STARR, Circuit Judge:

This case brings us once more into the legal battle between the banking and securities industries over the former's movement into business domains which the latter considers its own. The Investment Company Institute is a national association of investment companies, investment advisors and underwriters. The Institute brought an action for declaratory and injunctive relief in the District Court, challenging a decision by the Comptroller of the Currency allowing Citibank to establish and market a "Collective Investment Trust" for assets of Individual Retirement Accounts, commonly known as "IRAs." The Institute contended that Citibank's Trust is functionally equivalent to a mutual fund, and that Citibank could not operate the Trust without running afoul of the 1933 Glass-Steagall Act. That Act was designed to preserve public confidence in commercial banks like Citibank by keeping them out of the securities business.

The District Court granted the Comptroller's motion for summary judgment, Investment Co. Institute v. Conover, 596 F.Supp. 1496 (D.D.C. 1984), and the Institute appealed. The sole issue before us is whether "units of beneficial interest," or shares, in Citibank's Trust constitute "securities" within the meaning of sections 16 and 21 of the Glass-Steagall Act, 12 U.S.C. §§ 24 (Seventh), 378(a)(1) (1982) ("the Act"). The Comptroller concluded that the units are not "securities" for purposes of the Act, and that Citibank could therefore lawfully offer the units to the public. The District Court agreed with the Comptroller, thereby disagreeing with the then only other judicial decision on this issue, Investment Co. Institute v. Conover, 593 F.Supp. 846 (N.D.Cal. 1984) (Schwarzer, J.), appeal pending, Nos. 84-2622/2623 (9th Cir.). Guided by the deference principles enunciated in Chevron U.S.A., Inc. v. Natural Resources Defense Council, Inc., 467 U.S. 837, 104 S.Ct. 2778, 81 L.Ed.2d 694 (1984), we uphold the Comptroller's interpretation of the Act and affirm.

I

An understanding of the specific dispute before us requires familiarity with a portion of the laborate federal statutory scheme governing national banking activities and the newer provision of federal law permitting IRAs, an acquaintance with the structure of Citibank's Trust, and an appreciation of the procedural history of this case.

A

Enacted in the shadows of the 1929 stock market crash and the ensuing banking collapse, the Glass-Steagall Act prohibits national banks from engaging in the securities business. Section 16 of the Act forbids national banks to underwrite or deal in "securities or stock." 12 U.S.C. § 24 (Seventh). Section 21 in turn prohibits anyone in the business of underwriting, selling, or distributing "stocks, bonds, debentures, notes or other securities" from taking deposits. 12 U.S.C. § 378(a)(1). The term "securities," however, is left undefined by the Act. The leading Supreme Court case construing that term, Investment Co. Institute v. Camp, 401 U.S. 617, 91 S.Ct. 1091, 28 L.Ed.2d 367 (1971), overturned an earlier ruling by the Comptroller allowing Citibank to operate a "collective investment fund." Camp held, in brief, that the "units of participation" in Citibank's fund constituted "securities" within the meaning of the Act because, in the Court's view, contributions to Citibank's fund were made for "investment" purposes rather than true fiduciary purposes, and because Citibank's arrangement would likely pose several of the hazards to the banking system that Glass-Steagall was designed to eliminate. See 401 U.S. at 630-38, 91 S.Ct. at 1098-1102.

Also of relevance in this case is the Employee Retirement Income Security Act (ERISA), enacted by Congress at 1974. Among other things, ERISA added a provision to the Internal Revenue Code permitting qualified individuals to deduct investments in IRAs from taxable income up to a specified amount, currently $2,000 annually. 26 U.S.C. § 408(a)(1) (1982). Such contributions, together with earnings on IRA assets, are not taxed until distribution or withdrawal, id. § 408(d)(1), but a ten percent tax penalty is imposed if the account holder withdraws the funds before reaching the age of 59 1/2, id. § 408(f)(1). Section 408(a)(2), moreover, requires specific approval by the Secretary of the Treasury before a bank or other institution may serve as trustee for IRA trust assets. These assets may only be commingled in a common investment fund or trust fund. Id. § 408(a)(5).

B

In 1982, Citibank established a common trust fund for IRA investors. A common trust fund is a device, long known to the banking industry, by which a bank commingles the assets of individual trusts and invests those assets collectively. The assets of each participating trust are exchanged for an undivided ownership interest in the common trust fund, with the ownership interest being proportionate to the assets contributed by each trust to the fund's total assets. This device obviously permits each participating trust to enjoy greater diversification and wider investment opportunities than would otherwise exist. It also permits banks, acting as trustees, to manage trusts that are too small to be managed individually. National banks have operated such funds since at least 1936, when the Federal Reserve Board specifically authorized their establishment. See 1 Fed.Reg. 417, 420 (1936) (amending Regulation F).

Styled a "Collective Investment Trust," the Citibank trust fund at issue here consists of four separate investment portfolios into which an individual may instruct Citibank to place his or her IRA trust assets. This is accomplished by means of a written trust agreement, which is freely revocable save for the ten percent penalty imposed by I.R.C. § 408. An individual's ownership interest in the entire trust fund is expressed in terms of "units of beneficial interest," which are available only to Citibank IRA-holders and are non-transferable. IRA assets may, however, be freely transferred among the four portfolios, as well as among any of Citibank's other IRA programs. Citibank serves as trustee and investment advisor to the trust fund and receives a monthly fee based in part on the net value of the fund's portfolios. In order to comply with SEC requirements, however, Citibank has placed the fund under the ultimate control of an independent five-member "supervisory committee," which essentially serves as a board of directors. Citibank has marketed its fund as an alternative to, or a competitor of, similar programs now being offered by mutual funds.

C

Citibank first sought approval of its Collective Investment Trust from the Securities and Exchange Commission, which has in the past opined that ownership interests in collective trust funds are "securities" for purposes of the federal securities laws. Although not conceding the correctness of the SEC's interpretation, Citibank registered the Trust as an investment company under the Investment Company Act of 1940, and registered the units of beneficial interest as securities under the 1933 Securities Act.

Citibank then sought approval of the Trust from the Comptroller of the Currency, who has statutory responsibility for regulating fiduciary activities of national banks. 12 U.S.C. § 92a (1982). Citibank asked the Comptroller to determine that its operation of the Trust did not violate the Glass-Steagall Act; it also asked for exemptions from those regulations that Citibank feared might conflict with requirements imposed by the SEC pursuant to the Investment Company Act.

On October 28, 1982, the Comptroller issued a final ruling approving the Citibank Trust. J.A. at 174. The Comptroller found Citibank's operation of the Trust to be authorized by applicable trust regulations and granted the requested exemptions. Importantly for present purposes, the Comptroller provided a lengthy analysis of the applicability of the Glass-Steagall Act to the Trust, concluding that "there is no Glass-Steagall Act impediment to a bank collectively investing IRA trust assets." Id. at 180.

Much of the Comptroller's discussion of the Glass-Steagall Act centered on the Supreme Court's 1971 decision in Camp. After noting certain "general similarities" between the present Trust and the Citibank fund at issue in Camp, the Comptroller offered two reasons why the units of beneficial interest in the Trust should not be considered "securities" for purposes of Glass-Steagall. First, unlike the arrangement in Camp, Citibank receives the IRA assets in trust rather than in a "managing agent" capacity. The Comptroller observed that, as Camp had recognized, national banks have long enjoyed authority to commingle funds received in trust. See id. at 179. In light of this long established authority, the Comptroller concluded that "when, as in the present case, the `securities' merely represent the formal manifestation of a traditional banking service, . . . the prohibitions of the Glass-Steagall Act are not applicable." Id. at 180. According to the Comptroller, moreover, "the meaning of the term securities under the securities law [is] not necessarily synonymous with its meaning under the Glass-Steagall Act." Id.

Second, the Comptroller concluded that Citibank's operation of the Trust entails none of the "hazards and potential abuses" identified by the Supreme Court in Camp. The Comptroller explained in detail his reasons for so believing, concluding that "because the potential for abuse is non-existent or negligible here, and because this is essentially a traditional banking service, it would be inappropriate to conclude that there is any security here for purposes of the Glass-Steagall Act." Id. at 182.

The Institute then brought this action for declaratory and injunctive relief. Granting the Comptroller's cross-motion for summary judgment, the District Court held that the Comptroller's decision was entitled to "great weight," and that the Comptroller had acted reasonably in interpreting Glass-Steagall as not prohibiting Citibank's Trust. It is from this adverse decision that the Institute appeals.

The Institute's principal argument is that Citibank's Trust is flatly prohibited by Camp. We address this contention first. After concluding that the Trust is not forbidden by Camp's teaching, we analyze the Comptroller's ruling under the governing legal framework as set forth in Chevron. The latter discussion also address the Institute's two remaining contentions, namely that ERISA does not authorize Citibank's operation of its Trust and that the District Court improperly deferred to the Comptroller's statutory interpretation.

II

Despite our ultimate determination that Chevron's deference principles govern, we first examine the Institute's vigorously pressed contention that Camp itself disposes of this case, requiring invalidation of Citibank's latter-day efforts to compete with the mutual fund industry. As we understand its argument, the Institute asserts two independent grounds for this position. First, Camp makes it unlawful for national banks to operate a mutual fund, which, the Institute contends, is exactly what the Trust amounts to. Institute Brief at 1-2; Reply Brief at 1. Second, under Camp, the term "security" includes any "undivided and redeemable interest in the assets of a common trust fund." Institute Brief at 19; Reply Brief at 3.

A

The strength of the Institute's first argument lies in its intuitive appeal, for Camp indeed overturned the Comptroller's earlier attempt to authorize Citibank's operation of what amounted to a mutual fund. Now Citibank — again with the Comptroller's approval — is attempting to offer a service that, in the eyes of many customers and, apparently, Citibank's marketing and advertising departments, looks very much like a mutual fund. To call Citibank's fund anything but a mutual fund, according to the Institute, is to exalt form over substance.

In our view, however, Camp cannot fairly be read to prohibit any financial service that some actors in the marketplace view as functionally similar to a mutual fund. Camp nowhere states that all such services are prohibited. To be sure, the Court found significant the fact that the fund at issue there was "in direct competition with the mutual fund industry." 401 U.S. at 625, 91 S.Ct. at 1096. But the mere existence of such competition was not the decisive element in the Court's analysis.

More fundamentally, Citibank's Trust is, upon close analysis, quite distinct from the fund at issue in Camp. As the Comptroller found, the principal difference is that Citibank serves as trustee of its IRA funds rather than as managing agent, the situation before the Court in Camp. Although the Institute devotes considerable energy to arguing why this distinction is irrelevant, see, e.g., Institute Brief at 11-12; Reply Brief at 3, Camp itself suggests otherwise. After expressly observing that "the investment fund plan as established does not provide that the bank receives the investor's money in trust," 401 U.S. at 622 n. 8, 91 S.Ct. at 1094 n. 8, the Court stated:

For at least a generation . . . there has been no reason to doubt that a national bank can, consistently with the banking laws, commingle trust funds on the one hand, and act as managing agent on the other. No provision of the banking law suggests that it is improper for a national bank to pool trust assets, or to act as a managing agent for individual customers, or to purchase stock for the account of its customers. But the union of these powers gives birth to an investment fund whose activities are of a different character.

401 U.S. at 624-25, 91 S.Ct. at 1096. According to the Court, then, it was the Combination of the bank's commingling of trust assets and its acting as managing agent (rather than trustee) that spawned Glass-Steagall concerns. Here, by contrast, Citibank is seeking to exercise only one of the two powers which Camp held could not be joined: the power to commingle trust funds. The existence of a trust relationship is sufficient, by itself, to take this case out of Camp's express teaching.

But even if the investment fund at issue in Camp had employed a trust device, our case would still not be on all fours. Camp evinces the Court's concern with the risks to which a national bank might subject itself in offering a new, unconventional product. In the Court's analysis, whether such a product is a "security" depends to some extent on whether, from the bank's perspective, the product entails the relatively small risks presumably characteristic of commercial banking or, as in Camp, the relatively large risks (either real or apparent) usually associated with investment banking. See 401 U.S. at 629-33, 636-38, 91 S.Ct. at 1098-1100, 1101-02. This concern also appears to underlie the Camp Court's attempt to determine whether a financial service represents a "fiduciary" service or a run-of-the-mill "investment." See 401 U.S. at 621, 638, 91 S.Ct. at 1094, 1102. The Court in Camp was constrained to undertake an analysis of economic effects since, as discussed more fully below, the Comptroller in that case had not bothered to shed any light whatever as to Glass-Steagall's meaning. See 401 U.S. at 626-27, 91 S.Ct. at 1097. In order to determine whether Camp controls this case, we are obliged to examine some of the economic effects of Citibank's Trust.

We see four reasons why, from a banking perspective, the Citibank Trust is likely to be less risky than the investment fund at issue in Camp. First, customers of the Camp investment fund were permitted to invest between $10,000 and $500,000. 401 U.S. at 622, 91 S.Ct. at 1094. Here, by contrast, ERISA imposes a limit of $2,000 per person per year. 26 U.S.C. § 408(a)(1). This limitation, it seems to us (as it seemed to the Comptroller), makes it less likely that Citibank would be tempted to engage in unsound lending practices in order to attract or retain particularly valued customers. This was one of the "subtle hazards" identified in Camp. See 401 U.S. at 632, 637-38, 91 S.Ct. at 1099, 1102.

Second, Citibank's units may not be used as collateral, unlike the investment fund shares in Camp. This, again, is assured by ERISA. See 26 U.S.C. § 408(e)(4). One would expect this limitation to reduce Citibank's risk by preventing it from selling its units "on the margin," that is, financing a portion of the purchase price with a loan secured by the shares. Margin selling was, of course, one of the lending abuses widely thought to have contributed to the unhappy events of 1929.

Third, unlike the investment fund shares in Camp, Citibank's units are non-transferable. See, e.g., J.A. at 50 (Citibank prospectus). This restriction naturally impedes formation of an independent "market" in the units, and thereby allows Citibank more control over the aggregate size of the Trust. For example, if Citibank decides that its Trust is becoming too risky, or is perceived by the public as too risky, the bank can gradually scale down the program by the simple expedient of diminishing its marketing efforts and retiring any units presented for redemption. By contrast, if the units were transferable, owners wishing to trade their units for cash could do so by selling them to a third party rather than redeeming them at Citibank.

Finally, the ten percent income tax penalty imposed by ERISA for early redemption would seem to reduce the risk that Citibank will be called upon to redeem large numbers of units at any one time. This was another concern animating the Court in Camp. The Court feared that, if confronted with such circumstances, "a bank might find itself under pressure to rescue the fund through measures inconsistent with sound banking." 401 U.S. at 637, 91 S.Ct. at 1102.

For all these reasons, we believe the Trust at issue here is distinguishable in significant respects from the managing-agent fund in Camp.

B

The Institute's second argument under Camp is less plausible. The Institute cites that case for the proposition that the term "security" necessarily includes any "undivided and redeemable interest in the assets of a common trust fund." Institute Brief at 1-2; Reply Brief at 3. Camp will be searched in vain for any statement of this sweeping proposition. The Institute has arrived at this understanding by splicing together unrelated phrases from different parts of the Camp opinion. This effort, albeit creative, is unavailing.

The Institute's argument, moreover, proves far too much. If the term "security" were indeed interpreted as broadly as the Institute suggests, then any ownership interest in a bank's common trust fund would constitute a security, and all commingling of trust funds by national banks would be effectively prohibited. We are at a loss to determine how a national bank could pool the assets of various trusts without creating some kind of certificates or other instruments setting forth the ownership interests of each of the participating trusts in the common trust assets. But such commingling, as we have seen, has historically been permitted and is clearly sanctioned by Camp. See 401 U.S. at 624-25, 91 S.Ct. at 1095-96.

For the foregoing reasons, we conclude that Camp does not dictate the result demanded by the Institute here. This conclusion does not, of course, mean that the decision is in any wise to be ignored. To the contrary, as the discussion which follows demonstrates, Camp informs our analysis of whether the Comptroller's interpretation of Glass-Steagall can be upheld under the principles enunciated in Chevron. We turn now to that analysis.

III

Over the years, the Supreme Court has frequently stated that courts should defer to reasonable interpretations of statutes by the administrative agencies charged with their administration. A generation ago in the landmark case of NLRB v. Hearst Publications, Inc., 322 U.S. 111, 64 S.Ct. 851, 88 L.Ed. 1170 (1944), the Court held that "where the question is one of specific application of a broad statutory term in a proceeding in which the agency administering the statute must determine it initially, the reviewing court's function is limited." Id. 322 U.S. at 131, 64 S.Ct. at 860. The Court observed that the agency's interpretation should be accepted as long as it has "`warrant in the record' and a reasonable basis in law." Id. In another leading case, Red Lion Broadcasting Company Co. v. FCC, 395 U.S. 367, 89 S.Ct. 1794, 23 L.Ed.2d 371 (1969), the Court stated that an agency's interpretation "should be followed unless there are compelling indications that it is wrong," id. at 381, 89 S.Ct. at 1802.

Notably, the Camp majority reaffirmed this principle. The Court's declination to defer to the Comptroller signaled no departure from this settled principle, but was occasioned instead by the Comptroller's complete silence as to the meaning of the governing statute. That is to say, there was in Camp no explicit interpretation of the Glass-Steagall Act to which the Court could appropriately defer. The Court's comments in this respect are highly instructive:

It is settled that courts should give great weight to any reasonable construction of a regulatory statute adopted by the agency charged with the enforcement of that statute. The Comptroller of the Currency is charged with the enforcement of the banking laws to an extent that warrants the invocation of this principle with respect to his deliberative conclusions as to the meaning of these laws. . . .

The difficulty here is that the Comptroller adopted no expressly articulated position at the administrative level as to the meaning and impact of the provisions of §§ 16 and 21 [of the Glass-Steagall Act] as they affect bank investment funds. The Comptroller promulgated Regulation 9 without opinion or accompanying statement.

401 U.S. at 626-27, 91 S.Ct. at 1097 (citation omitted). This statement suggests a fundamental difference between this case and Camp. Here, in the administrative proceedings, the Comptroller provided a thorough, explicit interpretation of Glass-Steagall as it applies to this case. We are therefore duty bound, as Camp teaches us, to give "great weight" to that interpretation.

Subsequent to Camp, the Supreme Court has provided more specific guidance on the application of deference principles. The most important decision for our purposes is Chevron, which set forth a now familiar framework for analyzing interpretations of statutes by agencies charged with their administration. First, the court must determine whether Congress "has directly spoken to the precise question at issue." Id. 104 S.Ct. at 2781-82. This inquiry entails an examination of the statute and, if necessary, resort to the pertinent legislative history. If Congressional intent is unclear after this interpretative exercise, then the reviewing court is called upon to determine whether the agency's interpretation is "permissible," that is to say reasonable. Id. at 2783. If it is, then the agency's interpretation must stand.

A

In applying the first prong of the Chevron analysis, the inquiry before us is whether the framers of Glass-Steagall intended to prohibit national banks from engaging in activities like that at issue here. Although both the Comptroller and Citibank vigorously urge us to consider ERISA's legislative history, that history is not relevant here for two reasons: first, in enacting ERISA, Congress in no wise purported to amend Glass-Steagall; second, nothing in ERISA or its legislative history suggests that Congress as a whole had any view, let alone a clear intent, on the applicability of Glass-Steagall to the operation of IRA collective trusts by national banks. Any inferences that might be drawn from ERISA or its legislative history "cannot substitute for a clear expression of legislative intent" by the framers of the Glass-Steagall Act itself. See Southeastern Community College v. Davis, 442 U.S. 397, 411 n. 11, 99 S.Ct. 2361, 2370 n. 11, 60 L.Ed.2d 980 (1979).

1

In applying the first part of the Chevron analysis, we must first look to the language of the statute itself. Inasmuch as the term "security" is not defined in the Act, we must "`start with the assumption that the legislative purpose is expressed by the ordinary meaning of the words used.'" Securities Industry Ass'n v. Board of Governors, 468 U.S. 137, ___, 104 S.Ct. 2979, 2986, 82 L.Ed.2d 107 (1984) ( Becker) (quoting Richards v. United States, 369 U.S. 1, 9, 82 S.Ct. 585, 591, 7 L.Ed.2d 492 (1962)). The Institute urges us simply to borrow the definition used in the federal securities laws. Under this view, any instrument registered as a security with the SEC would be considered a security for Glass-Steagall purposes.

The Institute relies heavily upon the Supreme Court decision in Becker, supra, for this argument. There, the issue was whether the term "security" includes commercial paper, which all parties to that litigation agreed is simply an unsecured promissory note. The Court held that commercial paper falls within the plain meaning of the term "note," and since the term "security" explicitly includes all notes, commercial paper is therefore a "security." See 104 S.Ct. at 2986-88. The Institute interprets Becker as holding that SEC registration itself constitutes "considerable evidence" that the instruments at issue are "securities" within the meaning of Glass-Steagall. Institute Brief at 20.

Fairly read, however, Becker does not stand for this proposition. To be sure, the Court found that Congress' explicit inclusion of commercial paper as a "security" for purposes of the securities laws was an indication that Congress considered commercial paper to be a "security" for Glass-Steagall purposes as well. But the Court did not indicate that SEC registration by itself bears on the classification of a financial instrument as a "security" within the meaning of the banking statute. Nor did Becker, as the Institute suggests, hold as a general matter that the term "security" itself has an ascertainable ordinary meaning in all contexts. As already noted, the majority opinion in Becker relied primarily on the plain meaning of the term "note," not on the plain meaning of the term "security."

As we see it, the statutory language is of limited help in divining Congressional intent in this case. The term "security" clearly encompasses instruments like stocks, bonds, notes, and mutual fund shares, which have traditionally been underwritten and marketed by members of the securities industry. As Camp teaches, moreover, the term may include other financial interests that are functionally equivalent to these traditional kinds of securities. But, as we have also seen, the term cannot be interpreted to include all financial interests in assets or in a business venture, lest traditional common trust funds be rendered invalid. Although interests in Citibank's Trust share some of the features of traditional mutual fund shares — for example, they permit diversification of risk and can be widely marketed — they are, as we have seen, quite different from traditional mutual fund shares in several important respects. Obviously, interests in IRA trust funds did not exist in 1933 when Congress passed Glass-Steagall; the term Congress employed to identify the financial instruments it had in mind, moreover, is not sufficiently precise as to clearly encompass such interests.

2

Unfortunately, the legislative history is also of limited usefulness in illuminating Congress' intent. The Institute has uncovered only two relevant aspects of Glass-Steagall's legislative history, both of which have been discussed in prior Supreme Court decisions.

The first is that, as noted in Camp, the Glass-Steagall Act was in part the product of Congressional investigations into the activities of bank security affiliates, including their operation of investment trusts. See 401 U.S. at 635, 91 S.Ct. at 1101. The Court reasoned that, "[s]ince Congress generally intended to divorce commercial banking from the kinds of activities in which bank security affiliates engaged, there is reason to believe that Congress explicitly intended to prohibit a national bank from operating an investment trust." Id. (footnote omitted).

Camp did not, however, hold that the term "security" includes interests in any kind of fund consisting of pooled assets. It held, rather, that Congress intended to prohibit the kind of investment fund at issue there, which the Court believed was nearly identical to the investment funds operated by bank security affiliates in the pre-Depression era. IRA trusts, with all the limitations imposed by ERISA, did not exist, of course, in 1933. More to the point, as we have already discussed, Citibank's Trust is much less like a traditional mutual fund than was the investment fund at issue in Camp. We are not persuaded that Congress' concern with the activities of banks' securities affiliates would extend to the kinds of ERISA-related activities at issue here.

The second potentially relevant aspect of legislative history is the fact, noted in Becker, that in drafting the Glass-Steagall Act Congress rejected a regulatory approach in favor of a set of "flat prohibitions," thereby evincing a desire to "separat[e] as completely as possible commercial from investment banking." 104 S.Ct. at 2985. The Institute seems to interpret this as an indication that national banks should be prohibited from intruding onto what the securities industry considers its own turf. The Institute reasons that since many mutual funds have been offering their own IRA trusts, see Institute Brief at 10 n. 16, prohibiting Citibank's entry into the field would be in keeping with Congress' apparent desire to keep commercial banking separated as much as possible from investment banking.

The problem with this argument, however, is that it is not entirely clear who is intruding onto whose turf. It is true, as we have already observed, that Citibank's fund represents something of a departure from the traditional role of a commercial bank in that it allows small account holders to benefit from the diversification that was previously available only to relatively large customers of the bank's trust department. But operation of an IRA trust by a mutual fund, which necessarily puts the mutual fund in the position of serving as trustee of its customers' money, also would seem to represent something of an intrusion into the traditional domain of commercial banking. Citibank's Trust, then, like the IRA trust offered by mutual funds, falls somewhere between the traditional activities of commercial and investment banking.

B

Having concluded that Congress has not "spoken to the precise question" before us, we must determine whether the Comptroller's interpretation of the Glass-Steagall Act is reasonable. Although the Supreme Court has provided limited specific guidance on how a court is to go about making such a determination, we can discern in Chevron and other Supreme Court decisions some general principles to guide us.

The overriding principle is that as long as Congress has no clearly discernible intent on the point in question, it is the agency which is vested with primary responsibility for interpreting the statute. Chevron teaches that Congress may delegate interpretative authority implicitly — by failing to legislate in sufficient detail as to resolve a particular question of interpretation — or explicitly — by expressly leaving a gap for the agency to fill. See 104 S.Ct. at 2782. In either case, interpretative authority has been delegated to the agency, not to the court.

This broad principle has two corollaries. First, an agency interpretation may be reasonable even if it is one which the court would not have rendered had it been interpreting the statute in the first instance. 104 S.Ct. at 2782 n. 11; Chemical Manufacturers, supra, 105 S.Ct. at 1108. Obviously, the court must not simply "rubber stamp" the agency interpretation or transform deference into "judicial inertia." See Bureau of Alcohol, Tobacco, and Firearms v. FLRA, supra, 464 U.S. at 96-97, 104 S.Ct. at 443-44. But the reviewing court may not simply substitute its own reading for that of the agency.

Second, an agency interpretation may not be branded as unreasonable simply because the court doubts the wisdom of the policy choice embodied in the agency's interpretation. Chevron, 104 S.Ct. at 2793. It is true, as the Institute points out, that a court must reject administrative interpretations that "frustrate the policy Congress sought to implement." Becker, 104 S.Ct. at 2983 (quoting FEC v. Democratic Senatorial Campaign Committee, 454 U.S. 27, 32, 102 S.Ct. 38, 42, 70 L.Ed.2d 23 (1981)). But to be true to Chevron, a court can lawfully overturn an agency interpretation on this ground only if the Congressional policy is sufficiently clear, and the agency's interpretation so inconsistent with that policy, that the court can conclude that Congress' clear intent has been violated. Absent a violation of clear Congressional intent, it is for the agency, not the reviewing court, "to make such policy choices — resolving the competing interests which Congress itself either inadvertently did not resolve or intentionally left to be resolved." Chevron, 104 S.Ct. at 2793.

C

With these principles in mind, we turn in the final phase of our analysis to an examination of the Institute's attack on the reasonableness of the Comptroller's interpretation. We observe first that the Comptroller, in determining whether ownership interests in Citibank's Trust constitute "securities" for Glass-Steagall purposes, employed the analytical framework fashioned by the Supreme Court in Camp. The Comptroller first analyzed the transaction to determine whether Citibank would be receiving funds for a true fiduciary purpose; he then examined the transaction in light of each of the Congressional concerns identified in Camp. See Ruling at 8-11. On the basis of that analysis, the Comptroller concluded that the units of participation are not "securities" and that the Act's prohibitions are therefore inapplicable. Since the Comptroller's analytical framework was the very one employed in Camp, we find that framework, needless to say, eminently reasonable.

The Institute, however, challenges the Comptroller's application of this framework. It first contends that the Comptroller erroneously found Citibank's Trust to have a bona fide fiduciary purpose. Second, according to the Institute, the Comptroller erroneously concluded either that the Trust did not raise the Congressional concerns identified in Camp or, alternatively, that he could adequately address those concerns through the exercise of his regulatory powers.

1

The Institute advances two distinct reasons why it believes Citibank's Trust represents an investment rather than a fiduciary service. First, according to the Institute, the Comptroller granted an exemption to his own "fiduciary regulations," allowing Citibank to vest management of the fund in an independent supervisory committee. According to those regulations, a common trust fund must be under the "exclusive management" of the sponsoring bank. See 12 CFR § 9.18(12) (1984).

To be sure, the existence of a non-bank supervisory committee renders the Trust at issue here more akin to a traditional mutual fund. Yet the regulations at issue merely prescribe standards or requirements for collective investments by national banks; they do not purport to define when a particular service will be deemed "fiduciary" in nature. More fundamentally, vesting control over the Trust in an independent supervisory committee does not dilute the legal effects of the bank's fiduciary obligations to its IRA customers. This arrangement might make it more difficult for Citibank to discharge its trusteeship obligations, and Citibank's delegation of control might conceivably be alleged under some circumstances in this litigious age to constitute a breach of its fiduciary obligations. But the relationship between the bank, as trustee, and the IRA-holder, as beneficiary of the trust, remains intact. This feature, we are satisfied, does not take the Citibank Fund out of the domain of "fiduciary" services and propel it into the mutual fund business.

Second, the Institute points out that Citibank is not merely commingling assets of pre-existing trusts, but is actively marketing its Trust to the public as an investment. According to the Institute's interpretation of a 1940 Federal Reserve Board regulation cited in Camp, this aggressive marketing campaign disqualifies the Trust as a bona fide fiduciary service. See 26 Fed.Reserve Bull. 393, 394 (1940).

We agree with the Institute that, in light of Camp's concern with "promotional pressures," see 401 U.S. at 630-38, 91 S.Ct. at 1098-1102, Citibank's marketing strategy might provide one indication that its Trust falls on the wrong side of the line between investments and bona fide fiduciary services. We cannot agree, however, that this consideration is dispositive in this case. The passages of the 1940 regulation in which the Institute relies were not themselves quoted or cited in Camp, and we have discerned no indication that the Supreme Court intended to elevate this regulation to the status of binding precedent. Quite apart from the regulation itself, the Comptroller carefully considered each of the promotional pressures that Citibank's marketing strategy might arguably create, and concluded that in this case those pressures are not sufficiently strong to warrant concern. See J.A. at 138-84. This evaluation seems to us eminently reasonable.

Finally, we do not believe that Citibank's characterization of the Trust as an "investment opportunity" should bear on whether the Trust constitutes a bona fide fiduciary service. The two terms are not mutually exclusive. Any fiduciary service is also an investment opportunity if by that one means an opportunity to earn a return on one's money; very few customers would likely be satisfied with the services of a bank's trust department if the bank did no more than safekeep their funds. We agree with the Comptroller that the proper inquiry is whether the bank is offering a genuine fiduciary service in addition to an opportunity to earn a return. Banks can appropriately provide more than vaults and safety deposit boxes and still remain faithful to the demands of federal law.

In sum, we cannot say that the Comptroller unreasonably determined that Citibank's Trust represents a bona fide fiduciary service. Not only is the Comptroller much better situated than we are to make such determinations, but his thorough justification of his conclusion is quite reasonable.

2

The Institute contests the Comptroller's treatment of the potential economic hazards of Citibank's Trust on a number of grounds. It first advances an argument which it successfully asserted in the California litigation, see 593 F.Supp. at 856, namely that the Comptroller's findings exhibit a disagreement with Congress and the Court in Camp about the potential dangers of entry by commercial banks into the investment banking business. See Institute Brief at 21-22. In our view, this is not a fair reading of the Comptroller's opinion. Rather, that opinion reflects a belief that this particular arrangement does not raise the concerns identified in Camp, and that the arrangement is therefore not the kind of activity Congress intended to proscribe through Glass-Steagall's prohibitions.

The Institute next argues that the Comptroller simply dismissed all of the potential dangers identified in Camp based upon a belief that his general regulatory power under 12 U.S.C. § 92a and other statutes would be sufficient to curb any abuses that might arise. Institute Brief at 22. Upon analysis, this is likewise a mischaracterization of the Comptroller's ruling. The Comptroller nowhere even mentions 12 U.S.C. § 92a, and does not rely upon any general regulatory power in minimizing the importance of the dangers identified in Camp. Rather, the Comptroller points to specific, existing regulations that would obviate some of those dangers. For example, one regulation relied upon by the Comptroller, 12 C.F.R. § 9.18(b)(8)(i)(1982), prevents Citibank from lending money to the Trust itself or to any individual participant. Fairly read, the Comptroller's decision does not run afoul of Becker's teaching that a bank regulator may not rely upon its general regulatory powers in order to circumvent the "broad prohibition" of the Glass-Steagall Act. See 104 S.Ct. at 2988.

The Institute's remaining arguments reflect, at bottom, disagreement with the Comptroller's judgment about the potential economic effects of Citibank's Trust on Citibank and on the banking industry. We concede the apparent plausibility of the Institute's view. On the other hand, the Comptroller has advanced plausible reasons for his view that such concerns are misplaced in this particular instance given the entire set of circumstances before him. Any assessment of the dangers foreseen by the Institute would involve judgments about complex economic phenomena related to the banking industry. Congress has charged the Comptroller with making those judgments in the first instance. We may overturn the Comptroller's judgments only if they are unreasonable. We are unable to conclude that they are.

Affirmed.