TANNENWALD

George R. Abramowitz, Edward A. Scallet, and Diana E. Buckley, for petitioners. Albert L. Sandlin, Jr., Kim A. Palmerino, Frank M. McClanahan III, and S. Mark Barnes for respondent.

OPINION

TANNENWALD, Judge:

Respondent determined the following deficiencies in, and additions to, the Federal income taxes of The Taisei Fire and Marine Insurance Co., Ltd. (Taisei), The Nissan Fire and Marine Insurance Co., Ltd. (Nissan), The Fuji Fire and Marine Insurance Co., Ltd. (Fuji), and The Chiyoda Fire and Marine Insurance Co., Ltd. (Chiyoda):

+-----------------------------------------+ ¦ ¦ ¦ ¦Addition to Tax¦ +-------+----+------------+---------------¦ ¦ ¦Year¦Deficiency ¦Sec. 6661 ¦ +-------+----+------------+---------------¦ ¦ ¦ ¦ ¦ ¦ +-------+----+------------+---------------¦ ¦Taisei ¦1986¦$ 847.00 ¦0 ¦ +-------+----+------------+---------------¦ ¦ ¦1987¦295,134.00 ¦$ 73,784.00 ¦ +-------+----+------------+---------------¦ ¦ ¦1988¦2,363,924.00¦590,981.00 ¦ +-------+----+------------+---------------¦ ¦ ¦ ¦ ¦ ¦ +-------+----+------------+---------------¦ ¦Nissan ¦1987¦197,838.00 ¦49,460.00 ¦ +-------+----+------------+---------------¦ ¦ ¦1988¦2,272,534.00¦568,134.00 ¦ +-------+----+------------+---------------¦ ¦ ¦ ¦ ¦ ¦ +-------+----+------------+---------------¦ ¦Fuji ¦1986¦49,009.00 ¦12,252.00 ¦ +-------+----+------------+---------------¦ ¦ ¦1987¦256,173.00 ¦64,043.00 ¦ +-------+----+------------+---------------¦ ¦ ¦1988¦2,506,733.00¦626,683.00 ¦ +-------+----+------------+---------------¦ ¦ ¦ ¦ ¦ ¦ +-------+----+------------+---------------¦ ¦Chiyoda¦1986¦19,886.00 ¦4,972.00 ¦ +-------+----+------------+---------------¦ ¦ ¦1987¦662,135.00 ¦165,534.00 ¦ +-------+----+------------+---------------¦ ¦ ¦1988¦4,569,945.00¦1,142,486.00 ¦ +-----------------------------------------+

The principal issue in these consolidated cases is whether, during the years at issue, petitioners had a U.S. permanent establishment by virtue of the activities of a U.S. agent in accepting reinsurance on behalf of each petitioner. Depending on our resolution of this issue, there is a further issue concerning whether each petitioner's 1988 taxable income should include certain reductions in pre–1988 estimates of unpaid losses under excess of loss reinsurance contracts with non-U.S. insurers and reinsurers. Respondent has conceded that none of the petitioners are liable for the addition to tax under section 6661 for any of the years in issue.

FINDINGS OF FACT

Some of the facts have been stipulated and are so found. The stipulation of facts and accompanying exhibits are incorporated herein by reference. The facts found are those which, unless otherwise specified, existed during the years at issue.

Each of the petitioners is a Japanese property and casualty insurance company with its principal place of business in Japan. The stock of each petitioner is publicly traded on a Japanese exchange. There is no stock ownership relationship among petitioners.

The primary business of each petitioner is writing direct insurance in Japan. Each petitioner also assumes reinsurance ceded

to it by insurers and reinsurers, including U.S. insurers and reinsurers, through a reinsurance department located in Tokyo. Each petitioner obtains foreign reinsurance through foreign brokers that bring reinsurance proposals to it, and from foreign insurers and reinsurers with which each petitioner has a direct relationship.

Each petitioner has at least one representative office in the United States that provides information on the U.S. market to it and assists its clients in the United States, but which does not have authority to write any form of insurance. Taisei and Fuji do not have U.S. branches and do not have licenses to engage in the insurance business in the United States.

Chiyoda has a branch office in New York that has insurance licenses for California, Illinois, New Jersey, and New York. Nissan has a branch office in New York that has an insurance license for New York and California.

Fuji has a 100–percent owned U.S. subsidiary which holds an Illinois insurance license, and which participates in three insurance programs with a U.S. insurance company. The U.S. subsidiary retrocedes to Fuji, on a quota share basis, 50 percent of the business it receives under these programs.

In addition, each petitioner grants authority to two or three different U.S. agents, including Fortress Re, Inc., to underwrite reinsurance on its behalf and to perform certain activities in connection therewith.

Fortress Re, Inc. (hereinafter referred to as new Fortress or Fortress) is the successor to Fortress Reinsurance Managers, Inc., established in 1972 as a subsidiary of Penn General Agencies, Inc., which was owned in large part by Pennsylvania Life Company. Fortress Reinsurance Managers, Inc., acted as a reinsurance underwriting manager on behalf of various insurance companies with which it entered into management agreements. From its inception, the chief operating officer was Maurice D. Sabbah. In 1977, its name was changed to Fortress Re, Inc. (hereinafter referred to as old Fortress). In September 1979, certain of the assets and the name of old Fortress were sold to M.D. Sabbah and Company, a newly organized North Carolina corporation owned 66 2/3 percent by Mr. Sabbah and 33 1/3 percent by Kenneth Kornfeld. In October 1979, M.D. Sabbah and Company changed its name to Fortress Re, Inc. (new Fortress). New Fortress assumed the duties, but not the liabilities, of old Fortress with respect to treaty accounts underwritten by old Fortress and certain duties of old Fortress with respect to the facultative accounts underwritten by old Fortress. Since 1979, Mr. Sabbah has transferred some of his shares in Fortress to members of his family, so that Fortress is currently held as follows:

+----------------------------------------+ ¦Shareholder ¦Percent of Ownership¦ +-------------------+--------------------¦ ¦ ¦ ¦ +-------------------+--------------------¦ ¦Maurice D. Sabbah ¦33.14 ¦ +-------------------+--------------------¦ ¦Zmira Sabbah ¦23.62 ¦ +-------------------+--------------------¦ ¦Leeor B. Sabbah ¦9.91 ¦ +-------------------+--------------------¦ ¦Kenneth H. Kornfeld¦33.33 ¦ +----------------------------------------+

The directors of Fortress are Mr. Sabbah, his wife, and Mr. Kornfeld. Mr. Sabbah is the chairman of Fortress, and Mr. Kornfeld is the president and chief underwriter.

Mr. Sabbah handles contacts with insurance companies Fortress represents and has responsibility for reports provided to those companies, in addition to certain administrative responsibilities. Mr. Kornfeld's duties include underwriting the reinsurance entered into on behalf of the companies Fortress represents, establishing retrocession programs with respect to its reinsurance treaties, managing claims with respect to those treaties and managing the daily affairs of Fortress. Mr. Sabbah and Mr. Kornfeld have total control over the daily operations of Fortress, including the hiring and firing of employees and the assigning of responsibilities to them. Fortress has approximately 20 employees, whose duties include assisting underwriting, handling claims, data processing and computer operations, secretarial support, and accounting services.

Fortress maintains leased offices in Burlington, North Carolina, for which it pays the rent. Fortress purchases property and liability insurance in connection with its business. The operating costs of Fortress, including rent and salaries, are borne by Fortress.

Fortress is a reinsurance underwriting manager, which involves acting as an agent for insurance companies in underwriting and managing reinsurance on behalf of such companies. Fortress is not licensed to conduct insurance or reinsurance business in any jurisdiction. Fortress underwrites reinsurance and places retrocessions only on behalf of the companies with which it enters into management agreements. Fortress enters into reinsurance and retrocession contracts on behalf of the companies it represents only through brokers; Fortress itself does not act as a broker.

Fortress enters into a separate management agreement with each insurance company it represents. The agreements with petitioners are identical except for the net acceptance limit (see infra p. 10). Since its inception, Fortress has been involved in as many as 10 management agreements in a management year. A management year is defined as the annual period from July 1 to June 30. From inception through June 30, 1989, the management years have been designated as Years I through XVI, respectively.

Taisei and Chiyoda first entered into management agreements with old Fortress in November 1972. Nissan and Fuji first entered into management agreements with Fortress in May 1981. Nissan and Fuji had previously been quota share reinsurers with respect to the companies Fortress represented. In the agreements, Fortress is referred to as the “Manager”, and the insurance company is referred to as a “Member”. Each agreement authorizes Fortress, among other things, to act as agent of each company to underwrite and retrocede reinsurance on behalf of each company. Under the agreement, the liability of the member with respect to each reinsurance contract, underwritten by Fortress on the member's behalf, is several and not joint with any other member.

Under the agreement, it is contemplated that Fortress may enter into similar, or substantially similar, management agreements with other insurance or reinsurance companies or other insurers. Fortress does not need permission of, or even to consult, the companies with which it has agreements, before entering into a new agreement. Although in practice, when a member terminated a management agreement, Fortress offered to increase the participation of the companies it already represented, it was not obligated to do so. In 1988, another Japanese insurance company, Dai Tokyo, approached Fortress about entering into a management agreement, which would have had the effect of diluting the participation of petitioners. After declining to enter into this agreement, Fortress, at the insistence of Dai Tokyo, polled petitioners to determine and confirm whether petitioners would be receptive to Dai Tokyo's participation.

Each reinsurance contract underwritten by Fortress, or old Fortress, on behalf of the companies it represented, is assigned to a management year. All premiums and losses, including claims settled in later years, associated with a particular reinsurance contract are allocated to the management year to which the reinsurance contract was assigned. Fortress is responsible for the handling and disposition of all claims against the companies it represents. In many cases, claims relating to the reinsurance underwritten by Fortress on behalf of companies it represents are not fully settled for many years. Fortress has total control over the handling and disposition of claims on behalf of petitioners.

Pursuant to each agreement, Fortress regularly exercises the authority to conclude original reinsurance contracts and to cede reinsurance on behalf of each petitioner. Each agreement provides Fortress with underwriting authority on a continuous basis until the agreement is terminated. The agreements can be terminated by either party, but only with 6 months' notice, although in practice the notice period has been waived. After termination of an agreement, Fortress continues to have obligations with respect to reinsurance previously underwritten. During the years in issue, Fortress had continuing duties to 13 insurance companies, excluding petitioners, for contracts underwritten in prior management years.

The only material limitation on Fortress' authority under the agreement is a “net acceptance limit”, which is the maximum amount of net liability in respect of any one original reinsurance contract that Fortress can accept on behalf of a member. There is no gross acceptance limit in the agreements, so that Fortress can underwrite reinsurance contracts on behalf of a member that are greater than the net acceptance limit, provided that Fortress arranges for retrocessions of the excess over the net acceptance limit. In practice, Fortress sets its own gross acceptance limit, as to which it voluntarily advises petitioners. When approached by Chiyoda with regard to inserting a gross acceptance limit into its agreement, Fortress refused, and Chiyoda dropped its request.

Before each management year, Fortress provided each petitioner with an estimate of net premium income for the upcoming year, based on the gross and net participations of that petitioner. Net premium income equals the gross premiums received for all reinsurance contracts less retrocession premiums. Under the terms of the management agreement, Fortress is not limited on how many contracts it writes, only that no one contract can exceed the net acceptance limit, so in effect the total net premium income from reinsurance handled by Fortress is unlimited. When any of petitioners approached Fortress about lowering their net premium income, Fortress' advice was to increase the quota share cession to its affiliated quota share reinsurer, Carolina Reinsurance Limited (hereinafter referred to as Carolina Re). See infra pp. 18–19.

In 1986, Fortress anticipated a very favorable market and sought to increase its capacity. It did so by offering to increase the underwriting done for its existing members, and by soliciting four other Japanese insurance companies about entering into management agreements.

Each reinsurance contract underwritten by Fortress is executed with a “security stamp” that identifies each member and the percentage of the total liability assumed by each member under the contract. The percentage to be assumed by each member on each reinsurance contract entered into during the management year is determined before the start of the management year, after Fortress comes to terms with each member regarding the net acceptance limit for the year.

The percentage of each contract assumed is subject to that limit.

Mr. Kornfeld is the chief underwriter and, as such, decides what business Fortress will underwrite and retrocede on behalf of the members. The retrocession program for a management year was presented to each petitioner in advance, during an annual trip by Mr. Kornfeld to each petitioner's offices. However, Fortress does not need approval, and did not seek or receive input, from petitioners.

All original reinsurance contracts, as defined in the management agreements, are “excess of loss” contracts. The lines of reinsurance that were the subject of original reinsurance contracts and ceded reinsurance were aviation excess of loss, nonmarine catastrophic excess of loss (e.g., land-based risks such as hurricanes, tornadoes, earthquakes and fires), and marine excess of loss (e.g., water-based risks involving oil rigs and ocean liners). With excess of loss reinsurance, the reinsured pays a premium to a reinsurer for a layer of protection, whereby the reinsurer agrees to pay all losses above a certain amount (the retention), but only up to a certain limit. There may be several layers of protection where a different reinsurer “writes” each layer, the lowest layer being the first to bear a loss. Generally, Fortress wrote a percentage of a single layer of loss. In choosing which layer to write, Fortress tries to pick the first “true” catastrophic layer, i.e., a layer that would not bear ordinary losses but would be the first to bear a loss from a catastrophe or extraordinary loss. The rationale behind this strategy is that the reinsurer of lower layers receives a higher premium, and Fortress feels a true catastrophe would cause losses at all layers so that Fortress is getting a higher premium than reinsurers of higher layers, yet bears similar risk. In respect of each reinsurance contract, Fortress independently decides which layer it should write on behalf of petitioners.

As part of its retrocession program, Fortress cedes a part of the liability it writes, so that no reinsurance contract exposes its members to a direct liability greater than their net acceptance limit. In such a situation, each petitioner is liable in the event the reinsurers should default. The benefit of writing more reinsurance than it accepts on behalf of petitioners is that a commission is earned on the ceded reinsurance. Fortress transacted retrocessions through brokers, most of which are in London and the rest in New York.

The premiums and commissions were established by a lead underwriter of the reinsurance proposals and were followed by Fortress.

Petitioners utilize the services of Fortress because Fortress has good relationships with reinsurance brokers, has access to good business, and has a profitable business strategy.

Fortress has continued access to good business because it has a reputation for paying its claims promptly and, of immeasurable importance, because its clients are large insurance companies that represent good security. Brokers would not deal with Fortress if its clients were not as financially secure as petitioners. However, there are 21 insurance companies in Japan and hundreds in the world that meet the minimum capital requirements, and whom Fortress could have as clients and continue to obtain similar reinsurance business.

Fortress was compensated for its services pursuant to compensation schedules set forth in each agreement. During 1986 to 1988, Fortress' income was derived from management fees, contingent commissions, and override commissions payable under management agreements entered into for management years I through XVI. Also, Fortress earned investment income on its own funds, which was not related to the management agreements. Fortress' compensation structure is the same as other reinsurance underwriting managers, although its management fees are slightly lower and its profit commissions slightly higher than the norm.

Management fees are calculated as a fixed percentage of the gross earned premiums, with the percentage lower as the volume of the premiums increases. It was anticipated that the management fees would cover current operating expenses of Fortress, including salaries. The fees are established on the basis of experience rather than projections. For each of the years in issue, a portion of the management fees Fortress earned was related to reinsurance underwritten in prior management years.

Contingent profit commissions are based on the profitability of business underwritten. Each year, Fortress recomputes its contingent profit commission for the prior management years to reflect results as they unfold and to reflect revisions to the loss reserves with respect to reinsurance contracts assigned to that year.

Override commissions are amounts received on certain pro rata retrocessional treaties and are not a significant part of Fortress' compensation.

For 1986, 1987, and 1988, Fortress earned $1,819,152, $5,023,631, and $20,747,536, respectively, from management fees and profit commissions. Of these amounts, 74 percent, 78 percent, and 62 percent, respectively, were attributable to reinsurance underwritten in management years that ended before the particular calendar years.

Fortress has the authority, under the management agreements and its agreement with old Fortress, to control the investment of funds withheld pursuant to such agreements in its sole discretion. Distributions are made in accordance with the management agreements and are made over a period of years following the management year.

Fortress provides quarterly accounting reports to all insurance companies with which it or old Fortress has entered into management agreements. The quarterly reports contain a summary which reflects the results for the company to which the report is provided for the particular quarter for all management years for which Fortress underwrote reinsurance on behalf of such company. In addition, each report provides an individual summary of the results for each such management year.

Each quarterly report reflects estimates for unpaid losses, including reserves for incurred but not reported (IBNR) losses. The reserves are set by Fortress based on Mr. Sabbah's judgment, taking into account experience and information from the reinsureds.

Each report also contains a narrative summary of Fortress' overall underwriting results prepared by Fortress.

Each year Mr. Kornfeld separately visited the offices of each petitioner in Tokyo. At these meetings, he described the underwriting results for prior management years, as well as the retrocessional program. Communication during the year occurred via letter or telex. There were no communications by telephone. During the years in issue, a representative of each petitioner visited the offices of Fortress one or two times. There were also an additional few visits between representatives of Chiyoda and Fortress in connection with discussions of the sale of Fortress to Chiyoda. The other petitioners were not aware of these discussions between Fortress and Chiyoda.

Petitioners attend monthly industry meetings where each is present, but they have never met to discuss Fortress. With one exception relating to a profit commission, petitioners did not communicate with each other about their relationship with Fortress.

In 1984, the owners of Fortress caused the formation of Carolina Re, under the laws of Bermuda. In 1984, the stockholders of Carolina Re were as follows:

+-----------------------------+ ¦Stockholder ¦Shares ¦ +---------------------+-------¦ ¦ ¦ ¦ +---------------------+-------¦ ¦Fortress ¦120,993¦ +---------------------+-------¦ ¦Mr. Sabbah ¦1 ¦ +---------------------+-------¦ ¦Mr. Kornfeld ¦1 ¦ +---------------------+-------¦ ¦Zmira Sabbah ¦1 ¦ +---------------------+-------¦ ¦H.C. Butterfield ¦1 ¦ +---------------------+-------¦ ¦John A. Ellison ¦1 ¦ +---------------------+-------¦ ¦John Collis ¦1 ¦ +---------------------+-------¦ ¦Nicholas G. Trollope ¦1 ¦ +---------------------+-------¦ ¦Total ¦121,000¦ +-----------------------------+

The latter four shareholders were Bermuda residents and the shares were held as qualifying director's shares, and Fortress was the beneficial owner of each share. The original officers of Carolina Re were Mr. Sabbah, Mr. Kornfeld, Charles D. Edelman, Leon E. Nearon, and Alan L. Brown.

In 1986, Fortress sold its shares in Carolina Re for $1 each, as follows:

+--------------------+ ¦Purchaser ¦Shares¦ +-------------+------¦ ¦ ¦ ¦ +-------------+------¦ ¦Mr. Sabbah ¦46,095¦ +-------------+------¦ ¦Zmira Sabbah ¦34,572¦ +-------------+------¦ ¦Mr. Kornfeld ¦40,333¦ +--------------------+

Carolina Re acts as a quota share reinsurer of reinsurance underwritten by Fortress, meaning that it assumes a certain share of the reinsurance ceded by petitioners, for which it is paid a premium. Before forming Carolina Re, Fortress notified petitioners of its intentions although it did not need their approval.

Fortress requires that Carolina Re be retroceded a minimum percentage of the reinsurance contracts accepted on behalf of each petitioner, although each petitioner may increase the percentage. In 1988, at the insistence of Fortress, the minimum percentage ceded to Carolina Re was increased, despite the objections of three of four petitioners.

Fortress' income is subject to U.S. Federal income tax and is included on Fortress' Federal income tax returns. Each petitioner included the items of income and expense with respect to the reinsurance underwritten by Fortress on its behalf in its Japanese income tax returns. Each petitioner filed protective Federal income tax returns for the years in issue with the Internal Revenue Service Center at Philadelphia, Pennsylvania. After the issuance of the deficiency notices, each petitioner made payment of the tax and additions to tax asserted therein, together with accrued interest, for which each petitioner filed an amended petition asking for a determination of overpayment.

OPINION

Under the Convention Between the United States of America and Japan for the Avoidance of Double Taxation and the Prevention of Fiscal Evasion with Respect to Taxes on Income, Mar. 8, 1971, 23 U.S.T. (Part 1) 969 (hereinafter referred to as the U.S.–Japan Convention or Convention), the commercial profits of a Japanese resident are exempt from U.S. Federal income tax, unless such profits are attributable to a U.S. permanent establishment. Convention, art. 8(1). The relevant provisions of the Convention whereby a Japanese resident will be deemed to have a U.S. permanent establishment due to the activities of an agent are as follows:

(4) A person acting in a Contracting State on behalf of a resident of the other Contracting State, other than an agent of an independent status to whom paragraph (5) of this article applies, shall be deemed to be a permanent establishment in the first-mentioned Contracting State if such person has, and habitually exercises in the first-mentioned Contracting State, an authority to conclude contracts in the name of that resident, unless the exercise of such authority is limited to the purchase of goods or merchandise for that resident.

(5) A resident of a Contracting State shall not be deemed to have a permanent establishment in the other Contracting State merely because such resident engages in industrial or commercial activity in that other Contracting State through a broker, general commission agent, or any other agent of an independent status, where such broker or agent is acting in the ordinary course of his business. [Convention, art. 9.]

Initially, it is undisputed that Fortress had the authority, which it exercised, to conclude contracts on behalf of petitioners, so that unless Fortress is “a broker, general commission agent, or any other agent of an independent status” within the meaning of article 9(5), petitioners will be deemed to have U.S. permanent establishments. The parties are in agreement that Fortress was not a “broker” or “general commission agent”, and respondent concedes that Fortress was acting in the ordinary course of its business when acting on behalf of petitioners. Thus, the issue before us is whether, during the years at issue, Fortress was an “agent of an independent status” in respect of each petitioner. In this connection, we note that neither petitioner nor respondent has argued that any petitioner should be treated differently from any other petitioner in resolving this issue. Background

The U.S.–Japan Convention itself does not define an “agent of an independent status”. In applying a treaty definition, “Our role is limited to giving effect to the intent of the Treaty parties.” Sumitomo Shoji America, Inc. v. Avagliano, 457 U.S. 176, 185 (1982); see also Crow v. Commissioner, 85 T.C. 376, 380 (1985), quoting Maximov v. United States, 299 F.2d 565, 568 (2d Cir.1962) (“The goal of treaty interpretation is ‘to give the specific words * * * a meaning consistent with the genuine shared expectations of the contracting parties.’ ”), affd. 373 U.S. 49 (1963); Estate of Burghardt v. Commissioner, 80 T.C. 705, 708 (1983), affd. without published opinion 734 F.2d 3 (3d Cir.1984). Beyond the literal language, we must examine the Treaty's “purpose, history and context.” Crow v. Commissioner, 85 T.C. at 380.

Our examination shows that the relevant provisions of the Convention are not only based upon, but are duplicative of, article 5, comments 4 and 5, of the 1963 O.E.C.D. Draft [Model] Convention (hereinafter referred to as the 1963 Model).

See letter of transmittal from President Nixon to the Senate requesting ratification of the Convention, dated May 11, 1971, 2 Tax Treaties (CCH) par. 5222, p. 34,021–3; Senate Committee on Foreign Relations, Report on Tax Convention with Japan and Tax Protocol with France (Nov. 22, 1971), 1973–1 C.B. 642, 643. While the 1963 Model itself provides no more definition than the Convention, the model is explained in part by a commentary, which states in pertinent part:

15. Persons who may be deemed to be permanent establishments must be strictly limited to those who are dependent, both from the legal and economic points of view, upon the enterprise for which they carry on business dealings (Report of the Fiscal Committee of the League of Nations, 1928, page 12). Where an enterprise has business dealings with an independent agent, this cannot be held to mean that the enterprise itself carries on a business in the other State. In such a case, there are two separate enterprises.

* * *

19. Under paragraph 4 of the Article, only one category of dependent agents, who meet specific conditions, is deemed to be permanent establishments. All independent agents and the remaining dependent ones are not deemed to be permanent establishment. Mention should be made of the fact that the Mexico and London Drafts * * * and a number of Conventions, do not enumerate exhaustively such dependent agents as are deemed to be permanent establishments, but merely give examples. In the interest of preventing differences of interpretation and of furthering international economic relations, it appeared advisable to define, as exhaustively as possible, the cases where agents are deemed to be “permanent establishments”.

* * *

20. * * * In the Mexico and London Drafts and in the Conventions, brokers and commission agents are stated to be agents of an independent status. Similarly, business dealings carried on with the co-operation of any other independent person carrying on a trade or business (e.g. a forwarding agent) do not constitute a permanent establishment. Such independent agents must, however, be acting in the ordinary course of their business. * * *

* * *

The special problems which can arise in the case of insurance companies dealing by means of intermediaries or variously qualified representatives shall be further studied. [Commentary to art. 5 of the 1963 Model.]

Based on the above, petitioners argue that the test of independent status is one of both legal and economic dependence and that, if we find that Fortress was either legally or economically independent of petitioners, it will necessarily follow that Fortress was not a permanent establishment. Respondent argues that comment 15 erroneously phrased the standard in terms of “dependence” and the conjunctive “and” instead of the disjunctive “or”, thus allowing either legal or economic independence to satisfy the requirement for independent status. See Roberts, “The Agency Element of Permanent Establishment: The OECD Commentaries from the Civil Law View (Part Two),” Intertax 488, 490 (Oct.1993); Skaar, Permanent Establishment: Erosion of a Tax Treaty Principle, 506–507 (1991); Nitikman, “The Meaning of ‘Permanent Establishment’ in the 1981 U.S. Model Income Tax Treaty: Part 2,” 15 Intl.Tax J. 257, 267–268 (1989).

The basis for this argument is that comment 15 of the 1963 Model expressly refers to the Report of the Fiscal Committee of the League of Nations, 1928, p. 12, the commentary to which states “The words ‘bona-fide agent of independent status' are intended to imply absolute independence, both from the legal and economic points of view. ” (emphasis supplied);

see Roberts, supra at 490. Indeed, the commentary to the OECD model was changed in the 1977 revision so that both legal and economic independence is necessary.

Comment 36 to the OECD Revised Model Double Taxation Convention on Income and Capital—1977 (hereinafter referred to as the 1977 Model); see also comment 37 to art. 5 of the OECD Model Tax Convention on Income and On Capital (1992). Generally, we would have reservations about interpreting a convention, ratified in 1971, on the basis of a commentary, adopted in 1977, that contradicts the literal language of the commentary in effect at the time of ratification. However, in light of the extensive analysis by the previously cited commentators and the confirmation of such analysis by our own research, we are persuaded that the criteria in the later commentary reflects the original intention of the commentary to the 1963 Model and that the 1963 Model should be interpreted as having a disjunctive (“or”) meaning.

We note, however, that if we focus, as the parties have ultimately done, on the test for legal and economic independence set forth in comment 37 to article 5 of the 1977 Model, as applied to the facts herein, the issue of disjunctive versus conjunctive reading of the 1963 Model fades into the background. That comment provides:

37. Whether a person is independent of the enterprise represented depends on the extent of the obligations which this person has vis-a-vis the enterprise. Where the person's commercial activities for the enterprise are subject to detailed instructions or to comprehensive control by it, such person cannot be regarded as independent of the enterprise. Another important criterion will be whether the entrepreneurial risk has to be borne by the person or by the enterprise the person represents. * * * [Comment 37 to art. 5 of the 1977 Model.]

It is obvious that the tests of “comprehensive control” and “entrepreneurial risk”, as the determinants of legal and economic independence, involve an intensely factual inquiry, which does not lend itself to the articulation of a “definitive statement that would produce a talisman for the solution of concrete cases.” Commissioner v. Duberstein, 363 U.S. 278, 284–285 (1960).

Petitioners suggest that guidance can be found in the factors used in distinguishing employees from independent contractors. See Donroy, Ltd. v. United States, 301 F.2d 200, 205–206 (9th Cir.1962); Nitikman, supra at 266. We think the employee versus independent contractor analogy is of limited use. The fact that petitioners herein are clearly not employees (indeed, respondent does not contend that they are) and therefore would be considered independent contractors does not answer the question before us, namely, whether they are the kind of independent contractors who should be held to be “agent(s) of an independent status”. Nor are we prepared to accept respondent's argument that the quoted phrase should be given a narrow scope by virtue of the ejusdem generis rule in that it was intended to encompass only those agents who exhibited characteristics associated with a “broker” or “commission agent”. We think that the generality of the phrase “agent of an independent status” was intended to have an expansive rather than a confining scope, particularly since the words “broker” and “commission agent” themselves lack specificity. Respondent's reliance on Fleming (H.M. Inspector of Taxes) v. London Produce Co., 1 W.L.R. 1013, 2 All E.R. 975 (Ch.Div.1968), is misplaced. In that case the language, to which the doctrine of ejusdem generis was applied, was totally different (“In this subsection, ‘broker’ includes a general commission agent” (emphasis added)).

Against the foregoing background, we turn to the determination of Fortress' legal and economic independence. Legal Independence

The relationship between Fortress and petitioners is defined by the management agreement that Fortress entered into separately with each petitioner. Petitioners have no interest in Fortress, and no representative of any of the petitioners is a director, officer, or employee of Fortress.

The agreements grant complete discretion to Fortress to conduct the reinsurance business on behalf of petitioners. See Skaar, supra at 508.

Respondent agrees that Fortress had independence with respect to day-to-day operations, but then argues that its actions were restricted by gross acceptance limits and limits on net premium income. However, even if there were such restrictions, they would not necessarily constitute control. The gross acceptance limit and net premium income both relate to the total exposure of petitioners, and even an independent agent only has authority to perform specific duties for the principal. It is freedom in the manner by which the agent performs such duties that distinguishes him as independent.

In any event, the record is clear that the gross acceptance limits were set by Fortress as part of its strategy to limit risk through diversification. Fortress advised petitioners of the gross acceptance limits for informational purposes and changed the limits without the advice or consent of petitioners. Fortress refused to put gross acceptance limits in the management agreements in order to retain flexibility. Respondent implies that the limit forced Fortress to enter into many small contracts instead of being able to enter into a few large contracts, but the pattern is consistent with Fortress' strategy of limiting risk through diversification, a strategy which Fortress was clearly in a position to implement through a plethora of available contracts.

As to net premium income, there were no limits under the terms of the management agreements. If one of petitioners sought to lower its net premium income from U.S. sources, Fortress' advice was to cede a greater share to Carolina Re, operating in Bermuda. Petitioners could also terminate agreements with their other U.S. agents. Respondent places great weight on the estimates of net premium income Fortress provided to petitioners before each management year, but the estimates are clearly that, and nothing more, and were greatly exceeded for at least one of the years at issue. While Fortress was aware that at times petitioners wanted only to absorb a certain amount of net premium income, Fortress did not change its business to accommodate their concerns.

Respondent further argues there were restrictions on Fortress' corporate affairs not reflected in the agreements that gave petitioners comprehensive control of Fortress. As evidence, respondent relies on Fortress' consultations with petitioners in regard to the request of Dai Tokyo to become a client of Fortress, and to Fortress' intent to include Carolina Re in the reinsurance program. Respondent also points out that Fortress reported to petitioners more regularly than required by the agreements. However, these are actions of a company seeking to maintain good relations with longstanding clients, rather than one seeking approval. With respect to the Dai Tokyo and Carolina Re situations, Fortress had already made its decision before consulting with petitioners.

Lewenhaupt v. Commissioner, 20 T.C. 151, 162–163 (1953), affd. per curiam 221 F.2d 227 (9th Cir.1955), cited by respondent, not only involved a different test, i.e., whether the taxpayer was engaged in business in the United States through an agent, but involved continuous activity in managing U.S. real estate owned by the taxpayer which went beyond mere ownership or receipt of income. It is clearly distinguishable.

Respondent further argues that petitioners exercised “comprehensive control” over Fortress by acting as a “pool”. However, there is no evidence that petitioners acted in concert to control Fortress. In only rare and isolated instances did petitioners communicate with one another regarding Fortress. Further, there are references to a “pool” throughout the history of Fortress, which period covers relationships with 17 separate U.S. and Japanese insurance companies. The inferences respondent would have us draw from the fact that petitioners are all from Japan and that petitioners are among the participants in regular industry conferences in Japan are simply insufficient to establish the existence of control by a “pool”.

In a similar vein, we reject respondent's attempt to construct control from the fact that, during the years at issue, Fortress' activities were confined to the reinsurance it underwrote on behalf of petitioners. Pointing to article 2(2) of the U.S.–Japan Convention,

respondent attempts to support her position by drawing upon the phrase “other agent of independent status” in section 864(c)(5)(A) and the regulation thereunder, section 1.864–7(d)(3), Income Tax Regs. See S.Rept. 89–1707 (1966), 1966–2 C.B. 1055, 1072–1073. Obviously, the statute simply repeats the phrase used in the Convention. The regulations suggest two elements to be considered. The first is ownership or control, section 1.864–7(d)(3)(ii), Income Tax Regs., which the regulation specifically states is not determinative. The second, section 1.864–7(d)(3)(iii), Income Tax Regs., is whether the agent acts “exclusively, or almost exclusively, for one principal,” (emphasis added) in which event “the facts and circumstances of a particular case shall be taken into account in determining whether the agent, while acting in that capacity, may be classified as an independent agent.” Assuming without deciding that these regulations, implementing a particular statute, should be accorded interpretative effect in respect of a treaty provision, it has no application herein where we have concluded that Fortress acted separately in respect of each of four petitioners and where respondent concedes that Fortress was acting in the ordinary course of its business, a position that seems inconsistent with both the “pool” and “exclusively” concepts. Moreover, we note that the number of principals for whom Fortress acted varied over the years and that, even during the years before us, Fortress carried on a substantial amount of activity in handling claims, etc., for several other insurance companies.

Finally, we note that all four petitioners, while not their primary business, did have reinsurance departments. Thus, petitioners had the ability to give detailed instructions to Fortress, yet they did not.

As an agent, Fortress had complete discretion over the details of its work. As an entity, Fortress was subject to no external control. In sum, Fortress was legally independent of petitioners. Economic Independence

Fortress is owned solely by Mr. Sabbah and his family and Mr. Kornfeld. There was no guarantee of revenue to Fortress, nor was Fortress protected from loss in the event it had been unable to generate sufficient revenue. Fortress has management agreements with four separate clients, whereby any one of them can leave on 6 months' notice. If one of petitioners did end its relationship, Fortress would bear the burden of finding a replacement to subscribe to that client's share of reinsurance contracts.

Respondent argues that Fortress bore no entrepreneurial risk because its operating expenses were covered by a management fee, and because it was guaranteed business due to the creditworthiness of the reinsurers on whose behalf it acted, petitioners.

While the management agreements provided that Fortress earned a percentage of the gross premiums written which effectively covered Fortress' operating expenses, this did not mean that Fortress bore no risk. Fortress had to acquire sufficient business to produce the gross premiums. Further, it appears that this provision of the agreements is normal for an underwriting manager. That respondent's argument on this point misses the mark is illustrated, for example, by a large mutual fund that charges an annual management fee to cover operating expenses. Clearly, the mutual fund company would not be considered dependent on its thousands of investors. Under these circumstances, even with as few as four investors, Fortress cannot be considered dependent on petitioners to pay its operating expenses.

Nor do we agree with respondent's argument that Fortress is able to secure profitable reinsurance contracts only because its clients are petitioners. Although Fortress needs clients with a certain minimum capital to conduct its business, any of hundreds of other insurance companies worldwide would be adequate substitutes. Also, it cannot be denied that Fortress had access to the reinsurance contracts it considered good, in part because of Fortress' relationships and reputation in the industry. In fact, it appears that Fortress' access to profitable reinsurance contracts, as well as its experience and ability to choose profitable reinsurance contracts, attracted petitioners to Fortress, and would attract other insurance companies if Fortress needed another client to take a share of the contracts.

Finally, we think that the amount of Fortress' profits is significant. See Madole, “Agents as Permanent Establishments Under U.S. Income Tax Treaties,” 23 Tax Mgmt.Intl. 281, 293–294 (1994). For the 3 years in issue, Fortress was paid over $27 million. This is not the kind of sum paid to a subservient company. In addition, petitioners were in effect forced to share reinsurance profits with Carolina Re, an entity owned by the same people who owned Fortress, by permitting Fortress to cede reinsurance to Carolina Re even though Carolina Re was not as well known or financially secure as other potential quota share reinsurers. Conclusion

In sum, during the years at issue, Fortress was both legally and economically independent of petitioners, thus satisfying the definition of an agent of an independent status under article 9 of the U.S.–Japan Convention.

Two further items deserve comment. First, petitioners point to a decision of the Federal Republic of Germany Tax Court at Bremen, FG I 4/73, 10 EFG 467049 (1973), affd. Decision of the FRG Bundesfinanzhof, BFH IR 152/73, BStBl II (24) 626–629 (1975), regarding the application of the independent agent provision of the Germany–Netherlands Treaty to a German insurance agent. A Dutch company had engaged a German firm as its representative and principal agent and signed a standard form granting power of attorney to the German firm enabling the German firm to conclude contracts on behalf of the Dutch company. The German firm acted as an independent insurance agent for numerous domestic and foreign insurance companies. The court held the Dutch company did not have a permanent establishment in Germany by virtue of the performance of the German insurance agent. While the result reached in the German case is consistent with that which we reach herein, we think that its utility herein is limited by the clearly distinguishable facts. Nor does de Amodio v. Commissioner, 34 T.C. 894 (1960), affd. on other grounds 299 F.2d 623 (3d Cir.1962), provide petitioners with any sustenance. There, a resident of Switzerland owned U.S. rental property which was managed and operated through local real estate agents. We determined that the agents fell within the term “broker” or “independent agent”, in the income tax convention between the United States and Switzerland, but the discussion is limited.

Second, we note that, in the commentary to the O.E.C.D.'s 1977 Model, it is stated that an insurance company could do “large-scale business in a State without being taxed in that State on their profits arising from such business.” Comment 38 to art. 5 of 1977 Model; see also comment 21 to art. 5 of 1963 Model. The commentary goes on to suggest that contracting states may want to contemplate that an insurance company will be “deemed to have a permanent establishment in the other State if they collect premiums in that other State through an agent established there”, other than a dependent agent. Comment 38 to art. 5 of 1977 Model. However, the commentary notes that such a provision is not in the model and its inclusion should depend upon the factual and legal situation involved. Comment 38 to art. 5 of 1977 Model; see Avery Jones & Ward, “Agents as Permanent Establishments under the OECD Model Tax Convention,” European Taxation 154, 170–171 (1993).

The Convention between the United States of America and the Kingdom of Belgium for the Avoidance of Double Taxation and the Prevention of Fiscal Evasion with Respect to Taxes on Income, July 9, 1970, 23 U.S.T. 2687 (hereinafter referred to as the U.S.–Belgium Convention), does include such an insurance provision. It provides that the independent agent provision “shall not apply with respect to a broker or agent acting on behalf of an insurance company if such broker or agent has, and habitually exercises, an authority to conclude contracts in the name of that company.” U.S.–Belgium Convention, art. 5(6). Finally, we note that it was decided not to include reinsurance within the coverage of this provision. Technical Explanation by Treasury Department on the Convention Between the U.S. and Belgium for the Avoidance of Double Taxation and the Prevention of Fiscal Evasion with Respect to Taxes on Income, signed July 9, 1970, 1 Tax Treaties (CCH) par. 1350, p. 17,032. From the foregoing it appears that the resolution of the issue of the existence of an agent of independent status in the insurance arena turns, at least in part, upon the presence of a specific treaty provision. See Dept. of the Treasury, Report to Congress on the Taxation of Income Earned by Members of Insurance or Reinsurance Syndicates 45 (Feb.1989).

Given the absence of any provision dealing with insurance or reinsurance in the U.S.–Japan Convention, our holding herein that Fortress is not a permanent establishment of petitioners is consistent with the approach suggested by the O.E.C.D. Model

and the application thereof in the U.S.–Belgium Convention.

In view of our holding, we do not reach the further issue in respect of the inclusion in petitioners' 1988 taxable income of certain reductions in pre–1988 estimates of unpaid losses.

In order to permit the determination of the amounts of overpayments by petitioners in respect of the deficiencies determined by respondent,

Decisions will be entered under Rule 155.

+---------------+ ¦ ¦Percent¦ +-------+-------¦ ¦Chiyoda¦44.445 ¦ +-------+-------¦ ¦Taisei ¦22.222 ¦ +-------+-------¦ ¦Fuji ¦22.222 ¦ +-------+-------¦ ¦Nissan ¦11.111 ¦ +-------+-------¦ ¦TOTAL ¦100.000¦ +---------------+ U.S.Tax Ct.,1995.Taisei Fire and Marine Ins. Co., Ltd. v. C.I.R.104 T.C. No. 27, 104 T.C. 535, Tax Ct. Rep. (CCH) 50,620, Tax Ct. Rep. Dec. (RIA) 104.27

+---------------+ ¦ ¦Percent¦ +-------+-------¦ ¦Chiyoda¦40 ¦ +-------+-------¦ ¦Taisei ¦20 ¦ +-------+-------¦ ¦Fuji ¦20 ¦ +-------+-------¦ ¦Nissan ¦20 ¦ +-------+-------¦ ¦TOTAL ¦100 ¦ +---------------+