February 27, 2024

Investment Theory

Thinking Outside the BOXX

By Victor Haghani and James White 1

There’s been a lot of excitement and reporting about a new ETF: the Alpha Architect 1-3 Month Box ETF (ticker BOXX), designed to give investors the return of short-term US Treasury Bills with the tax character of long-term capital gains.

Long-term capital gains are taxed at lower rates than interest income – 20% versus 37% for the top US federal tax rates. With short-term interest rates at about 5%, this 17% difference in tax rates provides 0.85% more annual after-tax return, and you can add another 0.07% if you use BOXX to defer your tax bill for 5 years. That’s pretty good, and worth paying attention to if you can get it.

Two excellent Bloomberg articles were published on Feb 22nd that do a terrific job explaining how BOXX achieves its tax “magic.” The first is by Zachary Mider, “T-Bills Without Tax Bills? and hours later came “Put the Money in the BOXX” by Matt Levine (and a few days later also by Matt there’s “Maybe Don’t Put the Money in the BOXX?”).

Since BOXX went live on December 22, 2022, its annualized return has been 0.09% higher than that of State Street’s SPDR Bloomberg 1-3 Month T-Bill ETF (ticker BIL). BOXX has a stated expense ratio of 0.395%, but currently charges 0.195% and has about $1 billion of assets, while BIL sports an expense ratio of 0.136%, has $31 billion of assets and has been around for 17 years.

However, past returns are not necessarily indicative of future performance, and we think it’s reasonable to expect that BOXX’s prospective post-tax, risk-adjusted return relative to BIL (or direct ownership of Treasury Bills) does not justify owning BOXX. Before diving in, we should qualify what follows by noting that we are not tax experts.

Let’s start with an analysis of the expected return of BOXX, and then we’ll move to assessing the relative risk of BOXX versus Treasury Bills. First, BOXX has a higher expense ratio than BIL. The stated long-term expense ratio of BOXX is 0.395% per annum,2 compared to 0.136% for BIL. At the long-term expense ratio, that takes 0.21%3 out of the 0.85% of tax savings, leaving 0.64% of net tax savings versus BIL.

Next, there’s the question of what rate of interest will be set by market participants who provide the other side of the BOXX options trades. Historically, as discussed in this article from the NY Fed, the implied interest rate in options boxes has averaged about 0.35% above T-Bills (from January 1996 through April 2023). While this historical positive spread may continue into the future, it is worthwhile to ask yourself how you would price those options if you were being asked to be the counterparty to BOXX’s trades?

Your selling of the options box-spreads generates cash – the cash that the BOXX ETF wants to invest – but at the same time, you’ll need to post collateral to the clearing exchange. You need to do this to give comfort to your counterparty – the Options Clearing Corporation (OCC) – that if you disappear with the cash, they won’t have a loss.

You’ll probably go out and buy a T-Bill of equal maturity to the expiration of the options you traded and post that as collateral to the OCC. For this to be worthwhile, you’ll need to price the options using an interest rate lower than the T-Bills you had to buy to post as collateral, so you can earn a spread.

How much of a spread you’ll want to earn is a difficult question to answer beyond saying “as much as I can get.” Perhaps one place we can look for an indication of what market participants charge for nearly risk-free trades is the much-discussed Treasury bond basis trade. A number of recent articles (here and here) indicate that traders demand an expected return of at least 0.5% per annum on assets.

This required profit margin by counterparties supplying the options trades to BOXX plus the higher expense ratio of BOXX leaves 0.24% of tax benefit for BOXX versus holding BIL. We should also make an allowance for higher transactions costs in all the trades BOXX needs to do, not only in all the various options trades, but also the trades with the Authorized Participants (APs) needed to clean out all the capital gains from the primary options trades used to invest the capital of the ETF.

For investors living in a state with an income and capital gains tax,4 state taxes must be considered as well. US T-Bills are exempt from state income tax but the capital gains from BOXX would not be. An investor in a state with a roughly 10% income tax would lose an additional 0.5% of the potential tax benefit of BOXX. Combined with the reduced benefits described above, this would render BOXX significantly less attractive than owning T-Bills through an ETF such as BIL, and even less so versus owning T-Bills directly.

Turning to the risk side, we see one primary risk, and a number of smaller secondary risks. The most notable risk is that the ETF is not investing in T-Bills, but rather it holds a position in options contracts in an omnibus account at its options broker, who in turn is exposed to the OCC, an entity with a AA credit rating. The OCC is an exchange, with a wide range of counterparties of varying credit quality. And we know that exchanges can get into trouble, despite intense regulatory oversight and the requirement that counterparties post collateral to mitigate credit exposure.5 For example, the near-insolvency of the London Metal Exchange in the wild nickel price runup in March 2022 illustrates that investing in an exchange is riskier than investing in US T-Bills. How much spread an investor should require for these credit risks is hard to quantify precisely, but we’d suggest something around 0.2% – 0.5% per annum as a reasonable estimate. Note that 0.35% per annum of spread represents a 50% probability of at least one default with a 50% loss every hundred years, which doesn’t feel to us like an overestimate of this risk.

Adding the cost of credit risk to all the other costs already noted gives us total costs greater than the 0.85% potential tax benefit. We might stop here, but a number of other risks are also worth mentioning. These include a negative tax ruling from the IRS on the BOXX mechanics (see Matt Levine’s aforementioned “Maybe Don’t Put the Money in the BOXX?” or Steven Rosenthal’s “Tax Gimmick in a BOXX”), higher transactions costs in executing its strategy, and wider spreads required by the ETF market makers (APs) due to the complexity of the structure and the lower liquidity of BOXX relative to larger and higher volume Treasury Bill ETFs. The latter two risks become especially salient as BOXX increases in size. A further risk for a BOXX investor with a long-term horizon is that if interest rates are lower in the future, the potential tax benefits of BOXX will be proportionately smaller.

Investors are generally attracted to owning US T-Bills as a safe and highly liquid place to keep some of their savings, to be instantly available for unexpected emergencies or investment opportunities. In order to get the tax benefit of investing in BOXX, a holding period of more than one year is needed, since the short-term capital gains rate is equal to the tax rate on interest income, at the highest marginal rates. This seems at odds with the primary rationale of holding T-Bills in the first place. If you have a 30% chance of needing to sell BOXX to raise cash within one year, that reduces the gross tax benefit by 0.25%.6

So, we’re not at present planning on using BOXX for our Elm Wealth clients, especially those who are subject to high rates of state taxation. However, what concerns us most about BOXX is the potential harm it may do to the whole ETF marketplace by creating a feeling that it is taking advantage of an ETF “loophole.”

We believe that the tax treatment of ETFs is more correct and equitable for investors than the tax treatment of traditional mutual funds, which can unfairly accelerate capital gains on long-term investors and create more capital gains than are actually realized by the mutual fund. We explained this in some detail in a note we wrote in 2015, “ETFs: Better Than Mutual Funds for Long Term Investors too?” More than 16 million US households benefit from the diversification, liquidity and fair tax treatment offered by ETFs, according to Investment Company Institute estimates. If the BOXX ETF grows so large and attracts so much attention that it precipitates a change in the rules governing ETF taxation, the result would be a tremendous and lamentable decrease in investor welfare. We truly hope our worries are misplaced.


  1. Thank you to Larry Hilibrand, Charles Wright, Dave Blob and Jon Seed for their helpful comments, and to Wes Gray and Larry Lempert for discussing specifics of their ETF with us. The opinions expressed in the article are not necessarily shared by those who gave us help. Nothing in this note should be taken as tax advice, or investment advice, or an offer or solicitation to invest. Past returns are not indicative of future performance.
  2. Although currently, the sponsor has a fee waiver in place until January 31, 2025 so that the expense ratio of the fund is 0.195%.
  3. Expressed in after-tax terms, using a 20% tax rate. We make this adjustment throughout this note where appropriate.
  4. Every state except Alaska, Florida, New Hampshire, Nevada, South Dakota, Tennessee, Texas, and Wyoming.
  5. At the end of 2022, the OCC had about $14 billion of assets supported by about $700 million of equity. BOXX’s $1.4 billion of assets represents over 10% of the OCC’s Clearing Fund Deposits as of end of 2022. See here
  6. An exception to this analysis is an investor for whom capital gains are effectively tax-free because he has capital loss carryforwards so large that they will never be fully used.