Starr International Co. v. United States ( 2017 )

UNITED STATES DISTRICT COURT
FOR THE DISTRICT OF COLUMBIA
STARR INTERNATIONAL COMPANY,
INC.,
Plaintiff,
v. Case No. 14-cv-01593 (CRC)
UNITED STATES OF AMERICA,
Defendant.
UNITED STATES OF AMERICA,
Counterclaim-Plaintiff,
v.
STARR INTERNATIONAL COMPANY,
INC.,
Counterclaim-Defendant.
Table of Contents
I. Background ........................................................................................................................3
A. Statutory Background ...................................................................................................3
B. Starr’s History, Corporate Structure, and Previous Relocations .....................................7
C. Starr’s Move to Switzerland..........................................................................................9
D. Starr’s Request for Treaty Benefits Under Article 22(6)..............................................12
E. Administrative and Procedural History .......................................................................14
II. Legal Standards ...............................................................................................................16
III. Analysis ............................................................................................................................17
A. The Proper Legal Standard for Awarding Treaty Benefits Under Article 22(6) ............18
1. Starr’s Third-Country-Resident Argument .............................................................18
2. Starr Misunderstands the Technical Explanation’s Conception of “Residency” ..... 20
3. If It Were Valid, Starr’s Third-Country Rule Would Be a Mechanical
Test; It Is Not ........................................................................................................23
4. Starr’s Test Clashes With the Nature of Article 22’s Discretionary
Provision and Its Overriding Purpose..................................................................25
5. Starr’s Test Would Result In a Cramped Conception of Treaty Shopping ............... 27
B. The Competent Authority’s Application of the Article 22(6) Standard ........................30
1. Whether Certain, Purportedly Key Evidence Required a Contrary Result .............. 31
2. Whether the Competent Authority’s Analysis Rested on Irrelevant or
Incorrect Determinations ......................................................................................33
IV. Conclusion ........................................................................................................................37
MEMORANDUM OPINION
The bilateral tax treaty between the United States and Switzerland entitles Swiss-resident
entities to a reduction in the tax rate applied to dividends they receive from U.S. sources,
provided they meet one of a dozen or so objective criteria enumerated in the treaty. If none of
the listed requirements are satisfied, the Internal Revenue Service may still authorize a lower rate
if it determines that the Swiss entity was not established for a “principal purpose” of obtaining
treaty benefits. These rules are designed to limit treaty benefits to applicants that have a
sufficiently strong business or geographic connection to Switzerland.
Swiss-domiciled Starr International Company, Inc. (“Starr”), was once the largest
shareholder of the insurance giant AIG. In 2007, Starr, which did not meet any of the treaty’s
objective criteria for benefits, petitioned the IRS for a discretionary reduction in the rate applied
to some $191 million in dividends that Starr received from AIG during the 2007 tax year. After
a lengthy period of discussions between the two sides, the IRS ultimately denied Starr’s request
for treaty benefits on the ground that Starr’s historical selection of domiciles and its then-recent
relocation to Switzerland were motivated as much by tax reasons as by independent business
2
purposes. A “primary purpose” of the move, the IRS thus concluded, was to obtain treaty
benefits.
Starr now challenges the IRS’s denial of treaty benefits as arbitrary and capricious under
the Administrative Procedure Act. Starr’s primary contention is that the treaty’s primary
purpose test is designed to prevent the practice of “treaty shopping” and that the IRS applied an
erroneous definition of that term in concluding that the company’s relocation to Switzerland was
largely tax-driven. Starr argues that “treaty shopping” is a precise legal term, covering only
those instances where an on-paper resident of a country not party to the relevant tax treaty uses
an entity that is an on-paper resident of a treaty country in order to obtain treaty benefits.
Because Starr and its subsidiaries were on-paper Swiss residents and the majority of its voting
shareholders were U.S. citizens at the relevant time, Starr says it could not have been “treaty
shopping” under this definition. Starr’s legalistic conception of “treaty shopping,” however,
cannot be squared with the text of the U.S.-Swiss treaty or its accompanying agency guidance.
Instead, those authorities understand “treaty shopping” as encompassing situations where an
entity establishes itself in a treaty jurisdiction with a “principal purpose” of obtaining treaty
benefits. Because the IRS reasonably applied that standard in denying treaty benefits to Starr,
the Court declines to set aside its determination.
I. Background
A. Statutory Background
When foreign corporations receive dividends from U.S. sources, that income is generally
taxed at a 30% rate. 26 U.S.C. § 881(a)(1). To avoid double taxation and encourage cross-
border investments, the United States has entered into numerous bilateral tax treaties with other
nations. As a general matter, these treaties feature a reciprocal reduction in the tax rate on
3
foreign-source income for domestic residents of the contracting countries. For example, the
treaty at issue here, which the Court will refer to as the “U.S.-Swiss Treaty,” reduces the tax on
U.S.-source dividend income for Swiss residents from 30% to 5% or 15%, depending on the
Swiss entity’s percentage of ownership in the U.S. corporation. See Convention Between the
United States of America and the Swiss Confederation for the Avoidance of Double Taxation
with Respect to Taxes on Income art. 10(2), Oct. 2, 1996, S. Treaty Doc. No. 105–8,
https://www.irs.gov/pub/irs-trty/swiss.pdf [hereinafter “Treaty”].
Bilateral tax treaties, including the U.S.-Swiss Treaty, thus aim to benefit residents of the
two contracting states. But this “begs the question of who is to be treated as a resident of a
Contracting State for the purpose of being granted treaty benefits.” Dep’t of the Treasury,
Technical Explanation of the Convention Between the United States of America and the Swiss
Confederation for the Avoidance of Double Taxation with Respect to Taxes on Income 59,
http://www.irs.gov/pub/irs-trty/swistech.pdf [hereinafter “Technical Explanation”]. 1 The U.S.-
Swiss Treaty generally defines residency based on local tax liability: If a person “is liable to tax
. . . by reason of his domicile” or the like, that person is a resident of the taxing jurisdiction.
Treaty art. 4(1)(a). However, “[t]he fact that a person is determined to be a resident of a
Contracting State [under the treaty’s definition] does not necessarily entitle that person to the
benefits of the Convention.” Technical Explanation 10. As the treaty framers recognized, if on-
paper residency were enough to obtain treaty benefits, then it would be easy to skirt the system:
1
Technical explanations are created by the Treasury Department during treaty
negotiations and presented to the Senate for consideration during the ratification process. See
Xerox Corp. v. United States, 41 F.3d 647, 655 (Fed.Cir.1994). They serve as an analogue to
legislative history for treaty ratification, and courts properly consult these explanations when
construing treaty language. See, e.g., Haver v. Commissioner, 444 F.3d 656, 658 (D.C. Cir.
2006).
4
Any company—no matter its actual jurisdictional or geographic ties, or the location or identity of
its true beneficiaries—could simply establish itself or a subsidiary entity in one of the treaty
nations, and obtain treaty benefits. That company, in other words, could easily engage in treaty
shopping.
Enter “Limitation on Benefits” provisions. Common among bilateral tax treaties, and
housed in Article 22 of the U.S.-Swiss Treaty, these provisions aim to deny treaty benefits to
those who establish “legal entities . . . in a Contracting State with a principal purpose to obtain
[treaty] benefits.” Technical Explanation 59. Of course, unlike determining on-paper residency,
divining an entity’s “principal purpose” for establishing itself in a particular jurisdiction is no
easy task. Indeed, “it requires the tax administration to make a subjective determination of the
taxpayer’s intent.” Technical Explanation 59. To ease “the administrative burdens of such an
approach,” id., Article 22 spells out a number of objective, mechanical tests meant to identify
those treaty-country residents who are worthy recipients of treaty benefits. Generally, these
mechanical tests seek to identify entities with legitimate, non-tax-related motives (such as
business purposes) for their claimed state of residency: “The assumption underlying each of
[Article 22’s mechanical] tests is that a taxpayer that satisfies the requirements of any of the tests
probably has a real business purpose for the structure it has adopted, or has a sufficiently strong
nexus to the other Contracting State[.]” Technical Explanation 59. The idea is that, in order to
obtain treaty benefits, the entity’s “business purpose or connection [should] outweigh[] any
purpose to obtain [treaty] benefits.” Id.
Paragraph 1 of Article 22, for instance, provides that “a person that is a resident of a
Contracting State and that derives income from the other Contracting State may only claim the
benefits provided for in this Convention where such person” falls into one of seven categories.
5
Treaty art. 22(1). Those categories include an “individual,” id. art. 22(1)(a); an entity “engaged
in the active conduct of a trade or business in the . . . Contracting State” where it is a resident, if
the income it derives is connected to that trade or business, id. art. 22(1)(c); and a “family
foundation resident of Switzerland, unless the founder, or the majority of the beneficiaries, are
persons who are not [individuals] entitled to [treaty] benefits . . . or 50 percent or more of the
income of the family foundation could benefit persons who are not [individuals] entitled to
[treaty] benefits,” id. art. 22(1)(g). These criteria are proxies for intent: The treaty drafters
appreciated that individuals who actually live in Switzerland, who engage in an active trade or
business in Switzerland, or who set up a Swiss family foundation in Switzerland (primarily to
benefit Swiss individuals), probably are not residents of Switzerland because they want lower tax
rates on their U.S.-source income.
The drafters also recognized, however, that certain entities with legitimate reasons for
residing in a treaty nation might nevertheless fail Article 22’s rigid mechanical tests. See
Technical Explanation 60 (“[T]hese mechanical tests cannot account for every case in which the
taxpayer was not treaty shopping.”). Accordingly, in paragraph 6 of the Article, the treaty leaves
open the possibility for discretionary relief:
A person that is not entitled to the benefits of this Convention pursuant to the
provisions of the preceding paragraphs may, nevertheless, be granted the benefits
of the Convention if the competent authority of the State in which the income arises
so determines after consultation with the competent authority of the other
Contracting State.
Treaty art. 22(6). The Technical Explanation indicates that “[w]hile an analysis under paragraph
6 may well differ from that under one of the other tests of Article 22, its objective is the same: to
identify investors whose residence in the other State can be explained by factors other than a
6
purpose to derive treaty benefits.” Technical Explanation 60. The Technical Explanation
elaborates that
[t]he competent authority of a State will base a determination under [the
discretionary provision] on whether the establishment, acquisition, or maintenance
of the person seeking benefits under the Convention, or the conduct of such
person’s operations, has or had as one of its principal purposes the obtaining of
benefits under the Convention. Thus, persons that establish operations in one of
the States with a principal purpose of obtaining the benefits of the Convention
ordinarily will not be granted relief under paragraph 6.
Technical Explanation 72.
This Court previously concluded that the above guidance from the Technical
Explanation—at times referred to as the “principal purpose” test—provided the appropriate
standard for evaluating whether someone is entitled to relief under Article 22(6), both for the IRS
in the first instance and on judicial review. See Starr Int’l Co., Inc. v. United States (“Starr I”),
139 F. Supp. 3d 214, 229 (D.D.C. 2015).
B. Starr’s History, Corporate Structure, and Previous Relocations
Starr International was founded in 1943 by Cornelius Vander Starr. A.R. 44. Initially
incorporated in the Republic of Panama, the company aimed at developing a worldwide network
of insurance agencies that generated international business for U.S.-based insurance companies.
A.R. 45. By 1970, Starr was headquartered in Bermuda, and had come to operate over 100
offices in 40 countries. Id. That same year, Starr entered into a transformative agreement with
American International Reinsurance Company, Inc. (“AIRCO”), another insurance company
founded by Vander Starr. In essence, Starr swapped its insurance businesses for stock in
AIRCO. A.R. 45–46. At the same time, Starr reorganized. In a stated effort to keep the gains
from the 1970 stock acquisition “within [Starr] for future investment and use and ultimate
disposition to charity,” Starr’s voting shareholders created a new class of non-voting common
7
stock, with full residual rights to Starr’s assets (now mostly AIRCO stock), and issued it to a
charitable trust, whose ultimate beneficiary was a New York foundation. A.R. 47. Other 1970
amendments placed restrictions on the voting shareholders’ use of the so-called “Restricted
Surplus”—essentially, the economic gain that resulted from Starr’s acquisition of AIRCO stock.
See A.R. 46–48. In 1978, when AIRCO merged with AIG, the AIRCO stock became AIG stock,
and Starr became AIG’s largest shareholder. A.R. 48.
As the Government points out, Starr itself concedes, and a Southern District of New York
court has discussed at length, see Starr Int’l Co. v. Am. Int’l Grp., Inc., 648 F. Supp. 2d 546
(S.D.N.Y. 2009), the factors motivating the vesting of the corporation’s economic value in a
charitable trust were not wholly charitable in nature (to say the least). Beyond seeking “to build
value for eventual long range use and distribution to the common stock owners for charitable
purposes,” id. at 558, Starr’s charity ownership structure was also designed to prevent the hostile
takeover of AIG, id. at 555, and to fund compensation grants to AIG executives, id. at 571. See
also A.R. 148–49 (“[T]he [Charitable] Trust was set up as [a] long-term arrangement with
multiple goals. . . . . [A]mong the goals of the arrangement were the intentions to protect AIG
from unwarranted hostile bids for change in control and to permit [Starr], as AIG’s largest
shareholder, to make incentive compensation grants . . . to AIG employees.”). Motivations
aside, Starr maintained this capital structure—voting shares with virtually no economic value,
sitting atop valuable non-voting shares owned by a charitable trust—through the time period
relevant here. As of 2007, Starr had 10,000 shares of non-voting common stock owned by Starr
AG, the latest incarnation of Starr’s charitable trust; 120 shares of voting stock owned by ten
U.S. and two foreign individuals, largely current or former AIG executives; and 199 shares of
“first preferred stock,” owned by two U.S. individuals. A.R. 48–49.
8
Again, Starr was based in Bermuda at the time of its 1970 reorganization. It remained
there until 2004, when it moved to Ireland. As with the motives underlying the corporation’s
capital structure, its reasons for this move are a matter of some dispute. In a July 2009 meeting
with U.S. Competent Authority2 officials, Starr—through counsel—represented that Bermuda
was simply “too small of a place for a $20 billion charity,” and that “Bermuda has political
problems as well as a lack of skilled workers and professionals.” A.R. 173. However, as the
Government notes, there is abundant evidence that the move from Bermuda to Ireland was tax-
motivated. In a 2005 New York Times article, lodged in the administrative record, a Starr voting
shareholder and former AIG vice chairman explained that “Ireland taxes stock dividends at a 5
percent to 10 percent rate, compared with 30 percent in Bermuda, and that prompted the move to
Dublin.” A.R. 298. The Government also points to evidence brought out in the litigation
between Starr and AIG, including a 2004 memo from Starr’s then-president Michael Murphy,
recommending that Starr move to Ireland for tax reasons, and a payment schedule for Starr’s
AIG dividends, showing that those payments surged in 2003. Def.’s Reply Supp. Cross-Mot.
Summ. J. (“Def.’s Reply”) 19. (The payments increased by $16 million from the year before;
previous annual increases had ranged from $1 million to $7 million. Id.)
C. Starr’s Move to Switzerland
In 2005, roughly a year after Starr relocated to Ireland, the company began planning for
yet another move. The ostensible reason for this move—as Starr represented in a pre-filing
presentation to the Competent Authority on December 6, 2007, and again in its December 21,
2007 request for treaty benefits—was that Irish law was not amenable to its charitable objectives.
2
The U.S. Competent Authority is the official within the Treasury Department who is
authorized to interpret provisions of our bilateral income tax treaties. The post is occupied by
the IRS Deputy Commissioner for the Large Business and International Division.
9
See A.R. 23–42, 49. In particular, Starr indicated that Irish law prohibited it from making
donations to charities that were not approved as exempt charities by the Irish Revenue
Commissioners. A.R. 49. This, combined with Starr’s “related commitments to the Irish
authorities,” purportedly placed “severe practical limitations on the amounts that could be
distributed to donees outside of Ireland.” A.R. 50. Starr also said it wanted to amend the trust
deed so that the trustee—and not Starr itself—would have the power to direct the Trust’s
investments and distributions, but the deed “could not effectively be amended under Irish law.”
Id.
In a separate presentation and submission to the Competent Authority roughly eighteen
months later, however, Starr offered an additional (and wholly distinct) reason for its desire to
leave Ireland. The company explained that “beginning in September 2005, [Starr had become]
the target of a number of fiercely contested lawsuits with AIG” regarding the proper ownership
of Starr’s “most significant asset,” its AIG shares. A.R. 141–42; see also A.R. 152, 167 (July 23,
2009 presentation to the Competent Authority, explaining that “litigation risk was in fact an
immediate and pressing reason for [Starr’s] move to Switzerland”). 3 Starr was worried that
those assets were not sufficiently insulated from litigation in Ireland, and so it began looking for
a “jurisdiction [that] would provide the most safety in light of the assault being mounted by
AIG.” A.R. 143. Starr now maintains that these two broad concerns—flexibility for its charity,
3
Starr’s initial request to the Competent Authority, in December 2007, had omitted any
reference to the AIG litigation. This despite statements from Starr’s general counsel, under
penalty of perjury, that the “facts submitted in support of this request . . . are true, correct and
complete,” A.R. 66, and from Starr’s outside counsel at Skadden, Arps that “[t]here are no
relevant domestic or foreign judicial or administrative proceedings that involve the taxpayer or
related persons,” A.R. 65. It is unclear whether these obvious omissions had any impact on the
Competent Authority’s ultimate denial decision, or whether they ought to bear on the Court’s
review of that decision. See infra section III.C.
10
and protection for its assets—motivated its decision to relocate. See Pl.’s Mem. Supp. Cross-
Mot. Summ. J. (“Pl.’s Cross-MSJ”) 13–14.
But where should Starr move? The decisionmaking process Starr employed in answering
that question is summarized in a “decision matrix,” which both parties consider favorable
evidence. A.R. 151. The chart appears to have been created in 2009, in an effort to summarize
for the Competent Authority “the analysis the Voting Shareholders undertook in 2006.” A.R.
144. It includes eleven rows, one for each combination of jurisdictions considered as possible
homes for Starr and its charitable trust; and four columns, one for each criterion analyzed—
namely, “Local Tax,” “U.S. Tax,” “Litigation Risk,” and “Charities Regulation.” A.R. 151. For
example, the seventh row reads “Netherlands – Netherlands,” indicating the possibility of
relocating both Starr and its charitable trust to the Netherlands. Id. Reading from left to right,
the row for “Switzerland – Switzerland,” the jurisdictional combination ultimately selected,
indicates a “Very Good” “Local Tax” environment, “Low” “Litigation Risk,” and “Strong”
“Charities Regulation.” Id. The chart indicates that the “U.S. Tax” situation in Switzerland
would be “Bad (absent 22(6)),” i.e., without the discretionary exception to the Limitation on
Benefits provisions at issue here. Id.4
According to Starr, its voting shareholders first determined to locate both Starr itself and
the charitable trust in the same jurisdiction, as had been the arrangement for the previous 35
years. Keeping the two entities together would ostensibly provide “the strongest level of legal
protection” in the event of a legal challenge. A.R. 143. That choice narrowed the eleven options
to five: Bermuda, the United Kingdom, the Netherlands, Switzerland, and remaining in Ireland.
4
Somewhat counterintuitively, Starr considered charity regulations to be “strong” when
they were flexible, and “weak” when they were restrictive. See Hr’g Trans. 23–24.
11
A. 151. Of these, after Bermuda, which had no bilateral tax treaty with the United States,
Switzerland offered Starr the least favorable U.S. tax situation (with a rating of “Bad”).
Switzerland earned that rating because the tax breaks Starr would have received automatically in
the United Kingdom, the Netherlands, and Ireland, were available in Switzerland only by
application of Article 22(6)’s discretionary exception. A.R. 54. However, Starr claims to have
chosen Switzerland despite that rating because it was the only jurisdiction with both a “Low”
risk of litigation and “Strong” charities regulation. See Pl.’s Cross-MSJ 15–16. Starr began the
“migration” process in July 2006, and both Starr and Starr AG, the company’s new charitable
arm, were Swiss residents by December 11, 2006. A.R. 43, 50. As had been the case previously,
Starr AG owned the corporation’s non-voting common stock, which represented the lion’s share
of the company’s economic value. A.R. 57. Starr AG, in turn, was owned by a Swiss-resident
foundation, id., and Starr’s residual beneficiary, in the event of dissolution, was the Starr
Foundation, based in New York. A.R. 276.
D. Starr’s Request for Treaty Benefits Under Article 22(6)
Starr filed a request for treaty benefits under Article 22(6) on December 21, 2007. A.R.
43. In the request, Starr conceded that it was not entitled to benefits under any of Article 22’s
mechanical tests. A.R. 55. However, it sought discretionary relief under paragraph 6 primarily
on the ground that its move to Switzerland was motivated by charitable considerations, not tax
concerns. See A.R. 61 (“[Starr] chose to bear the risk of not obtaining relief under paragraph 6
of Article 22 because it believed that Switzerland was a better jurisdiction than available
alternatives from which to achieve the company’s charitable objectives, even though it was less
desirable than the available alternatives from a U.S. withholding tax perspective.”). Starr also
argued that its request was “within the spirit of the objective criteria” of Article 22(2), which
12
confers treaty benefits on certain non-profit organizations. A.R. 61–62. Finally, Starr indicated
that—given its Swiss residency, beneficial ownership by a Swiss non-profit, and majority-U.S.
voting power—it was “not aware of any policy reason” to deny it treaty benefits. A.R. 62.
Starr’s request prompted an extended, 34-month-long inquiry by the Competent
Authority. During that period, Starr responded to numerous requests for further information,
which included questions about the residencies of Starr’s voting shareholders, A.R. 78; the
company’s associations with Panama, A.R. 80; the historical withholding rate applicable to the
dividends Starr received on AIG stock, A.R. 85; and Starr’s tax benefits under the U.S.-Ireland
Treaty, A.R. 87. In June 2009, the Competent Authority informed Starr that it was “tentatively
adverse” to the request. A.R. 153. Shortly thereafter, Starr’s outside counsel prepared a detailed
memorandum responding to the Competent Authority’s concerns, which included the company’s
brief tenure in Ireland (suggestive of a tax motive) and its dubious charitable activities. A.R.
139–40. The memo attached the previously discussed “decision matrix,” and explained—for the
first time—that Starr had been in large part motivated by asset protection in moving to
Switzerland. A.R. 141–42. Two days later, Starr gave a presentation to the Competent
Authority with similar content. See A.R. 152–69.
On October 13, 2010, after further back-and-forth correspondence, the Competent
Authority issued a final determination letter denying Starr’s request for treaty benefits. A.R.
273–74. 5 The letter explained that, under the circumstances, the Competent Authority could not
“conclude that obtaining treaty benefits was not at least one of the principal purposes for moving
5
Although not in the administrative record, Starr’s former outside counsel has attested
that in November 2009, Starr was informed that the Competent Authority had tentatively decided
in favor of granting benefits, but that in March 2010, after the appointment of a new Competent
Authority, it was now tentatively adverse. See Declaration of Hal Hicks ¶¶ 7–9, ECF No. 48-2.
13
Starr’s management, and therefore its residency, to Switzerland.” A.R. 274. In support of that
conclusion, the letter highlighted the following four considerations:
• [Starr]’s original incorporation in Panama and its management and
control in Bermuda suggest the original corporate structure may
have been developed with tax avoidance purposes in mind and/or
with a purpose of avoiding the provision of information on [Starr]’s
activities to the Internal Revenue Service;
• [Starr]’s re-location to Ireland and its movement of management out
of Bermuda a relatively short time before the payment of dividends
to [Starr] further suggests that [Starr] was seeking to avail itself of
the treaty between the United States and Ireland to avoid U.S. tax on
those contemplated dividends;
• The transitory nature of [Starr]’s location in Ireland, which may or
may not have been intentionally transitory, and its subsequent
movement to Switzerland further suggests its intention of organizing
in a treaty jurisdiction to avail itself of a reduced rate of withholding
on U.S. source dividends;
• [Starr] is largely controlled by U.S. individuals and such control is
not in accord with recent development of U.S. policy on acceptable
corporate ownership for [Limitation on Benefits] purposes.
A.R. 274.
E. Administrative and Procedural History
Meanwhile, because Starr was a Swiss resident by the start of 2007, and yet did not
automatically qualify for U.S. tax benefits there, AIG withheld taxes at a rate of 30% on the
dividends it paid to Starr. In 2007, the amount withheld was roughly $57.3 million, and in 2008,
$42.3 million. Pl.’s Cross-MSJ 19. According to Starr, if the company had qualified for treaty
benefits, those amounts would have been about $19.1 million and $21.2 million, respectively.
14
Id. 6 In March 2010, while Starr’s request for treaty benefits was still pending with the
Competent Authority, the company filed a claim for a tax refund of the difference between the
2007 tax amount actually withheld and the amount that would have been withheld had Starr
qualified for treaty benefits. In February 2011, after the Competent Authority had issued a final
determination that Starr was not entitled to treaty benefits for the 2007 tax year, Starr filed a
similar refund claim for its 2008 taxes. Id. The IRS refunded the requested amount for 2008
(roughly $21.2 million), but not for 2007. Id. In a separate counterclaim, which is not the
subject of these cross-motions, the Government asserts that the 2008 refund was obtained
fraudulently and seeks its return. See Def.’s Am. Answer & Counterclaim, ECF No. 9.
Starr filed a tax-refund suit in this Court under 26 U.S.C. § 7422, seeking the 2007 refund
that the IRS had denied—largely on the grounds that the Competent Authority had abused its
discretion in denying treaty benefits. The Court ultimately dismissed that complaint without
prejudice, permitting Starr to bring a claim under the Administrative Procedure Act (“APA”).
See Starr I, 139 F. Supp. 3d 214, 219 (D.D.C. 2015); Starr Int’l Co., Inc. v. United States (“Starr
II”), No. 14-CV-01593, 2016 WL 410989 (D.D.C. Feb. 2, 2016). Starr did so, alleging that the
Competent Authority’s denial of treaty benefits to Starr was “arbitrary, capricious, an abuse of
discretion, and otherwise not in accordance with law” under the APA, and asking the Court for
“[a] judgment holding unlawful and setting aside” that denial. Am. Compl., ECF No. 43.
The parties have both moved for summary judgment, debating both what the proper
Article 22(6) standard is and whether it was reasonably applied. The Court held a hearing on the
cross-motions.
6
The withholding rates under the Treaty are based in part on the percent of the U.S.-
source company owned by the treaty beneficiary. Starr’s ownership of AIG was above 10% for
2007, and below 10% in 2008, triggering different withholding rates under the Treaty.
15
II. Legal Standards
When a court is tasked with reviewing final agency action under the APA, the summary
judgment standard set forth in Federal Rule of Civil Procedure 56 does not apply. Sierra Club v.
Salazar, 177 F. Supp. 3d 512, 527 (D.D.C. 2016). Instead, the APA “sets forth the full extent of
judicial authority to review executive agency action for procedural correctness.” FCC v. Fox
Television Stations, Inc., 556 U.S. 502, 513 (2009). Under the APA, a reviewing court must
“hold unlawful and set aside agency action, findings, and conclusions found to be . . . arbitrary,
capricious, an abuse of discretion, or otherwise not in accordance with law.” 5 U.S.C. § 706(2).
In making this determination, the statute further provides that “the court shall review the whole
record or those parts of it cited by a party.” Id. at § 706. This provision generally limits judicial
review of agency action to the administrative record. See Hill Dermaceuticals, Inc. v. FDA, 709
F.3d 44, 47 (D.C. Cir. 2013) (“[I]t is black-letter administrative law that in an APA case, a
reviewing court “should have before it neither more nor less information than did the agency
when it made its decision.”) (quoting Walter O. Boswell Mem’l Hosp. v. Heckler, 749 F.2d 788,
792 (D.C. Cir. 1984)).
This standard of review under the APA is “highly deferential” and “presumes the
agency’s action to be valid.” Defs. of Wildlife v. Jewell, 815 F.3d 1, 9 (D.C. Cir. 2016)
(citations omitted). The reviewing court is “not to substitute its judgment for that of the agency.”
Ark Initiative v. Tidwell, 816 F.3d 119, 127 (D.C. Cir. 2016) (quoting Motor Vehicle Mfrs.
Ass’n of U.S., Inc. v. State Farm Mut. Auto. Ins. Co., 463 U.S. 29, 43 (1983)). However, it must
nonetheless ensure that “the agency . . . examine[d] the relevant data and articulate[d] a
satisfactory explanation for its action including a rational connection between the facts found and
the choice made.” Id. (quoting State Farm, 463 U.S. at 43 (internal quotation marks omitted)).
16
Agency action is arbitrary and capricious, for example, if the agency “entirely failed to consider
an important aspect of the problem, offered an explanation for its decision that runs counter to
the evidence before the agency, or is so implausible that it could not be ascribed to a difference
in view or the product of agency expertise.” State Farm, 463 U.S. at 43. “[T]he party
challenging an agency’s action as arbitrary and capricious bears the burden of proof.” Oceana,
Inc. v. Pritzker, 24 F. Supp. 3d 49, 60 (D.D.C. 2014) (quoting San Luis Obispo Mothers for
Peace v. U.S. Nuclear Regulatory Comm’n, 789 F.2d 26, 37 (D.C. Cir. 1986)).
III. Analysis
In moving for summary judgment, Starr raises arguments related to both the proper
Article 22(6) standard and whether it was reasonably applied. Regarding the standard, Starr
argues that Article 22(6) was meant to provide relief to anyone not “treaty shopping,” and that
treaty shopping always involves a third-country resident—i.e., a resident of a country not party
to the relevant tax treaty. Because Starr was domiciled in Switzerland and its beneficial and
voting ownership was (largely) either Swiss or American, the argument goes, the company could
not have been a U.S.-Swiss Treaty shopper and therefore should have gotten relief under the
treaty. Regarding the application of the standard, Starr counters that, even assuming the Article
22(6) standard lacks a third-country-resident requirement, the Competent Authority arbitrarily
and capriciously relied on irrelevant facts and ignored material ones in reaching its
determination. The Court evaluates each set of arguments in turn.
17
A. The Proper Legal Standard for Awarding Treaty Benefits Under Article 22(6)
1. Starr’s Third-Country-Resident Argument
In holding that the Competent Authority’s decision was judicially reviewable, this Court
confirmed that the standard set forth in the Treaty’s Technical Explanation governs. Starr I, 139
F. Supp. 3d at 227–28. That standard, to repeat, is as follows:
The competent authority of a State will base a determination under [Article 22(6)]
on whether the establishment, acquisition, or maintenance of the person seeking
benefits under the Convention, or the conduct of such person’s operations, has or
had as one of its principal purposes the obtaining of benefits under the Convention.
Technical Explanation 72. In rendering its denial decision, the Competent Authority quoted this
standard, nearly verbatim, when it wrote that “the above-described circumstances make it
impossible . . . to conclude that obtaining treaty benefits was not at least one of the principal
purposes for moving Starr’s management, and therefore its residency, to Switzerland.” A.R.
274. Starr nevertheless argues that the Competent Authority’s determination should be set aside
because it identified and applied the wrong Article 22(6) standard.
Starr’s argument begins with two uncontroversial premises. First, Starr notes that Article
22, including its discretionary provision, was developed to combat treaty shopping. There is
abundant authority in support of that proposition, see Starr I, 139 F. Supp. 3d at 227 (Article 22
aims at “preventing companies from ‘treaty shopping’”); U.S.-Swiss Treaty, Letter of Submittal
(“Article 22 of the new Convention contains significant anti-treaty-shopping rules making its
benefits unavailable to persons engaged in treaty shopping.”); Technical Explanation 59
(explaining that Article 22 is designed to prevent treaty shopping), and the Government does not
dispute it. Second, Starr correctly cites the Technical Explanation’s discussion of treaty
shopping, which describes the concept as “the use, by residents of third States, of legal entities
18
established in a Contracting State with a principal purpose to obtain the benefits of a tax treaty
between the United States and the other Contracting State.” Technical Explanation 59.
From there, Starr’s assertions become more contentious. Starr argues that treaty
shopping is not just a policy evil that Article 22 was designed to foil, but also the legal standard
that the Competent Authority is required to apply when implementing Article 22(6). That is,
Starr urges that entities must be awarded benefits so long as they are not “treaty shopping.” See
Pl.’s Cross MSJ 23–24. 7 Starr next applies a formalistic reading to the Technical Explanation’s
discussion of treaty shopping—paying special attention to its references to third-country
residents—and arrives at its own treaty shopping definition, with two necessary conditions.
Under Starr’s definition, treaty shopping exists where: (1) “A Third-Country Resident . . .
establish[es] or use[s] a Treaty Resident to make an investment in the United States,” and (2) “A
principal purpose behind the establishment or use of the Treaty Resident [is] to obtain treaty
benefits for the Third-Country Resident.” Pl.’s Cross-MSJ 24. This definition, in other words,
adds a second prong to the Article 22(6) standard articulated in the Technical Explanation and
confirmed by this Court: To be denied benefits, says Starr, an entity must not only have a
“principal purpose” of obtaining treaty benefits, it must also be owned or established by a third-
country resident.
Finally, Starr applies its version of the Article 22(6) standard to its own case, and—not
surprisingly—the result is in Starr’s favor. Starr reasons that “when the relevant transactions
7
This was not what the Court held in its prior opinion. Rather, after quoting the
“principal purpose” standard and explaining why that standard was justiciable, the Court referred
back to that standard, in short-hand, as turning on “whether an applicant’s principal purpose was
treaty shopping.” Starr I, 139 F. Supp. 3d at 229. At the time, Starr had not yet developed the
formalistic definition of treaty shopping that it now advances, and the Court certainly did not
intend to adopt that definition as an element of the Article 22(6) standard.
19
were complete, Starr, Starr AG [the charitable entity housing Starr’s economic value], and the
[Swiss foundation that owns Starr AG] were all located in Switzerland.” Pl.’s Cross-MSJ 35.
Plus, “more than 83% of Starr’s Voting Shareholders were U.S. citizen individuals . . . and the
ultimate beneficiary of the [Swiss] [f]oundation was another U.S. entity, the [New York-based]
Starr Foundation.” Id. Accordingly, because “the vast majority of the control and essentially all
of the economic value of [Starr] . . . was held by entities or individuals resident in Switzerland or
the United States (i.e., the parties to the U.S.-Swiss Treaty),” the third-country resident prong of
Starr’s proposed Article 22(6) standard could not be satisfied. Pl.’s Cross-MSJ 34.
Starr’s argument is compelling on the surface, primarily because the Technical
Explanation does appear to contemplate that treaty shoppers will be third-country residents.
Upon closer analysis, however, Starr’s proposed third-country resident test simply cannot be
squared with Article 22’s text, its structure, or its accompanying Technical Explanation. At
bottom, the argument rests on a misconception—specifically, the conflation of two rather
different types of residency. Before explaining why Starr’s test cannot be reconciled with
Article 22, the Court first addresses that misconception.
2. Starr Misunderstands the Technical Explanation’s Conception of “Residency”
Starr is correct that Article 22’s Technical Explanation includes numerous references to
“third-country residents.” In addition to the treaty shopping definition cited above, the Technical
Explanation in several places expresses concern about the “misuse” of bilateral tax treaties by
“residents of third countries.” Technical Explanation 59; see also id. at 72. But these
observations about the ubiquity of the “third-country resident” term merely prompt the following
question: How exactly should “residency”—and correspondingly, the term “third-country
resident”—be defined for the purpose of awarding or denying treaty benefits?
20
A close reading of the U.S.-Swiss Treaty and its Technical Explanation reveals two
possible definitions. The first may be referred to as “on-paper” residency. Drawn from the text
of the treaty itself, this definition is formalistic, and is based on certain objective indicators like
“domicile, residence, nationality, place of management, place of incorporation, or any other
criterion of a similar nature.” Treaty art. 4(1)(a). This, of course, is the definition that applies
anytime the word “resident” is used in the text of the Treaty—including in the text of Article 22.
According to an on-paper conception of residency, a “third-country resident” would be an entity
without any of the formalistic indicators listed in Article 4 of the Treaty—i.e., one whose
“domicile, residence, nationality, place of management, place of incorporation, or any other
criterion of a similar nature” is not located in a contracting state (and likely is located in a third
country).
Although Starr never clarifies its own conception of “third-country resident,” it is clear
that the company largely relies on this on-paper definition when advancing its treaty shopping
argument. For example, Starr references the “place of incorporation” of its parent and subsidiary
entities when declaring that “[Starr], Starr AG . . . and the [Swiss foundation that owns Starr AG]
were all located in Switzerland” during the relevant period. Pl.’s Cross-MSJ 35. The company
cites its “place of management” when proclaiming that it “relocated its employees, operations,
offices and board meetings entirely to Switzerland.” Id. And Starr leans on the “nationality” of
its voting shareholders, and perhaps their “domicile” and “residence,” when it notes that “83% of
[them] were U.S. citizen individuals” at the relevant time. Id. Starr’s overriding point seems to
be: All of the individuals and entities affiliated with us have on-paper connections to either
Switzerland or the United States. That excludes us as treaty shoppers and entitles us to benefits.
21
This conception of residency, however, is generally not the one employed by the
Technical Explanation’s discussion of Article 22. That portion of the Technical Explanation
begins by noting that a bilateral tax treaty is properly “a vehicle for providing treaty benefits to
residents of the two Contracting states.” Technical Explanation 59. “This statement,” it
continues, “begs the question of who is to be treated as a resident of a Contracting State for the
purpose of being granted treaty benefits.” Id. (emphasis added). Right away, then, the
Technical Explanation signals that some on-paper residents will be “treated as . . . resident[s]”
when it comes to awarding treaty benefits, and that others will not be; see also Technical
Explanation 10 (on-paper residency “does not necessarily entitle [a] person to [treaty] benefits”).
Rather, in the words of the Technical Explanation, Article 22 “authorize[s] a tax authority to
deny benefits, under substance-over-form principles” to an individual or entity that does not
have a genuine connection to the jurisdiction, even when it resides there on paper. Id. (emphasis
added). As put succinctly by the Technical Explanation, the Article “effectively determines
whether an entity has a sufficient nexus to the Contracting State to be treated as a resident for
treaty purposes.” Id. at 60 (emphasis added). An on-paper resident with “a sufficient nexus to
the Contracting State” may be called a bona fide resident.
Given that Article 22’s Technical Explanation employs a language of bona fide
residency, it stands to reason that its references to “third-country residents” likely do not refer to
entities with mere on-paper ties to third countries (and without on-paper ties to treaty countries).
Instead, when the Technical Explanation refers to a “third-country resident,” it should be read in
light of “substance-over-form principles.” In other words, the term is best understood as
indicating an entity that is not a bona fide resident of a treaty country—that is, an entity without
“a sufficient nexus to the Contracting State.” Such a “third-country resident” would not
22
necessarily be a bona fide resident of any jurisdiction—and might actually reside on paper in a
contracting state.
Despite the fact that every reference to “third-country resident” that Starr identifies
appears in Article 22’s Technical Explanation—which speaks in a language of bona fide
residency—the company nevertheless applies an on-paper understanding of that term. It argues
that its lack of on-paper ties to third countries, and the prevalence of its on-paper ties to
Switzerland and the United States, are alone sufficient to entitle the company to treaty benefits.
That formalistic approach, however, is anathema to the “substance-over-form principles”
enunciated by Article 22’s Technical Explanation. And as explained further below, it cannot be
squared with the text or structure of Article 22.
3. If It Were Valid, Starr’s Third-Country Rule Would Be a
Mechanical Test; It Is Not
Perhaps the most conspicuous obstacle to the viability of Starr’s proposed third-country-
resident test is that it appears nowhere in Article 22. As discussed above, Article 22 is largely
comprised of roughly a dozen objective, “mechanical tests.” Even though these tests are
animated by “substance-over-form principles” in that they aim to uncover whether an entity has a
“sufficiently strong nexus” to the relevant jurisdiction, Technical Explanation 59, the tests
themselves are nevertheless objective and formalistic. They are based on such factors as an
entity’s non-profit status, its ownership structure, and the on-paper residency of its owners and
controlling shareholders. The tests vary in complexity, but implementing them requires little
discretion: All are objective enough that tax professionals are able to predict with near-certainty
whether an entity will qualify for benefits pursuant to the provisions. 8 For example, paragraph 1
8
As if to illustrate the tests’ objective, non-discretionary nature, Starr notes that the
company “automatically” qualified for benefits under the similar mechanical provisions of the
23
of the Article affords treaty benefits to any on-paper resident of a treaty nation who is, among
other things, “a recognized headquarters company for a multinational corporate group,” a
company “whose principal class of shares is primarily and regularly traded on a recognized stock
exchange,” and a “company, trust or estate” predominantly owned by persons entitled to treaty
benefits under the paragraph. Treaty art. 22(1). And paragraph 3 provides that “[a] company
that is a resident of a Contracting State shall . . . be entitled to” treaty benefits on dividends,
interest, and royalties if “the ultimate beneficial owners of more than 30 percent of the aggregate
vote and value of all of its shares are persons that are resident in that Contracting State [i.e., the
state of on-paper residency], and [if it] would qualify for benefits under [certain of paragraph 1’s
provisions].” Treaty art. 22(3)(a)(i). 9
These mechanical tests are strikingly similar to Starr’s proposed third-country resident
test. The latter is objective, rule-like, and mechanical, just like Article 22’s tests—and yet
Starr’s test appears nowhere in the Article’s text. Consider how simple it would have been for
United States’ tax treaties with Ireland, the United Kingdom, and the Netherlands. Pl.’s Cross-
MSJ 12, 16.
9
That same test also requires that:
ii) the ultimate beneficial owners of more than 70 percent of all such shares are
persons described in subparagraph i) and persons that are residents of member
states of the European Union or of the European Economic Area or parties to the
North American Free Trade Agreement that are described in subparagraph b); and
iii) the amount of the expenses deductible from gross income that are paid or
payable by the company for its preceding fiscal period (or; in the case of its first
fiscal period, that period) to persons that would not qualify for benefits under
subparagraphs a), b), d), e), f) or g) of paragraph 1, is less than 50 percent of the
gross income of the company for that period.
Treaty art. 22(3)(a).
24
the treaty drafters to include a provision along the following lines, among Article 22’s laundry
list of mechanical provisions: “[A] person that is a resident of a Contracting State [may claim
treaty benefits] where such person . . . is a company, unless the predominant share of its
aggregate vote or value is owned by persons who are not residents of a Contracting State.” The
drafters included no such rule. If treaty shopping, in the drafters’ view, covered only those
situations where an on-paper third-country resident owned or operated an on-paper treaty
resident, that omission is very difficult to explain.
4. Starr’s Test Clashes With the Nature of Article 22’s Discretionary
Provision and Its Overriding Purpose
Starr does not claim that its third-country test exists anywhere in Article 22’s text, where
one would expect it to be. Rather, Starr attempts to locate it in Article 22(6)’s discretionary
provision. This is an awkward fit, for two main reasons.
First, it bears repeating that Article 22(6) is a discretionary provision. It confers broad
discretion on the Competent Authority—so broad that the Government had at least a colorable
argument that determinations under Article 22(6) are entirely unreviewable. See Starr I, 139 F.
Supp. 3d 214. The text of provision merely provides that a person “may . . . be granted” treaty
benefits if the Competent Authority “so determines after consultation with the competent
authority of the other Contracting State.” Treaty art. 22(6); see also Starr I, 139 F. Supp. 3d at
226 (describing the provision as “us[ing] the language of open-ended permission rather than
command”). And yet, despite the discretionary nature of Article 22(6), Starr would have the
Competent Authority and the Court read into the provision a mechanical rule that leaves no room
for discretion. See Def.’s Sur-Reply Supp. Cross-Mot. Summ. J. (“Def.’s Sur-Reply”) 4–5
(noting that, under Starr’s rule, the Competent Authority would have “no discretion at all under
Article 22(6) in any case not involving [on-paper] third-country ownership”); Pl.’s Sur-Reply
25
Supp. Cross-Mot. Summ. J. (“Pl.’s Sur-Reply”) 8 (conceding that the Competent Authority
would still “retain[] . . . the discretion to allow benefits [only where] a taxpayer is owned,
maintained, or established by a Third-Country Resident”).10 Under the test, if an entity lacks on-
paper associations with on-paper third-country residents, that entity gets treaty benefits—period.
It makes no sense to read this non-discretionary rule into a discretionary provision.
Second, the Technical Explanation is clear about the standards that govern Article 22(6)
determinations. Those standards are concerned not with the existence of third-country residency,
but rather with an entity’s motivation for choosing to establish treaty-country residency. The
Technical Explanation directs that an Article 22(6) determination should be based “on whether
the establishment, acquisition, or maintenance of the person seeking [treaty] benefits . . . has or
had as one of its principal purposes the obtaining of [treaty] benefits.” Technical Explanation
72. At bottom, this “requires . . . a subjective determination of the taxpayer’s intent.” Technical
Explanation 59 (emphasis added); see also id. at 60 (describing the discretionary provision’s
objective as being the “same” as that of the mechanical tests: “to identify investors whose
residence in the other State can be explained by factors other than a purpose to derive treaty
benefits”) (emphasis added). In light of this clear guidance, it would be a strange result indeed if
an entity with a “principal purpose” of obtaining treaty benefits was nevertheless entitled to
benefits under Article 22(6). But Starr’s proposed test would in some circumstances require that
very outcome.
As a final note regarding Article 22(6), the Court observes that the provision’s stated
goals and its discretionary nature go hand in hand: Making a “subjective determination of . . .
10
Because the Government cited previously unaddressed authorities in its Sur-Reply, the
Court will grant Starr’s Motion for Leave to File a Sur-Reply.
26
intent” is not an activity that lends itself to precise, objective rules. And yet this is exactly the
kind of calculus that Starr would have the Court impose upon the Competent Authority via
Article 22’s discretionary provision. Such a reading would do violence to the structure and spirit
of the Article.
5. Starr’s Test Would Result In a Cramped Conception of Treaty Shopping
Starr’s proposed test is not only at odds with Article 22 and the Technical Explanation,
but it also depends on an unduly narrow definition of treaty shopping. Starr insists that “treaty
shopping is a well-defined legal standard,” which categorically excludes situations that do not
involve on-paper third-country residents. Pl.’s Cross-MSJ 2. But, tellingly, the term is nowhere
to be found in the U.S-Swiss Tax Treaty (or any bilateral tax treaty, for that matter). Rather, as
indicated by the commentaries, legislative testimony, and agency guidance cited by the parties,
there is a fair amount of imprecision surrounding the phrase. As used by these authorities, treaty
shopping does frequently involve the participation of a third-country resident, but it need not.
Rather, its essential characteristic is treaty abuse—manipulating on-paper residency for the
purpose of obtaining treaty benefits.
For example, the Government cites commentary from the Organisation for Economic Co-
operation and Development (“OECD”) on its model tax treaty, which expresses concern
regarding “the creation of usually artificial legal constructions, to benefit both from the tax
advantages available under certain domestic laws and the reliefs from tax provided for in double
taxation conventions.” OECD Committee on Fiscal Affairs, Model Double Taxation Convention
on Income and on Capital 47 (1977). 11 The commentary explains that such abuse would occur,
11
According to one tax scholar, “[t]he OECD model conventions have constituted the
principal bases for bilateral treaty negotiations among developed nations.” Robert Thornton
Smith, Tax Treaty Interpretation by the Judiciary, 49 Tax Law. 845 (1996).
27
“for example, if a person (whether or not a resident of a Contracting State), acted through a legal
entity created in a State essentially to obtain treaty benefits which would not be available directly
to such person.” Id. (emphasis added). Treaty abuse would also occur, according to the
commentary, if an individual “transferred his permanent home [from one Contracting State] to
the other Contracting State, where [capital] gains were subject to little or no tax.” Id. Clearly,
these examples do not turn on a third-country participant, and yet the OECD’s more current
commentary describes this treaty abuse as “the problem commonly referred to as ‘treaty
shopping.’” OECD Committee on Fiscal Affairs, Model Tax Convention on Income and on
Capital C(1)-26 (2014).
Similarly, the parties have cited and discussed testimony regarding the U.S.-Swiss
Treaty, presented before the Senate Committee on Foreign Relations, from former Deputy
Assistant Treasury Secretary Joseph H. Guttentag. See Bilateral Tax Treaties and Protocol:
Hearing Before the S. Comm. on Foreign Relations, 105th Cong. 354, at 11 (1997). Mr.
Guttentag explains that one
major objective of U.S. tax treaty policy is to . . . prevent abuse of the treaty by
persons who are not bona fide residents of the treaty partner. This abuse, which is
known as “treaty shopping,” can take a number of forms, but its general
characteristic is that a resident of a third state that has either no treaty with the
United States or a relatively unfavorable one establishes an entity in a treaty partner
that has a relatively favorable treaty with the United States.
Id. Guttentag elaborates that, while treaty shopping “general[ly]” involves a third-country
resident, it “can take a number of forms,” and it is primarily concerned with treaty abuse “by
persons who are not bona fide residents of the treaty partner.” Id.
Starr’s definition of treaty shopping, by contrast, would narrow the concept to such an
extent that even some persons who are not bona fide residents of a treaty nation—persons who
lack a “sufficient nexus” to either contracting state—would be entitled to benefits. Of course,
28
that is likely Starr’s reason for proposing such a standard, since it largely concedes that it was not
a bona fide resident of Switzerland or the United States at the relevant time. 12
Before proceeding to Starr’s remaining arguments, some qualifications are in order.
First, the Court does not mean to suggest that an entity’s on-paper residency (and that of the
individuals it associates with) is irrelevant to its bona fide residency. Surely, in exercising its
discretionary judgment under Article 22(6), it would be reasonable for the Competent Authority
to consider Starr’s lack of affiliations with on-paper third-country residents in evaluating the
company’s bona fide connections to treaty states. There is simply no per se Article 22 rule,
however, requiring the Competent Authority to reach such a determination. Second, clearly any
bilateral tax treaty is intended to benefit the legitimate residents of the two signatory nations.
Accordingly, it would also have been permissible as a matter of policy, but was not required as a
12
Rather than defending its bona fide connections to Switzerland, Starr focuses its
energies on asserting that bona fide residency is a “made-up” requirement, with “no basis in the
record or the law.” Pl.’s Reply Supp. Cross-Mot. Summ. J. (“Pl.’s Reply”) 19. That claim is
indefensible, given that the term “bona fide residency” is mere shorthand for the sufficient nexus
inquiry, which is outlined in Article 22’s Technical Explanation. In any event, defending its
bona fide Swiss residency would be an uphill battle in light of the company’s admissions that,
for most of the period between 2003 and 2008, Starr had only one salaried employee, who
followed the company from Bermuda, to Ireland, and finally to Switzerland. A.R. 90–91.
Despite its being controlled by U.S. citizens, Starr as an entity does not claim to be a
bona fide resident of the United States, either. Indeed, under the Treaty, it does not appear to be
possible for an entity to have bona fide residence in a jurisdiction without also having on-paper
residence in that jurisdiction. Because Starr and all of its subsidiary entities reside on paper in
Switzerland, it could therefore not establish bona fide residency in the United States. More to
the point, because the Treaty aims to provide taxation benefits to residents of one contracting
state on income sourced in the other, establishing residency in the United States under the U.S.-
Swiss Treaty would not benefit Starr, which seeks treaty benefits on U.S.-based income.
29
matter of law, for the Competent Authority to consider an argument that Starr’s majority-control
by U.S. citizens should counsel in favor of awarding it benefits under Article 22(6).13
In any event, the point remains that it is difficult to square Starr’s version of the Article
22(6) standard with Article 22’s text, structure, and accompanying Technical Explanation. The
Court therefore reaffirms that the proper standard for determining benefits under Article 22(6) is
“whether the establishment, acquisition, or maintenance of the person seeking [treaty] benefits
under the Convention, or the conduct of such person’s operations, has or had as one of its
principal purposes the obtaining of [treaty] benefits.” Technical Explanation 72. The Competent
Authority clearly applied this standard. A.R. 274.
B. The Competent Authority’s Application of the Article 22(6) Standard
In a separate line of argument, Starr contends that the Competent Authority’s
determination was arbitrary and capricious even assuming that it correctly framed the Article
22(6) standard. These arguments fault the Competent Authority’s analysis, i.e., the manner in
which it applied the “principal purpose” standard. At a broad level, the arguments make two
points: first, that the Competent Authority overlooked essential information that would have
compelled a different result; and second, that the Competent Authority heeded irrelevant or
incorrect facts, and thereby drew unreasonable conclusions. The Court now turns to each
argument.
13
Starr in fact did make that argument before the Competent Authority. See A.R. 62
(Starr arguing in its 2007 submission that it is “not aware of any policy reason to exclude [from
treaty benefits] a company with Swiss tax residency, more than 80% of whose voting power is
owned by U.S. citizens”). By contrast, Starr first advanced its third-country-resident test in this
litigation.
30
1. Whether Certain, Purportedly Key Evidence Required a Contrary Result
Starr argues that certain evidence shows, incontrovertibly, that obtaining tax benefits
under the Treaty was not one of its principal purposes in relocating to Switzerland. 14 Starr
emphasizes that its “U.S. tax position could not have been improved by relocating to Switzerland
because the most favorable withholding tax rate [it] could expect on dividends from AIG under
the U.S.-Swiss Treaty was the same rate Starr was entitled to under the U.S.-Irish Treaty.” Pl.’s
Cross-MSJ 37. Because that equivalent, favorable rate was automatic in Ireland but
discretionary in Switzerland, “the only change to [Starr]’s treaty position resulting from the
move from Ireland to Switzerland was a reduction in certainty that [Starr] would be entitled to a
reduction in the withholding tax rate,” meaning that tax benefits could not have been one of its
principal purposes in relocating. Id. at 38.
This argument, however, assumes a much narrower inquiry than is called for by the
Technical Explanation, which directs the Competent Authority to determine “whether the
establishment, acquisition, or maintenance” of a company in the relevant jurisdiction had a
principal purpose of obtaining treaty benefits. Technical Explanation 72. Notably, the
Explanation does not direct the Competent Authority to ask merely what made a company’s
current jurisdiction more favorable than its previous one—although that might be part of the
analysis—but rather why a company chose to “establish” or “maintain” itself where it did. In
other words, here the question was not simply why Starr chose Switzerland over Ireland, but
rather why Starr chose Switzerland over any other jurisdiction where it might have moved.
14
Starr argues that a principal purpose must be at least “a first purpose among equals,”
Pl.’s Cross-MSJ 36, and not simply an “important” purpose, Def.’s Reply 12. The Court finds
the distinction inconsequential as applied here. For the sake of argument only, then, the below
discussion assumes Starr’s more stringent definition.
31
Furthermore, the language employed by the Technical Explanation, along with analogous
guidance, suggests that the Competent Authority should engage in an historical, totality-of-the-
circumstances inquiry in evaluating a company’s reasons for choosing a jurisdiction. See
Technical Explanation 72; Treas. Reg. § 1.884-5(f)(2) (branch profits tax regulations, permitting
the IRS to consider “the continuity of the historical business and ownership of the foreign
corporation”). Presumably, that would include a consideration of Starr’s history of moves, not
just the company’s most recent relocation.
Relatedly, Starr argues that the “decision matrix” conclusively demonstrates that
obtaining treaty benefits was not of principal concern to the company when it decided to relocate
to Switzerland. See Pl.’s Cross-MSJ 39–40. Just as Switzerland offered a less favorable regime
than did Ireland, it also offered a less favorable tax regime than the other final contenders on
Starr’s list of potential new jurisdictional homes. In particular, the Netherlands and the United
Kingdom guaranteed Starr the low dividends withholding rate that it had in Ireland, and that it
could only attain in Switzerland under the discretionary exception. In Starr’s view, that shows
tax motives were not a chief concern when it made the move to Switzerland.
There are numerous problems with this argument. As an initial matter, the decision
matrix itself is suspect evidence: It appears to have been created in 2009, years after the move to
Switzerland, and in an effort to convince the Competent Authority that seeking treaty benefits
was not a principal motive behind Starr’s move. See A.R. 144. Second, even assuming the
matrix accurately summarizes Starr’s earlier decisionmaking process, it hardly shows that tax
benefits were low on the company’s priority list. In fact, as the Government points out, Starr’s
acknowledgment that “U.S. Tax” was one of four key criteria that the company analyzed in
deciding on a jurisdiction shows that it “constituted [a] principal consideration[]” in Starr’s
32
calculus. Def.’s Mem. Supp. Cross-Mot. Summ. J. (“Def.’s Cross-MSJ”) 21. Third, it is unclear
how Starr arrived at its list of five finalist jurisdictions, but presumably that narrowing process
would have some bearing on whether treaty benefits were a “principal” factor in its ultimate
selection. (For instance, if Starr’s most important criterion in selecting a list of five jurisdictions
was a favorable tax relationship with the U.S., then it would not be surprising if Starr focused
more on other factors in making its final cut.) Finally, Switzerland was rated as “Bad” for U.S.
taxes on the decision matrix. But it was definitely better in that regard than at least one of the
other finalists (Bermuda), and if the company had been afforded discretionary relief—which
Starr certainly seems to indicate was expected—then Switzerland would have tied for first place
in that category with the other jurisdictions.
For the above reasons, neither Starr’s decision matrix, nor its decision to leave Ireland in
favor of Switzerland, inevitably leads to the conclusion that treaty benefits were not a “principal”
concern for the company when it chose to relocate. Rather, if the matrix is accepted as reliable
evidence of Starr’s decisionmaking process, it actually reveals that “U.S. Tax” considerations
were a top priority in selecting Starr’s new home. For these reasons, this evidence does not
render the Competent Authority’s determination arbitrary and capricious.
2. Whether the Competent Authority’s Analysis Rested on Irrelevant or
Incorrect Determinations
In addition to faulting the Competent Authority for what it did not analyze, Starr takes
issue with what was analyzed, and how. The company argues that “each of the [Competent
Authority]’s justifications is either unsupported by the record or patently unreasonable.” Pl.’s
Cross-MSJ 40.
33
i. The Competent Authority’s Analysis Regarding the Treaty’s Silence on
Entities with Starr’s Particular Structure
During the period when its request for treaty benefits was under review, Starr argued to
the Competent Authority that the Treaty’s failure to specifically afford benefits to companies
with Starr’s particular structure—that is, for-profit companies owned by charitable entities—was
an inadvertent oversight. The argument was apparently grounded in a comparison with other
bilateral U.S. tax treaties, many of which do include specific provisions affording relief to
companies with such structures. In its determination letter, the Competent Authority stated that
it could not reach “a definitive conclusion” on Starr’s position. A.R. 273. It elaborated that “it is
not possible for us to conclude whether [a Limitation on Benefits] provision like that found in the
[U.S.-Netherlands Tax Treaty] or the [U.S.-United Kingdom Tax Treaty], which . . . may have
applied to permit [Starr] treaty benefits if included in the U.S.-Swiss Treaty, was not included in
the U.S.-Swiss treaty intentionally or by casual omission.” Id. In its briefing, Starr summarizes
what it characterizes as “strong evidence supporting the interpretation that the negotiators for the
U.S.-Swiss Treaty did not consciously exclude [companies like Starr] from the automatic
qualifying provisions in Article 22,” and it faults the Competent Authority for failing to
definitively reach that conclusion. Pl.’s Cross-MSJ 43. Then Starr calls this whole “justification
. . . irrelevant.” Id.
This argument is a nonstarter. Starr essentially asks the Court to find that the Competent
Authority acted arbitrarily and capriciously because it failed to definitively conclude that the text
of the U.S.-Swiss Treaty should be overwritten by text in other bilateral tax treaties, and because
there is no legislative history to the contrary. But “[t]he interpretation of a treaty, like the
interpretation of a statute, begins with its text.” Starr I, 139 F. Supp. at 226 (quoting Abbott v.
Abbott, 560 U.S. 1, 10 (2010)). So at the very least, it was not unreasonable for the Competent
34
Authority to decline to read into the treaty a provision that was not there. Moreover, it bears
emphasizing that the Competent Authority reached no conclusion one way or the other on the
matter, and therefore the analysis appears not to have grounded its final determination. It is
therefore misleading for Starr to characterize it as a “justification,” although perhaps accurate to
call it “irrelevant.” Pl.’s Cross-MSJ 43.
ii. Whether the Four “Circumstances” Cited by the Competent Authority Were
Inaccurate or Irrelevant to the Article 22(6) Standard
The Competent Authority cited four sets of “facts and circumstances” that it considered
“troubling” and that undergirded its principal-purpose determination: (1) Starr’s initial
incorporation in Panama and later residence in Bermuda, which suggested previous aims of tax
avoidance; (2) Starr’s move to Ireland shortly before major AIG dividend payments; (3) Starr’s
brief stay in Ireland, which “may or may not have been intentionally transitory”; and (4) Starr’s
majority control by U.S. individuals. A.R. 273–74. Starr argues that “[n]one of the[se]
circumstances relate to treaty shopping,” and that they are “unsupported by the record.” Pl.’s
Cross-MSJ 44. Again, these arguments are not persuasive.
As for the first three “circumstances,” which relate to Starr’s previous history of
residence and incorporation, Starr does not challenge their factual accuracy. Rather, Starr seems
to suggest that because they predate its move from Ireland to Switzerland, they are irrelevant to
the Article 22(6) inquiry. But as previously discussed, there is nothing in the Article 22(6)
standard that limits the historical lens through which the Competent Authority may make a
“principal purpose” determination. See supra section III.B.1. Rather, it stands to reason that a
company’s past jurisdictional moves—including its willingness to relocate in the interest of
obtaining tax benefits—might shed light on its most recent relocation. It was reasonable, not
capricious, for the Competent Authority to consider such historical data. Starr also takes issue
35
with the inferences the Competent Authority drew from such facts—most significantly, that the
move to Ireland was tax-motivated. But as recounted above, there was abundant evidence to
support that conclusion. See supra section I.B.
Starr also challenges the Competent Authority’s conclusion that Starr’s control by U.S.
individuals “is not in accord with recent developments of U.S. policy on acceptable corporate
ownership for [Limitation of Benefits] purposes.” A.R. 274. The Government explains that this
statement was “in reference to the [IRS’s] discovery of abusive structures whereby U.S.
individuals invest in the United States by structuring transactions through tax havens, including
treaty countries, which impose little or no tax on such arrangements.” Def.’s Reply 37. In
Starr’s view, “[a]llowing the [Competent Authority] to apply legal standards that were not
identified in the treaty itself or in the accompanying guidance would be tantamount to granting
the [Competent Authority] unfettered discretion in deciding whether to grant treaty benefits.”
Pl.’s Cross-MSJ 47. This greatly overstates the case. Article 22(6) bestows significant
discretion on the Competent Authority to sift treaty shoppers from non-treaty shoppers, and that
surely includes permission to take stock of current legal standards and policies. 15
iii. Whether the Competent Authority Improperly Relied on a “Narrow-Miss”
Understanding of Article 22(6)
After concluding that one of Starr’s principal purposes for relocating to Switzerland was
obtaining treaty benefits, the Competent Authority went on to state: “Moreover, in our view,
Article 22(6) of the U.S.-Swiss Treaty was designed to provide relief to a taxpayer that can make
15
Moreover, both the Treaty and the Technical Explanation suggest the legal standards
“in effect at the time the Convention is being applied, [and] not the law as in effect at the time
the Convention was signed,” should control in interpreting the treaty’s undefined terms.
Technical Explanation 9; see also Treaty art. 3(2) (providing that “any term not defined [in the
treaty] shall . . . have the meaning which it has under the laws of that State concerning the taxes
to which the Convention applies”).
36
a strong case that, while coming within the spirit of an available [Limitation on Benefits]
provision, it narrowly misses the mechanical tests associated with that provision.” A.R. 274.
Starr notes that “the U.S.-Swiss Technical Explanation and the U.S.-Swiss Treaty do not mention
or suggest a ‘near miss’ requirement,” and that even if such a requirement existed, “[Starr] would
satisfy it.” Pl.’s Cross-MSJ 48. As the Government explains, however, the Competent Authority
in making this statement was likely responding to arguments that Starr itself had made during the
review process. See Def.’s Reply 36. More to the point, the statement was made as an aside,
after the Competent Authority had applied the “principal purpose” standard in order to reach its
determination. Whether the “near-miss” concept accurately describes the scope of Article 22(6)
is therefore beside the point.
IV. Conclusion
In light of the above discussion, the Court sees nothing arbitrary or capricious in the
Competent Authority’s finding that “at least one of the principal purposes for moving Starr’s
management, and therefore its residency, to Switzerland” was to obtain tax benefits under the
U.S.-Swiss Tax Treaty. A.R. 274. By applying the correct legal standard to a host of
indisputably relevant facts, the Competent Authority satisfied its obligations under the APA to
“examine the relevant data and articulate a satisfactory explanation for its action including a
rational connection between the facts found and the choice made.” Ark Initiative, 816 F.3d at
127.
37
Accordingly, as specified in the accompanying Order, the Court will grant the
Government’s motion for summary judgment and deny Starr’s cross-motion. 16
CHRISTOPHER R. COOPER
United States District Judge
Date: August 14, 2017
16
Starr argues that the Chenery rule precludes the Government from raising certain
allegedly post hoc justifications for the Competent Authority’s denial decision. See SEC v.
Chenery Corp., 332 U.S. 192, 196 (1947) (A “simple but fundamental rule of administrative
law” is that a court reviewing agency action “must judge the propriety of such action solely by
the grounds invoked by the agency.”). In particular, Starr argues that the Court should not
consider the following: (1) Starr’s alleged failure to disclose its pending litigation with AIG; (2)
Starr’s supposedly non-charitable purposes; and (3) Starr’s purported failure to disclose its actual
motivations for various relocations. Pl.’s Cross-MSJ 50. Because the Court has found that the
Competent Authority’s denial decision was reasonable without the aid of these arguments, there
is no need to reach the question of whether they are barred by Chenery.
Similarly, because the Court has not relied in any way on the email attached to the
Government’s Sur-Reply, see Def.’s Sur-Reply, Ex. C, which forms the basis for Starr’s Motion
to Strike or Reopen Discovery, the Court will deny that motion as moot.
38