption is usually equal to at least several million dollars’ worth of a fund’s shares. In practice, however, the size of the creation unit tendered in an ETF redemption is generally far greater than the minimum—often exceeding $100 million worth of the ETF’s shares. So, to put it mildly, the volume of securities transferred in these exchanges is significant.

In Example 4, after ten sets of transactions in which the ETF delivers old ABC Corp. shares to its authorized participant in redemption of existing ETF shares (bundled into creation units) and thereafter issues new creation units to its authorized participant in exchange for new ABC Corp. shares, the ETF will eliminate all potential tax gain from the ABC Corp. stock in its portfolio. That is to say, the ETF will replace (1) all of its $100-basis ABC Corp. shares that had a $150 fair market value with (2) $150-basis ABC Corp. shares that each have a $150 fair market value. Consequently, if the ETF has to dispose of its 10,000 “new” ABC Corp. shares in a taxable sale or exchange, it will receive the same $15 million of proceeds that it would have received in a sale of its 10,000 “old” ABC Corp. shares, but it will avoid the $5 million of taxable gain that it would have had to recognize upon a sale of the “old” shares. Because the ETF’s $5 million of economic gain will not be recognized, the ETF will not have to distribute that gain to its shareholders to avoid an entity-level federal income tax. And, because the ETF’s shareholders thus will not receive $5 million in capital gain dividends, they will not be taxed on that gain either.

B. Using abusive “heartbeat trades” to avoid gain recognition on taxable exchanges of portfolio securities.

As beneficial as it plainly is, cherry-picking low-basis, high-value securities for delivery to authorized participants in day-to-day redemption transactions may not always be sufficient, on its own, to enable an ETF to avoid all taxable gain on nonredemption dispositions of portfolio assets. What if, for instance, an ETF that passively tracks a market index needs to rebalance its portfolio—in response to a change in the index that it tracks—during a period of lower-than-average investor outflows? Or, what if an actively managed ETF adds holdings to, and removes holdings from, its securities portfolio frequently as part of its investment strategy? In such cases, the ETF may not be able to utilize routine creation and redemption transactions to change out all of the appreciated portfolio assets that it needs to replace in order to avoid gain recognition on its taxable sales or exchanges of securities.

In response to this dilemma, ETF managers have devised a cynical complement to routine creation and redemption transactions—one that allows them to replace literally all of the appreciated portfolio securities that they do not turn over in transactions arising from day-to-day investor activity. This complementary strategy entails what industry watchers have dubbed as “heartbeat” trades. A heartbeat transaction culminates in what is essentially a sham redemption, orchestrated solely to allow an ETF to receive a tax break under section 852(b)(6).

Heartbeat trades are an even more flagrant abuse of section 852(b)(6) than ETFs’ other manipulations of the rule. In form, though not substance, there are two steps to a heartbeat trade. In the first formal step, the ETF acquires the securities it wants to add to its portfolio from an authorized participant in exchange for newly created ETF shares. In the second formal step—which occurs very shortly after the first—the authorized participant redeems those same ETF shares and receives, in satisfaction of its redemption demand, the appreciated portfolio securities that the ETF seeks to remove from its portfolio.

In substance, the two steps amount to the ETF’s exchange of the “old” portfolio securities for the “new” ones. But because the transaction is structured in form as an offsetting creation and redemption of ETF shares, it is not treated as a taxable exchange. Instead, thanks to section 852(b)(6), the ETF does not recognize any gain on the portfolio securities it exchanges away when the authorized participant “redeems” the newly created ETF shares back to the ETF.

Let us examine these abusive arrangements in a bit more detail: In a heartbeat trade, the first step is for the ETF to issue a new creation unit, which it delivers to a bank or other financial institution in exchange for a basket of securities that the ETF wants to add to its investment portfolio. Unlike a normal creation transaction, however, this step in a heartbeat transaction is not prompted by ordinary investor demand for new ETF shares. Indeed, in this kind of transaction, the participating financial institution has no intention of dividing up the creation unit and selling the constituent ETF shares in the market.

Instead, at the time of the creation step, the parties agree that the financial institution will redeem the creation unit back to the ETF shortly thereafter—in exchange for a basket of appreciated securities that the ETF wants to eliminate from its investment portfolio. This preplanned “redemption” is the second step in the heartbeat transaction, and it generally occurs within only a few days of the creation step. By structuring the trade in form as a couple of offsetting creation and redemption transactions, the ETF is able to achieve nonrecognition treatment for what is, in substance, a taxable exchange of appreciated portfolio securities for other securities of equal market value.

Imagine, once again, a fact pattern concerning an ETF that tracks a particular market index:

Example 5: The ETF in this scenario tracks a certain equity market index. On an upcoming date, ABC Corp. stock will be removed from the index and replaced by XYZ Corp. stock. On that date, the ETF will have to dispose of all of the ABC Corp. stock in its portfolio and add some XYZ Corp. stock to the portfolio instead. Assume that some or all of the ETF’s shares of ABC Corp. are highly appreciated. For whatever reason, the ETF had been unable to replace its low-basis ABC Corp. shares with higher basis ABC Corp. shares through creation and redemption transactions done in the ordinary course of its operations. The ETF now has to sell its low-basis ABC Corp. shares (or exchange those shares for XYZ Corp. shares), but it does not want to recognize the gain that it will realize in such a sale or exchange.

In this scenario, to avoid the gain recognition, the ETF manufactures a heartbeat trade. It delivers a creation unit to a friendly investment bank or other market-maker in exchange for a certain amount of XYZ Corp. shares. Pursuant to a pre-agreed plan, the bank or market-maker redeems the creation unit a day or two later—and receives, in exchange, a basket of ABC Corp. shares equal in value to the XYZ Corp. shares that it had delivered to the ETF. In the end, the ETF has simply exchanged its appreciated shares of ABC Corp. stock for an equivalent amount (by value) of XYZ Corp. shares. But because the second step of the trade is formally styled as an in-kind redemption of an ETF shareholder, the ETF avoids recognition of its gain on the ABC Corp. stock—courtesy of section 852(b)(6).

These sham transactions are nicknamed “heartbeats” because of the unusually large size of the trades when compared to the amounts typically involved in creation and redemption transactions that arise “organically” from investor buying or selling. The outsized amounts transferred in these manufactured exchanges create spikes on a chart of the fund’s trading activity that resemble the spikes on an EKG monitor. The redemption step in a heartbeat trade has no independent economic substance; its sole purpose is to allow the ETF to rid its portfolio of tax gain through the misuse of section 852(b)(6).

Yet, although heartbeat trades are blatantly abusive, virtually all ETFs appear to engage in them. A recent Bloomberg Businessweek report identified no fewer than 2,261 heartbeat trades—in an aggregate amount of $330 billion—that equity ETFs have effected since the year 2000. And the volume of such trades seems only to be increasing. The same report concluded that, in 2018 alone, equity ETFs engaged in 548 heartbeat transactions “worth a record $98 billion.”

It is important to note that heartbeat trades are utilized both by actively managed ETFs that rebalance their portfolios frequently, as part of their investment strategies, and by passively managed ETFs that simply track a particular market index. The fact that both passively managed funds with relatively lower portfolio turnover and actively managed funds with higher portfolio turnover use heartbeat trades as a tax-avoidance strategy “is a clear sign that virtually any ETF [benefits] from this practice.” It also indicates that not only actively managed ETFs, but also passively managed ETFs, dispose of significant amounts of portfolio securities for reasons other than to satisfy redemption demands. If that were not true, there obviously would be no need for such ETFs to engage in heartbeat trades. If such dispositions were not effected through heartbeat trades, they would be taxable transactions. The ETF’s would thus recognize gain on such dispositions (apparently despite their cherry-picking of low-basis deliverables for in-kind redemptions). This all suggests that stopping heartbeats—or treating them as the taxable transactions that they substantively are—would enable the Treasury to recover a significant portion of the tax on investment gains that both passively managed and actively managed ETFs—and their investors—currently avoid through their misuse of section 852(b)(6).

VII. The ETF Tax Break Under Section 852(b)(6) Forfeits Tax Revenue, Violates Fairness Norms, and Increases Wealth and Income Inequality

A. Because of the tax break under section 852(b)(6), ETFs recognize virtually none of their realized capital gains.

As a result of the ways in which ETFs manipulate section 852(b)(6), most ETFs recognize little or no capital gains in a typical year—even when they have substantial amounts of economic gain from dispositions of portfolio securities. Based on a review of “the annual reports of more than 400 funds,” Bloomberg Businessweek recently estimated that, in 2018 alone, “U.S. stock ETFs avoided tax on more than $211 billion in gains” as a result of “routine” in-kind redemptions and “special heartbeat trades.” And 2018 was hardly an anomaly. In 2015, for example, “the 25 largest equity ETFs distributed securities with [unrecognized gain] of almost $60 billion” in redemption transactions. Those ETFs, of course, “did not distribute any [of those unrecognized] capital gains to their shareholders.” Indeed, as I have described elsewhere, multiple “[d]ata from various periods since the inception of ETFs in 1993 uniformly indicate that equity ETFs make little or no capital gains distributions to shareholders in an average year.”

This is very different from the result that obtains in the case of a mutual fund with an equivalent investment portfolio. As but one example, as Bloomberg Businessweek pointed out in 2019, “State Street’s SPDR S&P 500 [ETF hadn’t] reported a taxable gain in 22 years. In contrast, a traditional mutual fund run by Fidelity Investments that tracks the same index had a taxable gain in ten of those years.” Similarly, a Morningstar study from 2012 found that, over the five-year period ending on December 31, 2011, the percentage of NAV that traditional mutual funds paid out to shareholders in taxable capital gains distributions was “more than 40-fold (4,000%) greater than the corresponding figure for ETFs” with equivalent investment portfolios. The same sorts of discrepancies also exist with respect to actively managed funds and equity funds focused on virtually any market sector. For confirmation, one need only turn to the ETF industry itself—which is all too happy to publicize what it dubs as ETFs’ greater “tax efficiency” relative to mutual funds. As touted on the website ETF.com, for instance, emerging market equity mutual funds distribute 6.46% of their NAV to their shareholders in capital gain dividends each year, on average. Emerging market equity ETFs, in contrast, pay out only a paltry 0.01% of their NAV, on average, as shareholder capital gain distributions annually.

This “greater tax efficiency” raises several serious policy concerns: It costs the federal government a presumable fortune; it violates basic notions of tax fairness; and it contributes to income and wealth inequality.

B. The ETF tax break under section 852(b)(6) results in substantial forgone tax revenue.

ETFs’ avoidance of fund-level gain recognition costs the federal Treasury an enormous amount in forgone tax revenue. Simply put, an ETF’s exploitation of section 852(b)(6) “cause[s] a problem for the tax collector” because “[i]t causes the ETF’s [unrecognized] capital gains to evaporate.”

The Joint Committee on Taxation (JCT) has traditionally classified the “[n]onrecognition of in-kind distributions by [RICs] in redemption of their stock” under section 852(b)(6) as a “[t]ax expenditure[] for which quantification is not available.” As defined in the Congressional Budget and Impoundment Control Act of 1974, “tax expenditures” are “revenue losses attributable to provisions of the [f]ederal tax laws which allow,” among other things, “a special exclusion . . . from gross income or which provide . . . a deferral of tax liability.” The JCT classifies a Code provision as a tax expenditure if such provision is expected to result in a federal revenue loss of at least $50 million over the upcoming five-fiscal-year period. Plainly, the revenue loss resulting from ETFs’ manipulation of section 852(b)(6) far surpasses that $50 million threshold. In fact, in connection with the Wyden Proposal, the JCT has reportedly provided Senator Wyden’s office with “preliminary estimates” indicating that an outright repeal of section 852(b)(6) could be “expected to generate just over $200 billion over a decade.” This suggests that section 852(b)(6) costs more than $20 billion in forgone tax revenue each year.

Other studies have gauged the revenue losses stemming from the ETF tax break under section 852(b)(6) to be even greater. Consider, for example, Bloomberg Businessweek’s estimate that 400 equity ETFs avoided more than $211 billion of gain recognition in 2018. As a result of the ETFs’ avoidance of that gain recognition at the fund level, the shareholders of those ETFs were able to avoid paying—for that one year alone—perhaps somewhere between $31.65 billion and approximately $50.22 billion in federal taxes on distributions of such gain that they never received. And those figures assume that all of the unrecognized gains in question were long-term capital gains. The forgone revenue could be higher still, if at least some of those gains were short-term capital gains instead. Although it may be difficult to quantify the extent of the revenue loss from section 852(b)(6) with precision, one thing is certain: The forgone federal tax revenue arising from ETFs’ routine use of section 852(b)(6) is very substantial.

C. The ETF tax break under section 852(b)(6) violates basic notions of tax fairness and increases income and wealth inequality.

In addition to the problem of forgone tax revenue, ETFs’ exploitation of section 852(b)(6) raises serious fairness concerns. The fact that ETF shareholders do not receive taxable distributions of their funds’ realized gains creates both horizontal and vertical equity problems. Simply put, horizontal equity is “the notion that all taxpayers with a like amount of income should be taxed alike.” Vertical equity, in turn, is “the principle that higher income taxpayers should be taxed more heavily than lower income taxpayers.” ETFs’ exploitation of section 852(b)(6) results in violations of both precepts. It also contributes to income and wealth inequality by enabling already-affluent investors who hold ETF shares in taxable accounts to amass even more wealth even more quickly.

1. The disparity in tax treatment between ETF investors and mutual fund investors is a violation of horizontal equity.

To see how ETFs’ use of section 852(b)(6) violates the principle of horizontal equity, consider Examples 2A through 2C above, in which a mutual fund investor and an ETF investor make equal investments in funds with identical investment portfolios and hold those investments for five years. In those examples, each investor has $1.00 per share of realized capital gains, each year for at least the first four years, from her fund’s dispositions of portfolio securities. The mutual fund investor receives a distribution of her $1.00 per share of capital gains for each such year and is taxed on those gains in the year of the distribution. The ETF investor, by contrast, receives no such capital gains distributions and pays no tax during years one through four on her share of her fund’s gains for those years.

Thus, the two taxpayers—who make essentially the same investment—are taxed very differently on their respective investment gains. Because her share of her fund’s realized gain is distributed to her each year, the mutual fund investor is taxed in each of the first four years on the entirety of her gains for that year. In contrast, the ETF shareholder is never taxed on any portion of her realized gains for years one through four until she sells her shares in year five. Because of the time-value of money, among other things, the ETF investor will effectively be taxed less than if her tax were due in the year in which her share of her fund’s gains was realized. This is true even if she ultimately is taxed in year five on all of her year-one-through-four gains. Indeed, deferring the tax on her gains for the first four years until year five is functionally equivalent to excluding a portion of those gains from taxation altogether.

Moreover, there are many scenarios in which the ETF shareholder literally is not taxed on the entirety of her year-one-through-four gains when she departs with her ETF shares in year five. In some such scenarios, she pays tax on only a portion of those year one-through-four gains in year five; in others, she is never taxed on any of those gains. In all of these cases, the disparities in tax treatment between the taxpayer who invests in an ETF and the taxpayer who makes an equal initial investment in a traditional mutual fund with an identical securities portfolio constitute obvious and substantial horizontal equity problems.

2. The deferral of tax on ETF investment gains exacerbates horizontal equity problems inherent in the realization rule.

Deferring any tax on an ETF investor’s share of her fund’s realized investment gains until the investor sells her ETF shares also exacerbates an even broader violation of horizontal equity, which stems from the realization rule. The U.S. income tax system has long contained a realization requirement, which defers taxation of asset appreciation until the occurrence of a realization event—usually, the taxpayer’s sale or exchange of the asset in question. The realization requirement does not arise from any constitutional mandate or tax theory precept; instead, as the Supreme Court has explained on more than one occasion, “the concept of realization is ‘founded on administrative convenience.’” In short, the realization rule exists largely because taxing asset appreciation as it accrues would raise problems with valuing a taxpayer’s assets every year and because of concerns about whether a taxpayer would have the means to pay the tax, absent a sale of the asset.

Regardless of the practical necessities that may justify the rule, tax scholars have long criticized the realization requirement on the ground that it deviates from the economic definition of income in a manner that results in taxing asset appreciation more favorably than income from other sources. Under “the widely accepted Haig-Simons definition” of the term, a person’s economic “income” for a given year “is the sum of [her] consumption plus [her] change in net worth” during the year. An ideally constructed income tax, therefore, would tax all increases in a taxpayer’s wealth as those increases occur. Accordingly, because “unrealized appreciation . . . [increases] a taxpayer’s net worth, such appreciation . . . should be included as it accrues in a tax base reflecting economic income.” In other words, an “ideal” income tax base would include the unrealized appreciation of a taxpayer’s assets during the taxable year in question.

Most categories of income are taxed in a manner that is closer to this ideal. That is to say, income from most other sources is taxed at the time when the taxpayer derives the economic benefit of the income. Compensation income, for example, is taxed when the service provider receives the economic benefit thereof (i.e., when she receives the compensation if she is a cash-basis taxpayer or when she obtains a fixed right to receive the compensation if she is an accrual-basis taxpayer). “Similarly, periodic income from capital (e.g., interest and dividends) is taxed on receipt.”

A taxpayer derives the economic benefit of asset appreciation “when the appreciation occurs because the taxpayer’s wealth increases at that time.” Accordingly, by parity of reasoning, asset appreciation should be taxed as it occurs. The realization requirement, in contrast, typically defers taxation of the economic benefit of asset appreciation until the later time of a realization event. A direct investor in stocks, for instance, is taxed on her gain when she sells or exchanges her shares. Likewise, a mutual fund investor is taxed when (1) the fund realizes gain by selling securities in which she invested through the fund and (2) the fund distributes that gain to the investor as a capital gain dividend.

Any taxation of investment gains realized through an ETF, however, is deferred until even later than the time of realization. Not only are ETF investors not taxed on their shares of their ETF’s unrealized investment gains as those gains accrue, they are not even taxed when the ETF realizes those gains on their behalf thanks to section 852(b)(6). Instead, even in a “best case” scenario in which an ETF investor is ultimately taxed on all of those gains, the investor is not required to pay the tax until the (often considerably) later time when she sells her ETF shares. Thus, as a result of ETFs’ exploitation of section 852(b)(6), the tax treatment of ETF investment gains strays even farther from the ideal income tax base than does the tax treatment of asset appreciation generally.

The realization requirement’s deferral of tax on asset appreciation produces basic violations of horizontal equity. Clearly, “[t]he ability to defer tax on asset appreciation confers a benefit on the taxpayer due to the time-value of money.” This benefit is “not enjoyed by recipients of other forms of income[,]” such as salary or interest income. Consequently, “[a] taxpayer with income in the form of gains from property bears less of a tax burden than a taxpayer with a like amount of income from other sources.” Because it creates a longer deferral of any tax on ETF investment gains—beyond the time of realization—section 852(b)(6) causes those gains to be treated even more favorably than asset appreciation generally, let alone income from other sources. As applied to ETFs, section 852(b)(6) thus exacerbates the realization requirement’s horizontal equity violations.

3. The deferral of tax on ETF investment gains also exacerbates vertical equity problems inherent in the realization rule.

Perhaps even more troublingly, ETFs’ exploitation of section 852(b)(6) also exacerbates violations of vertical equity that are inherent in the realization requirement. In general, the progressive rate structure of the U.S. income tax “causes the benefits of [tax] deferral to accrue most strongly to higher income taxpayers. Because they pay taxes at a higher marginal rate, deferring taxes saves them more money overall.” Moreover, in the particular case of asset appreciation, “the benefits of the deferral conferred by the realization requirement accrue disproportionately to the wealthy because wealthier taxpayers tend to own greater amounts of capital.” Given that “the majority of capital assets are owned by affluent taxpayers,” and that “higher income taxpayers are likely to have a greater percentage of their total income coming from capital than lower income taxpayers,” the realization requirement plainly produces “a tax benefit that is primarily skewed toward the wealthy.” This inordinate benefit from the realization requirement to wealthier taxpayers is an obvious violation of vertical equity. By extension, “the realization requirement is justly viewed as an inherently regressive feature that hinders the income tax’s wealth redistribution objectives.”

The inordinately greater advantage that wealthy taxpayers (compared to everyone else) derive from the realization rule is even more pronounced in the particular case of ETF taxation. Once again, because section 852(b)(6)’s nonrecognition rule results in the deferral of tax on ETF investment gains beyond the time when those gains are realized, the rule produces an even greater tax benefit than does the realization requirement generally. Meanwhile, wealthier investors own a disproportionately higher percentage (than less affluent investors) of ETF shares that are held in taxable investment accounts. As a result, the benefits of the ETF tax break under section 852(b)(6) redound disproportionately to higher income taxpayers. This, in turn, creates an even greater violation of vertical equity.

Several years ago, Professor Samuel Brunson predicted that “[t]ransforming RICs into a vehicle to defer taxes would also transform RICs from an investment primarily aimed toward low- and middle-income individuals to one aimed squarely at the wealthy.” In the case of ETFs, specifically, that prediction was prescient. Overall, “only a quarter of all listed equities are in taxable accounts.” The remainder, presumably, are held in tax-deferred or tax-exempt retirement accounts. At the same time, however, “as many as half of equity ETF units [are] held by taxable investors.” While traditional “[m]utual funds are more likely to be held in retirement accounts[,]” recent studies show that “[t]axable individual investors . . . have migrated to ETFs.” One such study, in particular, focuses on how “tax-sensitive investors, such as high-net-worth investors, gradually switched from mutual funds to ETFs over the last decade.” Not surprisingly, the authors of the study determined that the reason for this switch is precisely because “[h]igh-net-worth clients can benefit especially from the” tax advantages that ETFs provide. The authors found, for example, “that [the purchase] of ETFs is most pronounced among registered investment advisers with wealthy clients, who are most likely to be liable for capital gains tax.” Data from the study indicate that, as a result of their migration from mutual funds to ETFs, wealthy investors now own a substantially higher portion of the ETF shares that are held in taxable accounts than do investors “of lesser means.”

Against this backdrop, the distributional impact of the section 852(b)(6) tax break for ETFs is obvious: Given that most ETF shares “held in taxable accounts are . . . owned by high net-worth taxpayers,” the unduly favorable tax treatment of ETF investment gains “bestows an untoward benefit on [those wealthy] taxpayers.” Section 852(b)(6), in other words, “unfairly benefits high-net-worth owners of ETFs.”

4. The deferral of tax on ETF investment gains results in greater compounding of those gains for wealthy investors, which violates fairness norms and increases income and wealth inequality.

Just how big a benefit section 852(b)(6) provides to wealthy investors is manifest by the significantly greater compounding of ETF shareholders’ investment gains that results from the nonrecognition rule—relative, for instance, to the compounding that mutual fund shareholders experience. Compounding occurs as a shareholder reinvests (or when her fund reinvests on her behalf) her portion of the fund’s realized investment gains. Taxable capital gain dividends, however, have an adverse effect on the compounding of a mutual fund shareholder’s investment gains over time. The amount of those gains that the mutual fund shareholder could otherwise reinvest each year is reduced by the amount of tax that she must pay for the year, when she receives her distribution of those gains. This reduction in reinvestment resulting from the annual tax liability lowers the rate at which the shareholder’s year-over-year gains compound.

By contrast, thanks to section 852(b)(6), ETFs pay virtually no capital gain dividends to their shareholders. Therefore, during the entire time that she holds her ETF shares, a typical ETF shareholder will never have to allocate any of her portion of the ETF’s realized gains to tax payments on those gains. Instead, she will be able to reinvest (or have the fund reinvest on her behalf) the entirety of those gains in the ETF.

As a result—and as is illustrated in Example 2A above—the gain on an ETF shareholder’s investment will compound at a greater rate than the gain on an equal investment by a shareholder of a mutual fund with an identical investment portfolio. Over time, this discrepancy in the growth of their respective investment gains will increase. The longer that they hold their respective investments, the greater the amount by which the value of the ETF shares will exceed the value of the mutual fund shares.

Because of this difference in compounding, if the ETF shareholder and the mutual fund shareholder hold their respective shares for the same period and then sell their shares at the same time, the pretax gain of the ETF shareholder will be larger. Accordingly, as further demonstrated in Example 2A, if the two shareholders are taxed on their respective gains at the same nominal rate, the ETF shareholder will net a larger amount of after-tax gain on an original investment of the same size. As a result, even though the two taxpayers are subject to the same nominal rate, the ETF shareholder’s effective tax rate will be lower.

This is, of course, rightly viewed to be a major advantage for ETF investors. Indeed, the greater compounding of investment gains is touted as a central component of ETFs’ “tax efficiency.” But which ETF investors, exactly, receive the benefit of greater compounding specifically because of section 852(b)(6)’s nonrecognition rule?

The answer, of course, is that only investors who hold their ETF shares in taxable accounts depend on section 852(b)(6) for the deferral (or avoidance) of tax on their investment gains that allows them to compound those gains at a greater rate than mutual fund investors. Investors who hold ETF shares in tax-deferred or tax-exempt retirement accounts, by contrast, would continue to receive the same deferral or avoidance of tax on their investment gains—and the same resultant increase in compounding of those gains—even if section 852(b)(6) disappeared. An investor who holds ETF shares—or, for that matter, mutual fund shares or any other investments—in a tax-advantaged retirement account does not pay tax on any gains that accrue on her investments so long as the funds remain in her account. The rules that apply to retirement accounts, generally, thus facilitate the same enhanced compounding of investment gains that taxable investors in ETFs, specifically, obtain by virtue of section 852(b)(6). To put it another way, section 852(b)(6)’s nonrecognition rule provides taxable investments in ETFs with the same tax advantages that are otherwise reserved strictly for investments in retirement accounts. And what, then, is the general profile of the taxable ETF investor to whom the section 852(b)(6) tax break adheres?

The ETF industry would like to convince Congress (and the public) that the ETF tax break does not favor the wealthy but instead benefits “Main Street investors striving to build financial security.” To that end, in arguing against the Wyden Proposal to repeal section 852(b)(6), the Investment Company Institute (ICI) asserted that “[n]early 12 million U.S. households own ETFs as part of their financial planning, and the median income of those households is $125,000.” Importantly, however, the ICI did not disclose how many of those households hold their ETF shares in tax-advantaged retirement accounts rather than in taxable accounts. As noted above, only taxpayers who hold ETFs in taxable accounts would be affected by any revisions to section 852(b)(6). Therefore, income statistics for taxpayers with ETFs in retirement accounts are, at best, irrelevant to—and, at worst, misleading in—any policy discussion about the distributional effects of such potential revisions. In the absence of any statement to the contrary, it is reasonable to assume that the ICI’s income data include investors who hold ETFs in retirement accounts as well as those who hold ETFs in taxable accounts. If that is indeed the case, then the ICI data are likely to under-reflect substantially the incomes of the taxable ETF investors whom section 852(b)(6) helps.

Contrary to the statistics that the ICI has proffered, evidence from non-ETF industry sources indicates that ETF shares in taxable accounts are held disproportionately by higher income, higher net-worth taxpayers. This, in turn, signals that—by facilitating the greater compounding of investment gains for taxable ETF investors—section 852(b)(6) inordinately benefits those who are already wealthy. By increasing the rate at which taxable ETF investors are able to amass further wealth, section 852(b)(6) thus serves to increase income and wealth inequality.

VIII. The ETF Tax Break Must Be Rolled Back, But This Cannot Be Accomplished Simply By Repealing Section 852(b)(6)

A. Policymakers should not adopt any ETF taxation reform that eliminates the nonrecognition rule for in-kind redemption distributions.

Both revenue considerations and fairness concerns dictate paring back the tax advantages that ETF investors enjoy. A couple of commentators have thus recommended simply repealing section 852(b)(6)—à la the Wyden Proposal. That proposal, however, is fundamentally flawed. Up until now, at least, Congress has considered a traditional mutual fund’s ability to satisfy shareholder redemptions in kind to be a vital protection against portfolio deterioration in times of extreme market downturn. More to the point of this discussion, in-kind redemptions are also a core part of the arbitrage transactions that the SEC regards as necessary for ETFs to function as viable pooled investment entities. And, in both contexts, the government views the nonrecognition of gain on an in-kind redemption distribution to be vital for in-kind redemptions to serve their intended purpose. Eliminating section 852(b)(6) altogether would plainly conflict with those policy positions.

One commentator who favors total repeal, Lee Sheppard, questions whether the price-arbitrage activity that section 852(b)(6) supports is really necessary to the functioning of ETFs after all. Sheppard notes (with apparent disapproval) that “the SEC permit[s] creation units and in-kind redemptions” because “[i]t believes these practices keep market prices for ETF units close to NAV.” But she then postulates that the trading of ETF shares in the securities markets would be sufficient, on its own, to maintain ETF share-price parity with NAV-per-share without the need for continual creation and redemption transactions between ETFs and their authorized participants. “Aren’t we supposed to believe that the securities markets are the best price mechanism?” she asks rhetorically. In this particular case, the answer is no—at least if the experience of closed-end funds is any guide.

Closed-end funds are largely inviable as pooled investment vehicles because their shares often trade in the market at significant premiums or (more often) discounts to NAV-per-share. ETFs succeed where closed-end funds fail precisely because authorized participants’ price-arbitrage activity—which is made possible by in-kind creation transactions and tax-free in-kind redemption distributions—keeps ETF share prices in line with NAV. And, of course, the SEC knows this. Thus, no serious effort to reform the tax treatment of ETF investment gains can be founded on an argument that nonrecognition of gain on in-kind redemption distributions is unnecessary to the functioning of ETFs.

Another commentator, Professor Jeffrey Colon, blithely asserts that, “[i]f ETFs are only viable because of the tax subsidy of section 852(b)(6), they should not survive.” This, too, cannot form the basis of any serious reform effort. The U.S. ETF industry now has more than $5 trillion of assets under management, and the ETF market continues to grow each year. That being the case, it is inconceivable that the SEC, the Service, or Congress would undertake any reforms that might threaten to destroy ETFs. If section 852(b)(6) were simply repealed and ETFs became inviable as a result, the mass liquidation of ETF portfolio assets that would follow could cause sharp and protracted market disruptions. Given those stakes, policymakers cannot be expected to entertain Professor Colon’s suggestion to let the chips fall where they may. No approach to reforming the tax treatment of ETF investment gains can be premised on a notion that the extinction of ETFs is an acceptable outcome.

B. SEC Rule 6c-11 enables actively managed ETFs to make greater use of tax-free in-kind redemptions, and promises to accelerate the migration of high-net-worth investors from mutual funds to ETFs.

1. Rule 6c-11 standardizes and extends exemptive relief that allows ETFs to operate under the 1940 Act.

The SEC has recently reaffirmed its commitment to supporting ETFs—and to protecting the price-arbitrage mechanism that is facilitated through in-kind creation and redemption transactions—by adopting a comprehensive regulatory framework to govern ETF operations. ETFs do not fit squarely within the 1940 Act’s standard definitions of either an “open-end company” or a “closed-end company.” Accordingly, ETFs have always required exemptions (or variances) from certain 1940 Act provisions to operate as investment management companies under the statute. Until recently, every ETF sponsor had to apply for an individual exemptive relief order from the SEC to operate its fund(s) under the 1940 Act. In late 2019, however, the SEC finalized its long-awaited Rule 6c-11, which provides standardized exemptive relief for all ETFs and replaces the individual exemptive orders that ETF sponsors previously needed.

Rule 6c-11 permits ETFs to operate as modified versions of open-end companies under the 1940 Act, provided that they comply with specified operational and disclosure requirements. Both the scope of the exemptive relief provided to, and the concomitant requirements imposed on, ETFs under Rule 6c-11 encompass the terms of the SEC’s previous exemptive orders. But, compared to the old exemptive orders, Rule 6c-11 also imposes certain additional requirements on, and provides expanded relief or operational flexibility to, ETFs. For example, Rule 6c-11 imposes revised public disclosure obligations on ETFs. At the same time, Rule 6c-11 discards any distinction between passively managed and actively managed ETFs and includes provisions that make it easier for actively managed ETFs to operate.

2. Rule 6c-11’s allowance of “custom baskets” makes it easier for actively managed ETFs to take advantage of section 852(b)(6).

Perhaps the most significant expansion of exemptive relief or operational flexibility under Rule 6c-11 is the newly conferred permission for ETFs and their authorized participants to use so-called “custom baskets” of portfolio securities in in-kind creation and redemption transactions. As the SEC explained when it adopted Rule 6c-11, the composition of the “basket” of portfolio securities that an authorized participant delivers to an ETF in exchange for a newly issued creation unit, or that an ETF delivers to an authorized participant in exchange for a redeemed creation unit, “is an important aspect of the efficient functioning of the arbitrage mechanism.” For example, the specific composition of a basket that an ETF delivers to a redeeming authorized participant affects both the ETF’s “costs of assembling and delivering” the basket and the authorized participant’s “costs of liquidating” the securities comprising the basket.

As the SEC also explained when adopting Rule 6c-11, the composition of a basket that an ETF receives from, or delivers to, an authorized participant is “important to [the ETF’s] portfolio management, as each in-kind creation or redemption increases or decreases positions in the ETF’s portfolio, and allows portfolio managers to add or remove certain portfolio holdings.” The SEC observed that an ETF’s in-kind distribution of portfolio securities to a redeeming authorized participant “can be an efficient way for a portfolio manager to execute changes in the ETF’s portfolio because the manager can make the changes without incurring the additional expenses of trades in the market.” Chief among those “additional expenses,” of course, are the tax expenses that an ETF or its shareholders would incur if the ETF instead had to sell appreciated portfolio securities in the market to rebalance its portfolio. As the SEC additionally noted, however, “[w]hen an ETF does not have flexibility to manage basket composition . . . undesired changes to [its] portfolio may result . . . .”

Under the terms of the exemptive relief orders that preceded Rule 6c-11, an ETF was generally required to deliver (in a redemption transaction) or receive (in a creation transaction) baskets that corresponded pro rata to the ETF’s portfolio holdings. By mandating that an ETF only use pro rata baskets in in-kind creation or redemption transactions, the SEC had intended to preserve the interests of ETF shareholders. In particular, the pro rata requirement purportedly protected against the deterioration of an ETF’s investment portfolio that might otherwise occur in some cases if an authorized participant were able to deliver or accept non-pro-rata baskets to or from the ETF.

In crafting Rule 6c-11, however, the SEC reversed its previous pro rata rule because it determined that “there are many circumstances” in which “allowing basket assets to differ from a pro rata representation . . . could benefit [an] ETF and its shareholders.” Among other things, the SEC found that limiting flexibility in basket composition could make it “more difficult and costly for authorized participants . . . to assemble or liquidate baskets”—which, in turn, could “result in wider bid-ask spreads and potentially less efficient arbitrage.” In particular, the SEC expressed concern that the old “pro rata basket requirement” made it impracticable for some ETFs to take full advantage of in-kind redemptions and the related tax advantages. This was particularly true, for instance, in the case of actively managed ETFs.

Accordingly, Rule 6c-11 now permits ETFs to deliver “custom” baskets of securities from their investment portfolios to their authorized participants in redemption transactions and permits ETFs to receive “custom” baskets of securities for inclusion in their portfolios from their authorized participants in creation transactions. Basically, a “custom basket” is a group of securities to be removed from or added to an ETF’s investment portfolio, which does not correspond pro rata to the ETF’s portfolio holdings immediately prior to such removal or addition. Similarly, “if an ETF uses different baskets in transactions on the same business day, each basket after the initial basket would be a custom basket” as defined in Rule 6c-11.

An ETF’s permission to use custom baskets under Rule 6c-11 is contingent on the ETF’s implementation of certain operational policies and procedures for assembling the baskets. Those procedural requirements are intended to protect shareholder interests by guarding against portfolio deterioration. In the SEC’s view, Rule 6c-11’s allowance of custom baskets—and the rule’s accompanying procedural safeguards—“will provide ETFs with additional basket flexibility, which . . . could benefit investors through more efficient arbitrage and narrower bid-ask spreads,” while also providing “protections designed to address the risks that such flexibility may present.”

The newfound ability of ETFs to use custom baskets in in-kind creation and redemption transactions makes it far easier for actively managed ETFs to take advantage of the section 852(b)(6) tax break. An ETF with an active management strategy—as opposed to one that passively tracks the performance of a particular market index—can now use in-kind redemption distributions to rebalance its portfolio without recognizing gain on the appreciated securities in which it divests. As a result, “[e]ven [an ETF that pursues] a high-turnover [investment] strategy can now shield its owners from capital gains distributions.”

3. After Rule 6c-11, actively managed ETFs will likely replace actively managed mutual funds as the pooled investment vehicle of choice for high-net-worth taxable investors.

The flexibility to use custom baskets under Rule 6c-11 promises greatly to accelerate the proliferation and growth of actively managed ETFs. Now that active fund managers can better access the tax benefits of in-kind redemptions, affluent taxable investors who seek active investment strategies are likely to favor ETFs over traditional mutual funds. In fact, there are already signs that the demand for active ETFs is swiftly increasing. This stands to reason, of course, given that actively managed funds experience high rates of portfolio turnover and thus realize a significant portion of their accrued capital gains in a typical year. After all, if one holds actively managed investments in a taxable account (rather than a tax-advantaged retirement account), why would one choose “to own a mutual fund that faces annual distributions as high as 10-30%, when [one] can own an ETF that does the same thing and will not distribute capital gains?”

Rule 6c-11, in other words, will almost certainly result in even further migration of high-net-worth individuals from traditional mutual funds to ETFs. In response to the new rule, a number of fund managers have converted their actively managed traditional mutual funds to ETFs. And one high-profile mutual fund firm, which had never offered ETFs before, began to offer various active investment strategies “through the tax-efficient ETF format” in late 2020—even though the firm’s founder “personally opposes the [ETF] tax gambit” under section 852(b)(6). There is every reason to expect these trends to continue. Indeed, some industry watchers suspect that this is exactly what the SEC wants to see happen. One prominent securities attorney, for example, has speculated that, after adopting Rule 6c-11, “the SEC hopes that all mutual funds will move to an exchange-traded option”—with one significant caveat: “[T]hat depends on [the functioning of the] arbitrage mechanisms that the SEC views as necessary to keep unit prices close to NAV.”

C. Rule 6c-11 makes the need for ETF tax reform more urgent, even as it confirms policymakers’ commitment to the nonrecognition rule.

For anyone who seeks to eliminate—or, at least, curtail—the tax-avoidance opportunities that section 852(b)(6) provides to ETFs and their investors, the SEC’s adoption of Rule 6c-11 is instructive in two respects. First, it lends a new urgency to the effort to pare back the section 852(b)(6) tax break. As more wealthy investors flock from actively managed mutual funds to actively managed ETFs, even more realized investment gains will be sheltered from taxation if section 852(b)(6) remains in its current form. This, in turn, will increase both revenue losses to the Treasury and concentrations of income and wealth among the already-affluent. And, if predictions of ETFs eventually replacing mutual funds turn out to be true, section 852(b)(6)’s detrimental effects on tax revenue and on income and wealth inequality will become even greater still. In sum, unless the current tax break for ETFs and their investors is somehow curbed in short order, the changes that Rule 6c-11 is likely to bring to the pooled-investment-fund market will quickly exacerbate section 852(b)(6)’s harms.

Second, the adoption of Rule 6c-11 demonstrates the SEC’s ardor for the price-arbitrage mechanism facilitated by in-kind creation and redemption transactions, and the SEC’s intent to ensure that all ETFs and their authorized participants can engage in those transactions. It underscores that, to stand a chance of being seriously entertained by policymakers, any proposal to reform section 852(b)(6)—to reign in the ETF tax break—must protect that arbitrage mechanism. Rule 6c-11 confirms, in short, that any such proposal must preserve the nonrecognition rule for in-kind redemption distributions, which the SEC considers to be central to the mechanism’s functioning.

IX. Revisiting a Proposed Rule to Reduce an ETF’s Basis in Its Portfolio Assets by the Amount of Its Unrecognized Gain on In-Kind Redemption Distributions

A. How the proposed rule would work, and why it would make even more sense now.

Given that the nonrecognition rule for in-kind redemption distributions must remain intact, there is only one way to limit the tax avoidance by ETFs and their shareholders that section 852(b)(6) makes possible: The gain that an ETF realizes, but does not recognize, when it distributes appreciated portfolio securities to a redeeming authorized participant must be preserved within the ETF itself. Then, the ETF must be made to recognize that gain—or, at least, a substantial portion of it—when the ETF subsequently disposes of other portfolio securities in taxable (non-redemption-related) sales or exchanges. At that point, the ETF would have to distribute the recognized gain to its shareholders as a capital gain dividend to avoid entity-level tax.

This approach would ensure that an ETF’s shareholders will always be taxed on at least a significant part of the investment gains that were sheltered from tax at the time when the ETF satisfied its authorized participant’s redemption demand in kind. Even though such a mechanism might not capture all ETF investment gains that currently go untaxed, it would create two positive outcomes: It would considerably reduce the portion of such gains that escape taxation forever, and it would meaningfully shorten the period of deferral for such gains that ultimately are taxed. Both of those effects would increase tax revenue, reduce inequities in tax administration, and inhibit increases in income and wealth inequality.

To that end, I have previously proposed an amendment to section 852(b)(6) to provide for “a pro tanto reduction in the basis of [an ETF’s] remaining portfolio assets to the extent of any unrecognized gain that the [ETF] realizes in a distribution of appreciated assets to a redeeming shareholder.” Now, more precisely, I suggest an addition of the following italicized language at the end of section 852(b)(6) as currently enacted:

Section 311(b) shall not apply to any distribution by a regulated investment company to which this part applies, if such distribution is in redemption of its stock upon the demand of the shareholder. If a regulated investment company distributes property in a transaction to which this paragraph applies and the fair market value of such property exceeds its adjusted basis (in the hands of the distributing regulated investment company), then the basis of any securities or other assets or positions held for investment by such regulated investment company shall be reduced by the amount of such excess. The reduction of the basis of such properties shall be made in proportion to the relative bases of such properties.

Similar basis-reduction rules have been used in the Code to preserve (and ultimately tax) previously unrecognized gain in other contexts. In particular, this proposed rule for ETFs is patterned after section 362(c)(2), which provides for basis reductions in corporate assets following a nontaxable capital contribution from a nonshareholder.

Although I first proposed a basis-reduction rule for ETF assets in 2017, two developments since then make an even stronger case for implementing such a rule now. First, more recent reporting on the widespread use of heartbeat trades by virtually all ETFs signals that even passively managed ETFs need to dispose of significant amounts of portfolio securities for reasons other than to satisfy their authorized participants’ routine redemption demands. Second, following the implementation of Rule 6c-11 at the end of 2019, the actively managed ETF market is likely to grow rapidly in the near future—and those new active ETFs, of course, will experience higher rates of portfolio turnover. These facts indicate that (1) even in the current market, ETFs dispose of more portfolio securities than may previously have been thought in transactions that should be treated as taxable and (2) ETFs are likely to dispose of an even greater amount of portfolio securities in such transactions in coming years as active ETFs gain in market share. It follows that a basis-reduction mechanism—pursuant to which ETF redemption-related gains would be recognized upon a subsequent taxable disposition of ETF portfolio assets—would (1) be far more effective at taxing those gains, even under current conditions, than some skeptics have suggested and (2) become even more effective at taxing such gains in future years as the ETF market continues to evolve.

Here is a simplified example of how the ETF basis-reduction rule would work:

Example 6: An equity ETF has three shares of stock in its investment portfolio: The ETF’s share of X Corp. stock has a fair market value of $70, and its original cost basis in the X Corp. share is $60. The ETF’s share of Y Corp. stock has a fair market value of $60, and its original cost basis in the Y Corp. share is $40. The ETF’s share of Z Corp. stock has a fair market value of $50, and its original cost basis in the Z Corp. share is $20. Assume that the ETF distributes its share of Z Corp. stock to one of its authorized participants, in satisfaction of the authorized participant’s demand for redemption of an ETF creation unit.

In this example, the ETF does not recognize the $30 of gain that it realizes upon its distribution of the Z Corp. share because of section 852(b)(6). Under the proposed basis-reduction rule, however, the ETF’s basis in its X Corp. share is adjusted from $60 to $42, and the ETF’s basis in its Y Corp. share is adjusted from $40 to $28.

As a result, if the ETF subsequently disposes of its X Corp. share in a taxable sale or exchange—and receives, in return, cash or property of value equaling the X Corp. share’s $70 fair market value—the ETF will recognize tax gain of $28. This is $18 more than the $10 of gain that the ETF would have recognized on the disposition of the X Corp. share in the absence of the basis reduction. Thus, by virtue of the basis-reduction rule, when the ETF disposes of the X Corp. share, it recognizes $18 of the $30 in gain that it realized when it distributed the Z Corp. share to the redeeming authorized participant. The ETF will have to distribute a $28 capital gain dividend to its shareholders to avoid being taxed at the entity level on the recognized gain on the X Corp. share. When the ETF’s shareholders are taxed on the $28 capital gain dividend, they will be taxed on $18 of the $30 in previously unrecognized redemption-related gain on the Z Corp. share.

Similar results will obtain if the ETF disposes of its Y Corp. share for its fair market value of $60 in a taxable sale or exchange. The ETF will recognize $32 of tax gain on the disposition, instead of the $20 of gain that the ETF would have recognized were it not for the basis-reduction rule. Thus, when it disposes of the Y Corp. share, the ETF recognizes $12 of the $30 in gain that it realized when it made the in-kind redemption distribution of the Z Corp. share. By extension, when the ETF’s shareholders receive the $32 capital gain dividend that follows from the ETF’s gain recognition on the Y Corp. share, those shareholders will be taxed on $12 of the $30 in redemption-related gain on the Z Corp. share.

B. A response to a critic: Why the basis-reduction rule is still the best option for curbing ETF tax avoidance.

Shortly after I first proposed a basis-reduction rule for ETF portfolio assets, another critic of the ETF tax break under section 852(b)(6), Professor Jeffrey Colon, criticized the proposal on several grounds. A brief examination of Professor Colon’s criticisms, however, shows that a basis-reduction rule in fact continues to be the best option for curbing the tax avoidance that section 852(b)(6) currently makes possible for ETFs and their shareholders.

1. The basis-reduction rule appropriately balances tax policy goals and securities regulation concerns.

Professor Colon’s first objection was that the basis-reduction proposal “is inconsistent with sound tax policy principles.” In particular, he argued that, “[a]lthough the basis reduction proposal preserves fund-level gain, there does not seem to be a justifiable tax policy reason to defer the [recognition of an ETF’s] gains [on redemption-related distributions of appreciated portfolio securities] when the appreciated assets leave corporate solution.” It is true, as noted above, that section 852(b)(6)’s nonrecognition rule is not supported by any rationale rooted in tax policy or income tax theory. Yet it is equally true that Congress, the SEC, and apparently even the Service consider there to be significant nontax justifications for that rule. Professor Colon seems to dismiss the grounds on which policymakers support section 852(b)(6) as misguided or irrelevant, but it is Professor Colon’s dismissal of policymakers’ concerns that is truly irrelevant. Notwithstanding Professor Colon’s skepticism of the underlying policy rationales, Congress presumably would not want to remove the protections for traditional mutual funds that section 852(b)(6) was originally enacted to provide. Nor would Congress likely want to threaten the price-arbitrage mechanism which the SEC regards as necessary to ETFs’ existence, and which is facilitated largely by tax-free in-kind redemption distributions.

As I have written previously, “[i]n retrospect, as originally enacted, section 852(b)(6) should not have permitted RICs to effect tax-free in-kind redemptions in the normal course.” That is to say, when originally drafted, “section 852(b)(6) should have been circumscribed in a way that would have prevented the creation of ETFs in the first place.” Nevertheless, we are obviously now way past that point. Policymakers cannot be expected to entertain any tax reform that they think might threaten the viability of the $5-trillion-plus, and growing, U.S. ETF market. The recent adoption of Rule 6c-11 by the SEC plainly manifests a governmental policy to encourage, and not to impede, accelerated growth of the ETF market. If anything, the SEC may well wish to see ETFs replace traditional mutual funds altogether—as long as the price-arbitrage mechanism that maintains ETF share-price parity with NAV-per-share continues to function.

Against this backdrop, tax scholars who seek ways to curb ETF tax avoidance have a straightforward choice: We can call for an end to the section 852(b)(6) nonrecognition rule—knowing that it is unlikely to happen—and celebrate our fealty to “sound tax policy principles” in the pages of law reviews. Or we can try to a fashion an alternative reform that, while perhaps less theoretically “pure,” is more compatible with the policy goals of protecting mutual funds during severe market downturns and preserving ETFs’ viability as pooled investment vehicles. The proposed basis-reduction rule for ETF assets follows the latter approach.

2. The basis-reduction rule could ensure taxation of substantial ETF investment gains.

Professor Colon’s second objection was that a basis-reduction rule for ETF assets would “not eliminate . . . the tax competitive advantages of ETFs.” He predicted that “the deferral of unrecognized gains by ETFs” through the basis-reduction mechanism “would be tantamount for forgiveness” of tax on those gains. The reason for this, he maintained, was that “ETFs currently rarely sell assets, except in the case of mergers or changes in the index the ETF is tracking.”

These criticisms missed the mark when Professor Colon made them in 2017, and they are even further off-base now. If ETFs rarely sell assets, as Professor Colon suggests, it is only because they engage in abusive heartbeat trades instead. As described above, heartbeat transactions are, in substance, taxable exchanges of securities between an ETF and its authorized participant. These transactions occur when an ETF wants to rebalance its portfolio; they are not done in response to investor demand for ETF shares that would give rise to an ordinary creation or redemption transaction. Nonetheless, in form, heartbeats are structured as offsetting in-kind creation and redemption transactions. By using the heartbeat structure, ETFs are able to avoid gain recognition on dispositions of portfolio securities done solely to make changes in portfolio composition. Those dispositions should be treated as taxable, rather than as tax-free in-kind redemption distributions.

Nearly all ETFs reportedly engage in substantial amounts of heartbeat trades each year, and they have apparently been doing so for at least the last 20 years. This includes not only actively managed ETFs, but also passive, index-based ETFs. As explained above, these heartbeat transactions show that virtually all ETFs routinely dispose of considerable amounts of portfolio securities for reasons other than to satisfy routine (or “legitimate”) redemption demands. Therefore, even before taking account of any anticipated increase in the active ETF market, the proposed basis-reduction rule for ETF assets has the potential to trigger recognition of significant portions of previously excluded ETF redemption-related realized gains. Such recognition could occur whenever the ETF subsequently makes a routine disposition of portfolio securities to adjust the composition of its investment portfolio. For this to work, the only other requirement would be for the formal creation and redemption transactions in ETF heartbeat trades to be disregarded and for those trades instead to be treated as the taxable exchanges that they substantively are.

Of course, many ETFs that make in-kind redemption distributions will not dispose of all of their remaining portfolio securities in rebalancing transactions. A passively managed ETF, for example, typically rebalances its investment portfolio only in response to a change in composition of the particular market index that it tracks. As a result, the basis-reduction rule would not subject all redemption-related gains of all ETFs to taxation. To that extent, the basis-reduction rule is perhaps an imperfect solution to the problem of ETF tax avoidance—even though it is the best solution that is feasible, given the necessary constraint of retaining the nonrecognition rule for in-kind redemption distributions. Nevertheless, the basis-reduction rule could still ensure the taxation of a decidedly substantial portion of an ETF’s previously excluded gains on redemption distributions—even if the ETF in question disposes of only a modest percentage of its portfolio securities in rebalancing-related sales or exchanges.

As an illustration, consider the following example:

Example 7: An equity ETF has four shares of stock in its investment portfolio: The ETF’s share of W Corp. stock has a fair market value of $55, and its original cost basis in the W Corp. share is $50. The ETF’s share of X Corp. stock has a fair market value of $60, and its original cost basis in the X Corp. share is $50. The ETF’s share of Y Corp. stock has a fair market value of $70, and its original cost basis in the Y Corp. share is $50. The ETF’s share of Z Corp. stock has a fair market value of $80, and its original cost basis in the Z Corp. share is $50. Assume that (1) the ETF first distributes its share of Z Corp. stock to one of its authorized participants in satisfaction of the authorized participant’s demand for redemption of an ETF creation unit; (2) the ETF next distributes its share of Y Corp. stock in satisfaction of the authorized participant’s demand for redemption of an ETF creation unit; and (3) the ETF thereafter disposes of its share of W Corp. stock in a taxable sale or exchange for the W Corp. share’s fair market value of $55 as a part of a rebalancing of its investment portfolio. Also assume that, at all relevant times, the ETF retains its share of X Corp. stock in its portfolio.

In this example, the ETF realizes, but does not recognize, $30 of gain when it distributes the Z Corp. share to its authorized participant. Immediately thereafter, under the basis-reduction rule, the ETF’s respective basis in each of its W Corp. share, its X Corp. share, and its Y Corp. is reduced from $50 to $40. Then, when the ETF distributes the Y Corp. share to the authorized participant, the ETF realizes, but does not recognize, $30 of gain. Immediately after that distribution, the ETF’s respective basis in each of its W Corp. share and its X Corp. share is reduced again—from $40 to $25. When the ETF subsequently disposes of the W Corp. share in a taxable transaction, it thus recognizes $30 of gain. This is $25 more than the $5 of gain that the ETF would have recognized on the W Corp. share disposition in the absence of the basis reductions. At the same time, were it not for the basis reductions, the ETF’s previously unrecognized gains on the distributions of the Y Corp. share and the Z Corp. share would have been $20 and $30, respectively—for a total of $50 in unrecognized gain.

In Example 7, the ETF disposes of only approximately 21% (by value) of its portfolio securities in taxable transactions. Perhaps even more significantly, the securities that the ETF distributes in the redemption transactions have (in aggregate) ten times as much inherent gain as has the security that the ETF disposes of in the taxable sale. Nonetheless, the basis-reduction rule still causes the ETF ultimately to recognize 50% (or $25) of its $50 in redemption-related gains that otherwise would have gone unrecognized. This shows that the basis-reduction rule can be a powerful tool for recovering tax on ETFs’ redemption-related gains, even in the case of ETFs that engage in only limited portfolio rebalancing.

Moreover, the basis-reduction rule’s prospective efficacy at taxing ETF redemption-related gains will only increase as the active ETF market expands following Rule 6c-11. Actively managed ETFs have notably higher rates of portfolio turnover than index-based ETFs. Indeed, some actively managed funds have turnover rates of 100% or more—meaning that they replace literally all of the securities in their investment portfolios each year, if not even more frequently. In other words, active ETFs dispose of even more portfolio securities in transactions that are properly to be treated as taxable. Thus, as active ETFs gain in market share, the basis-reduction rule could cause the recognition of even more of ETFs’ previously excluded gains on redemption distributions. Accordingly, even Professor Colon conceded that his critique of the basis-reduction rule would “change if the [assets under management] of actively managed ETFs [were to] grow . . . .” Once again, however, the abusive practice of heartbeat trades would need to be shut down in order for the basis-reduction rule to work.

3. Taxing a RIC’s shareholders on gains that accrued prior to the shareholder’s investment in the fund.

Professor Colon’s third objection to the proposed basis-reduction rule was that it would “exacerbate[] the problem of matching the taxable gains of a RIC to the shareholders who have economically benefitted from those gains.” This problem arises directly from the Subchapter M requirement that a RIC must distribute to its shareholders the capital gains it recognizes on dispositions of its portfolio assets in order to avoid entity-level tax on such gains. In some cases, pursuant to that requirement, a shareholder receives a dividend containing gains that accrued on a RIC’s portfolio assets prior to the shareholder’s investment in the RIC. When the shareholder is taxed on her capital gain dividend in such a case, she is thus “taxed on fund income that accrued prior to [her] investment.”

When an investor purchases RIC shares, the accrued-but-unrealized gain inherent in the RIC’s portfolio assets at that time is reflected in the purchase price. Because the investor pays for her proportionate “slice” of that gain when she buys her shares, she is taxed on an amount that is—from her perspective—“noneconomic” or “phantom” income when she receives the gain in a dividend from the RIC. As another commentator puts it, the most obvious example is “the case of a mutual fund shareholder who unwittingly buys into a fund just before it declares a dividend and bears tax on income that accrued prior to her investment in the fund.”

The tax that the shareholder pays on the distribution of such noneconomic or phantom income is essentially an acceleration of tax that the shareholder will owe on her actual gain (if any) when she eventually sells her shares in the fund. This is because the reduction in the fund’s NAV following the distribution will reduce the gain (or increase the loss) that the shareholder would otherwise recognize when she subsequently sells her fund shares. To that extent, this is a timing issue—although, because of the time-value of money, the shareholder will never fully be made whole for the earlier tax on the phantom income when she later sells her shares. On the plus side, a prospective investor can at least mitigate—if not avoid entirely—this problem of tax acceleration by waiting to enter a fund until after “the fund’s ex-dividend date if the fund is planning to make a substantial distribution.”

As we have seen, section 852(b)(6) enables ETFs to avoid distributing virtually any capital gain dividends to their shareholders. Consequently, the occasional taxing of RIC shareholders on noneconomic gains is currently unique to traditional mutual funds. That notwithstanding, Professor Colon seems to have been curiously focused on this particular Subchapter M flaw when formulating his proposal for ETF tax reform. To that end, he has argued for “Congress [to] prohibit ETFs and perhaps all RICs that offer creation-type shares from being subject to Subchapter M and instead require them to be subject to certain partnership tax rules.”

4. The proposal to treat ETFs as partnerships is misguided and unworkable.

If current law were changed to permit RICs to be treated as partnerships for federal income tax purposes, as Professor Colon has proposed, ETF shareholders could be treated as partners in their ETF partnerships. Under this new regime, Professor Colon would have ETFs perform certain complex partnership tax accounting procedures to guard against taxing their shareholder-“partners” on gains that accrued prior the partners’ entry into the “partnership.” For one thing, whenever an ETF’s authorized participant acquires a creation unit from the ETF in a creation transaction, Professor Colon would have the ETF perform a revaluation of its assets pursuant to the section 704(c) regulations. Moreover, whenever an ordinary ETF shareholder-“partner” purchases ETF shares in the secondary market, Professor Colon would have the ETF make a section 754 election and perform a corresponding section 743(b) basis adjustment. According to Professor Colon, applying these partnership tax rules “to ETFs could eliminate many of the current deficiencies of Subchapter M, albeit with a concomitant, very significant increase in administrative complexity.”

The idea of treating ETFs as partnerships is, in a word, awful. As a threshold matter, and as other commentators have also observed, subjecting ETFs to the Subchapter K partnership taxation regime would cause intolerable levels of administrative complexity for both the funds themselves and their shareholder-“partners.” And the obstacles would be even more insuperable as a result of the revaluations and section 743(b) basis adjustments that Professor Colon recommends. This alone is an ample ground for dismissing Professor Colon’s partnership tax proposal for ETFs.

Even more germane to the instant discussion, however, is that treating ETFs as partnerships would destroy the price-arbitrage mechanism that ensures ETF share-price parity with NAV-per-share. A partnership does not recognize gain or loss on a distribution of partnership property to a partner. Instead, any such gain or loss generally remains inherent in the property following the distribution to the partner, by operation of special rules that govern the partner’s initial basis in the property. By virtue of those special basis rules, when and if the partner subsequently disposes of the property in a taxable sale or exchange, the partner generally recognizes the gain or loss that the partnership avoided at the time of the distribution.

Thus, if an ETF were treated as a partnership, it would still be able to avoid recognition of gain on a distribution of appreciated portfolio securities to a redeeming authorized participant—just as under section 852(b)(6). Under partnership tax rules, however, the authorized participant would generally have to recognize that gain when it resold those securities in the market. As explained above, if the tax burden on such ETF redemption-related gains were shifted to the ETFs’ authorized participants, those authorized participants would no longer have an economic incentive to engage in certain in-kind redemption transactions that are necessary to preventing ETF share prices from trading at a discount to NAV-per-share. And if authorized participants became unwilling to participate in those transactions, the price-arbitrage mechanism that the SEC views as vital to ETFs’ existence would simply fall apart. For this reason, especially, Professor Colon’s ETF partnership tax proposal has to be rejected.

5. The basis-reduction rule remains the most promising solution to the problem of ETF tax avoidance.

In the spirit of full disclosure, Professor Colon is correct that my proposed basis-reduction rule for ETF assets would cause ETF shareholders occasionally to be taxed on gain that accrued on the ETF’s portfolio securities prior to the shareholder’s investment. That would be an unavoidable consequence of finally requiring ETFs to recognize their realized investment gains and to distribute those gains to their shareholders as capital gain dividends. This flaw—which is intrinsic to Subchapter M—is hardly fatal to the basis-reduction proposal, however. Mutual fund shareholders have long lived with this imperfection, and the problem can be managed through the timing of new investments in a fund. To be clear, the acceleration of tax liability that results from dividend distributions of gain that accrued prior to a shareholder’s investment in a fund is plainly not optimal, and possible solutions to the problem within the Subchapter M framework are well worth exploring. Yet, as vexing as the issue of pre-investment-gain distributions may be, it is a decidedly less serious policy concern than allowing ETFs and their shareholders to avoid taxation of some or all of their investment gains altogether—which is what often happens under section 852(b)(6).

At the same time, neither of the other proposals for combating ETF tax avoidance that have been proffered thus far—repealing section 852(b)(6) or converting ETFs to a partnership tax regime—would work. Simply put, the basis-reduction rule remains the best option for curbing ETF tax avoidance under section 852(b)(6). It is the most practicable alternative for limiting the revenue losses and the exacerbations of income and wealth inequality that such tax avoidance currently engenders. Perhaps even Professor Colon has since come around on this point: In remarks during a 2019 conference, he reportedly suggested the basis-reduction rule as a viable approach to ETF tax reform.

X. The Starting Point of ETF Taxation Reform Must Be to Stop Heartbeat Trades

An essential first step in any effort to stop ETF tax avoidance is to put an end to abusive heartbeat trades. For one thing, as noted above, ending heartbeat transactions would be necessary to implementing the proposed basis-reduction rule for ETF assets. But even if the basis-reduction rule were never adopted, stopping heartbeat trades would nonetheless be crucial to any attempt at ETF tax reform. Heartbeat transactions enable ETFs to circumvent gain recognition on dispositions of appreciated securities done for portfolio rebalancing rather than for satisfying legitimate redemptions, and they account for a significant portion of the investment gains that ETFs realize but do not recognize each year. Putting an end to heartbeat trades—by recharacterizing them as the taxable exchanges of securities that they substantively are—would thus considerably limit ETFs’ opportunities to exploit section 852(b)(6) for tax avoidance.

Happily, eliminating heartbeat trades would be as easy as it is important. No act of Congress would be necessary. Instead, the Service could simply disallow the current treatment of heartbeat transactions—as separate in-kind creation and redemption transactions—immediately, if it wanted to.

A well-known and longstanding judicial doctrine provides that the substance, rather than the form, of a transaction determines the proper tax treatment of the transaction. The substance-over-form doctrine essentially requires that, where the formalities of a transaction exist solely to alter the participants’ tax liabilities, those formalities will be disregarded. Rather, in such cases, courts will look to the substance of the transaction to determine the proper tax consequences to the parties. One of the most frequently applied variations of the substance-over-form doctrine is the step transaction doctrine. This doctrine essentially holds that formally distinct separate transactions will be treated, for tax purposes, as steps in one integrated transaction if the parties’ purpose when taking the first step was to achieve the result that occurs after taking the last step.

The Service often asserts the step transaction doctrine “to attack allegedly artificial divisions of transactions by taxpayers trying to characterize recognition events—that is, events that result in a taxable sale or exchange—as nonrecognition events.” Indeed, the Supreme Court has held that the doctrine applies particularly when parties enter into a series of formally distinct transactions, the ultimate purpose of which is the purchase and sale (or exchange) of property. In such cases, the formalities of the transaction structure are to be disregarded and the arrangement is to be treated as a single transaction for acquisition of the property.

There are three primary formulations of the step transaction doctrine. Courts differ as to the particular formulation that they apply, and some courts apply more than one formulation (or effectively combine the formulations) in a given case. The first of the three formulations is the “binding commitment test.” Under this version, a formal series of transactions will be combined if, when the first step is taken, a binding agreement exists to take the subsequent steps. The second formulation is the “end result test.” Under this version, formally separate transactions will be combined if the court determines that those formal transactions were prearranged components of a single plan pursuant to which the parties intended from the outset to reach a particular end result. The third formulation of the doctrine is the “interdependence test.” This is essentially a variation of the end result test. Under the interdependence test, a court considers whether the formally separate transactions (or steps) are so interdependent that the legal relations created by the first step would have been fruitless without the later steps.

A legitimate creation transaction (i.e., one done outside the context of a heartbeat trade) happens when ordinary investors’ strong demand for an ETF’s shares (manifested by increased purchases in the market) temporarily drives the share price above the fund’s NAV-per-share. The ETF’s authorized participant enters into the creation transaction to take advantage of the arbitrage opportunity that the temporary price disparity creates. Similarly, a legitimate redemption transaction occurs when weakening investor demand (resulting in heavier selling of the ETF’s shares in the market) temporarily drives an ETF’s share price below the ETF’s NAV-per-share. The ETF’s authorized participant enters into the redemption transaction to profit, once again, from the arbitrage opportunity that the price disparity sets up. In sharp contrast, however, the formal “creation” and “redemption” steps of a heartbeat trade between an ETF and an authorized participant do not arise organically out of any such increases or decreases in investor demand for the ETF’s shares. Rather, those steps are preplanned by the parties solely to facilitate the ETF’s exchange of one basket of portfolio securities for another—in a way that allows the ETF to circumvent recognition of any gain on the securities of which it disposes in the exchange.

Given these economic realities, the Service could invoke any (or all) of the three step transaction tests to disregard the formal “creation” and “redemption” steps of a heartbeat trade and recharacterize the transaction as a taxable exchange of securities: The ETF’s counterparty in a heartbeat trade does not invest in a creation unit with any intent to resell the ETF’s shares at a profit. Instead, at the inception of the trade, the ETF and its counterparty enter into a binding agreement, pursuant to which the counterparty will redeem the same creation unit back to the ETF almost immediately thereafter. The intended end result of a heartbeat trade is an exchange of the basket of securities delivered by the counterparty in step one (the “creation” step) for the basket of securities delivered by the ETF in step two (the “redemption” step). The parties’ arrangement in the creation step of a heartbeat transaction would thus be fruitless without the arrangement in the simultaneously arranged redemption step. In other words, the two steps are inextricably interdependent. There are, of course, no nontax business reasons for the parties to structure such an exchange of securities in this manner. The offsetting creation and redemption of ETF shares in a heartbeat trade are superfluous steps done strictly to enable the ETF to avoid gain recognition on the exchange by virtue of section 852(b)(6).

To see the straightforward result of applying the step transaction doctrine to a heartbeat trade, let us return briefly to Example 5 above. If the Service invoked the doctrine in the Example 5 fact pattern, it would simply disregard both (1) the ETF’s exchange of its creation unit for the shares of XYZ Corp. stock delivered by the counterparty in the contrived creation step and (2) the ETF’s exchange of its shares of ABC Corp. stock for the return of that same creation unit, shortly thereafter, in the contrived redemption. The Service would recognize that the two steps were pre-agreed by the ETF and its counterparty, that the first step would not have occurred if the parties had not also arranged the second step, and that the ultimate purpose and result of the two steps was for the ETF effectively to have exchanged the ABC Corp. shares for the XYZ Corp. shares. Accordingly, the Service would recharacterize the two steps as one taxable exchange. The ETF would then be required to recognize the gain that it realized on the disposition of its appreciated ABC Corp. shares.

Although the Service has acknowledged its awareness of ETFs’ heartbeat trades, it has thus far refrained from challenging them as abusive. The likely reason for this inaction, in short, is a concern about doing anything that might impede ETFs’ operations. In justifying the new grant of permission for ETFs to use custom baskets, for instance, the SEC recently touted the use of in-kind creation and redemption transactions for portfolio rebalancing as more “efficient” than cash purchases and sales of portfolio securities. The Service may be reluctant to interfere with those “efficiencies” by attacking heartbeat transactions.

Treating heartbeat trades as taxable exchanges, however, would not harm ETFs’ ability to operate. In particular, disallowing the treatment of a heartbeat trade’s redemption step as a transaction to which section 852(b)(6) applies would not have any adverse effect on the price-arbitrage mechanism for maintaining parity of an ETF’s share price with the ETF’s NAV-per-share. As noted above, neither the creation step nor the redemption step of a heartbeat transaction arises in response to any changes in ordinary investors’ demand for an ETF’s shares. Therefore, heartbeat trades play absolutely no role in the price-arbitrage mechanism. Despite advocating for greater “efficiency” in ETF portfolio rebalancing, the SEC presumably is not endorsing an ETF’s manufacture of artificial creations and redemptions—unrelated to actual investor demand—done solely for the purposes of tax avoidance. In any event, as commentator Lee Sheppard helpfully reminds us, “[t]he SEC doesn’t enforce the tax law.” Instead, the Service is responsible for administering federal tax law, and it has a duty to thwart any opportunity for escaping capital gains taxation that heartbeat trades present. As an initial part of what will hopefully become broader ETF taxation reform, the Service should act on that duty.

XI. Conclusion

Congress enacted section 852(b)(6) to protect mutual funds from extraordinary portfolio depletion and deterioration in times of heavy redemptions during sharp market downturns. Not long thereafter, innovative financial-product structurers created a new pooled investment vehicle—the ETF—around the nonrecognition rule for in-kind redemption distributions. ETFs combine the tradability of closed-end funds with the share-price stability of open-end mutual funds. This attractive combination of features depends in large part on ETFs’ ability routinely to distribute appreciated portfolio securities to redeeming authorized participants.

The same arcane tax provision on which ETFs depend to ensure that the market price of their shares reflects the underlying value of their investment portfolios also provides an unwarranted tax break that disproportionately benefits high-income, high-net-worth shareholders. Section 852(b)(6) not only allows ETF investors to defer the payment of tax on their respective shares of an ETFs’ realized investment gains, but it also often permits investors permanently to avoid ever being taxed on those gains. This results in billions of dollars of losses in federal tax revenue each year, and it violates fundamental norms of tax fairness. Because it helps wealthy ETF shareholders to compound additional gains at a faster rate, section 852(b)(6) exacerbates a problem that is already inherent in the realization rule’s preferential treatment of capital gains: It furthers the undertaxation of income from capital and contributes to increases in income and wealth inequality.

Now that the ever-growing U.S. ETF market has surpassed $5 trillion in assets under management, it is far too late simply to repeal section 852(b)(6). Removing the nonrecognition rule for in-kind redemption distributions could destroy ETFs’ ability to operate, and the potential disruption to the capital markets if ETFs suddenly became nonfunctional is an unacceptable risk. Nonetheless, something needs to be done to stop—or, at least, reduce—ETF tax avoidance. That need has recently become even more urgent, now that Rule 6c-11’s allowance for “custom baskets” makes it easier for actively managed ETFs to take advantage of section 852(b)(6)—and, in turn, increases the prospects of ETFs supplanting mutual funds altogether.

As it happens, there are approaches to ETF tax reform that would curb ETF tax avoidance without threating ETFs’ continued existence. The first crucial step in any such reform would be for the Service to employ the step transaction doctrine to treat ETFs’ so-called “heartbeat” trades as the taxable exchanges of securities that they really are. Such nakedly abusive “heartbeat” transactions are simply sham tax-avoidance maneuvers that should be rejected out of hand. The Service could stop those abuses immediately without adversely affecting ETFs’ ability to maintain share-price parity with NAV-per-share. Once heartbeat trades are disallowed, the next step would be for Congress to amend section 852(b)(6) to add the basis-reduction rule for ETF assets that this Article recommends. The combination of these reforms would greatly reduce the portion of ETF investment gain that ultimately escapes taxation. Policymakers should be urged to implement these measures now.