Bank of America, N.A. v. Federal Deposit Insurance

CARNES, Circuit Judge: This appeal concerns the validity of the Federal Deposit Insurance Corporation’s regulation, codified as 12 C.F.R. § 327, which determines the insurance assessment rate applicable to the funds that result from the merger of a Banking Insurance Fund financial institution with a Banking Insurance Fund Oakar institution in certain circumstances. That specific question is one of those complicated and transitory regulatory banking issues that is of no immediate interest to anyone except those directly involved with it. The FDIC is interested in the issue because it is the FDIC’s regulation that is in question, Bank of America is interested because a lot of the Bank’s money is involved, and we are interested because it is our duty to decide the issue. We publish this opinion explaining our decision because the same regulatory issue may come up between the FDIC and other banks, and also because our decision turns to some extent upon a Chevron/Chenery issue of first impression in this circuit which may arise in other administrative law cases in the future. That issue is whether Chen-ery ’s prohibition on post-hoc agency arguments applies to arguments proffered under the first step of the Chevron analysis. More specifically, the issue is whether a regulation may be upheld based upon the agency’s Chevron authority to resolve statutory ambiguities when the agency had stated upon issuance of the regulation that the statute was not ambiguous at all. I. BACKGROUND Bank of America filed an action under the Federal Deposit Insurance Act, 12 U.S.C. § 1811 et seq., and the Administrative Procedure Act, 5 U.S.C. § 701 et seq., against the Federal Deposit Insurance Corporation. 1 The Bank sought judicial review of the FDIC’s determination that it owed the Savings Association Insurance Fund $28,000,000.00 in deposit insurance obligations on more than $3.7 billion of the Bank’s deposits. Specifically, Bank of America sought a declaratory judgment that the FDIC’s rate of deposit insurance assessments against it should have been determined solely on the basis of the Bank’s membership in the Bank Insurance Fund, and not to any extent on the basis of the higher Savings Association Insurance Fund rate. The district court granted the FDIC’s motion to dismiss under Federal Rule of Civil Procedure 12(b)(6) for failure to state a claim, and Bank of America brought the case here on appeal. In 1989, Congress adopted the Financial Institutions Reform, Recovery, and Enforcement Act, Pub.L. No. 101-73, 103 Stat. 183 (1989) (codified at 12 U.S.C. § 1811 et seq.) (“FIRREA”). Among other things, FIRREA abolished the Federal Savings and Loan Insurance Corporation and shifted its deposit insurance functions to the FDIC. See 12 U.S.C. § 1811. FIR-REA established two separate deposit insurance funds within the FDIC: the Bank Insurance Fund (“BIF”), to cover the deposits of commercial banks; and the Savings Association Insurance Fund (“SAIF”), to cover the deposits of savings and loan associations. 2 See id. § 1821(a)(5)-(6). To replenish the SAIF, which had been depleted by the failure of numerous savings and loan institutions in the 1980’s, FIR-REA imposed higher deposit premiums on SAIF member institutions than on BIF *1312 member institutions. See generally id. § 1817. In order to prevent banks’ evasion of the higher SAIF rate of assessment by the transfer of deposits from SAIF institutions to BIF institutions, Congress imposed several control measures. Of central importance to the present case was the five-year moratorium on inter-fund transfers, beginning on August 9, 1989, and the restrictions that applied during those five years. See id. § 1815(d)(2)(A)(ii). During the period of the moratorium, FIRREA limited the ability of depository institutions to engage in “conversion transactions” between BIF Institutions and SAIF institutions. FIRREA defines the term “conversion transaction” as: (i) the change of status of an insured depository institution from a Bank Insurance Fund member to a Savings Association Insurance Fund member or from a Savings Association Insurance Fund member to a Bank Insurance Fund member; (ii) the merger or consolidation of a Bank Insurance Fund member with a Savings Association Insurance Fund member; (v) the transfer of deposits— (I) from a Bank Insurance Fund member to a Savings Association Insurance Fund member; Id. § 1815(d)(2)(B). In this case we are concerned primarily with section 1815(d)(2)(B)(ii): “the merger or consolidation of a Bank Insurance Fund member with a Savings Association Insurance Fund member.” Congress included a limited number of exceptions to the moratorium on conversion transactions, the most important of which is contained in the so-called Oakar Amendment. 3 See id. § 1815(d)(3). The Oakar Amendment permitted the FDIC to approve conversion transactions between an SAIF institution and a BIF institution if the institution resulting from the transaction assumed continuing responsibility to make deposit insurance payments to the SAIF. See id. § 1815(d)(3)(B). The resulting bank is commonly known as an “Oakar institution” or an “Oakar bank.” If the acquiring bank was a BIF-insured institution, then the resulting BIF Oakar institution would have to pay deposit insurance premiums both to its primary fund, BIF, and to its secondary fund, SAIF. See id. 1815(d)(3)(B)(i). The resulting institution is commonly referred to as a “BIF Oakar institution” in order to differentiate it from an “SAIF Oakar institution,” which resulted when the acquiring bank was insured by the SAIF. See id. § 1815(d)(3)(B)(ii). The amount of deposit insurance premiums that the BIF Oakar institution paid to its secondary fund, i.e., to the SAIF, was determined by reference to its “adjusted attributable deposit amount” (“AADA”). The AADA is equal to at least the value of all of the deposits acquired from the SAIF institution. See id. § 1815(d)(3)(C). Essentially, the amount of insurance premiums an Oakar institution had to pay to the two funds was equivalent to the relative portions of BIF-insured and SAIF-insured deposits at the time of the conversion transaction. See generally id. § 1815(d)(3)(B), (D); Branch Banking & Trust Co. v. F.D.I.C., 172 F.3d 317, 322 (4th Cir.1999). 4 The Oakar *1313 Amendment also provided that the amount of SAIF assessments determined in accordance with section 1815(d)(3)(B)(i), the provision applicable to BIF Oakar institutions, “shall be deposited in the [SAIF].” Id. § 1815(d)(3)(D)®. Following the enactment of the Oakar Amendment, it was not immediately clear how the FDIC would treat a transfer of deposits between a BIF Oakar institution and an ordinary BIF member bank. Specifically, there was a question as to whether the FDIC would continue to assess the BIF bank at the SAIF rate for that portion of the deposits the bank had acquired from the Oakar institution which were SAIF assessable prior to the transfer. The FDIC answered that question in an interpretive letter issued in 1990 which became known as the Rankin letter. See FDIC Advisory Op. 90-22 (June 15, 1990). The FDIC stated in the letter that an Oakar institution which transfers its deposits will be considered to have transferred its primary fund deposits first and then, when the primary fund deposits are exhausted, to have transferred its secondary fund deposits. In more technical language, the Rankin letter spelled out the FDIC’s position this way: “if the aggregate of all deposits transferred to BIF-member banks reduces the Oakar bank’s total deposit base below the amount of its [AADA], any subsequent deposit transfer by the Oakar bank to another BIF-member bank would be regarded as a conversion transaction ....” Id. In December of 1996, the FDIC published a final regulation codifying the FDIC’s position, previously set forth in the Rankin letter, regarding deposit transfers by BIF Oakar institutions to BIF institutions. See 61 Fed.Reg. 64,960, 64,962 (Dec. 10, 1996) (codified at 12 C.F.R. § 327 (1997)). The regulation provides that if -a BIF institution merges with a BIF Oakar institution, and in the merger the BIF Oakar institution transfers all of its primary fund deposits, then “[to] the extent that the aggregate volume of deposits that is transferred ... exceeds the volume of deposits that is insured by its primary fund immediately prior to the transaction ... the ... deposits so transferred shall be deemed to be insured by the institution’s secondary fund....” 12 C.F.R. § 327.37(a)(2). After such a transaction, “[t]he deposits [held by the resulting institution] shall be deemed ... to be insured by the same fund or funds in the same amount or amounts as the deposits were so insured immediately prior to the transaction.” Id. § 327.37(b). That means a transfer of deposits from a BIF Oakar institution to another BIF bank would be considered a covered Oakar conversion transaction to the extent that the deposits transferred exceed the Oakar bank’s primary fund deposits. The resulting institution would continue to pay SAIF assessments on the original amount of the AADA transferred by the BIF Oakar institution. In other words, the obligation the BIF Oakar institution was under to pay the higher SAIF rate on some of its funds could not be washed out by merging it into a BIF institution. In late 1996, Congress authorized a onetime “special assessment on the SAIF-assessable deposits of each insured depository institution [that would] cause the [SAIF] to achieve the designated reserve ratio on [October 1, 1996].” Economic Growth and Regulatory Paperwork Act of 1996, Pub.L. No. 104-208, § 2702(a), 110 Stat. 3009-479 to 3009-480 (Sept. 30, 1996) (“EGRPA”). With the reserve ratio achieved, the moratorium on conversion transactions was lifted on the same date. As a result of the recapitalization, SAIF and BIF premiums have been set at the same level since October 1, 1996. That fact does not moot this case, because this case involves assessments that arose before that date. In mid-1996, Bank of America acquired through merger 29 sister bank subsidiaries of its parent holding company. Prior to the merger, Bank of America was a BIF *1314 member bank that had never engaged in an Oakar transaction. However, 17 of the banks that Bank of America acquired in the merger were themselves BIF Oakar institutions with AADA’s assessed at the higher deposit-insurance rate paid by SAIF members. On August 29,1996, approximately three months before the publication of its final rule, the FDIC notified Bank of America that it would be assessing SAIF premium fees on more than $3.7 billion of deposits corresponding to the AADA’s that had been previously reported by the 17 BIF Oakar institutions prior to their acquisition by Bank of America. As a result, the FDIC informed Bank of America that it owed two SAIF quarterly payments in the amounts of $1,440,251.65 and $2,163,237.98. In addition, the FDIC’s attribution of AADA’s to Bank of America caused the agency to impose the EGRPA one-time special assessment on these deposits in the amount of $25,296,967.00. Bank of America filed suit under the Federal Deposit Insurance Act, 12 U.S.C. § 1811 et seq., and the Administrative Procedure Act, 5 U.S.C. § 701 et seq., against the FDIC on the grounds that the SAIF assessments were beyond the agency’s authority because they were contrary to the plain meaning of the Federal Deposit Insurance Act. The Bank sought a declaratory judgment that the FDIC’s rate of assessments should be determined solely on the basis of Bank of America’s membership in the BIF, and that the FDIC’s assessment of a portion of the Bank’s deposits at the SAIF rate was unlawful. The FDIC filed a motion to dismiss, arguing that the interpretive rule contained in its regulation, which permits the agency to assess Bank of America’s deposits in the way it did, is valid. The district court granted the FDIC’s motion to dismiss, finding that the Federal Deposit Insurance Act was ambiguous and that the Agency’s interpretive rule “properly filled [a] gap” in the statute. II. DISCUSSION 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), the Supreme Court established the analytical framework for judicial review of an administrative agency’s interpretation of a statute. The reviewing court must first establish whether the intent of Congress is clear from the statute. .If it is, that intent controls, and the inquiry is over. Id. at 842-43, 104 S.Ct. at 2781. If, however, the reviewing court determines that the statute is ambiguous, the question becomes whether the agency’s resolution of the ambiguity is based on a permissible construction of the statute. Id. at 843, 104 S.Ct. at 2781-82. We turn now to the first question: whether the Federal Deposit Insurance Act is ambiguous in regard to how funds transferred from a BIF Oakar institution to an ordinary BIF institution should be treated for insurance fund assessment purposes. A. STATUTORY CONSTRUCTION 1. Bank of America’s Position Bank of America asserts that there is only one dispositive issue on appeal, and that is whether the merger between it and the 17 BIF Oakar banks is a conversion transaction under the Federal Deposit Insurance Act. Bank of America’s position is centered on the language in sections 1815(d)(2) and (3), which defines “conversion transaction” as a transaction occurring between a BIF “member” and an SAIF “member.” According to Bank of America, a merger between itself, a BIF bank, and a BIF Oakar institution could not satisfy the statutory definition of conversion transaction because a BIF Oakar bank is not technically an SAIF “member.” A BIF Oakar bank is not an SAIF member, Bank of America insists, because of what section 1815(d)(3)(B)(i) of the Act says. That provision, which is part of the Oa-kar Amendment, specifies that when “an *1315 acquiring, assuming, or resulting depository institution ... is a[BIF] member ... [a portion of its deposits] shall be treated as deposits which are insured by the [SAIF].” 12 U.S.C. § 1815(d)(3)(B)© (emphasis added). Bank of America argues that language must mean the resulting BIF Oakar institution is a member of only the Bank Insurance Fund. The language must mean that, the bank says, because while the statute defines an SAIF member as a “depository institution the deposits of which are insured by [SAIF],” id. § 1817(Z)(5), the language in section 1815(d)(3)(B)® refers to a BIF Oakar institution as having deposits that are only “treated as” insured by SAIF, not as an institution whose deposits “are insured by” the SAIF. As further support for its argument that the FDIC’s regulation is inconsistent with the unambiguous language of the statute, Bank of America contends that the regulation has altered the statutory method for computing the AADA. The statute provides, in essence, that the AADA should.be equivalent to “the amounts of any deposits acquired by the institution.” 12 U.S.C. § 1815(d)(3)(C)®. The regulation, on the other hand, provides that the AADA should be equivalent to “the aggregate amount of the transferred deposits minus that portion thereof that is equal to the institution’s primary-fund deposits.” 12 C.F.R. § 327.37(a)(2). Bank of America’s argument on this point is that the statutory method of computing the AADA indicates that conversion transactions were only intended to encompass deposit exchanges between BIF members and SAIF members. In a statutory Oakar transaction (between a BIF member and a SAIF member), the acquiring institution would acquire nothing but SAIF-insured deposits. Therefore, Bank of America says, it makes sense to define the size of the AADA by “the amounts of any deposits acquired by the institution,” § 1815(d)(3)(C)®, which is exactly what the statute does. On the other hand, in a regulatory Oakar transaction (between a BIF member and a BIF Oakar institution), the acquiring institution would acquire both BIF-insured deposits and SAIF-insured deposits, so it is necessary to subtract “that portion ... that is equal to the [Oakar institution’s] primary fund deposits.” 12 C.F.R. § 327.37(a)(2). Bank of America argues that the fact the statutory method for computing the AADA did not provide a method for differentiating between SAIF and BIF insured deposits acquired in a conversion transaction is evidence that for the regulation to include such transactions as Oakar conversion transactions goes beyond and is inconsistent with the statutory definition of Oakar conversion transactions. Finally, Bank of America argues that the FDIC itself has always taken the position that an Oakar bank, though a portion of its deposits are treated as insured by the SAIF, is still a “member” of only the BIF. See Preamble to the Final Rule, 61 Fed.Reg. 64,960, 64,962 (Dec. 10, 1996) (“[t]he FDIC accepts the proposition that an Oakar institution is a member of its primary fund only, and is not a member of its secondary fund even though it holds secondary-fund deposits.”). Based on all of these arguments, Bank of America asserts that the language of the statute clearly indicates that a merger of a BIF bank and BIF Oakar institution cannot satisfy the statutory definition of a conversion transaction because both institutions are members of the BIF only. If no conversion transaction occurred when Bank of America merged with the Oakar banks, none of its deposits can be subject to assessment at the SAIF rate. 5 Aecord- *1316 ingly, Bank of America argues that this Court should not defer to the FDIC’s interpretation of the statute under the first step of the Chevron analysis, because the agency’s interpretation expressed in the regulation is contrary to the unambiguous mandate of the statute. 2. The Fourth Circuit’s Construction The Fourth Circuit recently decided a case that presented precisely the same issues that are before us here. See Branch Banking & Trust Co. v. F.D.I.C., 172 F.3d 317 (4th Cir.1999). In rejecting arguments identical to those Bank of America has made to us, the Fourth Circuit reasoned the “provision stating that a resulting BIF Oakar institution merely holds deposits treated as insured by SAIF creates an ambiguity as to whether the resulting BIF Oakar institution is, in addition to being a member of BIF, a member of SAIF.” 6 Id. at 327 (emphasis in original). The Court then concluded that the ambiguity as to whether an Oakar institution could be considered an SAIF member rendered the definition of conversion transaction unclear as it applies to transactions between BIF institutions and BIF Oakar institutions, which gave the FDIC leeway to regulate as it did. Id. Although we reach the same result as the Fourth Circuit, we get there by a different route. 3. Our Analysis of the Statute The first step under the Chevron analysis is to examine the statute in order to determine whether “Congress has directly spoken to the precise question at issue.” Chevron, 467 U.S. at 842, 104 S.Ct. at 2781. The resolution of this threshold inquiry will be at least influenced, if not determined, by how broadly we frame the “precise question.” We agree with the Fourth Circuit’s framing of that question: [Wjhether a transaction whereby a BIF member institution merges with a BIF Oakar institution (an Oakar institution that is a member of BIF, its primary fund, but that holds deposits that are treated as insured by SAIF, its secondary insurance fund) is a “conversion transaction” under the Oakar Amendment such that the acquiring institution is required to pay insurance to both BIF and SAIF in accordance with the deposits being insured or treated as insured by each fund at the time of the acquisition. Branch Banking & Trust, 172 F.3d at 326 (emphasis in -original). Did Congress in the statute speak to that precise issue? There are two statutory provisions upon which Bank of America primarily relies: the definition of an Oakar institution as a BIF institution that has deposits “treated as deposits which are insured by the [SAIF],” 12 U.S.C. § 1815(d)(3)(B)©; and the statutory method for computing the AADA, which is arguably inconsistent with the method the FDIC prescribed in the regulation, see id. § 1815(d)(3)(C)©. We agree with Bank of America that these two provisions suggest that a BIF Oakar institution is not also a member of the SAIF. However, nothing in either of these provisions necessarily compels such a conclu *1317 sion. In other words, nothing in either provision speaks “directly” — as Chevron puts it — to that precise question. Bank of America also relies heavily on the FDIC’s prior position, stated in the preamble to the regulation, that a BIF Oakar institution is definitely not a member of the SAIF. See Preamble to the Final Rule, 61 Fed.Reg. 64,960, 64,962 (Dec. 10, 1996). However, as we will explain later in this opinion, the FDIC’s pri- or construction of the statute is not relevant under the first step of the Chevron analysis. And that is a fundamental flaw in Bank of America’s argument. The principal support for Bank of America’s construction of the statute as clearly precluding a BIF Oakar institution from also being a member of the SAIF is the FDIC’s previous statements to that effect. If we remove those statements from the analysis and focus exclusively on the relevant statutory provisions, Bank of America does not have a strong argument. The primary source of statutory ambiguity stems from a provision that received little attention from either the parties in this case or the Fourth Circuit in its Branch Banking & Trust opinion. That statutory provision provides that both funds will be liable for costs in the event that an Oakar institution defaults, with each fund being liable in proportion to the “amount of insured deposits of such acquiring, assuming, or resulting depository institution assessed by the [BIF] and the [SAIF], respectively, under [§ 1815(d)(3)(B)(i) ].” See 12 U.S.C. § 1816(d)(3)(G). Thus, an Oakar institution pays insurance premiums to the SAIF on its AADA deposits, id. § 1815(d)(3)(D)(i), and the SAIF — and not the BIF — is exclusively liable to the extent of those deposits. Construing sections 1815(d)(3)(G) and 1815(d)(3)(D)(i) together, an Oakar bank’s AADA deposits are “treated” like they are insured by the SAIF to such an extent that it appears as though they are in fact insured by the SAIF. We do not see (and counsel for Bank of America could not point out at oral argument) any difference at all between the way an Oakar bank’s AADA deposits are “insured” by the SAIF and the way its BIF deposits are “insured” by the BIF. Here is our point. Under the statute, a BIF Oakar institution holds some funds that are in every meaningful way and effect insured by the SAIF, and it holds other funds that are in every meaningful way and effect insured by the BIF. The statute defines an SAIF member institution as one whose funds “are insured by the [SAIF],” id. § 1817(0(5), and it defines a BIF member institution as an institution whose funds “are insured by the [BIF],” id. § 1817(0(4). Under these provisions and definitions, an Oakar institution can be a “member” of both funds. Thus, there is an ambiguity in the statute. That ambiguity exists because the provisions we have just discussed point in one direction (an Oakar institution being a member of both funds), while the provisions the Bank of America relies upon point in the other direction (an Oakar institution being a member of only the Banking Insurance Fund). So, although we have taken a different route than the Fourth Circuit did in Branch Banking & Trust, we have arrived at the same conclusion as it did regarding the first step of the Chevron analysis: the statute is ambiguous about whether the merger of a BIF member institution with a BIF Oakar institution is a “conversion transaction” under the Oakar Amendment; Congress did not speak directly and unambiguously to that precise issue. B. THE FDIC’s INCONSISTENT POSITIONS Before moving on to the second step of the Chevron analysis, we need to address another of Bank of America’s arguments relevant to the first step, an argument which goes to the validity of the regulation and the FDIC’s position on appeal. Bank of America makes much of the fact that, on appeal, the FDIC points to a different, and indeed antithetical, source of authority for adopting the interpretation of the Oakar Amendment than it did in the regulation. *1318 In the preamble to the regulation, the FDIC addressed a number of concerns that arose during the notice and comment period. Some commentators had suggested that the FDIC lacked the statutory authority to interpret the Oakar Amendment as it had in the Rankin Letter. In response, the FDIC never suggested that the source of its authority lay in the statute’s ambiguity about whether a BIF Oa-kar bank could also be an SAIF member. To the contrary, the FDIC stated: The FDIC accepts the proposition that an Oakar institution is a member of its primary fund only, and is not a member of its secondary fund even though it holds secondary-fund deposits ... [T]he original version of the Oakar Amendment .... made it abundantly clear that a BIF-member bank continued to be a BIF member after acquiring deposits from a SAIF member in an Oakar transaction .... The FDIC holds this view today. Preamble to the Final Rule, 61 Fed.Reg. 64,962 (Dec. 10, 1996). Bank of America charges that instead of taking the position that the term “member” was ambiguous, the FDIC simply (in the Bank’s words) “nullified the ‘membership’ requirement in the statutory definition of a ‘conversion transaction.’ ” Bank of America’s argument is not without force. In support of the FDIC’s position that it had the authority to promulgate the regulation at issue, the agency never once suggested that the term “membership” or any other provision of the statute was ambiguous. Instead, the FDIC simply ignored the plain language of the statutory definition of “conversion transaction” in favor of a more functional interpretation. As the FDIC stated in the preamble to the regulation: [T]he FDIC also takes the position that nominal fund membership is not the touchstone for determining whether a transaction is a conversion transaction within the meaning of [12 U.S.C. § 1815(d)(2) ], and accordingly does not determine whether a transaction comes within the scope of the Oakar Amendment. Id. Instead, the term “membership,” according to the agency’s original position, is only a “label that denotes the formal relationship of an insured institution to the FDIC as an insurer within the context of the two-fund system.” Id. The FDIC declared that instead of looking to fund membership it would look to the “substance of the relationship” between the Oakar institution and the FDIC as insurer in order to determine whether the transaction is a conversion transaction. Id. at 64,698. The agency said that it was giving the “label ‘member’ ... only that degree of significance that is appropriate to preserve the integrity of the two-fund structure.” Id. In this appeal the FDIC appears to have abandoned its initial, damn-the-statute-let’s-be-functional approach, in favor of its current, more persuasive position. The agency’s current position is that the statute is ambiguous about whether a BIF Oakar institution can be a member of both the BIF and the SAIF, and the existence of that ambiguity authorizes the FDIC to issue an interpretive regulation setting out its functional approach. Apparently, the FDIC first took this new position when it argued before the Fourth Circuit in Branch Banking & Trust. With respect to that decision, Bank of America argues it was “improper for the FDIC to contradict the position that it took in its rulemaking, and the Fourth Circuit should not have relied upon the agency’s litigation-induced construction of the statute.” 7 In support of that argu *1319 ment, Bank of America cites SEC v. Chenery Corp., 318 U.S. 80, 87, 63 S.Ct. 454, 459, 87 L.Ed. 626 (1943), which held that “[t]he grounds upon which an administrative order must be judged are those upon which the record discloses that its action was based.” Bank of America argues that the Fourth Circuit erred in deferring to the FDIC’s current interpretation of the Federal Deposit Insurance Act — that the statute is ambiguous as to whether a BIF Oakar institution is a member of both the BIF and the SAIF — because the FDIC had previously disavowed that interpretation. Instead of deferring to the agency’s litigation-induced interpretation of the statute, the Bank would have us hold the FDIC to its original position and evaluate the validity of the regulation, as well as the agency’s authority to promulgate it, without regard to the FDIC’s “new” statutory interpretation. The problem with Bank of America’s argument is that, while the FDIC’s new interpretation of the statute does appear to be post-hoe, or post-litigation, and is inconsistent with the position the agency took in the preamble to its regulation, those characteristics do not have any legal significance on the issue of whether the statute is ambiguous under the first step of the Chevron analysis. There is no support in either Chenery or its progeny for the proposition that Chenery’s prohibition on post-hoc rationales should apply to agency arguments proffered under the first step of the Chevron analysis. It is the duty of the courts to interpret statutory language, and courts should decide whether there is ambiguity in a statute without regard to an agency’s prior, or current, interpretation. We have found only a handful of decisions applying the Chenery principle in the context of a Chevron analysis. Most of those decisions apply Chenery during analysis involving Chevron’s second step, while the court was examining an agency’s resolution of the statutory ambiguity in order to determine whether that resolution wTas based on a permissible construction of the statute. See America’s Cmty. Bankers v. F.D.I.C., 200 F.3d 822, 835 (D.C.Cir.2000) (holding that a post-hoc rationale alone could not support the agency’s interpretation of a statute under the second prong of Chevron, stating that the Chenery principle “applies ... to our review of statutory interpretations under the second prong of Chevron ”); Environmental Def. Fund, Inc. v. Administrator of U.S. E.P.A., 898 F.2d 183, 189 (D.C.Cir.1990) (refusing to consider EPA’s post-hoc rationalization for its proffered interpretation of the statute under the second step of the Chevron analysis); Railway Labor Executives’ Ass’n v. Interstate Commerce Comm’n, 784 F.2d 959, 974 (9th Cir.1986) (refusing, under the second step of Chevron, to consider arguments in support of the agency’s interpretation of the statute that the agency had not stated contemporaneously with its issuance of the orders in question). That is the proper place for Chenery considerations to come into the Chevron analysis. A few decisions have addressed the question of whether the Chenery principle applies to agency post-hoc rationalizations involving the kind of pure questions of statutory analysis that are involved under the first prong of the Chevron analysis. All of those decisions conclude that it does not. See Arkansas AFL-CIO v. Federal Communications Comm’n, 11 F.3d 1430, 1440 (8th Cir.1993) (rejecting contention that Chenery applies to pure question of statutory construction, stating, “the Supreme Court clearly limited Chenery to situations in which the agency failed to make a necessary determination of fact or of policy.... The codification question we face involves neither a determination of fact nor a determination of policy”) (internal citation and quotation omitted); Railway Labor Executives’ Ass’n, 784 F.2d at 969 (“Generally, a reviewing court may only judge the propriety of an agency decision on the grounds invoked by the agency.... However, the court is not so bound when, as here, the issue in dispute *1320 is the interpretation of a federal statute.”) (internal citation omitted); North Carolina Comm’n of Indian Affairs v. United States Dep’t of Labor, 725 F.2d 238, 240 (4th Cir.1984) (“We do not ... perceive there to be a Chenery problem in the instant case because the question of interpretation of a federal statute is not a determination or judgment which an administrative agency alone is authorized to make.”) (citation and quotation omitted). We think that the Fourth, Eighth, and Ninth Circuits got it right in those decisions: the Chenery principle does not apply to agency justifications or positions put forward under the first step of the Chevron analysis. The Supreme Court articulated the basis of the Chenery principle as follows: [WJhere the correctness of the lower court’s decision depends upon a determination of fact which only a jury could make but which has not been made, the appellate court cannot take the place of the jury. Like considerations govern review of administrative orders. If an order is valid only as a determination of policy or judgement which the agency alone is authorized to make and which it has not made, a judicial judgement cannot be made to do service for an administrative judgment. Chenery, 318 U.S. at 88, 63 S.Ct. at 459. Under the first step of the Chevron analysis, we undertake an independent analysis of the statute in order to determine whether Congress has spoken directly to the precise issue before the agency. That inquiry does not involve, in Chenery’s terms, a “determination of policy or judgment which the agency alone is authorized to make.” Id. To the contrary, as the Supreme Court stated in Chevron: The judiciary is the final authority on issues of statutory construction and must reject administrative constructions which are contrary to clear congressional intent.... If a court, employing traditional tools of statutory construction, ascertains that Congress had an intention on the precise question at issue, that intention is the law and must be given effect. 467 U.S. at 843 n. 9, 104 S.Ct. at 2781-82 n. 9 (citations omitted). Put another way, it is ultimately the function of the judiciary, not the administrative agency, to decide whether Congress spoke directly to the issue in question. See id., at 843 n. 9, 104 S.Ct. at 2782 n. 9 (“The judiciary is the final authority on issues of statutory construction ...”); Gulf Power Co. v. Federal Communications Comm’n, 208 F.3d 1263, 1272 (11th Cir.2000) (“Under Chevron step one, we determine whether Congress has spoken unambiguously to the question at issue.”); National Mining Assoc. v. Secretary of Labor, 153 F.3d 1264, 1267 (11th Cir.1998) (“when applying Chevron’s first step, we do not need to defer when the issue is a ‘pure question of statutory construction’ ”). We have found only one decision that has applied Chenery during the first step of a Chevron analysis. In The Business Roundtable v. Securities and Exchange Comm’n, 905 F.2d 406, 407 (D.C.Cir.1990), the issue was whether the agency had exceeded its jurisdiction when it promulgated a regulation impacting the rules of self-regulatory organizations. The relevant provision of the Securities Exchange Act establishing the limits of the SEC’s jurisdiction provided that the SEC could amend the rules of a self-regulatory organization if it was in furtherance of the purposes of the Act. See id. at 408-09. The SEC pointed to a number of provisions of the Exchange Act, arguing that each revealed a “purpose” of the Act that the new rule could be said to be “in furtherance” of. See id. at 410-16. The court rejected the SEC’s arguments and decided, under Chevron step one, that the Exchange Act could not be read as granting the SEC jurisdiction to promulgate the rule at issue in that case. Id. at 417. The court concluded its opinion by stating: “Even if other statutory provisions could support the Commission’s asserted authority, we cannot supply grounds to sustain *1321 the regulations that were not invoked by the Commission below.” Id. at 417 (footnote omitted) (citing Chenery). The Business Roundtable decision is distinguishable from this case. There, the court rejected the possibility of a post-hoc suggestion that an entirely different statute could support the agency’s action, see id. n. 10 (“Some commentators argued that the Commission could ground its authority in the Williams Act, § 14(d)-(f), 15 U.S.C. § 78n(d)-(f).”). Here, by contrast, the agency’s post-hoc rationale involves a different construction of the same statute. The distinction is important. An administrative agency should attempt to conduct its actions, whether they be rulemaking or adjudication, within the statutory limits that Congress has placed on its authority. At the same time, an agency should have the opportunity to take full advantage of that authority, and to do so the agency must be able to recognize its outer limits. However, when, as was the case in Business Roundtable, the agency so fundamentally misunderstands the source of its authority (assuming it had any to begin with) that it purports to rely on the wrong statute, then any action taken by the agency pursuant to that statute would presumably not utilize a Congressional grant of authority. In such a case, the proper remedy is not to adopt the agency’s litigation-induced rationale, but to vacate the agency action and allow the agency to take full advantage of the “new” or “actual” basis of its authority, if there be one. See Michael Herz, Deference Running Riot: Separating Interpretation and Laumaking under Chevron, 6 Adm. L.J. Am. U. 187 (1990) (“Where an agency underestimates its freedom of movement, the proper relief is to remand to the agency, not to uphold the agency action. This reflects not only standard principles of administrative law settled since Chenery ... but also the basic theory of Chevron, which is that agencies should be left free to make policy determinations where Congress has not.”) (footnote omitted). As the court in Business Roundtable noted, even if it had been persuaded that a different statute would have provided the agency with authority to act as the agency did, “a change in the jurisdictional basis would almost certainly alter the substantive content of the final regulations.” See Business Roundtable, 905 F.2d at 417; see also Women Involved In Farm Econ. v. U.S. Dep’t of Agric., 876 F.2d 994, 998-99 (D.C.Cir.1989) (“By adopting a specific argument in support of agency action offered by counsel in the litigation] process — but not relied oh by the agency — the courts might actually restrict improperly the agency’s future freedom of action to make policy under a particular statute.”). No such concerns are present in this ease. There has been no “change in the jurisdictional basis” for the agency action. Moreover, while the FDIC did not originally believe that the statute was ambiguous, the agency’s basis for adopting the regulation is entirely consistent with the basis of our holding. The original grounds the FDIC identified for promulgating the regulation was that nominal fund membership did not determine whether a conversion transaction had taken place; instead, the FDIC would look to the substance of the relationship between the Oakar institution and the FDIC as insurer. Similarly, our conclusion is that the statute is ambiguous in part because the “substance of the relationship” between an Oakar institution and the two funds to which it pays insurance premiums makes it seem as though it is insured by (and thus a member of) both funds. Because the basis for the FDIC’s interpretation of the source of its authority to adopt the regulation and our subsequent interpretation are substantively the same, there is no risk that the FDIC “underestimate[d] its freedom of movement” under the FDI Act. One additional consideration supports our conclusion. Our decision is only that the statute is ambiguous about whether an Oakar institution is a “member” of both the SAIF and the BIF. The FDIC re *1322 mained (and remains) free to exercise its power to make policy judgments and interpret that ambiguity as it deems appropriate, so long as its interpretation is permissible under the second step of the Chevron analysis. See Herz, supra (“One concern underlying the Chenery rule is that to uphold an agency action on a ground invented by Department of Justice lawyers or the court itself could lock the agency into a position that it had never in fact espoused and with which it disagrees. This concern is present only if the court’s ruling results in final and binding conclusions of law. If the conclusion of the court is, as in Chevron, that the agency was free to do what it did, but was and is also free to do the opposite, then the court is not limiting agency options.”). Thus, there is no risk that our decision “might actually restrict improperly the agency’s future freedom of action to make policy under a particular statute.” See Women Involved In Farm Econ., 876 F.2d at 999. Chenery’s prohibition on litigation-induced, post-hoc rationalizations does not apply here, and the fact that the FDIC identified a different source of its perceived authority to adopt the regulation is of no moment under the first prong of Chevron. C. CHEVRON STEP TWO We agree with the Fourth Circuit that the -FDIC’s interpretation of the statute, which is embodied in the regulation, was permissible under the highly deferential second step of the Chevron analysis. 8 See Branch Banking & Trust, 172 F.3d at 327-29. As the Fourth Circuit stated: We find the FDIC’s interpretation consistent with the purpose and intent of FIRREA as a whole — to recapitalize the failing FSLIC-insured institutions — and the purpose of the Oakar Amendment— allowing certain conversion transactions between institutions that are BIF members and SAIF members, while protecting SAIF by requiring the resulting institution to continue to pay assessments to SAIF. Id. at 328. Moreover, despite the fact that the FDIC has taken inconsistent positions on the basis for its authority to adopt the interpretive regulation at issue, the policy reflected in the regulation itself, embodied originally in the Rankin letter and later in the regulation, has always remained consistent. Specifically, the FDIC determined that Congress intended to protect the SAIF by ensuring either: (1) that no SAIF-insured institution switched to the BIF (as reflected in the moratorium on conversion transactions), or (2) that an SAIF institution that engaged in an Oakar conversion transaction would continue to pay SAIF assessments until the moratorium was lifted. This interpretation by the FDIC involved “[t]he power of an administrative agency to administer a congressionally created ... program” and it “necessarily requires the formulation of policy and the making of rules to fill any gap left, implicitly or explicitly, by Congress.” Chevron, 467 U.S. at 843, 104 S.Ct. at 2782 (citation and quotation omitted). III. CONCLUSION We AFFIRM the district court’s dismissal of Bank of America’s claim. Footnotes [1] Barnett Bank was the original party to this suit. However, after the initiation of the present appeal, Bank of America acquired Barnett Bank. Accordingly, all of our references to the plaintiff will be to Bank of America. [2] A more complete history of the development of the contemporary two-fund deposit insurance scheme is contained in Branch Banking & Trust Co. v. F.D.I.C., 172 F.3d 317, 319-21 (4th Cir.1999), and Great W. Bank v. Office of Thrift Supervision, 916 F.2d 1421, 1423-24 (9th Cir.1990). [3] In addition to the conversion transactions authorized by the Oakar Amendment, institutions could also engage in conversion transactions that either: (1) affected an insubstantial portion of the total deposits of each institution; or (2) that occurred in connection with the acquisition of a failing SAIF institution or a failing BIF institution, if the FDIC determined that the loss of assessments to the fund from which the institution was departing was outweighed by the financial benefits to that fund. See 12 U.S.C. § 1815(d)(2)(C)(i)-(iii). Institutions participating in a conversion transaction pursuant to either of these exceptions were required to pay both an "exit” fee to the insurance fund being exited and an “entrance” fee to the fund being entered. See id. § 1815(d)(2)(E). [4] The Oakar Amendment also provided that upon expiration of the moratorium, an Oakar institution would be permitted to convert all of its deposits to its primary fund upon pay *1313 ment of entrance and exit fees. See 12 U.S.C. § 1815(d)(3)(H)(ii). [5] As Bank of America pointed out in the district court, BIF or SAIF assessments are made solely on the basis of a financial institution's membership in one fund or another, with the limited exception of the Oakar Amendment. Thus, if Bank of America’s interpretation of the Oakar Amendment is correct (meaning that Bank of America did not engage in a covered Oakar conversion transaction and cannot be assessed SAIF fees on any of its newly acquired deposits), there is no other statutory basis for imposing the SAIF assessments on it. [6] The Fourth Circuit based its finding of ambiguity on the following analysis: the prohibition on inter-fund transfers, 12 U.S.C. § 1815(d)(3)(E)(ii), seems to indicate that after an Oakar transaction, the resulting Oakar institution must be a member of both the BIF and the SAIF (because if the resulting institution holds BIF-insured, as well as SAIF-in-sured deposits, then no inter-fund transfer actually took place. In other words, rather than interpret an Oakar conversion as resulting in the transfer of SAIF-insured deposits from a SAIF-insured institution to an exclusively BIF-insured institution, under the Fourth Circuit’s interpretation, one institution now holds deposits insured by both funds). However, the language contained in the assessments provision belies that interpretation, because it says that a portion of the deposits of a BIF Oakar institution are to "he treated as deposits which are insured by the [SAIF]." Id. § 1815(d)(3)(B)(i) (emphasis added). Thus, according to the Fourth Circuit, it is not clear whether a BIF Oakar institution merely holds deposits treated as insured by SAIF, or is in fact insured by the SAIF in addition to being insured by the BIF. Id. at 327. [7] Bank of America also argues that it appeared as though the Fourth Circuit reached its conclusion "without any apparent awareness of” the FDIC’s conflicting views. However, since the Fourth Circuit quoted extensively from the preamble of the regulation, see Branch Banking & Trust, 172 F.3d at 323-24, we think it had to have been aware of the FDIC's change in position. Apparently, the Fourth Circuit simply chose not to address that aspect of the case. [8] Bank of America did not contest the second step before us, instead aiming all of its argu-menl at the first step.