The Tax Trap Inside the BOXX

In recent weeks, an exchange-traded fund called Alpha Architect 1-3 Month Box ETF (trading under the ticker symbol “BOXX”) has garnered attention in the financial press for its supposedly tax-efficient method of generating nearly risk-free returns.¹ BOXX purports to produce Treasury bill-like yields while deferring any income tax liabilities until shareholders sell their ETF shares. And when shareholders do sell, their gains — according to BOXX promoters — will be taxed as capital gain rather than ordinary income, an additional advantage over ordinary income-generating Treasury bills.

If all that sounds too good to be true, it’s because it most likely is too good to be true. BOXX’s strategy runs headlong into two code provisions — section 1258 and section 1092 — that are designed to prevent the sorts of maneuvers that BOXX seeks to exploit. In light of these provisions, BOXX shareholders in most states are likely to end up in a worse position than if they had simply purchased Treasury bills. For BOXX shareholders, an ostensibly tax-efficient investment may turn into a tax trap.

This article considers the application of the section 1258 conversion transaction rules and the section 1092 straddle rules to BOXX’s too-clever-by-half tax strategy. After examining BOXX’s disclosures to reconstruct the fund’s trades, it evaluates both the fund-level and shareholder-level tax consequences of BOXX’s maneuvers. In contrast to other commentary emphasizing the possibility that capital gains on sales of the BOXX ETF could be recharacterized as ordinary income at the shareholder level,² this article concludes that the IRS would likely succeed in recharacterizing gains from BOXX’s trades as ordinary income at the fund level. Whereas shareholder-level recharacterization would leave BOXX’s deferral benefits in place, fund-level recharacterization would erase both the deferral and rate-arbitrage benefits that BOXX advertises.

I. Building the BOXX

Here’s how BOXX is supposed to work. The fund engages in two parallel trades — a “box spread” on S&P 500 index futures (hence the fund’s name) and a straddle on an individual security (usually the stock of the travel technology company Booking Holdings). To evaluate the fund-level and shareholder-level tax consequences of BOXX activities, we need to delve into the details of these two trades.

A. Box Spread

A box spread is a combination of long and short positions on the same security or index that is designed to generate fixed income-like returns. To illustrate, let’s imagine the S&P 500 index is at 5,000. Let’s also imagine that S&P 500 index options are priced so that one index point equals $1. An investor might make the following four moves in combination:

• buy an S&P 500 index call option with a strike price of 4,500 points, or $4,500 (an “in-the-money call”);

• sell an S&P 500 index put option with a strike price of 4,500 points, or $4,500 (an “out-of-the-money put”);

• buy an S&P 500 index put option with a strike price of 5,500 points, or $5,500 (an “in-the-money put”); and

• sell an S&P 500 index call option with a strike price of 5,500 points, or $5,500 (an “out-of-the-money call”).

“In-the-money” refers to the fact that the in-the-money options would be valuable if the options expired today when the S&P 500 index is trading at 5,000 (that is, it would be attractive to buy the S&P 500 index at 4,500 or sell it at 5,500). “Out-of-the-money” refers to the fact that those options would be worthless today but may become valuable if the S&P 500 index moves up or down before the expiration date.

For the box spread, the fund typically uses European-style options, which cannot be exercised or assigned before the expiration date (as opposed to American-style options, which can be exercised or assigned in advance of expiration). BOXX’s materials indicate that the fund typically purchases options with one-month and three-month terms.

The magic of the box spread is that whatever value the S&P 500 trades at when the options expire, the investor will have a return of $1,000. For example, if the S&P 500 remains at 5,000, in a month the investor will have a gain of $500 on the in-the-money call and a gain of $500 on the in-the-money put, for a total gain of $1,000. Likewise, if the index is at 4,000, the investor will have a loss of $500 on the out-of-the-money put and a gain of $1,500 on the in-the-money put, for a total gain of $1,000. And if the index is at 6,000, the investor will have a gain of $1,500 on the in-the-money call that it bought and a loss of $500 on the out-of-the-money call that it sold, again for a total gain of $1,000.

In a highly liquid market like the S&P 500 index options market, the net price of the four positions (that is, the cost of the options that the investor must buy minus the proceeds from the options that it will sell) will typically be slightly less than $1,000. The reason is that the box spread delivers $1,000 at expiration, and given the time value of money, a guarantee of $1,000 in one to three months is worth slightly less than $1,000 now. BOXX’s strategy relies on the assumption that the yield on its net investment in the options basket will match or slightly exceed the yield on Treasury bills with similar maturity dates. (The yield on the box spread might slightly exceed the yield on Treasury bills to compensate for the remote risk that the options exchange will default.)

For tax purposes, S&P 500 index options are typically treated as section 1256 contracts, which give rise to capital gain or loss. Under section 1256, 60 percent of gain or loss is treated as long-term capital gain or loss, and the remaining 40 percent is treated as short-term capital gain or loss.

B. Booking Holdings Straddle

In addition to the box spread, BOXX buys and sells call and put options on an individual security. For this purpose, BOXX appears to be using calls and puts on Booking Holdings, the owner of Booking.com, Priceline.com, OpenTable, and several other travel and hospitality websites. According to its holding disclosures, the fund appears to be buying calls and selling puts with identical strike prices.³ For example, as of this writing, BOXX owned 6,000 call options on Booking Holdings with a strike price of approximately $3,430 and an expiration date of April 19, and it had sold 6,000 put options on Booking Holdings with the same $3,430 strike price and April 19 expiration date. It also owned 6,000 put options and had sold 6,000 call options with a similar strike price and the same April 19 expiration.

When those options approach their expiration date, either the call options or put options owned by BOXX will be “in the money” (depending on whether Booking Holdings is trading above or below the strike price). BOXX plans to deliver the in-the-money options to an “authorized participant,” or AP, in exchange for shares of BOXX. An AP is a financial institution that is authorized to buy and sell ETF shares in exchange for other securities. Under code sections 311 and 852(b)(6), an in-kind redemption of an AP’s shares in a regulated investment company (such as an ETF) is treated as a nonrecognition event for the regulated investment company. After BOXX disposes of its in-the-money options, it will realize losses on its out-of-the-money options.

C. The ‘Gift’ Inside the BOXX

Here’s how BOXX hopes that it and its shareholders will be taxed. The box spread strategy will give rise to a combination of long-term and short-term capital gain under section 1256. The Booking Holdings straddle will give rise to no gain (because the in-the-money options will be delivered to the AP before expiration) but will generate capital losses. The fund can use the capital losses on the Booking Holdings straddle to offset the long-term and short-term capital gains on the box spread, resulting in a net capital gain of zero.

Thus, BOXX shareholders supposedly will recognize no gain or loss as long as they hold onto their shares. Only when they sell their shares will they recognize any gain, and that gain will be long-term capital gain as long as they hold their shares for more than one year. Voilà, BOXX has transformed Treasury bill returns into preferential-rate capital gains that shareholders can defer as long as they’d like.

Or, at least, that’s the idea.

II. Analyzing the Tax Consequences

BOXX is a regulated investment company, and RICs typically pass their tax liability along to their shareholders through ordinary dividends (which give rise to dividends paid deductions at the fund level) and capital gain distributions. Nonetheless, RICs still must compute taxable income as a corporation would; they still must calculate net capital gain and net capital loss before distributing capital gain to shareholders; and they still must pay tax at the corporate level if their dividends and capital gain distributions do not offset their income.⁴ In theory, the IRS could attack BOXX’s tax strategy at the fund level or the shareholder level, even though most RICs give rise to tax liability only at the shareholder level. As we shall see, a fund-level attack would be especially devastating, as it would destroy all of BOXX’s purported tax benefits.

A. Tax Consequences at the Fund Level

1. The section 1258 conversion transaction rules.

The first problem with BOXX’s tax strategy arises out of section 1258, enacted in 1993 as part of a sweeping deficit reduction package. That provision recharacterizes most or all of the capital gain from a conversion transaction as ordinary income. A “conversion transaction” is defined as a transaction in which (1) “substantially all of the taxpayer’s expected return” is attributable to the “time value of the taxpayer’s net investment in such transaction,” and (2) any one of four additional conditions applies.⁵ Two of those four conditions are immediately relevant here: the case in which the transaction is an “applicable straddle”⁶ and the case in which the transaction is “marketed or sold as producing capital gains” from the expected return on the time value of money.⁷

Based on BOXX’s disclosures, the fund’s box spreads appear to satisfy the first prong of the conversion transaction test. The entire point of the box spread is to achieve a nearly risk-free return reflecting the time value of the net investment in the basket of options. As of this writing, the average yield to expiration on the fund’s options is approximately 5.7 percent, compared to 5.5 percent for Treasury bills with similar maturities. Thus, more than 95 percent of the return on BOXX’s spreads is attributable to the Treasury bill return (a proxy for the risk-free return). While Treasury and the IRS have adopted various definitions of “substantially all” in other regulations (ranging from 70 percent to 95 percent),⁸ BOXX’s spreads surpass the “substantially all” threshold by any definition.

Moreover, the box spread meets at least one of the four conditions in section 1258(c)(2) — it is an “applicable straddle,” defined as “any straddle (within the meaning of section 1092(c)).”⁹ A straddle within the meaning of section 1092(c) arises when a taxpayer holds “offsetting positions” for personal property if there is a “substantial diminution” of the taxpayer’s risk of loss from holding any position by reason of the taxpayer holding other positions.¹⁰ The whole idea of the box spread is that even though any one of the four legs would expose the investor to a risk of loss, the other three legs substantially diminish — indeed, virtually eliminate — that risk of loss.

There is, concededly, one wrinkle in this analysis. Section 1256(a)(4) provides that “if all the offsetting positions making up any straddle consist of section 1256 contracts” — which would be the case for the box spread — then section 1092 “shall not apply with respect to such straddle.” This carveout raises the question of whether a straddle consisting entirely of section 1256 contracts still qualifies as a straddle “within the meaning of section 1092(c)” even though section 1092 does “not apply to” the straddle. As a linguistic and logical matter, the answer would seem to be yes — meaning and application are distinct concepts. However, Treasury and the IRS do not appear to have addressed this question in any regulation, revenue ruling, or even private letter ruling, and one might want to hang one’s hat on more than linguistics and logic.

Fortunately, we can look nearby for guidance. Section 1256(d)(4) states that “the term ‘mixed straddle’ means any straddle (as defined in section 1092(c)),” but only if “at least 1 (but not all) of the positions . . . are section 1256 contracts” and the taxpayer makes an appropriate election. Section 1256(d)(4) thus contemplates that the term “any straddle” under section 1092(c) — on its own and without further qualification — includes a straddle consisting entirely of section 1256 contracts. Otherwise, there would be no need to exclude those straddles from the definition of “mixed straddle” by adding the words “but not all.”¹¹

Moreover, the construction used in section 1258 — in which the phrase “within the meaning of” is paired with a cross-reference to a definition in another provision — is deployed dozens more times throughout the code. These parallel usages shed light on the interpretation of that construction in the section 1258 context. For example, section 45F states that a taxpayer cannot claim the employer-provided child care credit for costs of acquiring a “principal residence (within the meaning of section 121) of the taxpayer.”¹² Section 121 — the exclusion of gain from the sale of a principal residence — allows a taxpayer to elect out of the application of that provision to a specific transaction.¹³ Common sense tells us that Congress did not mean to allow taxpayers to opt out of the “principal residence” limitation in section 45F — a purely elective limitation would be no limitation at all. Congress clearly sought to borrow a definition spelled out in another code provision without bringing along all the qualifications and limitations that might apply to that other code provision. There is no sign that Congress intended for the phrase “within the meaning of” to play a different function in section 1258.

This conclusion is bolstered further by the legislative history of section 1258. In their reports accompanying that provision, the House and Senate tax-writing committees both explained that their goal was to provide for ordinary income treatment of “transactions the economic substance of which is indistinguishable from loans in terms of the return anticipated and the risks borne by the taxpayer.”¹⁴ Neither committee offered any indication that it desired a different result for straddles consisting entirely of section 1256 contracts.¹⁵ To the contrary, both committee reports included an illustration involving a straddle composed entirely of gold futures contracts — contracts that generally would fall within the scope of section 1256.¹⁶ Although the straddle in the illustration did not qualify as a conversion transaction because the taxpayer did not earn an interest-equivalent return, the committees’ careful analysis of the composition of the return would have been unnecessary if straddles consisting entirely of section 1256 contracts were exempt from the conversion transaction rules.¹⁷

Even if BOXX’s spreads somehow escaped classification as a straddle under section 1258(c)(2)(B), the fund would face a further problem under section 1258(c)(2)(C), which — as noted above — applies when a transaction is “marketed or sold as producing capital gains” on a time-value-of-money return. While the box spread is a transaction executed by the fund and not a transaction “sold” to investors, BOXX’s prospectus prominently claims that the box spread will generate capital gains subject to the 60/40 rules for section 1256 contracts.¹⁸ Thus, one might say that the box spread is “marketed . . . as producing capital gains” even though the targets of the marketing (ETF investors) do not buy into the box spread directly.¹⁹ The IRS presumably would prefer to prevail on a section 1258(c)(2)(B) straddle theory rather than a section 1258(c)(2)(C) marketing theory because the latter would leave room for a future fund to skirt the conversion transaction rules by being more circumspect in its marketing materials. But BOXX has done itself no favors — at least on the legal front — by broadcasting its tax arbitrage aims so loudly.

Finally, if this conclusion regarding the application of the conversion transaction rules to box spreads is incorrect, Treasury and the IRS could make quick work of BOXX’s tax arbitrage strategy by invoking their delegated authority under section 1258(c)(2)(D). Under that provision, Treasury and the IRS may — by regulation — classify a transaction as a conversion transaction even if the transaction does not fall within one of section 1258(c)(2)’s other three prongs, so long as “substantially all” of the taxpayer’s expected return reflects the time value of money.²⁰ To be clear, the IRS would have a very strong case — even without any additional regulations — that BOXX’s spreads are conversion transactions under existing law. Further regulatory action under section 1258(c)(2)(D) would add a belt and suspenders to a conclusion that already stands on its own.

When a transaction is classified as a conversion transaction, capital gain is recharacterized as ordinary income to the extent that gain does not exceed 120 percent of the applicable federal rate, compounded semiannually.²¹ For March, the applicable short-term federal rate, compounded semiannually, is 4.66 percent, and 120 percent of the applicable federal rate is 5.59 percent.²² Thus, if the box spread is a conversion transaction, all or nearly all of the fund’s capital gains on the transaction will be recharacterized as ordinary income. In that case, the fund’s capital losses on the Booking Holdings straddle cannot be used to erase that income because corporations cannot offset ordinary income with capital losses.

2. The section 1092 straddle rules.

A second — and independent — problem with BOXX’s tax strategy arises from the application of the section 1092 straddle rules to the fund’s Booking Holdings trades. Note that the capital losses purportedly produced by these trades only become useful to BOXX if the fund escapes ordinary income treatment of its box spread gains. Still, it is worth examining the tax treatment of these trades because other ETFs may be tempted to use similar maneuvers to offset actual capital gains.

There is little doubt that the fund’s offsetting long and short positions in Booking Holdings constitute a straddle when the fund enters into those positions. (Recall that a straddle arises when a taxpayer holds offsetting positions if there is a substantial diminution of the taxpayer’s risk of loss on any position by reason of the taxpayer holding one or more other positions.) The harder question arises when the fund exits the straddle by distributing its gain position to an authorized participant and realizing the loss position. A close reading of the text of section 1092 can shed light on this question.

When enacted in 1981, section 1092 provided that any loss should be taken into account only to the extent that the loss exceeds the “unrealized gain” for offsetting positions.²³ For example, imagine that a taxpayer enters into offsetting long and short positions for the same property. After 364 days, assuming that the property moves up or down in value, one of those positions will have a built-in gain and the other will have a built-in loss. The taxpayer might be tempted to realize the built-in loss position for a short-term capital loss while holding onto the built-in gain position until it becomes eligible for long-term capital gain treatment. Section 1092, by deferring the loss until the gain is realized, would prevent this result.

As part of the Technical Corrections Act of 1982 (which became law in January 1983),²⁴ Congress expanded the scope of section 1092 so that it covers not only unrealized gain but also unrecognized gain. The term “unrecognized gain” is defined in section 1092(a)(3)(A)(ii) to reach realized gain “in the case of any position with respect to which, as of the close of the taxable year, gain has been realized but not recognized.” Thus, if a taxpayer enters into a straddle and then disposes of the built-in-gain position through a nonrecognition transaction (that is, a transaction in which gain is realized but not recognized), the taxpayer can claim a loss on the loss position only to the extent that the loss exceeds the realized-but-not-recognized gain.

Section 1092(a)(1)(B) mitigates this loss-suspension result by allowing the taxpayer to carry the suspended loss over to the next tax year. Still, the taxpayer cannot take the suspended loss into account unless and until the loss exceeds the unrecognized gain from positions that were formerly offsetting positions.²⁵ So if a taxpayer enters into a straddle position and disposes of the built-in gain position through a nonrecognition transaction, the loss on the loss position will be suspended indefinitely.

Section 1092’s loss-suspension rule ensnares BOXX’s Booking Holdings straddle. When a regulated investment company distributes a built-in gain position to an authorized participant as part of an in-kind redemption, the company realizes a gain within the meaning of section 1001. As the Supreme Court held in the landmark Cottage Savings case, a taxpayer realizes income when it exchanges property for “materially different” property.²⁶ Surely, a call or put option on Booking Holdings stock is materially different from a share in an ETF that generates a fixed income-like return. The fund avoids recognition of the gain only because section 311(a) provides for nonrecognition of in-kind redemptions of corporate stock and section 852(b)(6) shields RICs from an otherwise applicable recognition rule for distributions of appreciated property.²⁷

Returning to the text of section 1092, the key operative provision states that “any loss with respect to 1 or more positions shall be taken into account for any taxable year only to the extent that the amount of such loss exceeds the unrecognized gain (if any) with respect to 1 or more positions which were offsetting positions with respect to 1 or more positions from which the loss arose.”²⁸ Thus, when BOXX seeks to claim losses on the losing legs of its Booking Holdings bets, it must contend with the fact that it has unrecognized gain for other positions (the gain positions) “which were offsetting positions” for the loss positions. Note that section 1092 uses the past tense, “were offsetting positions.” BOXX’s disposal of the gain positions before it realizes the loss positions does not change the fact that the gain positions were offsetting positions.

Granted, there is little evidence that lawmakers had BOXX’s specific strategy in mind when they enacted the 1982 technical corrections to section 1092. In its explanation of the 1982 technical corrections, the staff of the Joint Committee on Taxation highlighted a different problem. Under the original 1981 statute, cash-method taxpayers that sold offsetting positions at the end of the year would be able to recognize losses when they executed the sales contract while deferring recognition of gain until the receipt of sale proceeds the following year.²⁹ But even though lawmakers were not thinking about BOXX’s specific strategy, they did have in mind the general principle that taxpayers should not be able to recognize capital losses until they recognize capital gains on offsetting positions. When the language of a statute straightforwardly applies to a transaction and that application is entirely consistent with the statute’s overall purpose, BOXX has no viable argument for wiggling its way out of the plain text.³⁰

B. Tax Consequences at the Shareholder Level

In the unlikely event that BOXX escapes ordinary income treatment of its box spread gains at the fund level and manages to dodge the application of the straddle rules to its Booking Holdings trades, the IRS still could seek to apply the section 1258 conversion transaction rules at the shareholder level. As Steven Rosenthal observes, there is “no question” that BOXX’s shareholders expect a time-value-of-money return — and no shortage of evidence in BOXX’s marketing materials that the fund has been touted as generating capital gain upon disposition.³¹

But importantly, the section 1258 rules apply only when gain is recognized.³² If BOXX’s strategy succeeds at the fund level, shareholders will recognize gain only when they sell their shares. Thus, if recharacterization occurs only at the shareholder level, BOXX’s shareholders still will achieve a deferral benefit — akin to the benefit of holding an original issue discount bond while avoiding application of the OID interest imputation rules.³³

BOXX shareholders should not bank on this deferral benefit, though. The much likelier result is that BOXX’s tax strategy will fail at the fund level and the fund will be forced to distribute ordinary dividends to shareholders as income on the box spreads accrues.³⁴ And if that happens, then both of BOXX’s supposed tax benefits — the transformation of an interest-like return into capital gain and the deferral of tax until the time of sale — will evaporate.

For BOXX shareholders in states with an income tax, that’s not the end of the story. Interest on Treasury securities is exempt from state income tax,³⁵ but ordinary income from a box spread is not. Thus, by buying into BOXX, shareholders in those states likely will have increased their state income tax liability. For top-bracket taxpayers in California and New York, the extra state income tax (and the extra local income tax for New York City residents), when combined with fund expenses, may reduce their after-tax yield by nearly a full percentage point.³⁶

III. Lessons Learned

Perhaps the most surprising aspect of the BOXX episode is not that the fund’s tax strategy is likely to fail but rather that the fund has succeeded in raising large sums from credulous investors. Bloomberg reported that the fund — formed at the end of 2022 — had surpassed $1 billion in assets in February.³⁷ As of March 5, the fund’s website displayed holdings of more than $1.65 billion.³⁸ With its rapid rise, BOXX is unlikely to remain below the IRS’s radar for long. It appears to be emulating Icarus’s flight path.

For most of BOXX’s shareholders, the fallout will probably be limited. The federal income tax consequences from BOXX’s ordinary dividend distributions will be similar to the consequences of holding short-term Treasury bills. Most shareholders will end up paying additional state income taxes, along with extra expenses, making BOXX a tax-inefficient investment but not a financially disastrous one. For some, it may be a price worth paying for a valuable lesson: Promises of fantastic tax savings are often just that — fantasies.