Tax Notes logo

Why the United States Needs a GAAR

Posted on Mar. 4, 2024
Reuven S. Avi-Yonah
Reuven S. Avi-Yonah

Reuven S. Avi-Yonah ( is the Irwin I. Cohn Professor of Law at the University of Michigan Law School. He thanks Karen Burke, Jonathan Choi, Nir Fishbien, and David Hariton for helpful comments.

In this installment of Reflections With Reuven Avi-Yonah, Avi-Yonah advocates for Congress to adopt a general antiabuse rule to combat loopholes formed by the voluminous tax code.

The Internal Revenue Code of 1986, as amended, has over 1 million words and more than 5,600 pages. It is by far the longest and most complicated law in the U.S. code. The regulations add over 4 million words. Why?

The standard answer — going back to Stanley Surrey — is that Congress added many provisions to the code that have more to do with regulating taxpayer behavior than raising revenue. That is true, and there is a lively debate on whether tax expenditures belong in the code or should be delivered through direct subsidies. I agree with David Weisbach and Jacob Nussim that the answer depends on which agency is best suited to administer the relevant provision, and that some non-revenue laws, like the Affordable Care Act, appropriately make use of the IRS’s expertise and knowledge about levels of income to administer non-revenue programs.1

However, the length of the code also derives from three other features of tax law.

First, Congress rarely deletes sections, even when they are effectively replaced by new sections. For example, if asked whether some foreign-source income qualifies for deferral or exemption, the taxpayer must address the rules for not just global intangible low-taxed income (added in 2017) but also passive foreign investment companies (1986), subpart F (1962), personal holding companies (1935), and accumulated earnings tax (1921). The only anti-deferral rule that was deleted was the foreign personal holding company regime (1937), which was deemed unnecessary in 2004 because of PFIC rules and the application of personal holding company rules to foreign corporations.

Second, Congress does not delete sections when their rationale has been eliminated by other changes. For example, now that the dividend and capital gains rates are the same and almost all C corporations are publicly traded (and therefore can always engage in redemptions that qualify for capital gains treatment rather than pay dividends), sections 302, 304, 305, and 306 are arguably obsolete.

Finally, when confronted with a loophole (that is, unintended consequences), Congress frequently reacts by narrowly closing it, leading taxpayers to find a workaround that requires more legislation. For example, Congress reacted to the tax shelter transactions in Compaq2 and IES3 by enacting section 901(k) (requiring a holding period for receiving the foreign tax credit for dividend withholding taxes), but taxpayers engaged in similar transactions involving interest and royalty withholding, so Congress had to enact section 901(l).

If Congress wants to simplify the code, it should delete sections that have been replaced by other sections, as well as sections whose rationale has expired. In addition to complexity, the problem with keeping obsolete rules in the code is that they invite game playing. For example, in the foreign leveraged investment partnership transaction devised by KPMG, a U.S. corporation (USCo) seeking to shield a capital gain from tax acquired an option on the shares of a foreign corporation (ForCo) that invested in UBS. UBS then redeemed its shares from ForCo, and at the same time, USCo invested the same amount in UBS. The redemption was treated as a dividend to ForCo because there was no reduction in ownership because of the option. And because it was a complete termination, ForCo’s basis in the UBS shares was shifted to USCo’s shares in UBS, which were then sold and created a capital loss.4

This shelter would not have been possible if Congress had repealed section 302 because it is almost never applied to create taxable dividends; in the public corporation context, redeeming shareholders can always satisfy the section 302(b) conditions for capital gains treatment, and the stakes are low because the dividend and capital gains rates are the same. Alternatively, Congress could at least provide that section 302 only applies to taxable U.S. individuals.

What can be done about loophole closers? Here, the problem is that Congress cannot anticipate how taxpayers would react, even in a simple case like substituting interest and royalty withholding for dividend withholding.

The solution is for Congress to rely on the courts to close loopholes. Two provisions could, in principle, help: the codified economic substance doctrine (section 7701(o)) and the partnership antiabuse rule (PAAR). But there are issues with both.

The problem with the economic substance doctrine is that it focuses on taxpayer purpose, not Congressional purpose.5 Taxpayers can usually satisfy both the subjective and objective prongs of section 7701(o) by (1) producing contemporaneous documentation of their nontax business purpose and (2) inserting a low-likelihood but high-profit element into the transaction. Also, if the economic substance inquiry is framed as whether the transaction as a whole (rather than each of its parts) had a business purpose, the government will lose.6 The district court decision in Liberty Global attempted to overcome this issue in a clear case of unintended consequences by relying on congressional purpose, but that runs counter to the textualist tendency of most judges.7 The case is on appeal in the Tenth Circuit.

Regarding the PAAR, professor Karen Burke has recently analyzed the Tax Court opinion in Tribune Media,8 which the government has appealed to the Seventh Circuit.9 Burke argues that the PAAR could be used to invalidate the transaction. She writes that:

The Tribune Media litigation involves high stakes for the government as well as for taxpayers. While district courts have applied reg. section 1.701-2 “without questioning its validity,” leading partnership tax practitioners have described the antiabuse rule as a “stunning departure from existing law.” Despite the controversy surrounding its initial promulgation, the general antiabuse rule appears to be consistent with the prevailing trend of judicial decisions. The Seventh Circuit need not reach the general antiabuse rule if it determines that the transaction violates the specific antiabuse rule of reg. section 1.752-5(j). Absent that determination, the Seventh Circuit will be forced to deal squarely with Tribune’s claim that reg. section 1.701-2 is invalid. Given the apparently abusive nature of the Tribune transaction, finding the antiabuse rule invalid under these facts would invite even more aggressive tax planning. Contrary to Tribune’s claim that it engaged in a routine debt-financed distribution that complied with the literal requirements of reg. section 1.707-5(b)(1), this highly engineered transaction sought permanent elimination of gain by exploiting the provisions of subchapter K in a manner unintended by lawmakers.

Regardless of whether the specific antiabuse rule applies, the general antiabuse rule remains viable as an alternative ground for challenging the transaction. Even under reg. section 1.701-2, however, the government could lose if Tribune’s disguised sale is framed broadly as the overall Cubs transaction rather than the tax-motivated debt-financed guarantee and distribution. Focusing the business purpose requirement on the “function of the partnership as a whole,” as the Tax Court did, would undercut the antiabuse rule and allow validly formed partnerships to engage in tax-avoidance transactions with impunity. Despite the restrictive view articulated by the Tax Court in Tribune Media and Countryside, the substantial business purpose requirement appears to impose a more rigorous standard than a merely plausible business purpose. Indeed, the noncommittal view of the antiabuse rule espoused in Countryside should perhaps have prompted the Tax Court to scrutinize more closely the carefully contrived appearance of risk created by the guarantee in Tribune Media. Moreover, the section 701 regulations require that the business purpose for the transaction be weighed against the claimed tax benefit. If the Seventh Circuit compares Tribune’s tax-motivated guarantee with the claimed tax benefit — gain elimination rather than mere deferral — the government should prevail.10

Burke is clearly worried that (1) the circuit court would hold that the taxpayer complied with the literal language of the code, and (2) the circuit court would deprive the PAAR of any meaning by just requiring that the transaction as a whole (that is, the sale of a business) have a business purpose, or invalidate the PAAR for having no statutory basis and require courts to investigate the intent of subchapter K. This inquiry runs counter to the textualist tendency of most federal judges and the majority of the Supreme Court.

The solution is for Congress to enact a general antiabuse rule, modeled on the PAAR.11

When the PAAR was first proposed, it was met with an almost universal rejection by the tax bar, except for the New York State Bar Association Tax Section. The argument against it was that it would create uncertainty and chill legitimate transactions.12 But the PAAR has now been in effect for over 25 years, and there is no evidence that it has chilled legitimate transactions using partnerships, including aggressive (but arguably legitimate) transactions like the Up-C method of avoiding the dual tax on income earned through a partnership while doing an initial public offering of a corporation that is a partner. Up-C transactions also allow the company to step up the basis of the partnership’s underlying tax assets when the pre-IPO owners exit the company and sell their partnership units to the corporation. That step-up generally saves each Up-C (and costs the government) hundreds of millions of dollars in taxes over the life of the company. Unsurprisingly, rather than leaving that value with the company, the founders put a contract in place that requires it to pay that value over to them. Despite these tax advantages, the IRS has not challenged the Up-C method.13

I would suggest that Congress explicitly adopt a GAAR modeled after the PAAR but applying to the entire code, not just subchapter K.14 Such a GAAR may give textualist judges who wish to strike down a transaction the statutory basis to do so without relying on purposivism.

Many common-law countries, like Australia, Canada, India, Israel, New Zealand, and the United Kingdom, have adopted the GAAR, frequently with procedural limitations on how it may be deployed (for example, approval by the chief counsel). The result has been that these countries have far fewer tax shelters than the United States, even though some of them (for example, the United Kingdom) also tend to interpret statutes literally. Also, these countries have much shorter tax laws than the U.S. code, because of the GAAR.


1 See Weisbach and Nussim, “The Integration of Tax and Spending Programs,” 113 Yale L.J. 955 (2004); Reuven S. Avi-Yonah and Yoseph M. Edrey, “Constitutional Review of Federal Tax Legislation,” 1 Illinois L. Rev. 3-52 (2023).

2 Compaq Computer Corp. v. Commissioner, 277 F.3d 778 (5th Cir. 2001).

3 IES Industries Inc. v. United States, 253 F.3d 350 (8th Cir. 2001).

4 Calvin Johnson, “Tales From the KPMG Skunk Works: The Basis-Shift Shelter,” Tax Notes Int’l, Aug. 1, 2005, p. 435.

5 See Avi-Yonah, “Why Did the IRS Win? A Remarkable Year in Tax Litigation, Part 2,” Tax Notes Int’l, Jan. 15, 2024, p. 349.

6 Karen Burke, “Reframing Economic Substance,” 31 Va. Tax Rev. 271 (2011); David P. Hariton, “The Frame Game: How Defining the ‘Transaction’ Decides the Case,” 63 Tax Law. 1 (2009).

7 Liberty Global Inc. v. United States, No. 1:20-cv-03501-RBJ (D. Colo. 2022).

8 Tribune Media Co. et al. v. Commissioner, T.C. Memo. 2021-122.

9 Burke, “How Abusive Was Tribune Media’s Disguised Sale?Tax Notes Federal, Jan. 29, 2024, p. 855.

10 Id. (internal citations omitted).

11 The following is based on Avi-Yonah, “A Response to Professor Choi’s Beyond Purposivism in Tax Law,” 108 Iowa L. Rev. 69 (2023).

12 See Monte A. Jackel, Alison L. Chen, and James M. Maynor Jr., “Time to Revoke the Partnership Antiabuse Regulation,” Tax Notes, Jan. 29, 2018, p. 669; and Jackel, Chen, and Maynor, “Proving That the Partnership Antiabuse Reg Has No Place,” Tax Notes, May 14, 2018, p. 1027. See also Lee A. Sheppard, “Final Partnership Antiabuse Rule: Grudging Acceptance,” Tax Notes, Feb. 6, 1995, p. 776.

13 See Gladriel Shobe, “The Substance Over Form Doctrine and the Up-C,” 38 Va. Tax Rev. 249 (2018); Shobe, “Supercharged IPOs and the Up-C,” 88 U. Colo. L. Rev. 913 (2017).

14 For a proposal to codify the PAAR, see Jackel, “Two Peas in a Pod: Partnership Tax Reform and the Antiabuse Rule,” Tax Notes Federal, Feb. 5, 2024, p. 1039.


Copy RID