Court Orders Enforcement Of A Summons Request From Abroad Allegedly To Harass Prominent Members Of Opposition Political Party

Tax agencies face problems enforcing tax laws when their citizens have investments located abroad. One tool that countries use to address that problem is to include exchange of information provisions in tax treaties.  Another tool is a specific exchange of information treaty, or a tax information exchange agreement. The general effect of these treaties and agreements is that a requesting country can rely on the other country’s process to gather information that be relevant for ascertaining compliance with the requesting country’s tax obligations.

Recent appellate opinions address non US country requests that lead to the IRS issuing a third party summons to obtain information about US-sourced investments. I discuss the issue extensively in Saltzman and Book Chapter 13, but the upshot is that courts have essentially held that a challenge to the enforcement of the summons on the basis that the requesting nation is seeking the information for an improper purpose is not relevant to the inquiry.

What is relevant?

read more…

The case law, essentially applying the well-known Supreme Court Powell standard, requires that the IRS must make a prima facie showing of its “good faith” in issuing the summons.  That is a minimal burden, and the IRS need only demonstrate its own good faith, not that of the requesting party.

Once that minimal hurdle is cleared the taxpayer then has the burden to challenge the government’s evidence or otherwise show that there are grounds to quash the summons.

This framework does not sit well with taxpayers who allege that the requesting country may have improper or nontax related motives in getting the US documents. For example, in Puri v US, a Ninth Circuit  nonprecedential case from earlier this year, the taxpayer alleged that the Indian tax authority sought investment information from Citibank to harass their family, who are prominent members of the Indian opposition political party.

Under the framework I discuss above, the court held that those allegations were not relevant, resulting in an order to enforce the summons.

One aspect of the Puri opinion that stood out is that it proceeded to analyze the merits of the allegations, stating that “in any event” Puri “does not present any plausible evidence suggesting that the Indian tax authorities acted in bad faith.” The opinion notes that Puri failed to connect how the requested bank statements could be used for harassment.

That discussion engendered a concurring opinion that joined in all but the “in any event” discussion:

Because courts are categorically forbidden from inquiring into the bad faith of a foreign government when deciding whether to quash an IRS summons, I see no need for us to reach the “in any event” argument in paragraph two

As the concurrence notes, an inquiry of any kind into a requesting state’s motives raises separation of powers concerns:

I have serious separation-of-powers concerns for even raising the prospect that courts can look through the Executive branch’s decision to comply with an international treaty and surmise the motives of a foreign government. That is clearly beyond our competence. 

Conclusion

We generally take for granted that US tax administration is a nonpartisan business, though recent events are starting to show that the norms are shifting.  Questions about the perhaps coincidental research audits of prominent opponents of the Trump Administration and allegations of pressure to audit from the former President’s Chief of Staff John Kelly have brought attention to that topic.  

There is a sense, perhaps more intuitive than empirical, that other tax administrators are less able to insulate their actions from political pressure. As Puri and other cases reflect, the norm is that US courts are to stay out of the messy inquiry of motive.

A Second Review of Ninth Circuit Argument in Altera v. Commissioner

Today we welcome back guest blogger Stu Bassin for his take on the argument in the Altera case. Stu has blogged with us on several occasions. Because of the importance of the case, we are providing two views of the argument in Altera today. Keith

The Ninth Circuit held the long-awaited argument on the Government appeal of the Tax Court’s ruling in Altera Corp. v. Commissioner, 145 T.C., No. 3 (2015), on Wednesday, October 11. The case arose out of an IRS notice of deficiency which invoked Section 482 (and, specifically, Treas. Reg. §1.482-7(d)(2)) to redetermine the transfer prices employed for intra-group transactions amongst Altera corporate affiliates.   The Tax Court’s ruling, which invalidated the regulation under the Administrative Procedure Act (the “APA”) because of defects in the rulemaking process, has drawn wide-spread interest amongst practitioners involved in both transfer pricing and regulation validity matters.

read more...

Before the Tax Court, the parties agreed that the law generally requires participants in intra-group transactions to determine transfer prices in accordance with the prices comparable unrelated parties employ in arms-length agreements. The parties disagreed, however, regarding the proper allocation of stock-based compensation costs amongst the affiliates. The IRS supported its deficiency notice with a regulation which specifically required affiliates to share stock-based compensation costs in computing the transfer price, while the taxpayer contended that the regulation was invalid under the APA because it deviated from the comparable arms-length transaction test traditionally employed in computing transfer prices.

According to the taxpayer, during the rule-making process, commenters submitted substantial evidence supporting the proposition that, in practice, cost sharing agreements amongst unrelated entities operating at arms-length do not require sharing of compensation costs. The IRS did not identify any instance of a cost sharing agreement which provided for sharing of compensation costs in the preamble to the final regulations. Instead, it asserted an economic theory-based policy analysis to support its determination that cost sharing agreements must provide for sharing of compensation costs. The taxpayer, therefore, argued that the regulation was invalid because its requirement of sharing compensation costs in computing transfer prices was arbitrary, capricious, and inconsistent with the evidence before the Service during the rulemaking process.

The Tax Court unanimously ruled in favor of the taxpayer, invalidating the regulation and rejecting the proposed Section 482 adjustment.   The Tax Court’s analysis focused upon the second stage of the regulation validity inquiry mandated by Mayo Foundation v. United States— whether the determinations reflected in the regulation were arbitrary and capricious. The opinion criticized the IRS for failing to engage in actual fact-finding, failing to provide factual support for its determination that unrelated parties would share compensation costs in their cost-sharing agreements, failing to respond to significant comments, and acting contrary to the factual evidence before Treasury. Accordingly, the regulation failed to satisfy the reasoned decision-making standard established by Supreme Court precedent under Mayo and related cases.

On appeal, Altera was heard by a panel consisting of Chief Judge Thomas, Judge Reinhardt (the dissenter in the Ninth Circuit’s earlier Xilinx decision in favor of the taxpayer in a similar Section 482 case), and Judge O’Malley of the Federal Circuit. All three judges were appointed by Democratic presidents. Arthur Catterall, one of the top appellate lawyers from the Justice Department’s Tax Division, argued the case on behalf of the Government.   Donald Falk, a general appellate litigation specialist from Mayer Brown, argued the case on behalf of the taxpayer.  Appellate junkies familiar with appellate arguments in tax cases where the panel is largely silent may be surprised to learn that all three judges actively questioned both lawyers and that the argument extended to a full hour.

The Government focused its argument upon the first stage of the Mayo analysis—the agency’s statutory authority to issue a regulation which departed from the comparable arms-length standard for evaluating transfer pricing arrangements. It argued that the Treasury had authority to regulate on the treatment of cost-sharing agreements because of statutory ambiguity produced by tension between the two sentences of Section 482. The text of the statute provides—

“In any case of two or more organizations . . . owned or controlled directly or indirectly by the same interests, the Secretary may distribute, apportion, or allocate gross income, deductions, credits, or allowances between or among such organizations, trades, or businesses, if he determines that such distribution, apportionment, or allocation is necessary in order to prevent evasion of taxes or clearly to reflect the income of any of such organizations, trades, or businesses. In the case of any transfer (or license) of intangible property (within the meaning of section 936(h)(3)(B)), the income with respect to such transfer or license shall be commensurate with the income attributable to the intangible.”

The first sentence, which has been part of the Code for decades and is consistently reflected in many tax treaties, has historically been construed by Treasury and the courts to incorporate a requirement that a taxpayer’s transfer prices be evaluated based upon their comparability to the arrangements negotiated by unrelated entities operating at arms-length. The second sentence, added in 1986 and focusing upon transfers of intellectual property, requires that the income from the transfer be apportioned in a manner “commensurate with income.” According to the Government, the differing results occasionally produced by a commensurate with income standard and comparable arms-length transaction standard create an ambiguity which allows Treasury to issue regulations which deviate from the arms-length standard for cost allocation.

The taxpayer acknowledged that the arms-length comparability standard and the commensurate with income standard are somewhat different and can produce different results in some cases. That difference, however, did not authorize Treasury to abandon the arms-length comparability standard for allocation of stock-based compensation costs. According to the taxpayer, both the statutory language and the legislative history of the 1986 amendment support a far narrower role for the commensurate with income standard. While the legislative history demonstrates that Congress was concerned about problems which had arisen with arms-length comparability analyses employed in connection with intellectual property transfers, the legislative history contains many references endorsing arms-length comparability analysis in other contexts. Similarly, the statutory language of the commensurate with income provision only applies to intellectual property transfers. Ultimately, the taxpayer contended the commensurate with income statutory language did not support abandonment of arms-length comparability in evaluating the allocation of compensation costs under the taxpayer’s cost-sharing agreement.

Virtually all of the panel’s questions focused upon the statutory construction questions and their implications for the scope of Treasury’s authority to promulgate regulations inconsistent with the arms-length comparability standard. The panel appeared to recognize the tension between the arms-length comparability standard and the commensurate with income standard. It questioned, however, the scope of the tension and the range of costs which Treasury could allocate without regard to arms-length comparability analysis. The government contended that the tension allowed Treasury to promulgate regulations governing all aspects of cost sharing agreements, while the taxpayer tried to limit such regulations to the intellectual property transfer arena.

Interestingly, the argument gave relatively little attention to the second stage of the Mayo analysis—the arbitrariness of Treasury’s determination.   The government did not challenge the Tax Court’s conclusions that the regulation was contrary to the evidence regarding comparable arms-length transactions. Instead, it argued that Treasury had almost unlimited discretion to prescribe the allocation of costs if the court agreed that Treasury had authority to prescribe rules contrary to the arms-length comparability evidence. To the contrary, the taxpayer argued that the absence of any arms-length comparability evidence rendered the regulation arbitrary and capricious. The panel, however, did not pursue this line of argument, notwithstanding the Tax Court’s focus on the issue.

The panel gave no indication of when it would render its decision in Altera. Full opinions on appeals to the Ninth Circuit tend to take a long time, so it seems likely that it will be several months before a decision is issued.

 

 

 

 

 

 

 

The Altera Oral Argument

We welcome back guest bloggers Professor Susan C. Morse from University of Texas School of Law and my colleague Senior Lecturer on Law Stephen E. Shay from Harvard.  Professors Morse and Shay, build on their post last week to fill us in on what happened before the 9th Circuit in Altera. Keith

At the Ninth Circuit on Wednesday October 11, government counsel carefully threaded the needle of statutory and regulatory interpretation in Altera, a case about transfer pricing and administrative law. Taxpayer counsel appeared to overreach. It refused to concede that Treasury has any authority to regulate the pricing of intercompany intellectual property sharing under qualified cost sharing arrangements (QCSAs) unless the guidance proceeds from the starting data point of unrelated party dealings, otherwise known as comparability analysis.

read more...

The panel included Ninth Circuit Chief Judge Sidney Thomas, Ninth Circuit Judge Stephen Reinhardt, and D.C. Circuit Judge Kathleen O’Malley, sitting by designation. Reinhardt joined the first Ninth Circuit Xilinx decision overturning the Tax Court decision, which interpreted the prior cost sharing regulation to allow the IRS to include stock option costs in the pool of shared costs. After a rehearing, Reinhardt dissented in the superceding Ninth Circuit Xilinx opinion that upheld the Tax Court. In Xilinx, he would have allowed the government to require inclusion of stock option costs in a cost-sharing pool even under earlier regulations that did not explicitly address stock options. The final regulations at issue in Altera, the current case, plainly say that stock option costs must be included in a QCSA cost pool, to the disadvantage of U.S. multinational groups which as a result may take fewer tax deductions resulting from the exercise of stock options. Billions of dollars of tax revenue are at stake in Altera.

The oral argument featured three important threads: The imposition of an administrative law framework with a Chevron starting point; the argument that “arm’s length” is not synonymous with “comparability analysis”; and the idea that the second sentence of section 482, which refers to “commensurate with income” payment for intellectual property “transfers”, specifically envisioned transfer pricing not tethered to unrelated party data points.

Judge O’Malley, who brought seven years’ worth of D.C. Circuit administrative law experience    to the hearing, repeatedly insisted on a textbook administrative law analysis. She asked both parties whether there is statutory authority for these regulations under Chevron. Yes, replied the government. Chief Judge Thomas asked whether the government has the statutory authority to “eliminate” comparability analysis altogether, for all transactions. No, replied the government, here trying to thread the needle. The statute does not say “arm’s length,” let alone comparability. Both are described in regulations. But there is “too much history.”

Well, then if the government cannot erase the arm’s length standard, how can it write regulations that set aside unrelated party data, like the agreements taxpayers point to under which unrelated parties develop technology together without mentioning stock options? Judge Reinhardt suggested that the validity of the regulation had to do with its subject: the sharing of intangible assets. Perhaps comparability analysis is not relevant for transactions involving intangibles in particular, he suggested. Agree, with respect to cost-sharing arrangements, replied the government.

But why doesn’t the departure from comparability analysis for intangibles violate the arm’s length standard? In response to prompts from the panel, the government agreed that arm’s length and comparability do not “go hand in hand” and are “not synonymous.” There are several “means to [the] end” of an arm’s length result. In the case of QCSAs, unrelated party data is “inherently not comparable” and cannot support clear reflection of income.

Taxpayer counsel, in contrast, contended that “it has to be an empirical analysis” and that “you have to take comparables as far as they will go,” and appeared to argue that this approach was required by the statute itself. “What if [the comparables] don’t go anywhere?” asked Chief Judge Thomas. Well, replied taxpayer counsel, then the government should “erase” regulations’ reference to an arm’s length standard. In rebuttal, the government further argued that the term “arm’s length standard” is a “term of art” and that Treasury’s interpretation is entitled to deference.

The second sentence of Section 482, added in 1986, allows the government to adjust related parties’ inclusions from “transfer” of intangibles so that they are “commensurate with income.” As the government pointed out, the legislative history clearly explains that unrelated party data points – i.e., comparability analysis – are not sufficient to allow clear reflection of income in these situations involving intangibles. This is strong evidence of statutory authority for the government to write regulations that depart from comparability analysis. Taxpayer counsel suggested that a QCSA might not qualify as a “transfer” under this sentence of the statute, so that perhaps it was not statutory authority at all. But government counsel disagreed, arguing that the word “transfer” was broad enough to encompass QCSAs and noting that this issue was apparently briefed, and ignored, in Xilinx.

The Tax Court cited State Farm, which requires reasonable explanation of policy changes, in its decision to set aside the Treasury’s regulations. Other reasonable explanation cases include Fox Television and Encino Motorcars, both of which came up during oral argument. O’Malley asked government counsel why the regulations requiring cost sharing were not a change; the government replied that the policy of requiring stock options costs to be included in pools had existed since 1997, years before the regulations were promulgated. Later, taxpayer counsel pushed the State Farm argument, insisting that some of the government’s arguments in litigation were “not what they said” in the preamble. But the panel did not pursue the specifics of the preamble’s language. And taxpayer counsel’s assertion that Chevron should be the “last step” of the analysis of regulatory validity was met with silence by the court.

Stay tuned for our analysis of the Ninth Circuit’s Altera decision – we’ll blog it here in due course.

 

 

 

 

 

 

Court Sustains Competent Authority Decision to Not Grant Treaty Relief

The other day I discussed Starr International v US, and the lead up to an opinion earlier this month concerning the application of the US –Swiss income tax treaty. Before the court was able to resolve the matter on the merits, the district court addressed its jurisdiction to hear a challenge to the US Competent Authority’s decision to not grant discretionary relief under the treaty. While concluding that it could not order monetary relief, its prior opinion opened the door to Starr challenging the Competent Authority’s decision not to grant a lower withholding rate under the APA.

read more...

The amount at issue was substantial. Starr was one of the largest shareholders in AIG. It received about $190 million in dividends. The US has treaties with many countries; those treaties generally provide exceptions or reduction to the default 30% withholding on some US sources of income, including dividends. The US Swiss treaty reduced withholding to either 5% or 15 %, depending on the Swiss entity’s ownership of the US corporation.

Most treaties have some form of anti-treaty shopping provisions that are meant to ensure that only bona fide residents of contracting states can take advantage of the treaty. The Swiss treaty has such a provision, Article 22, which, according to the treaty’s technical explanation (sort of treaty analogue to legislative history, which Treasury staffers draft and present to the Senate during the Senate’s treaty ratification process), denies treaty benefits to those who establish “legal entities . . . in a Contracting State with a principal purpose to obtain [treaty] benefits.”

Article 22 has a number of objective tests; an entity can establish that it is a bona fide resident if it meets any of the objective tests. The treaty recognized, however, that a party might be entitled to treaty relief even if it were unable to satisfy any of the objective tests. To effectuate that policy, the treaty provides:

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.

It was this discretionary benefit position that was at issue in Starr International. For international tax folks, the opinion has an important discussion of the precise contours of the anti-treaty shopping provision; Starr wanted a more mechanical approach to the issue but the opinion agreed with the government that the test is one that pivots off of a finding that the party seeking the benefits “has or had as one of its principal purposes the obtaining of [treaty] benefits.”

After agreeing that the treaty rule revolves around a principal purpose analysis, the court turned to the APA. 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 USC 706(2). As we have discussed, this is generally a deferential standard; the opinion, citing the Supreme Court State Farm decision, notes that by way of example that “[a]gency action is arbitrary and capricious…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.’”

In arguing that the Competent Authority decision was arbitrary and capricious, Starr essentially made two main points: the Competent Authority considered irrelevant information and failed to consider the relevant information.

In finding that the Competent Authority did not consider irrelevant information, the opinion squarely addresses how much it should consider Starr’s prior moves. One of the main points Starr made was that its move to Switzerland was from Ireland, which had an automatic treaty reduction; in other words, it could not have had a principal purpose to get treaty benefits if it moved from a jurisdiction where it already was entitled to benefits. The government noted that Starr had moved previously, and wanted to consider the entity’s history of moving as to show that tax was often a if not the main reason for location.

The court agreed with the government, looking mainly to the treaty’s explanation:

[Starr’s] 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.

This broader inquiry required not just a look at Starr’s recent move, but also its overall history of moves.

Starr also argued that the Competent Authority failed to consider the implications of how other bilateral treaties automatically allow treaty benefits to similarly structured entities (a for profit company owned by a charitable entity). Starr argued that there was strong evidence that the US and the Swiss did not consciously exclude that structure from benefits. The opinion gives short shrift to this point:

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 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.”

Conclusion

On the merits, the opinion is a major victory for the government. Yet it is  an important procedural victory for taxpayers. It is another defeat of the reflexive government argument that some of its decisions are completely insulated from court review. It also is a roadmap for showing how parties can use the APA to challenge the somewhat murky world of Competent Authority decisions under treaties.

Starr v US and The Power to Confer Discretionary Treaty Benefits: Part 1

It is not often that the courts wrestle with the application of discretionary treaty provisions. Earlier this month, in Starr International v US , a DC federal district court found that the Competent Authority did not act arbitrarily or capriciously in denying discretionary relief under the U.S.-Swiss Treaty. In today’s post, I will  discuss the jurisdictional battle that led to last week’s opinion. I will follow up in Part 2 with a discussion as to the court’s application of the APA to the treaty claims.

read more...

In 2015, in District Court Hands IRS Loss in its Bid to Exclude Discretionary Treaty Benefits From Judicial Review I discussed the fight between The Starr International Group and the IRS over Starr’s efforts to get the benefits of discretionary treaty relief that would have reduced US withholding on AIG dividends. Much of the post discussed the government’s unsuccessful efforts to convince the court that the decision was not subject to court review. That opinion, and my post, discusses why there is a strong presumption against unfettered agency discretion.

Following that setback, the government asked the court to reconsider. In a 2016 opinion, the government did get a victory, of sorts. Despite reaffirming its earlier conclusion about the court’s power to review the IRS’s decision to not grant discretionary treaty relief, the District Court held that separation of powers principles meant that it could not order a monetary refund for Starr even if it felt that the US competent authority improperly applied that the anti-treaty shopping provisions of the US –Swiss tax treaty.

The 2016 opinion explored the  limits of the court’s powers. In so doing, it discusses how in treaties there is a consultation process between the contracting states following one state’s Competent Authority determination that a party is entitled to discretionary treaty relief.  That consultation is an executive branch function. As such, the 2016 opinion concludes that the courts are unable  to mandate that a party is entitled to receive a refund based on a claim that there was an improper application of a discretionary tax treaty provision. On that point, the opinion was clear:

To determine that Starr is entitled to a certain sum of benefits, the Court would be forced to step into the shoes of the IRS and its Swiss counterparts and effectively preordain the outcome of any consultation between the two. This a court may not do.

Yet in that 2016 development the district court concluded that Starr could bring a claim under the APA seeking to set aside the U.S Competent Authority’s determination and that if Starr “prevailed on that claim, [it] would be entitled … to have the matter remanded to the U.S. Competent Authority for further action” consistent with the Court’s opinion.

The 2016 opinion nicely summarizes how the APA provided jurisdiction over Starr’s dispute:

The APA makes “final agency action for which there is no other adequate remedy in a court … subject to judicial review.” 5 U.S.C. § 704. The government concedes that the Competent Authority’s decision constitutes final agency action. And if the Court were to hold that Starr could not challenge the decision through a claim brought under the tax-refund statute, then no other adequate remedy would exist, and review under the APA would be proper. Cf. Cohen v. United States, 650 F.3d 717, 736 [108 AFTR 2d 2011-5046] (D.C. Cir. 2011) (en banc) (directing the district court to consider the merits of an APA claim against the IRS when plaintiffs had “no other adequate remedy at law”).

In allowing Starr to bring a claim under the APA, the 2016 opinion acknowledged that Starr’s ultimate goal was a refund, not just an academic finding that the Competent Authority acted improperly. Yet, the opinion paved the way for the Starr Company to amend its complaint to bring the APA claims and suggested that such a finding might in fact lead to a refund (or at least a consultation about a refund):

As the Court has explained, however, monetary relief of any sort is unavailable to Starr without improper judicial intervention into the consultation process….

The Court declines to assume, however, that Starr would forgo an opportunity simply to have the Competent Authority’s decision set aside as arbitrary, capricious, or an abuse of discretion. Indeed, as the government recognizes, remanding to the agency for further consideration is the norm when a court sets aside an agency’s action. And this relief is not illusory. Regardless of whether the Court possesses the authority to order the IRS to engage in consultation, counsel for the IRS has represented—and the Court would fully expect—that the IRS would not decline to consult with the Swiss in the event that the Court found that the IRS abused its discretion and remanded to the IRS, and the IRS otherwise preliminarily determined that Starr qualified for treaty benefits. Hr’g Tr., ECF No. 34, 42:8-18. The Court thus will not deprive Starr of the opportunity to seek this form of relief under the APA. It will grant Starr leave to amend its complaint to bring such a claim.

Starr dutifully amended its complaint to include APA claims. This led the district court in an opinion earlier this month to apply the APA to the Competent Authority’s decision to deny treat benefits. In Part 2 of this post, I will discuss the court’s analysis as to why under the APA the Competent Authority did not act improperly in finding that the discretionary treaty benefits did not apply to reduce withholding on the AIG dividends.