273 F3d 1135 Saba Partnership v. Commissioner of Internal Revenue

273 F.3d 1135 (D.C. Cir. 2001)


No. 00-1328, 00-1385


Argued October 2, 2001
Decided December 21, 2001

Appeals from the United States Tax Court (No. IRS-1470-97; 1471-97).

Thomas C. Durham argued the cause for appellants/crossappellees. With him on the briefs was Joel V. Williamson.

Richard Farber, Attorney, U.S. Department of Justice, argued the cause for appellee/cross-appellant. With him on the briefs was Edward T. Perelmuter, Attorney. Stuart L. Brown, Attorney, Internal Revenue Service, entered an appearance.

Before: Edwards, Rogers and Tatel, Circuit Judges.

Opinion for the Court filed by Circuit Judge Tatel.

Tatel, Circuit Judge:

view counter

Through an elaborate scheme involving partnerships with a foreign bank operating in a tax-free jurisdiction, a diversified U.S. company generated over $190 million worth of tax losses while incurring an actual loss of only $5 million. The Tax Court found that because certain of the partnerships' transactions lacked economic substance, they created no gains or losses for federal tax purposes. At the same time, the Tax Court declined to address the government's alternative contention that both partnerships were shams for federal tax purposes. The partnerships, together with the company, now appeal, and the government crossappeals. We vacate and remand to the Tax Court for reconsideration in light of our recent decision in ASA Investerings Partnership v. Commissioner, 201 F.3d 505 (D.C. Cir. 2000), where we invalidated what appears to be a similar--perhaps even identical--tax shelter on the grounds that the entire partnership, not merely the specific transactions at issue, was a sham for federal tax purposes.



This case involves the legality of a tax shelter marketed by Merrill Lynch to a small number of U.S. corporations. Designed for corporations anticipating large capital gains, the shelter takes advantage of certain Internal Revenue Code provisions and related Treasury Department regulations that govern installment sales where the taxpayer lacks advance knowledge of the installment payments' value. See 26 I.R.C. § 453; Temp. Treas. Reg. § 15A.453-1(c)(3)(i) (1984). To build such a shelter, the Merrill Lynch client forms a partnership with a foreign corporation operating in a tax-free jurisdiction. This partnership then buys and immediately sells a debt instrument on an installment basis. Although the transaction is basically a wash, generating hardly any economic gain or loss, Merrill Lynch's lawyers' interpretation of the relevant provisions allows the partnership to claim a massive tax gain, which is allocated to the foreign partner, and a massive tax loss, which the U.S. corporation keeps for itself. A detailed description of this shelter and the code provisions on which it depends appears in ASA, 201 F.3d at 506-8. In that case, we affirmed a Tax Court determination that another Merrill Lynch client that had adopted the shelter, AlliedSignal, had "not entered into a bona fide partnership" for federal tax purposes. Id. at 515.


The facts of this case appear similar to those of ASA. In 1990, appellant Brunswick Corporation, a diversified manufacturer, decided to divest itself of certain business groups. Because the sales would generate massive capital gains, Merrill Lynch proposed that Brunswick generate compensatory paper losses by forming a partnership with a foreign bank. In an extensive memorandum, Judith P. Zelisko, an attorney and Brunswick's Director of Taxes, laid out step by step how Merrill Lynch's proposal would "generate sufficient capital losses to offset the capital gain which w[ould] be generated on the sale of [certain divisions]." Saba P'ship v. Comm'r, 78 T.C.M. (CCH) 684, 689. Because of this document's significance, we quote it in substantial part:Step 1:


BC [Brunswick] and an unrelated foreign partner [FP] would form a Partnership no later than March 1, 1990 with BC contributing $20 million in cash and the FP contributing $180 million in cash. The Partnership would have a fiscal year-end of March 31st since that would be the year-end of the FP, the majority Partner.

Step 2:


Partnership buys a private placement note for $200 million with the cash in the Partnership and holds the note for one month.

Step 3:


Before March 31, 1990, the Partnership would sell the $200 million private placement note for $160 million in cash and five-year contingent note with an assumed fair market value (fmv) of $40 million. Under this contingent note, payments would be made to the Partnership over a five-year period equal to [a variable interest rate] times a fixed notional principal....


The Partnership would recognize gain on the sale of the private placement note calculated as follows:

view counter

Cash                                            160.0
Basis                                            33.3
(1/6 of 200) 
Gain                                            126.7
BC's Gain                                        12.67
FP's Gain                                       114.03

Total Gain                                      126.70

BC's share of the gain equals its 10% ownership in the Partnership for a taxable gain to BC of $12.67 million in 1990.

Step 4:


In April 1990 or later, (i.e., until there has been some movement in the value of the contingent note) BC buys 50% of FP's interest in the Partnership for $90 million, assuming that the fmv of the contingent note is still $40 million. With this purchase, BC's basis in its Partnership interest is $122.67 million calculated as follows:


BC's initial investment                         $20.0 million 
Gain 						 12.67 
Purchase of 50% of FP's 
interest                                         90.00

Step 5:


The Partnership distributes the contingent note to BC assuming a fmv of $40 million. In addition, the Partnership would distribute approximately $32.72 million in cash to FP which is the equivalent cash distribution to FP given its percentage ownership.

Step 6:


BC sells the contingent note for cash. This sale of the contingent note by BC generates the capital loss.


BC's basis in the note                         $122.67
FMV of the note                                  40.00
Capital loss                                     82.67
Net Gain on sale of FP note                      12.67

Net Capital loss                                 70.00

After the sale of the note, BC's tax basis in the Partnership is zero and the Partnership still has 127.28 in cash (160-32.72).

Step 7:


In April 1991, the Partnership will be terminated ...


Id. at 689-90. The Zelisko memorandum also notes that Merrill Lynch would earn a fee of "5-10% of the tax savings"; that the fee "would not be due if the tax law changed prior to implementation"; that "[l]egal fees for BC and operating expenses of the Partnership ... would be paid by BC"; and that the foreign partner would earn "40-75 basis points on the FP's equity investment." Id. Finally--and ironically-the memorandum reminds Zelisko's superiors that "[t]here cannot have been any agreements, negotiations, or understandings of any kind among the Partners or their representatives regarding the possible liquidation of the Partnership or the assets to be distributed to each respective Partner upon termination and liquidation of the Partnership or the transactions described in Steps 4 and 5." Id.


To execute the scheme, Merrill Lynch enlisted the same Dutch bank, Algemene Bank Netherlands, N.V. (ABN), that had served as the foreign partner in the shelter at issue in ASA. See ASA, 201 F.3d at 508. Merrill Lynch drafted a "credit proposal" for ABN that, like the Zelisko memorandum, outlined the partnership's investment steps, including (1) the purchase of highly rated private placement notes (PPNs); (2) the sale of the PPNs for cash and contingent notes; and (3) Brunswick's gradual buy-down of ABN's partnership interest. Id. at 692. In a separate memorandum to ABN, Merrill Lynch confirmed that "ABN will receive ... an upfront fee [of] around $600,000." Id. Previously, Merrill Lynch had assured ABN that in these types of deals, it would face "virtually no credit risk [since] the paper invested in [would] be of the highest credit quality and [would] have short term maturities," and that interest rate risk would be eliminated by a series of "perfect hedges." "Legal and tax risk," Merrill Lynch assured ABN, "will be covered by opinions of legal and tax counsel." Id.


Because Brunswick ultimately sold more assets than originally anticipated, it implemented the scheme outlined above on two separate occasions using two separate partnerships: Saba Partnership and Otrabanda Investerings Partnership.

Saba Partnership


On February 26, 1990, Brunswick contributed $20 million, and ABN contributed $180 million to the newly formed Saba Partnership (Zelisko memorandum, Step 1). Id. at 693. Saba immediately--in fact, the very same day--bought $200 million worth of 5-year PPNs (Step 2). Merrill Lynch then began to negotiate the sale of the notes and, on March 6, transmitted a summary of terms to two potential buyers. On March 23, just prior to the close of Saba's first taxable year, Saba sold the PPNs, worth $200 million, for an immediate cash payment of $160 million and four indefinite debt instruments, known as LIBOR (London Interbank Offering Rate) notes, worth approximately $38.5 million (Step 3). Id. at 695. Saba could have reduced its $1.5 million loss by selling the PPNs to a money market fund and then purchasing LIBOR notes, but because such funds cannot issue LIBOR notes, Saba eliminated them from consideration. Id. Saba could also have eliminated the loss by investing in LIBOR notes directly instead of first purchasing, then immediately selling, PPNs. Id. at 696.


Because LIBOR varies, Saba could not determine with certainty the aggregate selling price of the PPNs and thus reported the sale as an installment sale. Assuming it would receive payments over a period of six years (year of sale plus five years of LIBOR payments), Saba calculated its annual basis at $33,333,333, yielding a massive tax gain of $126,666,667, ninety percent of which it allocated to ABN, the foreign, tax-free "partner." Id. Pleased with the result, Brunswick's Vice-President of Finance, William R. McManaman, prepared a "Foreign Partnership Tax Update" predicting that Brunswick would ultimately realize capital losses of $80 million from the Saba deal and nearly $60 million in its second, yet-to-be formed partnership, Otrabanda. Id. at 696-97.


In July 1990, Brunswick purchased 50% of ABN's interest, giving Brunswick a 55% stake (Step 4). Id. at 697. Around the same time, it entered into a "consulting" agreement with ABN under which Brunswick paid ABN $750,000. In August, Saba distributed the three LIBOR notes to Brunswick and cash to ABN (Step 5). Id. In determining the amount of cash owed to ABN, Saba valued the LIBOR notes in such a way as to eliminate ABN's portion of the $1.5 million loss from the sale of the PPNs. In addition, Brunswick added a $535,000 "fee." Id. Brunswick then sold the three LIBOR notes at a slight discount for $26 million (Step 6). Brunswick calculated the tax basis for all four notes as $166,666,667 ($200 million minus the already used basis of $33,333,333). Since it had only received three of the four LIBOR notes, Brunswick multiplied this number by 3/4 to obtain its actual basis of $125 million, which it then reduced to $123 million for reasons not here relevant. Id. at 697-98. Using this basis, Brunswick calculated its paper loss at $84 million (basis in the three notes minus the notes' sale price minus its small share of the paper gain reported by Saba from the sale of the PPNs), even though it actually lost probably no more than $2.5 million. Id. at 698.


In September, Brunswick purchased an additional partnership share from ABN, bringing its total share to approximately 90%. Id. In June, Brunswick dissolved Saba and, after a series of complicated transactions not here relevant, ended up with the remaining LIBOR note, which it sold (Step 7). Id. at 698-99. After computing its tax basis as described above, Brunswick recorded a tax loss of $32 million, even though it actually lost only about $700,000. Id. at 700. Throughout all these transactions, Brunswick protected itself from LIBOR volatility by a series of hedges. Id. at 701.

Otrabanda Investerings Partnership


In June 1990, Brunswick began discussing the possibility of forming another partnership. Once again, Merrill Lynch, acting as Brunswick's agent, sent ABN a memorandum promising "total remuneration [of] $600,000" to ABN and outlining a specific "calendar" for the transaction, beginning with the purchase of PPNs and ending with a proposed termination of the partnership in July 1991. Id. at 701-02. With the confidence of one whose "[l]egal ... risk" is "covered by opinions of ... tax counsel," id. at 692, ABN's Risk Management Department approved the transaction solely on the condition that ABN reserved the right to liquidate the portfolio if its interest in the partnership was not reduced according to the proposed schedule, id. at 702.


On June 25, ABN and Brunswick contributed $135 million and $15 million respectively to the newly formed Otrabanda Investerings Partnership. Id. at 703. Four days later, Otrabanda bought four certificates of deposit for $100 million, selling them three weeks later for $80 million in cash and four LIBOR notes, which it then hedged. Id. at 703-04. Again, had Otrabanda invested directly in the LIBOR notes, it would have avoided the $750,000 loss attributable to the CDs' lack of liquidity. Id.


The remaining steps are familiar: The use of LIBOR notes allowed the transaction to be reported as a six year "installment sale," yielding an annual basis of $16,666,666 and concomitant paper gain of $63,333,333, the vast majority of which Saba allocated to ABN. Id. at 705. Following Brunswick's purchase of 50% of ABN's interest, Otrabanda distributed the LIBOR notes to Brunswick and an amount of cash to ABN equal to the fair value of the LIBOR notes plus ABN's share of the loss associated with the CDs' sale. Id. Brunswick sold the notes for only about $17.5 million. Because it calculated its basis as $83,333,333, the sale yielded a tax loss of approximately $60 million (tax basis minus sales price of the notes and Brunswick's share of the original tax gain on the sale of the CDs). In actuality, Brunswick lost only about $1.7 million--the difference between the value of the LIBOR notes and their sale price. Id. at 705-06. In December, Brunswick bought a further portion of Otrabanda bringing its total interest to approximately 90%, amended the agreement to give it total control over Otrabanda and paid ABN a $645,000 "control premium." Id. at 706. In June 1991, one month ahead of schedule, Brunswick terminated Otrabanda. Id.


Based on the Saba and Otrabanda transactions, Brunswick claimed tax losses of nearly $195 million even though it actually lost only about $5 million. These massive paper losses piqued the interest of the Commissioner, who proposed various adjustments to the partnerships' returns based on two alternative theories: that the installment sale transactions should be disregarded for federal tax purposes because they had no rational economic motivation except tax avoidance--that is, they lacked "economic substance"; or that Brunswick formed Saba and Otrabanda solely for tax reasons and therefore both partnerships, in their entireties, should be disregarded as "shams" for federal tax purposes. Id. at 710.


Brunswick, as Saba and Otrabanda's "tax matters partner," filed a petition for readjustment in the Tax Court. Because our opinion in ASA had not yet been released, the Tax Court followed the Third Circuit's opinion in ACM Partnership v. Commissioner, 157 F.3d 231 (3d Cir. 1998), which struck down a virtually identical Merrill Lynch tax shelter on the grounds that the installment sale transactions lacked economic substance. Mirroring the Third Circuit's reasoning, the Tax Court held the installment sales in this case to be "economic shams." Saba P'ship v. Comm'r, 78 T.C.M. (CCH) 684, 722.


Brunswick and the partnerships now appeal, arguing that the transactions in fact had economic substance. The Commissioner, relying on ASA, cross-appeals the Tax Court's refusal to find Saba and Otrabanda invalid partnerships.



All parties agree that the sham transaction and sham partnership approaches yield different results. See Appellee's Br. at 27 n.12 ("alternative theor[y] would produce adjustments to the returns ... not identical to the adjustments resulting from the Tax Court's decision"); Appellant's Reply Br. at 43 ("The Commissioner is asking this Court to reach a different result from the Tax Court's judgment.") Although the Commissioner seemed to concede at oral argument that under either approach, Brunswick could deduct actual losses from the transactions, we assume (without deciding) that different adjustments would result from the two approaches. Thus, we may not simply affirm on an alternative ground. Cf. EEOC v. Aramark Corp., Inc., 208 F.3d 266, 268 (D.C. Cir. 2000) ("[B]ecause we review the district court's judgment, not its reasoning, we may affirm on any ground properly raised."). At a minimum, we would simply remand to the Tax Court for reconsideration in light of ASA.


Urging us to go further, the Commissioner argues that we should apply ASA, find the partnerships to be shams, and remand for the limited purpose of making any necessary adjustments. In response, Brunswick argues that the question of whether "an entity should be regarded as a partnership for federal tax purposes is inherently factual." Appellant's Reply Br. at 38. Although we agree that whether a valid partnership exists for tax purposes is a fact-intensive inquiry, it is a determination that we may make in cases where the Tax Court has "ma[d]e adequate fact findings." Transp. Mfg. & Equip. Co. v. Comm'r, 374 F.2d 173, 177 (8th Cir. 1967) (cited in Appellant's Reply Br. at 39).


According to Brunswick, the Tax Court's findings are inadequate because of "significant differences" between the actions of Brunswick in this case and those of Allied Signal in ASA. Appellant's Reply Br. at 44. Though we agree that the differences Brunswick points to could be significant if they existed, we have our doubts that they do. Brunswick claims that it "did not promise ABN a 'specified return,' " but the record demonstrates that Brunswick, through its agent Merrill Lynch, promised ABN $600,000 for participating in Saba and the same for participating in Otrabanda. See supra at 1137-38, 1139. Brunswick claims that it "did not agree to pay Saba's and Otrabanda's expenses," but the Zelisko memorandum makes clear that Brunswick understood that it would pay "operating expenses." See supra at 1137. Brunswick claims that "ABN understood it would share in Saba's and Otrabanda's losses," but record evidence demonstrates that Saba and Otrabanda eliminated ABN's share of the loss occasioned by the sale of the PPNs and CDs through the valuation of the LIBOR notes. See supra at 1138, 1139. Brunswick claims that "there were no agreements between the parties contrary to their written [partnership] agreement" like the "Bermuda Agreement" in ASA, but the record demonstrates not only that Merrill Lynch, Brunswick's agent, sent written proposals to ABN outlining the future actions of Saba and Otrabanda, but that ABN approved these proposals. See supra at 1137-38, 1139. As far as we can tell, the only difference between this case and ASA is that Brunswick and ABN did not meet in Bermuda.


In any case, ASA makes clear that "the absence of a nontax business purpose is fatal" to the argument that the Commissioner should respect an entity for federal tax purposes. ASA, 201 F.3d at 512.


Here, the Tax Court specifically found "overwhelming evidence in the record that Saba and Otrabanda were organized solely to generate tax benefits for Brunswick." Saba P'ship, 78 T.C.M. (CCH) at 718 (emphasis added). Arguably, this broader finding subsumes any factual differences that might exist between this case and ASA.


In the end, however, we will not, as the Commissioner urges, affirm on the basis of ASA. Although the present record might strongly suggest that Saba and Otrabanda were sham partnerships organized for the sole purpose of generating paper tax losses for Brunswick, fairness dictates that we ought not affirm on this ground. In particular, in presenting its case in the Tax Court, Brunswick may have acted on the mistaken belief that the Supreme Court's decision in Moline Properties, Inc. v. Commissioner, 319 U.S. 436 (1943), established a two-part test under which Saba and Otrabanda must be respected simply because they engaged in some business activity, an interpretation that ASA squarely rejected, see ASA, 201 F.3d at 512 ("[C]courts have understood the 'business activity' reference in Moline to exclude activity whose sole purpose is tax avoidance.... Thus, what the petitioner alleges to be a two-pronged inquiry is, in fact, a unitary test ... under which the absence of a nontax business purpose is fatal.").


The Tax Court's judgment is vacated and the case remanded for reconsideration in light of ASA.

So ordered