Per Curiam

Matt G. Kenna argued the cause and filed the briefs for appellant.

Peter C. Pfaffenroth, Assistant U.S. Attorney, argued the cause for appellees. With him on the brief were Jessie K. Liu, U.S. Attorney, and R. Craig Lawrence, Assistant U.S. Attorney.

Before: Garland, Chief Judge, Katsas, Circuit Judge, and Randolph, Senior Circuit Judge.

Opinion filed by Senior Circuit Judge Randolph concurring in the judgment.

Per Curiam

Under the Freedom of Information Act (FOIA), a court "may" award attorney’s fees to a requester "in any case under this section in which the complainant has substantially prevailed." 5 U.S.C. § 552(a)(4)(E)(i). In the OPEN Government Act of 2007, Congress amended FOIA to clarify that "a complainant has substantially prevailed if the complainant has obtained relief" through either of the following: "(I) a judicial order, or an enforceable written agreement or consent decree; or (II) a voluntary or unilateral change in position by the agency, if the complainant’s claim is not insubstantial." Id. § 552(a)(4)(E)(ii). This case concerns Grand Canyon Trust’s eligibility for fees under the second prong, known as the "catalyst theory." Brayton v. Office of the U.S. Trade Representative , 641 F.3d 521, 524-25 (D.C. Cir. 2011).

In August 2016, the Trust requested records from the Bureau of Land Management (BLM) and the Office of the Secretary of the Interior. All parties agree that the Trust received the lion’s share of the records it requested only after it filed suit. But the parties draw different conclusions from more or less the same timeline.

The Trust maintains that it brought about a change in the agencies’ positions, at least to the extent that its suit caused a "sudden acceleration" in the processing of its requests. Trust Br. 16 (quoting EPIC v. U.S. Dep’t of Homeland Sec. , 218 F. Supp. 3d 27, 41 (D.D.C. 2016) ). The agencies answer that the Trust’s suit caused no such change: they produced all the requested documents on approximately the schedule they had predicted before the suit was filed. The district court sided with the agencies, finding that the Trust failed to show that its suit caused the agencies to change their positions. See Grand Canyon Trust v. Zinke , 311 F. Supp. 3d 381, 390 (D.D.C. 2018). The Trust now appeals.

I

We begin our analysis by resolving the parties’ dispute over our standard of review.

The Trust maintains that we must review de novo the district court’s finding that it did not cause the release of the requested documents, whether more quickly or at all. The agencies’ view is that the question of causation is reviewed only for clear error. The agencies are correct. To explain why, we start with a brief retelling of the history of the attorney’s fees provision of the Freedom of Information Act.

For much of FOIA’s history, this court held that a plaintiff could show that it "substantially prevailed," and thus was eligible for fees under then § 552(a)(4)(E), either by pointing to a favorable action by a court (now codified in the first prong of § 552(a)(4)(E)(ii) ), or through the catalyst theory (now codified in the second prong). See Brayton , 641 F.3d at 524-25. In the 2001 Buckhannon case, the Supreme Court disagreed, concluding that "the ‘catalyst theory’ is not a permissible basis for the award of attorney’s fees" under the comparable language of the Americans with Disabilities Act and the Fair Housing Amendments Act. Buckhannon Bd. & Care Home, Inc. v. W. Va. Dep’t of Health & Human Res. , 532 U.S. 598, 610, 121 S.Ct. 1835, 149 L.Ed.2d 855 (2001). Rather, fees could only be obtained by litigants who were "awarded some relief by [a] court." Id. at 603, 121 S.Ct. 1835. Taking the hint, we shelved the catalyst theory for FOIA actions as well. See Oil, Chem. & Atomic Workers Int’l Union v. DOE , 288 F.3d 452, 456-57 (D.C. Cir. 2002).

In the OPEN Government Act of 2007, however, Congress disagreed with the Supreme Court and amended FOIA. As we have recounted several times, "[t]he purpose and effect of this law ... was to change the ‘eligibility’ prong back to its pre- Buckhannon form," Brayton , 641 F.3d at 525, and thus to "reinstate[ ] the catalyst theory in FOIA actions," Judicial Watch, Inc. v. F.B.I. , 522 F.3d 364, 370 (D.C. Cir. 2008). We have therefore returned to our original understanding, whereby a plaintiff can prove fee eligibility by showing that its lawsuit "substantially caused the government to release the requested documents before final judgment." Brayton , 641 F.3d at 524-25.

We have not revisited our standard of review since Congress restored the catalyst theory. Before Buckhannon , however, we repeatedly held that whether a plaintiff’s suit caused the production of documents "is, of course, a question of fact entrusted to the District Court and the appellate court is to review that decision under a clearly-erroneous standard." Weisberg v. Dep’t of Justice , 745 F.2d 1476, 1496 (D.C. Cir. 1984) ; see also, e.g., Weisberg v. Dep’t of Justice , 848 F.2d 1265, 1268 (D.C. Cir. 1988) ; Crooker v. Dep’t of the Treasury , 663 F.2d 140, 142 (D.C. Cir. 1980) ; Cox v. Dep’t of Justice , 601 F.2d 1, 6 (D.C. Cir. 1979). That should come as no surprise. Appellate courts review findings of fact only for clear error, see Pierce v. Underwood , 487 U.S. 552, 558, 108 S.Ct. 2541, 101 L.Ed.2d 490 (1988) ; FED. R. CIV. P. 52(a)(6), and actual causation is as much a question of fact in the FOIA context as it is in any other, cf. Pub. Citizen Health Res. Grp. v. Young , 909 F.2d 546, 549 (D.C. Cir. 1990) (reviewing but-for causation for clear error under the Equal Access to Justice Act); Hitchcock v. United States , 665 F.2d 354, 357-58 (D.C. Cir. 1981) (same in a negligence case).

The Trust notes that, in several cases decided after Buckhannon , we reviewed a district court’s conclusion about fee eligibility de novo. We did -- but only to the extent that it "rest[ed] on an interpretation of the statutory terms that define eligibility for an award." Edmonds v. F.B.I. , 417 F.3d 1319, 1322 (D.C. Cir. 2005) (quoting Nat’l Ass’n of Mfrs. v. Dep’t of Labor , 159 F.3d 597, 599 (D.C. Cir. 1998) ). Where parties dispute a question of law -- such as the meaning of a statutory term or of a judicial precedent like Buckhannon -- we apply the de novo standard. See Pierce , 487 U.S. at 557, 108 S.Ct. 2541. In Edmonds , for example, we reviewed de novo whether a plaintiff who had won a court order requiring expedited processing had "been awarded some relief by [a] court" within the meaning of Buckhannon . Edmonds , 417 F.3d at 1322 (quoting, inter alia, Buckhannon , 532 U.S. at 603, 121 S.Ct. 1835 ); see also Judicial Watch , 522 F.3d at 367 (same regarding a court-approved stipulation to disclose specified documents by dates certain); Davy v. C.I.A. , 456 F.3d 162, 164 (D.C. Cir. 2006) (same). But in none of those cases did we purport to alter the established clearly-erroneous standard for questions of fact. Accordingly, we apply that standard here.

II

As the plaintiff acknowledges, "the question under th[e] ‘catalyst theory’ is whether the ‘institution and prosecution of the litigation cause[d] the agency to release the documents obtained.’ " Pl.’s Mot. for Attorney’s Fees and Costs, at 2 (ECF No. 17) (quoting Church of Scientology v. Harris , 653 F.2d 584, 587 (D.C. Cir. 1981) ). Accord Brayton , 641 F.3d at 524 ; see also Buckhannon , 532 U.S. at 610, 121 S.Ct. 1835 (noting that, under the "catalyst theory," the test was "whether the lawsuit was a substantial ... cause of the defendant’s change in conduct"). Here, the parties agree that the Trust received the bulk of the documents responsive to its request only after it filed suit. But "the mere filing of the complaint and the subsequent release of the documents is insufficient to establish causation." Weisberg , 745 F.2d at 1496 ; see also Cox , 601 F.2d at 6 ("[A]n allegedly prevailing complainant must assert something more than post hoc, ergo propter hoc."). Rather, as the parties again agree, the plaintiff has the burden of showing "that it is more probable than not that the government would not have performed the desired act absent the lawsuit." Pub. Citizen Health Res. Grp. , 909 F.2d at 550 ; Trust Br. 16; Sec’y Br. 35.

As the district court’s opinion observed, the plaintiff’s own evidence "makes clear that both [agencies] had begun processing the plaintiff’s request well before this lawsuit was initiated and that both agencies had even made partial releases ... before the complaint was filed." Grand Canyon Trust , 311 F. Supp. 3d at 388. Neither agency suggested it would fail to comply with the request; to the contrary, both gave the plaintiff their predictions as to when production would be completed. Moreover, both "completed their disclosures within four months of the start of litigation, and these disclosures were satisfactory to the plaintiff." Id. ; see also Trust Br. 12 (noting that "the Trust determined that it would not challenge any of the agencies’ redactions"). These facts were sufficient for the district court to find that the "plaintiff has failed to show that this suit ‘cause[d] the agenc[ies] to release the documents.’ " Id. at 388-89 (quoting Church of Scientology , 653 F.2d at 587 )).

The Trust does not dispute this point on appeal. Rather than claim that its lawsuit caused the agencies to release documents they otherwise would not have released, it argues that the lawsuit "caus[ed] the Government to accelerate its final determinations and productions of documents." Trust Br. 13 (emphasis added). "The record shows," the Trust maintains, that its "lawsuit prompted the agencies to produce the requested documents more quickly than the agencies planned or said they would prior to when the suit was filed." Id.

We need not decide whether a "sudden acceleration" of production can, of itself, represent a "change in position" within the meaning of the statute. Here, even accepting the plaintiff’s own characterization of the agencies’ pre-suit predictions about their compliance timelines, those predictions were reasonably close to the final outcomes. The plaintiff alleges that the BLM said in October of 2016 before suit was filed that it would take "at least a year to compile and produce the responsive documents." Compl. ¶ 54 (App. 15). In fact -- without any court-imposed deadlines -- the agency wrapped production with the release of a final 57,112 pages on August 31, 2017, just six weeks before the earliest date it had predicted. See Grand Canyon Trust , 311 F. Supp. 3d at 385-86 ; Joint Status Report (App. 40-41); Levine Decl. ¶ 10 (App. 113).

As for the Office of the Secretary, before the onset of litigation it represented that it would need at least two more months -- and probably some undefined further amount of time -- to finish processing its remaining 8,100 pages because of the need for attorney review. See Trust Br. at 18; Compl. ¶ 43 (App. 13). In the end -- again without any deadlines imposed by the court -- it completed its work just two weeks earlier than predicted. See Grand Canyon Trust , 311 F. Supp. 3d at 385-86 ; Joint Status Report (App. 40-41).

These facts show that the agencies produced all of the requested documents roughly within the schedules that they had estimated before the litigation began. At most, the timeline reflects a modest acceleration from those earlier predictions. But predictions, by their very nature, are not perfect. The routine administrative and legal tasks required before agencies can release documents readily explain the minor timeline discrepancies here, particularly given the thousands of documents involved. Whatever we might have made of those discrepancies if we were deciding the question in the first instance, the district court certainly did not clearly err in finding that the plaintiff’s lawsuit did not cause a change in the agencies’ positions.

III

In order to establish eligibility for attorney’s fees, a FOIA plaintiff must show that its lawsuit caused a change in the agency’s position regarding the production of requested documents. We review a district court’s fact-finding regarding causation only for clear error. Because we find no such error here, the judgment of the district court is

Affirmed .

Randolph, Senior Circuit Judge, concurring in the judgment:

Every case about the meaning of a statute begins with a question. Here the initial question is whether, under the 2007 amendment to the Freedom of Information Act, the government changed its position after the plaintiff filed its lawsuit. My colleagues answer that the government did not change its position and so the plaintiff was not eligible for attorney fees. If they restricted their opinion to that issue I would be with them. But their opinion goes further and perpetuates what I consider to be a misunderstanding about the meaning of the 2007 FOIA amendment.

The misunderstanding, stated in one early case and then leapfrogged into three more of our opinions, is that the 2007 FOIA amendment reinstated the so-called "catalyst theory" rejected in Buckhannon v. West Va. Dep’t of Health & Human Resources , 532 U.S. 598, 601, 121 S.Ct. 1835, 149 L.Ed.2d 855 (2001). The catalyst theory is as follows. Even if the FOIA plaintiff obtained relief without a favorable judgment, the plaintiff could still recover attorney fees by proving that its lawsuit "caused" the government to change its position by disclosing previously withheld documents or producing documents on an accelerated timeline.

The statements in these four opinions about the catalyst theory appear to be casual, offhand. No analysis, rigorous or otherwise, backs them up. The opinions do not even attempt to square their statements with the words of the 2007 amendment. Still worse, none of these four cases even turned on an application of the 2007 amendment. To put it in another word, each of the statements was dicta.

Yet the majority opinion here cites these four opinions and, again without analysis, endorses their dicta that the 2007 amendment incorporated the essential requirement of the catalyst theory – namely, that the plaintiff must prove that its lawsuit caused the government to change positions. As in the other four opinions, the majority’s statement is dictum because the government did not change positions.

I am tempted to stop here. But extraneous pronouncements in a growing number of our cases seem to be taking hold. And so I believe it appropriate to counter dicta with my own dictum. Which brings me to the language of the 2007 amendment, otherwise known as the OPEN Government Act.

FOIA plaintiffs are eligible for attorney fees if they have "substantially prevailed." The 2007 amendment defined "substantially prevailed" to mean that the plaintiff "has obtained relief through either– (I) a judicial order, or an enforceable written agreement or consent decree; or (II) a voluntary or unilateral change in position by the agency, if the complainant’s claim is not insubstantial." 5 U.S.C. § 552(a)(4)(E)(ii).

We are concerned with subsection II. It contains three and only three conditions. The first is that the plaintiff "has obtained relief through" the government’s "change in position." Second, the government’s change in position was "voluntary or unilateral." Third, the plaintiff’s lawsuit must not have been "insubstantial."

The conclusion is inescapable – subsection II does not embody the catalyst theory. It does not do so because the provision requires only correlation not causation. Absence of statutory language supporting a theory (here the catalyst theory) is not evidence that Congress enacted the theory. Just the opposite.

Consider also Congress’s use of the word "unilateral" to describe the government’s action that qualifies the plaintiff for an award of attorney fees. "Unilateral" action signifies an undertaking by one side only, without involvement of the other side, a description quite at odds with the dicta coming from this court.

The statute thus does not require a plaintiff to show a causal connection between its lawsuit and the government’s capitulation. It is fair to ask why Congress would have enacted such a provision. See Antonin Scalia & Bryan A. Garner, Reading Law: The Interpretation of Legal Texts at 34 (2012). There is a ready answer. FOIA plaintiffs who have not received a judgment face obstacles in proving that their lawsuits caused the government to release previously withheld documents. Only government employees know with certainty why an agency changed its position. Those employees will doubtless be in the agency’s general counsel office and the appropriate division of the Department of Justice representing the agency. So the most plausible way for a plaintiff to prove that its lawsuit actually caused the government’s about-face would be to take depositions and seek documents from FOIA officers and other government employees. As the Grand Canyon Trust points out, this undertaking would effectively elevate the attorneys’ fee litigation to a "second major litigation," a result the Supreme Court has warned against. See Pierce v. Underwood , 487 U.S. 552, 563, 108 S.Ct. 2541, 101 L.Ed.2d 490 (1988) (citing Hensley v. Eckerhart , 461 U.S. 424, 437, 103 S.Ct. 1933, 76 L.Ed.2d 40 (1983) ). In addition, the most important communications would involve government attorneys and so would be protected by attorney-client privilege, making it even more challenging for litigants to make the required showing.

In all of this I am impressed by Judge Berzon’s comprehensive and well-reasoned opinion on this subject, making many more points against our colleagues’ opinion here and in the four prior cases I have discussed. See First Amendment Coalition v. Dep’t of Justice , 878 F.3d 1119, 1130 (9th Cir. 2017) (Berzon, J., concurring).