dissenting:
I respectfully dissent. In my view, the court’s formalistic interpretation of ERISA’s section 303(b), 29 U.S.C. § 1083(b) (1982), unduly burdens the Internal Revenue Service in its treatment of proposed amendments and will do substantial harm to the workability of the statutory scheme crafted by Congress.1
I
The court’s interpretation of section 303(b)’s four-factor analysis is intuitively appealing, if only because it is straightforward. Based upon a perceived statutory command, the court requires IRS, in its decision to approve or disapprove proposed plan amendments, to engage (at minimum) in a boilerplate recitation of the impact of the four listed factors. In the court’s not entirely unpersuasive view, the plain meaning of “shall” makes further analysis unnecessary.
I readily concede that my colleagues’ reading of the statute is quite natural. But upon further analysis, especially of the statute’s structure and purposes, the court’s natural reading turns out to be the wrong reading. The Supreme Court has taught time and again that broader statutory purposes and the statute’s structure may overcome even the most seemingly clear statutory command. See Steelworkers v. Weber, 443 U.S. 193, 201, 99 S.Ct. 2721, 2726, 61 L.Ed.2d 480 (1979) (“It is a ‘familiar rule, that a thing may be within the letter of the statute and yet not within the statute, because not within its spirit, nor within the intention of its makers.’ ” (quoting Holy Trinity Church v. United States, 143 U.S. 467, 459, 12 S.Ct. 511, 512, 36 L.Ed. 226 (1892))); cf. Young v. Com*900munity Nutrition Institute, 476 U.S. 974, 106 S.Ct. 2360, 90 L.Ed.2d 959 (1986) (statutory command that agency “shall promulgate regulations” sufficiently ambiguous to allow agency to proceed by informal “action levels” rather than formal “regulations”). This is one instance in which, upon careful analysis, the structure and purpose of a statute speak more clearly than some of the words within it.
First, section 303(b)’s four enumerated factors are not even keyed to review of amendments to plans. The “substantial business hardship” test primarily governs waivers of minimum funding responsibilities for employers, see 29 U.S.C. § 1083(a), and was adopted only by reference to apply to amendments to plans. See 29 U.S.C. § 1082(c)(8) (1982). And the majority commendably concedes that applying section 303(b)’s enumerated factors to plan amendments makes no sense without “judicial construction.” Maj. op. at 891 n. 83. The language of section 303(b) clearly contemplates analysis focusing on individual employers. The first factor, section 303(b)(1), refers to whether “the employer is operating at an economic loss,” a consideration which is inapposite when the IRS is inquiring into the viability of a multi-employer plan. In order to make sense in the amendment context, the reference to “employer” must be read in the plural, thus turning section 303(b)(l)’s inquiry into something more akin to an industry-wide analysis. But this commonsense interpretation necessarily renders indistinct the four factors that the court requires IRS to examine. In the plan amendment context, the first three enumerated factors — all relating to the economic health of the industry — constitute premises for the ultimate conclusion, contained in the fourth factor, on which approval of the amendment properly hinges: whether “it is reasonable to expect that the plan will be continued only if the [amendment] is granted.” 29 U.S.C. § 1083(b)(4). As it should be, this fourth factor seems to be the ultimate consideration Congress had in mind. Separate consideration of each of the factors, it seems to me, would likely be repetitive and unnecessary in many instances.2
In addition to the difficulty in applying section 303(b)’s four-factor analysis to the amendment context, the structure of section 303(b) itself contemplates that the four factors will not necessarily be determinative. As the court candidly recognizes, Congress explicitly did not require that each of the four factors be present; what is more, Congress explicitly provided that IRS may consider whatever other factors it deems relevant. See 29 U.S.C. § 303(b) (stating that the factors to be considered include “but shall not be limited to” the four-factor analysis); see also H.R.Rep. No. 807, 93rd Cong., 2d Sess. 82 (1974) U.S.Code Cong. & Admin.News 1974, p. 4639 (“The determination of substantial business hardship is not to be limited to an examination of these factors, ... nor .must all these factors be met for there to be a finding of substantial business hardship.”)3 As a result, today’s decision creates a situation in which IRS may approve proposed amendments on bases entirely irrelevant to the four enumerated factors, but it nonetheless must indicate explicitly — most likely in a boilerplate fashion — that it considered each of the four factors. Failure to jump through these formalistic hoops, even if compliance would be burdensome for an agency already faced with a 90-day approval deadline, constitutes reversible error. This is an odd as well as unfortunate *901result of the court’s rigid approach to statutory interpretation.
Along with the provisions of section 303(b) itself, other parts of the statute evince Congress’ intent that IRS review be flexible and informal. Congress has, after all, put IRS in the position of having to take action within strict time constraints. See 29 U.S.C. § 1082(c)(8). Ninety days is not a very long time. The need for speed is no small matter, especially when, as must be the case if approval is warranted, the very viability of the plan is threatened if the amendment is not accomplished. The court disagrees, arguing that the burden of investigating and considering each of the four factors is slight, and discounting the significance of the normal review period of 90 days. In the majority’s view, if 90 days is not long enough to do the agency’s job, then regulations are in place for an extension until the job is properly done. See Rev.Proc. 79-18, § 7, 1979-1 C.B. 525, 527.
To be sure, IRS regulations provide for extensions, but they were certainly not intended to be routine, and until today were not so. I fear that today’s holding will go a long way toward transforming the 90-day approval time in the ERISA context into what the 10-day reply time in the Freedom of Information Act context, see 5 U.S.C. § 552a(d)(2)(A) (1982), has become: a practical nullity. This transformation would be an unfortunate and unnecessary development, inflicted only by a wooden interpretation of the statute.
II
With respect to the court’s treatment of the intervention issue, I am satisfied that it is correct. The District Court’s conclusion that ERISA preempted the otherwise applicable APA provision for intervention, see 5 U.S.C. § 555(b) (1982), cannot stand. The APA states explicitly that a “[subsequent statute may not be held to supersede or modify this subchapter ... except to the extent that it does so expressly.” 5 U.S.C. § 559 (1982). There being no express statement of preemption in ERISA, the agency is bound to allow interested parties to intervene “so far the orderly conduct of public business permits.” 5 U.S.C. § 555(b). Under such a statutory command, the agency may not adopt a blanket rule, however reasonable, prohibiting intervention.
In light of my agreement with the court’s judgment in this respect, I write only to emphasize the narrowness of today’s holding: the court holds only that a blanket prohibition of intervention is impermissible. It does not hold that intervention is required in all or even most instances. In my view, by virtue of the constraints placed upon it by the 90-day deadline, the IRS in most instances will likely be well within its discretion in denying intervention. In each case, however, the agency is obliged to supply an adequate explanation of why intervention would be unduly burdensome.
Ill
On a different front, I am troubled by one less obvious aspect of the court’s treatment of this case, and thus add a brief comment. In Footnote 102 of its opinion, see Maj. op. at 895 n. 102, the court properly rejects IRS’ argument that since the injury complained of was caused by plan trustees and not by IRS’ approval of the plan amendments, the case presents no case or controversy under Article III. In IRS’ view, because the amendments would have gone into effect had the agency not acted, the agency’s affirmative steps did not work any harm to the plaintiffs. The court properly rejects this argument, for the harm alleged to have been caused by the agency’s affirmative steps is sufficient to satisfy Article Ill’s standing requirements (and Congress has provided by statute for judicial review). See 29 U.S.C. § 1132(a)(3) (1982).
To this point, I have no quarrel. But the court goes further and suggests that judicial review would have been available even if the agency had not acted within 90 days, and the amendments therefore had taken effect by operation of law. The majority states “[bjecause the amendment could not be enforced absent submission to IRS and its explicit or defacto approval, appellants’ *902injury cannot reasonably be construed as springing solely from the actions of the trustees.” Maj. op. at 895 n. 102 (emphasis added). The majority thus apparently believes that agency inaction resulting in approval of amendments is amenable to judicial review, and seemingly so holds.
For my part, the issues of whether Congress provided for judicial review of de facto approval of amendments or whether such a cause of action can properly be implied, are fraught with difficulty. However, because the agency affirmatively approved the amendments in this instance, these questions are not part of the case before us. As a result, any statements as to the reviewability of de facto approvals of plan amendments constitute no part of today’s holding.
IV
A final observation. If the administrative process, as Congress has fashioned it, takes time, so be it. But in the context of ERISA plan amendments, the substance of the statutory scheme is thwarted, not advanced, by delay. Beneficiaries are adequately protected by judicial review — easily obtained — following IRS approval. IRS approval is intended to be expeditious, focusing primarily on whether a proposed amendment is justified by economic necessity, while judicial review properly focuses on substantive legal issues, such as questions concerning fiduciary duties. Today, the court frustrates this sensible structure by imposing delay at the administrative level to the detriment of all concerned. The judicial branch is well-advised to give credence not only to individual words within a statute, but to reflect upon and give life to the statute’s structure and purpose. Because the court fails to discern what I am satisfied is Congress’ intent, I am constrained respectfully to dissent.
. I agree that the case is properly before us, and therefore concur in part III of Judge Robinson’s opinion.
. The court also notes that Congress recognized the unworkability of the four-factor analysis in the multi-employer context. The majority cites as "significant” that the House Report, H.R.Rep. No. 779, 93rd Cong., 2d Sess. 25 (1974), omitted the first factor in its recitation of the IRS’s responsibilities in determining substantial business hardship in this context. See Maj. op. at 891, n. 83.
. That other factors may be considered is particularly relevant in this case. It scarcely requires stating that the asbestos industry is beset with the prospect of enormous potential liabilities, which, although not directly relevant to any of the four enumerated factors, undeniably spell deep trouble for the industry. It takes no great prescience to recognize that the very future existence of this industry is in substantial doubt, and along with the industry, its pension plans.