General Electric Co. v. Browner

Opinion

UNITED STATES DISTRICT COURT FOR THE DISTRICT OF COLUMBIA

GENERAL ELECTRIC COMPANY

Plaintiff,

v. Civil Action No. 00-2855 (JDB)

LISA JACKSON,1 Administrator, United States Environmental Protection Agency, and the UNITED STATES ENVIRONMENTAL PROTECTION AGENCY

Defendant.

MEMORANDUM OPINION

This case presents a broad constitutional attack on a significant federal environmental program. Plaintiff General Electric Company ("GE" or "plaintiff") challenges the U.S. Environmental Protection Agency's ("EPA") administration of section 106 of the Comprehensive Environmental Response, Compensation, and Liability Act ("CERCLA" or "the Act"), 42 U.S.C.§ 9601 et seq. GE contends that EPA's "pattern and practice" of administering section 106, 42 U.S.C. § 9606, violates the Due Process Clause of the Fifth Amendment. Am. Compl. at ¶ 1. Before the Court are the motions for summary judgment of GE and EPA and its administrator, Lisa Jackson.2 For the reasons that follow, the Court rejects GE's novel, but

1 Former acting administrator Stephen L. Johnson was previously named as the lead defendant in this case. Pursuant to Federal Rule of Civil Procedure 25(d), the Court automatically substitutes his successor, Lisa Jackson, as the new lead defendant.

2 EPA styled its motion as a motion for judgment on the pleadings, or alternatively, a motion for summary judgment. The same legal standard applies for each kind of motion. See Greenhouse v. Green, 574 F. Supp. 2d 57, 63-64 (D.D.C. 2008). Because the record in this case contains material from the parties' extensive discovery, the Court will treat EPA's motion as one

2

substantial, challenge. On the record presented, the administrative order regime under section 106 of CERCLA, as administered by EPA, does not offend due process.

BACKGROUND

I. The CERCLA Framework3 CERCLA was enacted to ensure efficient and expedient clean-up of hazardous waste sites resulting from industrial pollution. See Key Tronic Corp. v. United States, 511 U.S. 809, 814 (1994); see also United States v. Bestfoods, 524 U.S. 51, 55 (1998). Congress intended the parties responsible for polluting a site to clean it up themselves, or at least to pay the costs of clean-up. See Control Data Corp. v. S.C.S.C. Corp., 53 F.3d 930, 936 (8th Cir. 1995). Under CERCLA, once EPA identifies a hazardous site, it seeks to identify "potentially responsible parties" ("PRPs"). EPA may then initiate negotiations with PRPs to clean up a site. See 42 U.S.C. § 9622.

EPA has several options if negotiations fail. One option is to perform the clean-up itself, paying for it out of the so-called "Superfund" established by Congress for this purpose. EPA would then bring an action in federal district court under section 107 of CERCLA to recover clean-up costs from responsible parties. Id. §§ 9607(a)(4)(A), 9611(a). Another option is to seek an order from a federal district court compelling a responsible party to clean up a hazardous site. Id. § 9606(a). EPA's third option is to issue a "unilateral administrative order" ("UAO") under

for summary judgment.

3 The Court set forth a more comprehensive discussion of the CERCLA framework in its previous opinions in this case. See General Electric Co. v. Whitman, 257 F. Supp. 2d 8, 12-14 (D.D.C. 2003) (General Electric I); General Electric Co. v. Johnson, 362 F. Supp. 2d 327, 330-31 (D.D.C. 2005) (General Electric III); see also General Electric Co. v. EPA, 360 F.3d 188, 189-90 (D.C. Cir. 2004) (General Electric II).

3

section 106, ordering PRPs to clean up a site. Id. EPA may issue a UAO upon a finding "that there may be an imminent and substantial endangerment to the public health or welfare or the environment because of an actual or threatened release of a hazardous substance from a facility." Id. EPA's practice of issuing UAOs under section 106 is at the heart of GE's due process challenge in this case.

Under CERCLA, a PRP that believes it is not responsible for the clean-up faces a choice whether or not to comply with a UAO. If the PRP complies, it may seek reimbursement from other PRPs or from EPA itself upon completion of the clean-up. Id. § 9606(b). If the PRP does not comply, then EPA must file a civil action in federal district court to enforce the UAO. Id. § 9606(a). The court reviews EPA's remedy selection (i.e., its selection of a UAO) under the Administrative Procedure Act's "arbitrary and capricious" standard. See id. § 9606(b)(2)(D). The court reviews EPA's selection of the responsible party de novo. See Kelley v. EPA, 15 F.3d 1100, 1108 (D.C. Cir. 1994).

EPA may seek two kinds of monetary penalties if a PRP does not comply with a UAO.

First, if the noncomplying PRP lacks "sufficient cause" and willfully violates the order, then EPA may seek $32,500 for each day of noncompliance. Id. § 9606(b). Second, EPA may seek punitive damages up to three times "the amount of any costs incurred by the [Superfund] as a result of such failure to take proper action." Id. § 9607(c)(3). But these penalties are not automatic under CERCLA. If the reviewing court finds that the PRP had sufficient cause for noncompliance, then it cannot impose either form of penalty. Id. §§ 9606(b), 9607(c)(3). Even absent a finding of sufficient cause, moreover, the court has complete discretion as to the imposition and amount of any penalty. Id. § 9607(c)(3).

4

II. Procedural History GE filed its Complaint on November 28, 2000, and amended it on March 14, 2001. GE challenges CERCLA in two ways. First, GE alleges that the text of CERCLA itself facially violates the Due Process Clause. Am. Compl. at ¶¶ 21-28, 50, 54. Second, GE alleges that EPA's "pattern and practice" of administering the UAO regime under section 106 violates GE's due process rights. Id. at ¶¶ 17-20, 51-52, 54.

In March 2001, EPA moved to dismiss GE's Amended Complaint on jurisdictional grounds. EPA argued that this Court lacked subject matter jurisdiction because section 113(h) of CERCLA bars judicial review of a section 106 order until the clean-up is complete. This Court agreed and granted EPA's motion to dismiss. General Electric I, 257 F. Supp. 2d at 12. On appeal, however, the D.C. Circuit reversed and instructed this Court to consider the merits of GE's due process challenge on remand. General Electric II, 360 F.3d at 194. The court of appeals agreed that section 113(h) bars review of "as-applied, or particularized challenges," but concluded that section 113(h) does not bar "facial, or 'systemic'" challenges to CERCLA. Id. at 192. The court of appeals relied primarily on two cases to reach its holding: Johnson v. Robison, 415 U.S. 361, 373-74 (1974), and McNary v. Haitian Refugee Center, Inc., 498 U.S. 479, 492-94 (1991). Robison was a facial constitutional challenge to the Veterans' Readjustment Benefits Act, 38 U.S.C. § 211(a). Section 211 of that Act barred review of individual benefit determinations, but the Supreme Court held that section 211 did not extend to facial constitutional challenges to the statute. Robison, 415 U.S. at 367. McNary was a challenge to the Immigration and Naturalization Service's "pattern and practice" of administering the Special Agricultural Workers (SAW) provisions of the Immigration Reform Control Act, 8 U.S.C. §

5

1160. Section 210(e) of that Act barred judicial review "of a determination respecting an application" for SAW status, but the Supreme Court held that section 210(e) did not apply to "general collateral challenges to unconstitutional practices and policies used by the agency in processing applications." McNary, 498 U.S. at 492. Because Robison and McNary permitted facial and pattern and practice claims to proceed in the face of provisions similar to CERCLA's section 113(h), the D.C. Circuit instructed this Court to consider GE's due process challenges to CERCLA on remand. 360 F.3d at 194.

In May 2004, EPA filed a motion for summary judgment. EPA argued that GE's challenge was solely a facial one and that it had to be analyzed under the Salerno doctrine. See United States v. Salerno, 481 U.S. 739 (1987). Under Salerno, a plaintiff pursuing a facial challenge must establish that a statute is unconstitutional in every application. Id. at 745. EPA argued that, at the very least, CERCLA is constitutional in emergency situations and hence GE's facial challenge had to fail. This Court agreed and granted EPA's motion for summary judgment on GE's facial (or textual) challenge to CERCLA in March 2005. General Electric III, 362 F. Supp. 2d at 330. Considering only the text of CERCLA, this Court held that section 106 does not deprive PRPs of property interests without a sufficient judicial hearing. Id. at 338-42. Nor is section 106 as written so coercive as to deprive PRPs of procedural due process under Ex parte Young and its progeny. Id. at 342-43. This Court held that "[t]he ability of the PRP to choose, under the language of the statute, whether to comply with a section 106 order is key." Id. at 339. By choosing not to comply, a PRP receives a pre-deprivation hearing in a federal district court. That court also has the sole discretion to impose penalties for noncompliance, thereby releasing any unconstitutional pressure that might otherwise be inherent in the statute. Finally, the Court

6

held that because section 106 would be constitutional in a true emergency situation even if a PRP were deprived of a property interest before an adequate judicial hearing, GE's facial challenge failed under Salerno. Id. at 343-44.

This Court disagreed, however, that GE's facial and "pattern and practice" claims were one and the same. See id. at 333-36. EPA argued that when the court of appeals remanded the case to this Court, it had held that only facial claims cleared the section 113(h) jurisdictional hurdle. Therefore, EPA reasoned, only GE's facial challenge was before this Court on remand. To be sure, this Court recognized that some isolated language from the D.C. Circuit's opinion did support EPA's position. But the D.C. Circuit's opinion, read as a whole, firmly undercut that position. This Court noted that the court of appeals clearly distinguished between facial, systemic, and particularized challenges. Id. at 334 (citing General Electric II, 360 F.3d at 192). Moreover, the court of appeals held that facial and systemic challenges are treated similarly in analyzing jurisdictional provisions like section 113(h). The rationale, the court of appeals reasoned, is that neither facial nor systemic challenges impair the policy considerations underlying such jurisdictional bars. General Electric II, 360 F.3d at 194. Section 113(h), and provisions like it, are meant to ensure prompt responses to agency action and to avoid litigation until the responses are complete. Because neither facial nor systemic challenges would delay the clean-up at any particular site, they are treated the same for purposes of section 113(h). Id.

The Court found this key point inescapable given the cases that the court of appeals relied upon in reaching its holding. General Electric III, 362 F. Supp. at 335 (citing General Electric II, 360 F.3d at 192-93). For example, the court of appeals relied in significant part on McNary, a pattern and practice case challenging INS's administration of certain immigration law provisions.

7

In McNary, the Supreme Court held that such a pattern and practice claim cleared a jurisdictional hurdle analogous to section 113(h). See McNary, 498 U.S. at 492. Because the Supreme Court allowed such a claim to proceed in McNary, the court of appeals reasoned, GE should be permitted to proceed with its claim here as well. See General Electric II, 360 F.3d at 193. Bound by the court of appeals's decision to remand GE's pattern and practice claim, this Court thus rejected EPA's argument that GE's facial and pattern and practice challenges were exactly the same. General Electric III, 362 F. Supp. 2d at 334-36. At the time of the Court's decision, GE had not yet been able to develop the factual record to support its claim that EPA's administration of section 106 denies PRPs procedural due process in fact, even if the statute does not do so in theory. The Court, therefore, permitted GE to proceed with discovery on that claim.

Discovery lasted over two years, closing in August 2007. During the course of discovery, the Court issued rulings on motions to compel in September 2006 and February 2007. See General Electric v. Johnson, No. Civ.A.00-2855, 2006 WL 2616187 (D.D.C. Sept. 12, 2006) (General Electric IV); General Electric v. Johnson, No. Civ.A.00-2855, 2007 WL 433095 (D.D.C. Feb. 5, 2007) (General Electric V). The parties produced tens of thousands of documents, conducted numerous fact and expert depositions, and exchanged reports and rebuttal reports for six experts. GE Memorandum in Support of its Motion for Summary Judgment ("GE Mem.") at 9-10. The factual record in this case is now, to put it bluntly, extensive.

Before the Court at this time are the parties' cross-motions for summary judgment based on that record. The crux of GE's argument is that EPA's administration of section 106 violates the Due Process Clause because, applying the framework of Mathews v. Eldridge, 424 U.S. 319 (1976), section 106 as administered deprives PRPs of protected liberty and property interests

8

without a hearing, and because PRPs are forced to comply with UAOs, thereby offending Ex parte Young, 209 U.S. 123 (1908). EPA counters that GE has not suffered the pre-hearing deprivations it alleges and that such deprivations do not require greater process than EPA already provides consistent with CERCLA. EPA further argues that an Article III court conducts de novo review of penalties available for UAO noncompliance before any penalties can attach, thereby curing any coercive practices allegedly undertaken by EPA.

LEGAL STANDARD

Summary judgment is proper when the pleadings and evidence demonstrate that "there is no genuine issue as to any material fact and that the moving party is entitled to judgment as a matter of law." Fed. R. Civ. P. 56(c). The moving party may successfully support its motion by "informing the district court of the basis of its motion, and identifying those portions of 'the pleadings, depositions, answers to interrogatories, and admissions on file, together with the affidavits, if any,' which it believes demonstrate the absence of a genuine issue of material fact." Celotex Corp. v. Catrett, 477 U.S. 317, 323 (1986) (quoting Fed. R. Civ. P. 56(c)). The court must regard the non-movant's statements as true and accept all evidence and make all inferences in the non-movant's favor. See Anderson v. Liberty Lobby, Inc., 477 U.S. 242, 255 (1986). The moving party may point to the absence of evidence proffered by the non-movant to succeed on summary judgment. See Celotex, 477 U.S. at 322. Where, as here, the Court would be the trier of fact if the case were to proceed to trial, "the 'Court is not confined to deciding questions of law, but also may . . . draw a derivative inference from undisputed subsidiary facts, even if those facts could support an inference to the contrary, so long as the inference does not depend upon an evaluation of witness credibility.'" OAO Alfa Bank v. Ctr. for Public Integrity, 387 F. Supp. 2d

9

20, 39 (D.D.C. 2005) (quoting Cook v. Babbitt, 819 F. Supp. 1, 11 & n.11 (D.D.C. 1993)); see also Ramallo v. Reno, 931 F. Supp. 884, 888 (D.D.C. 1996).4 ANALYSIS

I. Applicability of Salerno.

EPA argues that the standard from United States v. Salerno, 481 U.S. 739 (1987), should apply to GE's pattern and practice claim. EPA Memorandum in Support of its Motion for Summary Judgment ("EPA Mem.") at 17-19. Under Salerno, a plaintiff facially challenging a statute's constitutionality must establish that the statute is unconstitutional in every application. Salerno, 481 U.S. at 745. Notwithstanding some debate within the Supreme Court on this "no set of circumstances" prong of Salerno, see City of Chicago v. Morales, 527 U.S. 41, 55 n.22 (1999), the D.C. Circuit continues to apply Salerno to facial challenges based on statutory text. See Nebraska v. EPA, 331 F.3d 995, 998 (D.C. Cir. 2003); Chemical Waste Mgmt. v. EPA, 56 F.3d 1434, 1437 (D.C. Cir. 1995). Indeed this Court did so earlier on GE's textual challenge to CERCLA. See General Electric III, 362 F. Supp. 2d at 343-44. EPA argues now that GE's pattern and practice claim is tantamount to a facial challenge, and that the remedy GE seeks through the present claim -- a remedy "that would apply universally to EPA's conduct" -- is no different than the remedy GE sought in the facial challenge this Court considered (and rejected)

4 Discovery has closed in this case and the factual record is complete. At the motions hearing before this Court on October 14, 2008, counsel for GE and EPA recognized that some "details" of the factual record remain in dispute. See Transcript of Motions Hearing ("Tr.") at 75-77, 89-91. But neither counsel maintained that these disputes would stand in the way of summary judgment. Indeed, both counsel stated that the disputed facts are not material. Moreover, both counsel recognized that the legal and factual issues in this case are not wellsuited for resolution at trial. See Tr. at 75 ("I'm not sure the Court is ever going to have much more than it has right now."); id. at 89-90 ("[W]ell, is there a need for a factual trial here[?] [T]he nightmare of continuing this already long proceeding much longer, I think there is no need. . . .").

10

previously. EPA Mem. at 17. EPA assures this Court that other federal courts apply Salerno to similar facial challenges to agency rules and policies. Therefore, EPA argues, Salerno should apply to GE's pattern and practice claim just as it applied to GE's previously-rejected facial claim based on the text of CERCLA.

The Court will not apply the Salerno standard to GE's pattern and practice claim.5 Salerno applies to challenges based on hypothetical applications of a statute, not to challenges based on the particular facts of a case. See Salerno, 481 U.S. at 745 & n.3; Reno v. Flores, 507 U.S. 292, 300-01 (1991) (applying Salerno because plaintiffs did "not challenge [an INS regulation's] application in a particular instance; [the regulation] had not yet been applied in a particular instance -- because it was not yet in existence -- when [plaintiffs'] suit was brought."). But GE's challenge here does not hypothesize how section 106 could be applied in the abstract. Rather, GE's pattern and practice claim is based on how EPA actually administers section 106. GE supports its claim -- drawn from an extensive record after considerable discovery -- with evidence about its own experiences and the experiences of other PRPs, including a raft of deposition testimony, expert reports, and EPA manuals. GE's claim is therefore more analogous to the McNary line of cases, which involved a pattern and practice claim based on INS's actual administration of the SAW provisions of the Immigration Reform Control Act. See McNary, 498 U.S. at 487-90. No court in that case -- beginning with the District Court and ending with the Supreme Court -- even cited to Salerno. This is not the type of generally-discouraged facial challenge to the text of a statute to which Salerno is generally applied, see General Electric III,

5 The Court previously indicated, but did not decide, that Salerno might not apply to GE's pattern and practice claim. See General Electric III, 362 F. Supp. 2d at 343 & n.9.

11

362 F. Supp. 2d at 343 n.9, and hence the Court will not apply Salerno to GE's pattern and practice claim, which has already been distinguished from a pure facial challenge. II. Standing EPA next argues that GE lacks standing to raise injuries based on noncompliance with a UAO.6 See EPA Reply Memorandum in Support of its Summary Judgment Motion ("EPA Rep.") at 9-11. Standing "requires a plaintiff to demonstrate the now-familiar elements of injury in fact, causation, and redressability." Lance v. Coffman, 549 U.S. 437, 439 (2007) (citing Lujan v. Defenders of Wildlife, 504 U.S. 555, 560-561 (1992)). Here, EPA focuses on injury in fact. EPA points out that GE has complied with every UAO it has been issued but that GE nevertheless concentrates its Mathews v. Eldridge analysis, discussed infra at § IV, on deprivations triggered by noncompliance with a UAO. EPA argues that "[i]f such injuries never occur to GE because GE never disobeys a UAO, then GE does not have standing to base a due process challenge upon them." EPA Rep. at 11.7 The injuries relevant to standing are not, as EPA asserts, injuries arising from an act of noncompliance. The thrust of GE's argument is that EPA's administration of section 106 so amplifies the risks and costs of noncompliance that no rational PRP would choose not to comply with a UAO. The argument thus responds to this Court's holding in General Electric III, where the Court rejected GE's textual challenge to section 106 in part because a PRP theoretically can

6 EPA does not argue that GE lacks standing to challenge EPA's pattern and practice of issuing UAOs as unconstitutionally coercive under Ex parte Young.

7 This standing argument is distinct from a different challenge by EPA to GE's standing to assert a pattern and practice claim at all, which the Court previously rejected. See General Electric III, 362 F. Supp. 2d at 336-37.

12

choose not to comply with a UAO. 362 F. Supp. 2d at 339. If a PRP chooses not to comply, EPA cannot compel compliance without judicial intervention, so a PRP can always obtain a pre- deprivation hearing. Id. The Court recognized, however, that if PRPs are "deprived of any meaningful choice to refuse to comply with [a UAO], that might pose a different situation in assessing whether a deprivation of property has occurred." Id. Responding to the Court's caveat, GE now points to noncompliance-based deprivations -- like a PRP's decreased market value -- as evidence that PRPs are effectively forced to comply with UAOs. Moreover, even if PRPs are not forced to comply, GE uses noncompliance-based deprivations as evidence that PRPs suffer pre- hearing deprivations once a UAO is issued, no matter what they choose to do.

The injury inquiry for standing is limited to whether GE has "set forth" sufficient "specific facts" to support its claim that EPA deprives GE of liberty or property without due process by issuing UAOs to GE. See Lujan, 504 U.S. at 561 (establishing a plaintiff's burden of establishing standing at the summary judgment stage).8 EPA does not dispute GE's showing that EPA has issued 68 UAOs to GE and that GE has complied with all of them. See EPA Memorandum in Opposition to GE's Summary Judgment Motion ("EPA Opp.") at 20. And EPA concedes that "GE [can] establish injury for standing purposes where it does comply with a UAO and incurs response costs." EPA Rep. at 11 (emphasis in original). Therefore, GE has standing to bring this pattern and practice claim and may rely on noncompliance-based injuries to the extent it uses such injuries to establish that PRPs are deprived of a meaningful choice not to

8 Courts do not assess the merits of a plaintiff's claim when considering standing. See Emergency Coalition to Defend Educ. Travel v. Dep't of Treasury, 545 F.3d 4, 10 (D.C. Cir. 2008) ("In considering standing, we must assume the merits in favor of the party invoking our jurisdiction."). Therefore, at this stage of the analysis, the Court presumes that the deprivations GE asserts can form the basis for a due process claim.

13

comply with a UAO or that PRPs suffer pre-hearing deprivations regardless of whether they comply. III. Ex Parte Young GE's first claim is that EPA's pattern and practice of administering section 106 violates procedural due process under Ex parte Young, 209 U.S. 123 (1908). Under Ex parte Young and its progeny, a statutory scheme that imposes penalties on those seeking judicial review is unconstitutional if "the penalties for disobedience are by fines so enormous . . . as to intimidate the company and its officers from resorting to the courts to test the validity of the legislation." Id. at 147. Statutes prescribing significant penalties for violators are not per se unconstitutional, however. Rather, if a party challenging a penalty or a statute in "good faith" may not be penalized, then the statute may be constitutional under Ex parte Young. See Reisman v. Caplin, 375 U.S. 440, 446-47 (1964). Here, under section 106(b)(1), a district court must review any penalty EPA seeks for noncompliance with a UAO. The court may only impose penalties if a PRP has failed to comply "without sufficient cause"; moreover, even if the court finds that the noncomplying PRP lacks "sufficient cause," the court still has full discretion in deciding whether to impose a civil penalty, or punitive damages, and if so, in what amount. See 42 U.S.C. §§ 9601(b)(1), 9607(c)(3). Courts considering Ex parte Young challenges to sections 106 and 107 have held that the sufficient cause defense operates as a good faith safe harbor, adequately curing any constitutional problems that steep CERCLA fines and penalties could otherwise create. See Employers Ins. of Wausau v. Browner, 52 F.3d 656, 664 (7th Cir. 1995); Solid State Circuits, Inc. v. EPA, 812 F.2d 383, 391 (8th Cir. 1987); Wagner Seed Co. v. Daggett, 800 F.2d 310, 316 (2d Cir. 1986); United States v. Capital Tax Corp., Civ.A.No. 04-4138, 2007 WL 488084 at *6

14

(N.D. Ill. Feb. 8, 2007). Applying this reasoning in General Electric III, this Court rejected GE's facial Ex parte Young challenge based on the text of CERCLA's UAO provisions, although leaving room for GE to further develop an Ex parte Young challenge in its pattern and practice case. See 362 F. Supp. 2d at 342-43.

GE's pattern and practice claim under Ex parte Young fails for the same reasons previously explained. GE points to four aspects of EPA's UAO policy that supposedly intimidate PRPs into compliance: EPA seeks maximum penalties for noncompliance; EPA seeks multiple penalties for violations at a single UAO site; EPA rejected Justice Department advice that EPA should impose a cap on daily penalties; and EPA labels noncomplying PRPs as "recalcitrant." But no matter what EPA arguably does or seeks, a judge ultimately decides what, if any, penalty to impose.9 Indeed, courts routinely exercise their discretion in determining what kinds of daily or punitive penalties to impose for violations of section 106. See, e.g., United States v. Capital Tax Corp., Civ.A.No. 04-4138, 2007 WL 2225900 at *13 (N.D. Ill. Aug. 1, 2007) (awarding $750 daily penalty and zero punitive damages); United States v. Barkman, Civ.A.No. 96-6395, 1998 WL 962018 at *18 (E.D. Pa. Dec. 17, 1998) (awarding $100 daily penalty).10

9 In a footnote, GE cited to three cases for the proposition that a statute may run afoul of Ex parte Young even if a penalty scheme is not automatic or self-executing. See GE's Memorandum in Opposition to EPA's Summary Judgment Motion ("GE Opp.") at 46 n.29 (citing Chicago, M. & St. P. Rwy. Co. v. Polt, 232 U.S. 165, 167-68 (1914); Missouri Pac. Rwy. Co. v. Tucker, 230 U.S. 340, 346-51 (1913); Missouri Pac. Rwy. Co. v. Nebraska, 217 U.S. 196, 207- 08 (1910)). But at the motions hearing, counsel for GE conceded that none of these cases stand squarely for that proposition. See Tr. at 48.

10 GE cites to United States v. Tannery, Civ.A.No. 91-693, 1992 WL 1458802 (N.D. Tex.

Dec. 7, 1992), and United States v. LeCarreaux, Civ.A.No. 90-1672, 1992 WL 108816 (D.N.J. Feb. 19, 1992), to support its argument that courts do not exercise their judgment in awarding punitive or treble damages. GE Rep. at 32-33 & n.34. In Tannery, the court noted in a one-page opinion that the defendant had repeatedly refused to participate in proceedings before the court.

15

Nor does EPA's pattern and practice regarding UAOs prevent federal courts from exercising their discretion. In General Electric III, this Court affirmed as satisfying due process requirements CERCLA's judicial review provisions, which provide for de novo review of liability and review of remedy selection under an "arbitrary and capricious" standard. 362 F. Supp. 2d at 341-42. Now, GE argues that EPA's pattern and practice of compiling the administrative record creates an incomplete basis for a reviewing judge's decision, thereby impairing the judge's discretion. Specifically, GE argues that EPA's practice of excluding irrelevant and pre-decisional deliberative documents from the administrative record results in a skewed record. See GE Mem. at 57. But irrelevant documents should be excluded from the record -- the record should only include documents that the agency "directly or indirectly considered." See Maritel, Inc. v. Collins, 422 F. Supp. 2d 188, 196 (D.D.C. 2006); see also James Madison Ltd. by Hecht v. Ludwig, 3 F.3d 1085, 1095 (D.C. Cir. 1996). And "an agency generally may exclude material that reflects internal deliberations." Maritel, 422 F. Supp. 2d at 196. Moreover, GE has not demonstrated that EPA actually has a pattern and practice of excluding documents that should be included in the record. Although GE has provided examples

Indeed, the court noted that "even if the summary judgment were not to be granted, the court would grant default judgment in favor of plaintiff as a sanction . . . ." 1992 WL 1458802 at *1. Similarly, in LeCarreaux, the court noted that because it had "received no opposition," it adopted the recommendation of a Magistrate Judge, who himself adopted the government's proposed findings of fact and conclusions of law in their entirety. 1992 WL 108816 at *1. Unsurprisingly, the government proposed the maximum damages allowed under CERCLA. To be sure, both cases represent instances of courts adopting EPA's suggested treble and punitive damages. But in each case, the defendants either failed to oppose EPA's requested damages or altogether failed to participate in the enforcement proceedings. These cases, therefore, are unique in both the brevity of their analysis and the lack of engagement displayed by the defendants, and do not represent generally a failure by the courts to exercise discretion regarding the imposition of penalties for section 106 violations.

16

from two sites where an administrative record was arguably incomplete, two anecdotal examples do not form a pattern and practice. Cf. Coser v. Moore, 739 F.2d 746, 751-52 (2d Cir. 1984); Int'l Bhd. of Teamsters v. United States, 431 U.S. 324, 336 (1977). And even if the record was incomplete in those two instances, safeguards within CERCLA ensure that a reviewing court could have supplemented the record. See 42 U.S.C. § 9613(j)(1). The Court is not persuaded, then, that EPA abuses its stewardship of the administrative record and hence rejects GE's assertion that courts are prevented from exercising their discretion in awarding or denying penalties and fines.

Finally, GE argues that EPA has purposefully muddied the contours of the sufficient cause defense, thus keeping PRPs guessing whether they have sufficient cause in deciding not to comply with a UAO. To be sure, Ex parte Young problems may arise when the imposition of penalties turns on the interpretation of an imprecise legal standard. See Solid State Circuits, 812 F.2d at 391. GE contends that Solid State Circuits provided an "unambiguous call on EPA to provide the type of guidance necessary to provide the sufficient cause defense with . . . requisite certainty." GE Mem. at 55.11 GE argues that EPA has ignored this judicial command, choosing instead to keep the sufficient cause defense vague. Solid State Circuits, however, does not go as far as GE wishes. There, the Eighth Circuit recognized that because CERCLA had only recently been enacted, the term "sufficient cause" had not yet been developed by EPA or the courts. 812 F.2d at 391. The court noted that "as the EPA and the courts face concrete cost recovery and

11 GE's citation to Pearson v. Shalala, 164 F.3d 650, 660 (D.C. Cir. 1999), in support of this same point is unhelpful. Pearson considered an agency's failure to articulate reasons for taking action, not an agency's failure to give definitional content to an analogous "safe harbor" term. GE does not argue that EPA routinely fails to articulate reasons for issuing UAOs.

17

treble damage cases, section 107(c)(3) of CERCLA [i.e., the sufficient cause defense] will develop accordingly." Id. at 391 n.11. In the meantime, the court bypassed potential Ex parte Young problems by shifting the burden -- rather than requiring the noncomplying PRP to show that it had sufficient cause not to comply, the court required EPA to show that the challenging PRP lacked sufficient cause. Id. at 392. Although the court noted that EPA could have avoided this kind of burden-shifting by enacting detailed regulations regarding the sufficient cause defense, the court did not issue the kind of "unambiguous call" that GE suggests. In the end, courts, not agencies, provide the conclusive interpretations of imprecise legal terms like "sufficient cause." See Marbury v. Madison, 1 Cranch 137, 177 (1803) ("It is emphatically the province and duty of the judicial department to say what the law is."). Indeed, since Solid State Circuits, district courts have given meaning to the term. See, e.g., Capital Tax Corp., 2007 WL 2225900 at *13; Barkman, 1998 WL 962018 at *17. Hence, while EPA's unwillingness to issue guidance regarding the meaning of sufficient cause may be poor policy, it does not constitute a coercive pattern and practice. In sum, GE has not demonstrated that EPA practices -- whether considered individually or together -- are unconstitutionally coercive under Ex parte Young, and that due process challenge must accordingly fail. IV. Mathews v. Eldridge That brings us to the primary due process challenge GE asserts, which must be assessed within the framework of Mathews v. Eldridge. Both EPA and GE propose broad rules for certain categories of deprivations or government actions. But courts approach due process claims with scalpels, not cleavers. See Hannah v. Larche, 363 U.S. 420, 442 (1960) ("'Due process' is an elusive concept. Its exact boundaries are undefinable, and its content varies according to specific

18

factual contexts."). Moreover, the parties' proposed rules are not adequately supported by the cases they cite, and hence the Court will not adopt them.

GE argues that due process requires a trial-type hearing for non-emergency, "adjudicatory" agency decisions. GE Mem. at 10; GE Reply Memorandum in Support of its Summary Judgment Motion ("GE Rep.") at 23. GE insists that this Court held in General Electric IV that issuance of a UAO is an adjudicatory decision. See 2006 WL 2616187 at *15 n.3. Leaving aside whether issuance of a UAO is "adjudicatory," GE's argument fails because it is a legal proposition without legal support. Certainly, adjudicative decisions require more individualized process than do rule-making decisions. See United States v. Fla. East Coast Ry., 410 U.S. 224, 244 (1973). And, to be sure, some cases contain dicta suggesting that the more an agency's decision resembles a judicial decision, the more the agency's procedures should resemble judicial procedures. See, e.g., Hannah, 363 U.S. at 442; Logan v. Zimmerman Brush Co., 455 U.S. 422, 433-34 (1982). But GE has not cited to any cases holding that trial-type hearings are required for adjudicatory decisions, and this Court therefore will not adopt GE's proposed rule.

EPA, on the other hand, argues that "consequential" deprivations cannot form the basis for a due process challenge at all. EPA Opp. at 9-10. EPA points out that some of the deprivations GE alleges only occur because the marketplace values a PRP less after EPA issues a UAO to that PRP. In support of its proposed rule, EPA relies on O'Bannon v. Town Court Nursing Center, 447 U.S. 773 (1980), and Blum v. Yaretsky, 457 U.S. 991 (1982). O'Bannon rejected a claim by nursing home residents that they were entitled to a hearing before the government decertified a nursing home. Once the government decertified the home, residents

19

could no longer use their government benefits there. The Court reasoned that the residents did not have a due process claim because they were only incidentally affected by the government's decision. 447 U.S. at 787. Unlike the incidentally-affected residents in O'Bannon, however, UAOs are issued directly to PRPs, who are themselves affected by government action. Blum rejected the claim of Medicaid recipients that they were entitled to greater process before a state- subsidized hospital discharged them. The Court reasoned that the recipients did not have a due process claim because actions by the hospital did not constitute "state action." 457 U.S. at 1004- 05. But UAOs are issued by EPA itself and indisputably constitute state action. Hence, neither O'Bannon nor Blum is applicable here.

Connecticut v. Doehr, 501 U.S. 1 (1991), on the other hand, does apply, and it requires this Court to reject EPA's argument. In Doehr, a statute provided for ex parte prejudgment attachment of real estate without notice to the affected party. The Supreme Court noted that attachment "clouds title [and] impairs the ability to sell or otherwise alienate the property" and then held that these "consequences" were sufficient to merit due process protection. Id. at 11-12. Doehr therefore settles the question whether consequential injuries may warrant due process protection, even if courts must carefully assess whether any particular consequential deprivation is actionable.

Finally, EPA argues that mere "diminution in value" does not give rise to a due process claim. EPA Mem. at 24-25. EPA cites only to Andrus v. Allard, 444 U.S. 51, 65-66 (1979), in support of this proposition. In Andrus, the Supreme Court rejected a takings claim after the government banned the sale of eagle feathers, reasoning that "loss of future profits . . . provides a slender reed upon which to rest a takings claim." Id. at 65 (emphasis added). But GE does not

20

bring a takings (i.e., just compensation) claim; it brings a due process claim. "'Property' as used in the Just Compensation Clause is defined much more narrowly than in the due process clauses." Corn v. City of Lauderdale Lakes, 95 F.3d 1066, 1075 (11th Cir. 1996). Andrus is therefore inapposite and the Court will not adopt this EPA proposal, either.

Having rejected the parties' proposed easy solutions, the Court must apply the well-settled Mathews v. Eldridge framework for due process claims. Under Mathews, courts balance the private interests, the governmental interests, and the risk of error to determine whether existing procedures are adequate to satisfy constitutional due process requirements. 424 U.S. at 335. Before reaching the balancing stage, however, this Court must determine that issuance of a UAO deprives PRPs of property or liberty. See American Mfrs. Mut. Ins. Co. v. Sullivan, 526 U.S. 40, 59 (1999) ("The first inquiry in every due process challenge is whether the plaintiff has been deprived of a protected interest in 'property' or 'liberty.'").

PRPs arguably suffer deprivations without adequate due process under the section 106 UAO regime in three ways. First, the mere issuance of a UAO -- before the PRP decides whether to comply -- arguably impacts the PRP's stock price and brand value. Second, if a PRP chooses not to comply with a UAO, then the PRP arguably suffers increased damage to its stock price and brand value. Noncompliance also may impair a PRP's relationships with regulatory personnel and stakeholders, potentially depriving the PRP of a liberty interest. Finally, if a PRP chooses to comply with a UAO -- and GE argues here that PRPs are forced to do so -- then the PRP must incur the costs of clean-up. The parties agree that clean-up costs constitute a protected property

21

interest; the parties disagree, however, on whether PRPs are forced to comply.12 A. Mathews Prong One 1. Deprivations at UAO Issuance The parties dispute whether GE is required to show that a PRP suffers a deprivation when a UAO is issued or when a PRP elects not to comply with it. EPA contends that GE is required to show that PRPs suffer injuries at issuance. EPA Opp. at 12. GE argues that the relevant deprivations are those at noncompliance because the parties agree that there is a deprivation at compliance. GE Rep. at 12. Nevertheless, GE submits that even though it focused on costs at noncompliance, its expert, Dr. John Geweke, has established that "issuance of a UAO would cause an immediate deprivation, whether or not a PRP complied." Id. Issuance of a UAO arguably causes two kinds of property deprivations: (1) reduced stock price and increased cost of financing; and (2) damage to the PRP's brand value.

Neither stock price/cost of financing impacts nor brand value depletions are protected property interests upon mere issuance of a UAO. Damage to stock price and brand value could occur anytime any agency takes any action that the market interprets as adverse against a company. For example, if the Securities and Exchange Commission announced an investigation of a company for securities fraud, that company's stock price and brand value might suffer. The same would likely occur if the Food and Drug Administration announced that it was recalling a company's drug because of health concerns. But surely neither of those situations triggers due

12 GE also argues in its briefs that the cost of simply responding to a UAO deprives PRPs of a protected interest. See GE Mem. at 23-25. At the motions hearing, however, counsel for GE conceded that GE does not rely on the cost of responding as a deprivation under the Mathews analysis. See Tr. at 33-34.

22

process concerns. Indeed, announcement of the very pre-issuance process that GE seeks would adversely impact a PRP's stock price and brand value under GE's argument. The Court would be inviting a host of unfounded due process claims were it to conclude that stock price and brand value damage upon mere issuance of a UAO constitute actionable property deprivations. Counsel for GE recognized the need for line-drawing at the October 14, 2008 motions hearing before the Court, but counsel proposed drawing the line at noncompliance, not at issuance. Tr. at 17-19. Accordingly, GE has not shown that PRPs suffer any deprivation of a protected property interest simply upon issuance of a UAO.

2. Deprivations if a PRP Does Not Comply GE argues that PRPs are deprived of property and liberty if they elect not to comply with a UAO. GE points to three kinds of deprivations at noncompliance. GE again asserts that PRPs suffer injuries relating to their stock price and cost of financing. GE also reprises its argument that PRPs incur brand value damage. Finally, GE argues that PRPs' relationships with stakeholders and EPA itself are impaired if PRPs do not comply, thereby depriving PRPs of a protected liberty interest.

As a threshold matter, GE's stock price/cost of financing and brand value reduction arguments do not suffer from the same line-drawing problems at the noncompliance stage that doomed them at the issuance stage. As counsel for GE put it, "the axe falls" at noncompliance. See Tr. at 17. That is true because a PRP is normally branded a recalcitrant actor once it decides not to comply -- a label that enhances the harm to stock price and brand value, and also supposedly exposes PRPs to greater penalties, increases permitting times, bars PRPs from certain EPA programs, and impacts PRPs' relationships with certain stakeholders. The deprivations are

23

thus both qualitatively and quantitatively different upon noncompliance than upon issuance of a UAO. These differences are sufficient to distinguish the impact of noncompliance with a UAO from the theoretical impacts of other types of adverse government action.

GE bases its stock price/cost of financing argument on the expert reports from Drs. John Geweke and Jason Johnston. Dr. Geweke, faced with a dearth of data regarding the impact of noncompliance on stock price, instead examines the impact of "special notice letters" on stock price for 290 publicly-traded companies in 1994. Geweke Report at ¶ 13. Dr. Geweke also looks at the impact of special notice letters on bond yields for 89 companies from 1995 to 1998. Id. EPA usually sends special notice letters to PRPs to initiate settlement discussions regarding clean-up of a hazardous site. EPA's Statement of Facts ("EPA SOF") at ¶¶ 19-20. If settlement fails, then EPA normally issues a UAO. Id. at ¶ 27. Therefore, Dr. Geweke concludes, special notice letters associate a PRP with a contingent liability and negatively impact stock price and cost of financing. Geweke Report at ¶ 6. Using two multivariate regression analyses, Dr. Geweke then extrapolates what the impact on stock price and cost of financing would be if a PRP chose not to comply with a UAO. Logically, contingent liabilities associated with noncompliance are greater than those associated with a special notice letter because EPA may seek treble and punitive damages if a PRP chooses not to comply. Therefore, Dr. Geweke concludes that "[a] firm's decision not to comply with a UAO . . . subjects the firm to a loss in market value that is 5.9 times greater than its loss in market value arising from a special notice letter." Id. at ¶ 19. He calculates that the average decrease in market value based on noncompliance would be $76.4 million, compared with an average $12.9 million decrease in market value based on receipt of a special notice letter. Id. at ¶ 158 & Figure 7.

24

Dr. Johnston takes a different approach. Like Dr. Geweke, Dr. Johnston does not have sufficient data to analyze the impact of noncompliance on stock price. Rather than focusing on special notice letters, however, Dr. Johnston also looks at the impact of other "negative environmental events" on a company's stock price, such as EPA's filing of a complaint against a company or a company's settlement of a Justice Department lawsuit. See Johnston Report at ¶¶ 34-37. Dr. Johnston does not conduct regression analyses. Instead, he surveys four existing "event studies" that test the relationship between negative environmental news and a stock price decrease. See id. at ¶ 29 (explaining the event study methodology). Dr. Johnston concludes that these event studies provide "strong evidence" that noncompliance with a UAO will result in "statistically significant negative abnormal stock market returns" for a targeted PRP. Id. at ¶ 33. Because Dr. Johnston does not conduct his own event study, he does not project the size of a potential noncompliance-based stock price decrease.

EPA asserts that Dr. Geweke's analysis fails under the standard governing the admissibility of expert opinion evidence. See Daubert v. Merrell Dow Pharms., Inc., 509 U.S. 579 (1993). The proffering party must demonstrate the admissibility of expert testimony by a preponderance of the evidence under Daubert and its progeny. See Meister v. Med. Eng'g Corp., 267 F.3d 1123, 1127 n.9 (D.C. Cir. 2001). EPA first challenges Dr. Geweke's methodology. EPA argues that he should have considered empirical evidence concerning actual instances of UAO noncompliance rather than conducting a "counterfactual" analysis based on the impacts of special notice letters. The Court certainly agrees that analysis of hard data is generally preferable to projected data, and would be here. But empirical evidence concerning actual instances of UAO noncompliance is scarce because very few publicly-traded firms have chosen not to comply

25

with UAOs. EPA's own rebuttal expert, Dr. Donald Siegel, testified that making inferences based on related events was appropriate absent sufficient data. See Deposition of Donald Siegel at 173:19-175:3. Moreover, regression analyses are commonly used as evidence in cases requiring "extrapolati[on] from a sample of known data." League of United Latin Amer. Citizens v. Perry, 548 U.S. 399, 467 (2006); see also Daniel L. Rubinfeld, Reference Guide on Multiple Regression, in Reference Manual on Scientific Evidence 181-85 (Federal Judicial Center 2d ed. 2000). Therefore, a regression analysis is an acceptable method of demonstrating the impact of UAO noncompliance on stock price absent sufficient empirical data.

EPA next questions the assumptions underlying Dr. Geweke's analysis. See EPA Opp. at 13-14. EPA highlights three supposedly flawed assumptions: that firms that are issued UAOs at multiple sites will decide not to comply with any UAO; that firms that choose not to comply with a UAO will be assessed maximum daily penalties for a five-year period (i.e., that EPA will not bring an enforcement action until the statute of limitations is set to expire); and that the markets do not discount contingent liabilities based on the likelihood of enforcement, the chance that a PRP will successfully defend an enforcement action, or the possibility that a court will impose a smaller penalty than EPA requests. Although not without some force, none of these alleged flaws is sufficient to disqualify Dr. Geweke's report. His model does calculate potential impacts if a PRP receives multiple UAOs and does not comply with any of them, but he reaches his main conclusion -- that the average decrease in market value based on UAO noncompliance is $76.4 million -- by assuming that a PRP decides not to comply with a single UAO. See Geweke Report at ¶ 158. Moreover, EPA does not meaningfully dispute GE's assertion that "[i]t is the government's practice in bringing suit for UAO noncompliance to seek the statutory maximum in

26

penalties and treble damages." See GE's Statement of Undisputed Facts ("GE SUF") at ¶ 125 and EPA's Response to GE SUF at ¶ 125. While the Court is persuaded that EPA has a policy of seeking maximum damages, GE has not offered any evidence to suggest that EPA routinely waits as long as possible before filing a cost recovery or enforcement action. The Court, then, will not hazard a recalculation of Dr. Geweke's formula. But it is clear that shorter delays in filing cost recovery or enforcement actions would result in smaller potential penalties, smaller contingent liabilities, and, therefore, an average market value reduction of less than $76.4 million. Similarly, while Dr. Geweke specifically acknowledges that the market may discount potential liabilities based on possible defenses or a court's decision to impose smaller penalties than EPA seeks, he argues that "the difference would be one only of magnitude." Geweke Report at ¶ 27. True enough, but because he did not attempt to quantify how much markets would discount potential liabilities, his $76.4 million calculation is necessarily too high. The question at this stage of the inquiry, however, is whether a deprivation exists, not how large that deprivation is. See Sullivan, 526 U.S. at 59. Although imperfect, Dr. Geweke's analysis demonstrates that PRPs are deprived of some portion of the market value of their stock if they decide not to comply with a UAO.

EPA also asserts that Dr. Johnston's report fails the Daubert standard. EPA Opp. at 14-

15. EPA again contends that Dr. Johnston should have examined hard data rather than similar "negative environmental events." But Dr. Johnston, like Dr. Geweke, could not examine actual instances of noncompliance with UAOs because there are not enough to comprise a reliable data set. EPA next argues that Dr. Johnston should have conducted his own event study rather than relying on previously-published ones. That argument has merit. Although none of the cases

27

EPA cites support its argument that Daubert requires experts to conduct their own event studies,13 Dr. Johnston's reliance on other economists' analyses of other (albeit comparable) events certainly diminishes the probative value of his expert report. Dr. Johnston equates UAO noncompliance to these other events -- such as EPA's announcement of a settlement or a newly- filed lawsuit -- by reasoning that "UAOs are 'one of the primary enforcement tools to obtain a . . . response by PRP's,' and are used to 'provide an incentive for PRP's to settle' or to 'force commencement of work at the site when settlement cannot be reached.'" Johnston Report at ¶ 32 (quoting USEPA, OSWER Directive No. 9833.0-1a-3 (March 1990)) (ellipsis in original). But Dr. Johnston does not explain why UAOs are like any of the specific events that the other event studies examine. His conclusion -- that negative environmental events are associated with large stock market decreases and that a UAO is a negative environmental event -- is equally vague. Thus, Dr. Johnston's report largely duplicates Dr. Geweke's conclusions, only in a more general and imprecise way. The Court will not give significant weight to vague, cumulative evidence. See Fed. R. Evid. 403. Hence, the Court finds that Dr. Johnston's report does not advance GE's argument that noncompliance with a UAO causes a decrease in stock price.

The second proffered deprivation is GE's contention that a PRP's brand asset is damaged if it does not comply with a UAO. Intangible assets, like a brand name, may receive due process protection. See Delaware, Lackawanna & W. R.R. Co. v. Pennsylvania, 198 U.S. 341, 359 (1905). GE again relies on Dr. Johnston, who, without the benefit of adequate data, once more offers a general, conclusory opinion:

13 See EPA Opp. at 15 (citing In re N. Telecom Ltd. Secs. Litig., 116 F. Supp. 2d 446 (S.D.N.Y. 2000); Garay v. Mo. Pac. R.R., 60 F. Supp. 2d 1168 (D. Kan. 1999); and In re Oracle Secs. Litig., 829 F. Supp. 1176 (N.D. Cal. 1993)).

28

For firms that have made substantial investments in establishing a corporate brand name for strong environmental performance, events that weaken the firm's brand name as an environmentally responsible actor cause a definite and potentially large decrease in the value of the corporate environmental brand asset . . . .

Johnston Report at ¶ 15. This time, Dr. Johnston does not apply any scientific methodology to test his conclusion. His report, however, is salvaged because no one disputes his conclusion. Not even Dr. Siegel, EPA's rebuttal expert, takes issue with it. See, e.g., Siegel Dep. at 70:11- 70:14, 72:10-72:16, 73:6-73:10, and 79:13-79:17. More controversial is how much noncompliance with a UAO decreases the value of the brand asset. While Dr. Johnston opines that the decrease is "potentially large," EPA correctly points out that Dr. Johnston has no data to support his conclusion. But, again, at this stage of the Mathews analysis, only the existence of a pre-hearing deprivation matters, not its magnitude. See Sullivan, 526 U.S. at 59. Therefore, although Dr. Johnston's report does not help the Court assess how much noncompliance impacts brand value, his report, and Dr. Siegel's response to it, sufficiently demonstrates that PRPs are deprived of a protected property interest -- the value of their brand -- when they do not comply with a UAO.

Finally, GE argues that PRPs suffer pre-hearing liberty deprivations under the D.C.

Circuit's "reputation-plus" standard. This standard has its origins in Paul v. Davis, 424 U.S. 693 (1976), where the Supreme Court "recognized the serious damage that could be inflicted by branding a government employee as 'disloyal,' and thereby stigmatizing his good name." Id. at 706. The Court noted, however, that it had "never held that the mere defamation of an individual, whether by branding him disloyal or otherwise, was sufficient to invoke the guarantees of procedural due process absent an accompanying loss of government employment."

29

Id. (emphasis added). Under the D.C. Circuit's "reputation-plus" standard, collateral effects accompanying reputational injury must be "sufficiently formal or sufficiently broad" to merit due process protection. See Taylor v. Resolution Trust Corp., 56 F.3d 1497, 1506 (D.C. Cir. 1995); see also Kartseva v. Dep't of State, 37 F.3d 1524 (D.C. Cir. 1994) (explaining that a plaintiff could only succeed under the reputation-plus standard if she showed reputational damage that "automatically exclud[ed] her from a definite range of employment opportunities" or that "broadly preclud[ed] her from continuing in her chosen career").

Collateral consequences are "sufficiently formal or sufficiently broad" when the reputation-injuring statement automatically triggers some adverse effect or precludes the injured party from pursuing chosen categories of activities in a way "equivalent in every practical sense to formal debarment." See Trifax Corp. v. District of Columbia, 314 F.3d 641, 643-44 (D.C. Cir. 2003). For example, in Nat'l Council of Resistance of Iran v. Dep't of State, 251 F.3d 192 (D.C. Cir. 2001), the court considered the Anti-Terrorism and Effective Death Penalty Act, 8 U.S.C. § 1189, which automatically bars designated "foreign terrorist organizations" from holding bank accounts or receiving material support. The court held that an organization was entitled to process before the State Department could issue a "foreign terrorist organization" designation. Id. at 203-04. Similarly, in Old Dominion Dairy Products., Inc. v. Sec'y of Defense, 631 F.2d 953 (D.C. Cir. 1980), the Defense Department placed a written finding of "nonresponsiblity" in a contractor's permanent file. Under the applicable acquisition regulations, a "responsibility" finding was required before a contract could be awarded. The court held that the contractor was entitled to process before being labeled as "nonresponsible" because the label would broadly, if not formally, preclude the contractor from obtaining other government contracts. Id. at 963.

30

GE argues that PRPs that do not comply with UAOs suffer reputational damage along with four collateral consequences: EPA seeks greater penalties for PRPs that routinely do not comply with UAOs; noncompliance could make PRPs ineligible for participation in EPA programs like "Performance Track" and "Corporate Leaders"; noncompliance could lead to increased permitting times; and noncompliance impairs PRPs' relationships with some stakeholders. The Court can easily dispose of the latter three collateral consequences GE alleges, as they are entirely hypothetical. The few courts willing to speculate that a collateral consequence will follow only do so when the consequences are inevitable. See, e.g., Reeve Aleutian Airways, Inc. v. United States, 982 F.2d 594, 598 & n.1 (D.C. Cir. 1993) (holding that labeling an airline "unsafe" would inevitably cost the airline future government contracts). The conclusion that noncompliance leads to greater permitting times, ineligibility for EPA programs, and impairment of relationships with stakeholders is not nearly so inescapable. Indeed, despite years of discovery, GE has not a single example of any of these collateral consequences occurring to any PRP.14 Absent any evidence of such hypothetical collateral consequences, the Court will not assume they in fact occur.

Remaining, then, is GE's claim that noncompliance is accompanied by greater EPA penalties. The parties agree that EPA can and does take past UAO noncompliance into account when determining what penalties it will seek for a fresh instance of UAO noncompliance. See GE SUF at ¶ 132, EPA's Response to GE SUF at ¶ 132, and GE's Reply to EPA's Response to

14 GE argues that "green" investors calculate that GE's overall environmental performance impacts its financial performance. GE SUF at ¶ 19. GE does not, however, provide any evidence to suggest that noncompliance with a UAO impacts PRPs' relationships with such investors.

31

GE SUF at ¶ 132. But greater penalties are not an automatic consequence of noncompliance. See Deposition of Walter Mugdan at 228:4-12; Deposition of Kenneth Patterson at 91:4-92:22. Therefore, the logic of cases like Nat'l Council of Resistance of Iran does not apply. Rather, the question is whether the likelihood of greater penalties will "broadly preclude" PRPs from carrying out their chosen line of business. See Kartseva, 37 F.3d at 1527. And the answer is that they do not. An "unsafe" airline may be broadly precluded from receiving air transport contracts, see Reeve, 982 F.2d at 598; an "unreliable" military contractor may be broadly precluded from obtaining future contracts, see Old Dominion, 631 F.2d at 963; and a Russian translator with "counterintelligence concerns" -- before the collapse of the Soviet Union -- may be broadly precluded from a chosen career as a public employee, see Kartseva, 37 F.3d at 1529-30. Increased penalties do not have the same impact on a PRP -- a typical PRP is not broadly precluded from its chosen line of business when it does not comply with a UAO, even if the penalties are substantial. To be sure, GE argues that UAOs can at times be so onerous as to put PRPs out of business. Geweke Report at ¶ 21. Accepting that assertion arguendo, it still does not rise to the level needed to trigger the reputation-plus standard: a "won some and lost some" record is not "remotely close" to "broad preclusion." See Trifax, 314 F.3d at 644. Because GE has not pointed to "sufficiently formal or sufficiently broad" collateral consequences of being labeled a noncomplying PRP, then, noncompliance does not cause PRPs to suffer a liberty deprivation under the reputation-plus standard.

Still, GE has alleged sufficient pre-hearing, noncompliance-based deprivations for the Court to proceed to prong two of the Mathews analysis. Although GE has not pointed to a reputation-plus liberty deprivation, GE has pointed to two deprivations of protected property

32

interests resulting from noncompliance with a UAO: damage to stock price and damage to brand value. The Court discounts Drs. Geweke's and Johnston's conclusions with regard to the size of those deprivations, but the Court accepts their conclusions as to the existence of those deprivations.

3. Deprivations if a PRP Complies The question before the Court is not whether the response costs a PRP incurs if it complies with a UAO are protected property interests under the Due Process Clause. The parties agree that they are. See EPA Mem. at 21. Rather, the question is whether PRPs have a meaningful choice not to comply with a UAO once it is issued. In rejecting GE's facial challenge to section 106 in General Electric III, the Court explained that:

The ability of the PRP to choose, under the language of the statute, whether to comply with a section 106 order is key. Upon a refusal to comply with a section 106 order, EPA is powerless to deprive the PRP of property without judicial intervention. On the other hand, were the PRP to be deprived of any meaningful choice to refuse to comply with an order, that might pose a different situation in assessing whether a deprivation of property has occurred.

362 F. Supp. 2d at 339. GE argues that EPA's pattern and practice of administering section 106 does indeed deprive PRPs of a "meaningful choice" to refuse to comply with UAOs. Therefore, the argument goes, PRPs are denied the very safeguard the Court previously held necessary -- the opportunity for a pre-deprivation hearing by choosing not to comply. GE's argument raises two questions. First, as a legal matter, if a PRP is not unconstitutionally coerced under Ex parte Young, may it nonetheless be deprived of a meaningful choice not to comply? Second, if some lesser coercion standard does apply, are PRPs in fact deprived of a meaningful choice not to comply? The answer to both questions is no.

33

GE has not proposed any legal framework to analyze coercion other than the Ex parte Young standard. As discussed supra, GE has not demonstrated that EPA's pattern and practice of administering section 106 violates Ex parte Young by unconstitutionally intimidating or coercing PRPs into compliance whenever a UAO is issued. GE has not cited to any precedent suggesting that a statute or agency policy, although not unconstitutionally coercive under Ex parte Young, can still be sufficiently coercive so as to deny one a meaningful opportunity to elect not to comply. The Court will not invent a new "lesser coercion" legal framework from scratch. Therefore, GE's argument that PRPs are deprived of a meaningful choice -- even if they are not coerced into compliance under an Ex parte Young analysis -- fails as a matter of law.

GE's argument is belied by the facts as well. GE represented at both the motions hearing and in its brief that EPA achieves a near-perfect record of compliance with UAOs. See Tr. at 5; GE Mem. at 50-51. GE's expert, Dr. Rouhani, however, suggests that EPA's record reflects a modest rate of noncompliance. Of the 5,422 PRPs who were issued UAOs between 1982 and 2006 that he examined, there were 189 instances of PRPs not complying -- a rate of 3.5 percent. See Rouhani Report at 6.52 & Table 8(a). And of the 1,638 PRPs who have been issued UAOs most recently, there were 75 instances of noncompliance -- a rate of 4.6 percent. Id. at Table 8(b). These rates of noncompliance mirror the kind of "acceptable rate of error" tolerated by courts, see Shands v. Tull, 602 F.2d 1156, 1160 (3d Cir. 1979) (holding that a four percent error rate constitutes "substantial compliance" with a statute), not the "remarkable record" of "self- confidence" GE claims, see Tr. at 29:9.

The costs of clean-up that follow compliance, paired with the pre-hearing stock price and brand value deprivations caused by noncompliance, demonstrate that PRPs suffer pre-hearing

34

deprivations whether or not they comply. But because the Court will not invent a new and lesser legal standard for coercion under Mathews -- and GE has not demonstrated that EPA's pattern and practice is unduly coercive under Ex parte Young -- the Court holds that PRPs are not deprived of a meaningful opportunity not to comply. This conclusion is bolstered by GE's own expert, Dr. Rouhani, who demonstrates that instances of noncompliance are sufficiently numerous to suggest that PRPs are not, in fact, forced to comply.

B. Mathews Prong Two GE has demonstrated that PRPs are deprived of at least some property interests whether or not they comply with a UAO, and hence the Court must proceed to the second prong of the Mathews analysis. At this stage, the Court must first identify the private interest impacted by government action, the government interest in avoiding additional pre-decision process, and the risk that the current process will result in error. See Mathews, 424 U.S. at 355. Once the Court has identified those three factors, it must weigh them against one another to determine whether additional process is constitutionally required before EPA issues a UAO. See id.

The process that EPA now provides before issuing a UAO bears on the identification and balancing of these factors. Once EPA identifies a "Superfund" site, it searches for PRPs by issuing information requests and conducting interviews. EPA SOF at ¶ 5. PRPs are permitted to provide information regarding both liability and remedy selection. Id. at ¶ 6. EPA provides notice to potential PRPs through a "general notice" letter. Id. at ¶ 8. PRPs then again have the opportunity to respond to general notice letters and provide information regarding liability and remedy selection. Id. at ¶ 9. Before selecting a remedy, EPA solicits public comments and compiles an administrative record, which normally happens over the course of weeks or months.

35

Id. at ¶¶ 13-14, 18. PRPs are, of course, permitted to contribute to that record. If the remedy selected is settlement -- EPA's preferred remedy -- then EPA either issues a "special notice letter" or provides other written notice to PRPs. Id. at ¶¶ 19-20. Settlement negotiations take a "significant period of time" and PRPs are normally represented by counsel. Id. at ¶¶ 21-22. PRPs are permitted to, and often do, challenge EPA's determinations regarding liability and remedy selection during the negotiation process. Id. at ¶ 23. Finally, at the end of this lengthy process and only if negotiation fails, EPA will normally issue a UAO.

1. Private Interest The type and length of the pre-hearing deprivation inform the constitutional relevance of the private interest. See City of Los Angeles v. David, 538 U.S. 715, 717 (2003); Mathews, 424 U.S. at 341-42. Deprivations that can be recouped after a hearing -- like purely financial deprivations -- are less significant than irreparable deprivations. See David, 538 U.S. at 717 (holding that a temporary deprivation of a car was more significant than a deprivation of money in the form of impoundment and towing fees). A substantial financial deprivation bearing collateral consequences -- for example, an effect on a company's operations -- is not purely financial and is a more significant private interest. See Mathews, 424 U.S. at 341-42. No matter the type of the deprivation, moreover, "there is a point at which an unjustified delay in completing a post-deprivation proceeding 'would become a constitutional violation.'" Federal Deposit Ins. Corp. v. Mallen, 486 U.S. 230, 242 (1988) (quoting Cleveland Bd. of Educ. v. Loudermill, 470 U.S. 532, 547 (1985)).

Here, the significance of the pre-hearing deprivation varies depending on whether or not a PRP complies with a UAO. GE relies on the expert opinion of Dr. Shahrokh Rouhani to

36

establish the deprivations faced by PRPs who comply with a UAO. Dr. Rouhani obtained data regarding EPA's CERCLA enforcement activities from the Comprehensive Environmental Response, Compensation and Liability Information System ("CERCLIS"), an EPA database. Rouhani Report at ¶¶ 1.2, 4.1-4.4. Based on his assessment of that data, Dr. Rouhani opines that, on average, complying with a UAO costs a PRP $4 million. Id. at ¶¶ 2.1.3, 6.2-6.3.15 For the costs of noncompliance with a UAO, GE relies on the expert reports from Drs.

Johnston and Geweke. As discussed earlier, neither expert report properly quantifies brand value reduction or market value decline. Dr. Johnston, who addresses brand value reduction, does not attempt to quantify how much brand value drops for an average noncomplying PRP. Dr. Geweke, who addresses market value decline, opines that PRPs suffer an average $76.4 million decline in market value if they do not comply with a UAO, but he acknowledges that markets may discount potential liabilities, thereby "lessen[ing] somewhat the immediate financial impacts of a decision not to comply with a UAO." Geweke Report at ¶ 27. Yet Dr. Geweke does not assess how much markets discount potential liabilities. Moreover, his $76.4 million figure is based on the unfounded assumption that EPA waits as long as possible before bringing cost recovery or enforcement actions. Therefore, the Court is left without a reliable quantification of the economic deprivations noncomplying PRPs suffer. Rather than dismissing the expert reports entirely, however, the Court is persuaded that noncomplying PRPs suffer a significant decrease in

15 Dr. Rouhani also calculates the average costs if a PRP chooses not to comply with a UAO. Assuming that EPA cleans up a site itself, brings an enforcement action against the noncomplying PRP, and persuades a court to award punitive damages and the maximum daily fines for the average period of completion, Dr. Rouhani calculates that the average costs of noncompliance are more than $54 million. Rouhani Report at ¶ 6.3.5. But because PRPs receive a full judicial hearing before any of those costs may be imposed, those costs of noncompliance do not factor into an assessment of whether more process is needed before a UAO may be issued.

37

brand and market value, albeit something less than $76.4 million. The Court will proceed with the Mathews prong two assessment based on that estimate of the private interest impacted by noncompliance.

The deprivations that GE alleges are primarily financial ones. Financial deprivations are less troubling because money can be recouped in a post-deprivation hearing. See David, 538 U.S. at 717. But deprivations caused by UAOs are potentially so large -- on average $4 million for complying PRPs and some substantial, unidentified amount for noncomplying PRPs -- that they may have collateral effects. UAOs could put some PRPs out of business. See Geweke Report at ¶ 21. For other PRPs, UAOs may affect operations, like whether to bid for new projects or to hire additional employees. See id. at ¶ 25. Yet for other companies, like GE, UAOs are "not material to . . . financial positions, results of operations, or liquidity." See EPA SOF at ¶ 45 (quoting GE Annual Reports, 1997-2000). The variety of collateral effects, if any, precludes the Court from reaching a single conclusion about the significance of UAOs for all PRPs. Still, a general conclusion is possible: although the private interests are less constitutionally significant because they are primarily financial, they are sufficiently large and have enough potential collateral effects to constitute weighty private interests.

GE argues that PRPs are deprived of these private interests for lengthy periods of time.

Under section 106(b), a complying PRP may seek reimbursement "upon completion of the required action." On average, it takes three years to fully comply with a UAO. Rouhani Report at ¶ 6.2. Moreover, GE argues, EPA claims the power unilaterally to delay "completion." In proceedings before the Environmental Appeals Board ("EAB"), the body that reviews section 106(b) reimbursement petitions, EPA has at times successfully argued that a petition should be

38

dismissed because of a PRP's failure to obtain an EPA "certificate of completion." GE SUF at ¶ 91. In the current proceedings, however, EPA contends -- correctly -- that the statute does not require an EPA determination that the clean-up is complete before a complying PRP may file a reimbursement petition. See 42 U.S.C. § 9606(b); EPA Response to GE SUF at ¶ 91. If a complying PRP believes that it has completed a clean-up but the EAB declines the PRP's reimbursement petition for failure to obtain a certificate of completion, the PRP may appeal the EAB decision to an Article III court. See Employers Ins. of Wausau, 52 F.3d at 662. There, the PRP can argue that the EAB made legal errors (like accepting EPA's argument that a certificate of completion was required) or factual errors (like finding the clean-up incomplete). Therefore, although Dr. Rouhani's conclusion that UAOs take, on average, three years to complete is persuasive, the Court is not persuaded that EPA can unilaterally delay reimbursement proceedings by refusing to issue a certificate of completion.

GE argues that PRPs are also indefinitely "left in limbo" if they choose not to comply with a UAO. EPA has two ways to enforce UAOs under CERCLA. First, EPA can bring an enforcement action against a noncomplying PRP. Enforcement actions must be brought within five years of a "violation." See 28 U.S.C. § 2462. Second, EPA can complete the UAO itself and file a cost recovery petition against the PRP. Cost recovery petitions must be filed within three years of completion for removal actions and within six years of completion for remedial actions. 42 U.S.C. § 9613(g)(2). GE argues that these statutes of limitations permit an unconstitutionally long deprivation because they are paired with a "removal or remedial action [that] may itself take years to complete" such that "the government may take its own sweet time before suing." See GE Mem. at 38-39 (quoting Reardon v. United States, 947 F.2d 1509, 1519

39

(1st Cir. 1991)). In Reardon, an as-applied challenge, the First Circuit held that the CERCLA statutes of limitations contributed to the unconstitutionality of a different section of CERCLA in a specific setting. 947 F.2d at 1521. But no court has rejected these CERCLA statutes of limitations in a facial challenge. Indeed, this Court denied GE's facial challenge to the statutes of limitations in General Electric III. See 362 F. Supp. 2d at 363. The Court invited GE to revisit this argument in its pattern and practice claim by presenting evidence that EPA indeed unreasonably delays enforcement proceedings. Id. But GE has not offered any evidence to suggest that EPA in fact delays enforcement or cost recovery proceedings. Therefore, the Court confirms its prior rejection of a challenge to the CERCLA statutes of limitations and hence also rejects GE's argument that PRPs are deprived of property for extended periods of time if they choose not to comply with a UAO.

In short, the relevant private interest at issue is primarily a financial one. The magnitude of the private interest depends on whether or not a PRP complies, and UAOs have collateral effects on operations for some PRPs but no material impact on others. While in theory a PRP could be deprived of that private interest for several years, GE has not demonstrated that EPA in fact routinely waits as long as the CERCLA statutes of limitations allow before bringing enforcement or cost recovery actions.

2. Government Interest Two factors are paramount in identifying the government interest. The first is whether the government has "a special need for very prompt action." See Fuentes v. Shevin, 407 U.S. 67, 91 (1972). The greater the government's need for prompt action, the greater the government interest in avoiding additional pre-deprivation process. Id. The second factor is the financial and

40

operational cost of additional process and the need to conserve "scarce fiscal and administrative resources." See Mathews, 424 U.S. at 347-48. The cost of additional process will vary according to its nature. Whereas a full judicial hearing carries substantial financial and administrative costs, notice and an informal opportunity to be heard is far less burdensome. The cost of additional process also varies according to its volume. The more often the government is required to provide additional process, the greater the burden on the government. See id. at 348.

EPA lacks a "special need for very prompt action" in issuing UAOs under section 106 of CERCLA. See Fuentes, 407 U.S. at 91. To be sure, EPA may only issue a UAO upon a finding of "imminent and substantial endangerment to public health or welfare or the environment." 42 U.S.C. § 9606(a). Nonetheless, the parties agree that EPA does not issue UAOs in true emergency situations.16 See GE SUF at ¶¶ 37-38. In true emergency situations, EPA cleans up a site itself and later files a cost recovery action. Id. at ¶ 38. EPA nevertheless argues that UAOs are more time-sensitive than wholly non-emergency situations -- that is, UAOs occupy a gray area in between emergency and non-emergency situations. But GE responds that whatever the contours of that gray area, EPA's practices demonstrate that UAOs do not require "very prompt action." GE points to Dr. Rouhani's report, which demonstrates an average eight year lag-time between identification of a hazardous waste site and issuance of a UAO and a four year lag-time between remedy selection and UAO issuance. Rouhani Report at ¶ 6.1.3. Although EPA insists that it is busily investigating and negotiating during that lengthy period, Dr. Rouhani's report demonstrates that the government lacks a "special need for very prompt action" in issuing UAOs.

16 Whether or not "imminent and substantial" is synonymous with "emergency" as a legal matter is an issue this Court need not decide.

41

The second consideration is the cost of the additional process GE seeks. Although GE has not outlined exactly what that additional requested process is, GE argues that some sort of neutral decision-maker is required. See GE Opp. at 24-25. The "most obvious solution" in GE's view is a full judicial hearing before an Article III judge. See Tr. at 9:8-10:17. Judicial proceedings, however, are fraught with considerable expense and delay, and courts rarely import the judicial model into administrative decision-making. See Mathews, 424 U.S. at 348. Other methods of providing additional process are less costly and time-consuming. For example, courts have held that an administrative law judge ("ALJ"), see Rosen v. NLRB, 735 F.2d 564, 570 (D.C. Cir. 1984), and a detached "presiding officer" within an agency, see Chemical Waste Mgmt., Inc. v. EPA, 873 F.2d 1477, 1485 (D.C. Cir. 1989), are both adequate neutral decision- makers. Although the government has a high interest in avoiding the costs and delays of a full judicial hearing, it has a lower interest in avoiding the less formal process before an ALJ or agency presiding officer.

The cost of the additional process also depends on how often the government must provide it. See Mathews, 424 U.S. at 348. Between August 16, 1982 and May 25, 2006 -- a period of 285 months -- EPA issued 1,705 UAOs to more than 5,400 PRPs. See Rouhani Report at ¶¶ 4.3, 6.5.2. On average, then, EPA has issued approximately six UAOs to nineteen PRPs every month.17 Because EPA generally issues UAOs only when negotiations with a PRP have failed, most PRPs would contest the UAO through the pre-issuance hearing GE seeks. To be sure, EPA might issue fewer UAOs if more pre-issuance process were required. But because

17 Dr. Rouhani points out that UAOs have been issued more frequently in the past 15 years than they were for the first ten years. See Rouhani Report at 6.7. The Court's calculation, therefore, underestimates the number of UAOs EPA currently issues per month.

42

UAOs involve high stakes and complex technical judgments, any hearing before a neutral decision-maker would involve significant fiscal and administrative burdens; hence, even a fraction of the current monthly rate of UAO issuance would generate a substantial impairment of the government's interest measured in the financial and administrative costs of that additional process.

3. Risk of Error Analysis of the risk of error is different for a pattern and practice claim than it is for either a facial or an as-applied challenge. Case law provides scant guidance, however, on just what that difference is. The few courts that have decided pattern and practice procedural due process claims have expounded little on the risk of error analysis. See, e.g., Haitian Refugee Ctr., Inc. v. Nelson, 694 F. Supp. 864 (S.D. Fla. 1988), aff'd 872 F.2d 1555 (11th Cir. 1989), aff'd sub nom. McNary v. Haitian Refugee Ctr., Inc., 498 U.S. 479 (1991). GE urges the Court to adopt the same kind of abstract approach that courts use in both facial and as-applied challenges. See Tr. at 42:5-42:11; GE Mem. at 48 (citing Carey v. Piphus, 453 U.S. 247, 266 (1978)). Under this approach, courts "look generically at whether the kind of inquiry that is involved is the kind that is error-prone and the kind that would be benefitted by procedural fairness." Tr. at 42:5-42:11. But the Court is not persuaded that the abstract approach is the correct one here. With the benefit of a full factual record -- in this case, following years of discovery -- plaintiffs in pattern and practice claims cannot rely on abstract legal concepts alone. Although abstract legal concepts may inform the Court's analysis regarding the risk of error, GE must also demonstrate that the current procedures in fact result in an unacceptable rate of error. It would make little sense to ignore the empirical record altogether in favor of a hypothetical concern that the risk of error may

43

be too high.

a. Abstract Concepts According to GE, the crucial element missing from EPA's pre-UAO issuance process is a neutral decision-maker. Although the Constitution does not require a neutral decision-maker, see, e.g., Old Dominion, 631 F.2d at 968 (holding that an opportunity to be heard by a contracting officer, the very person deciding whether the plaintiff is a "responsible" contractor, satisfies due process), ample case law supports GE's point that a neutral decision-maker is an important constitutional safeguard, see, e.g., Propert v. District of Columbia, 948 F.2d 1327, 1333 (D.C. Cir. 1991). Therefore, the lack of a neutral decision-maker, as an abstract principle, supports GE's argument that EPA's pre-issuance process is error-prone.

Several other abstract principles suggest the pre-decisional process may result in error.

The risk of error is greater when "the State stands to benefit" by taking action that deprives someone of property. United States v. James Daniel Good Real Property, 510 U.S. 43, 55-56 (1993) (quoting Harmelin v. Michigan, 501 U.S. 957, 979 n.9 (1991)). Although there is no direct financial incentive at work here, EPA arguably stands to benefit by issuing UAOs because EPA has an interest in conserving Superfund resources. See EPA SOF at ¶ 2. The risk of error is also greater when junior or regional agency staff, without senior or centralized oversight, make deprivation-causing decisions. See James Madison Ltd. by Hecht, 82 F.3d at 1100. UAOs are issued by regional EPA officials without review and approval from EPA headquarters. EPA SOF at ¶ 42. Therefore, although nothing prevents PRPs from contacting EPA headquarters to register complaints about UAO issuance, the authority of regional EPA officials to issue UAOs also supports GE's argument that EPA procedures could result in a high rate of error.

44

On the other hand, some principles that GE cites do not suggest that EPA's process is error-prone. GE argues that ex parte procedures, like those in Doehr, 501 U.S. at 14, enhance the risk of error. In Doehr, the Supreme Court took issue with "secret, one-sided determination[s]" that provided affected parties with no notice or opportunity to be heard. Id. (quoting Joint Anti- Fascist Refugee Comm. v. McGrath, 341 U.S. 123, 170 (1951) (Frankfurter, J., concurring)). The informal pre-issuance process that EPA provides here, however, is neither secret nor one-sided. PRPs are given multiple kinds of notice and have several opportunities to be heard before a UAO is issued. See EPA SOF at ¶¶ 5-23. GE also argues that the highly factual, highly technical nature of UAO issuance decisions elevates the risk of error. Certainly, the simpler the question, the lower the risk of erroneous decision. Mitchell v. W.T. Grant Co., 416 U.S. 600, 609 (1974). And the more "highly factual" a question, the greater the risk of erroneous decision. Doehr, 501 U.S. at 8. But at the same time, courts often defer to agencies' highly technical decisions, reviewing such decisions only on an "arbitrary and capricious" standard. See Thomas Jefferson Univ. v. Shalala, 512 U.S. 504, 512 (1994). Because the decision to issue a UAO is both factual and technical, this abstract principle neither helps nor hurts GE's case.

Still other abstract principles suggest that EPA's process is not error-prone. Most notably, representation by counsel during the pre-decisional process sharply reduces the risk of error. See Reeve Aleutian, 982 F.2d at 600. PRPs may be and usually are represented by counsel in pre- UAO issuance negotiations with EPA. EPA SOF at ¶¶ 21-22. This is, in the Court's view, of great significance. Therefore, although several aspects of the pre-UAO issuance process suggest, in the abstract, a high risk of error, others suggest neither a high nor a low risk of error, and still others suggest a low risk of error.

45

b. Concrete Evidence

Overlaid on this abstract assessment of the risk of error is the voluminous empirical record in this case. GE points to four categories of evidence drawn from that record to argue that EPA's process in fact results in errors in issuing UAOs. First, GE proffers cases demonstrating that EPA makes mistakes in other, non-UAO contexts. Next, GE provides an expert report from Dr. Kip Viscusi, who examines whether EPA's remedy selection is erroneous under a cost-benefit, cancer- prevention standard. Third, GE relies on declarations cataloging the experiences of several PRPs, including some sites where GE itself was issued a UAO. Finally, GE provides evidence from reimbursement proceedings filed by complying PRPs against the government.

The cases demonstrating that EPA has made mistakes in other CERCLA contexts are not informative as to EPA's rate of error in issuing UAOs. GE cites to seven cases.18 But none involve UAOs. Rather, they merely demonstrate that EPA has made some mistakes in administering other sections of CERCLA. That EPA makes some mistakes is unsurprising; every agency, especially one as large as EPA, makes mistakes. The APA's "arbitrary and capricious" standard for the review of agency action anticipates that some agency errors will not warrant judicial reversal. All that courts can expect is that agency procedures "achieve an acceptable rate of error." See Califano v. Boles, 443 U.S. 282, 285 (1979). Even if these seven cases were relevant to determining EPA's rate of error in issuing UAOs -- which they are not -- the Court has

18 United States v. Burlington N. R.R. Co., 200 F.3d 679 (10th Cir. 1999); In re Bell Petroleum Servs., Inc., 3 F.3d 889 (5th Cir. 1993); United States v. Ottati & Goss, Inc., 900 F.2d 429 (1st Cir. 1990); United States v. B & D Elec., Inc., No. Civ.A.05-0063, 2007 WL 1395468 (E.D. Mo. May 9, 2007); United States v. Broderick Inv. Co., 963 F. Supp. 951 (D. Colo. 1997); United States v. Wedzeb Enters., Inc., 844 F. Supp. 1328 (S.D. Ind. 1994); and United States v. Hardage, 663 F. Supp. 1280 (W.D. Okla. 1987).

46

no context whatsoever to determine what kind of "rate of error" these seven cases reflect.

The expert report from Dr. Viscusi is similarly unhelpful.19 Dr. Viscusi examined EPA's clean-up efforts under CERCLA from 1993 to 1999 to analyze whether EPA effectively prevents cancer from a cost-benefit perspective. Viscusi Report at ¶¶ 7, 10-13. But Dr. Viscusi's report is at once too broad and too narrow. His report is too broad because it examines EPA's overall administration of CERCLA. Even if Dr. Viscusi's study reveals that EPA makes errors in its overall administration of CERCLA, that does not show that EPA errs in issuing UAOs because Dr. Viscusi did not conduct a UAO-specific analysis. The report is too narrow because it focuses only on cancer prevention and cost-benefit analysis -- just two of the nine factors EPA must consider in selecting a remedy. See 40 C.F.R. § 300.430(e)(9)(iii); see also Ohio v. EPA, 997 F.2d 1520, 1531 (D.C. Cir. 1993) (noting that cost is given less weight in remedy selection than some of the other nine factors). Whatever weight Dr. Viscusi's report may have colloquially or politically, it is of little value in assessing whether EPA's current administrative process erroneously deprives PRPs of protected property interests by issuing UAOs.

GE's best evidence is in its declarations. In five declarations,20 GE describes nine sites21 where EPA supposedly erred in issuing a UAO. Five of these sites do not advance GE's case. For

19 Indeed, GE's counsel indicated at the motions hearing that Dr. Viscusi's report is primarily intended to show that political factors influence EPA's decision-making process, not that EPA in fact makes errors in issuing UAOs. See Tr. at 46; see also GE SUF at ¶¶ 68-69.

20 Declaration of Richard Krablin; Declaration of John Baker; Declaration of Daniel Uyesato; Declaration of Randall McAlister; and Supplemental Declaration of Randall McAlister.

21 The Grand Street Mercury Superfund site; the Fletcher's Paint Works site; the former GE electrical transformer manufacturing facility in Rome, Ga. ("Rome site"); the Vega Alta Superfund site; the Solvent Savers Superfund site; the Li Tungsten site; the Woolfolk Chemical Works site; the Herington Army Airfield site; and the Palmerton Zinc Pile Superfund site.

47

the Li Tungsten site, the Grand Street Mercury site, the Fletcher's Paint Works site, the Herington Army Airfield site, and the Woolfolk Chemical Works site, the declarants provide only their unsubstantiated assertions that a UAO either was issued to the wrong PRP or contained some improper remedy. For the Fletcher's Paint Works site, for example, Randall McAlister, a GE Senior Program Manager, states that GE was unfairly targeted and that EPA selected an improper remedy. See McAlister Decl. ¶¶ 69-98. Without some independent verification that EPA in fact made errors in issuing a UAO to GE at the Fletcher's Paint Works site, McAlister's declaration is self-serving and uncorroborated. Such declarations are of little value at the summary judgment stage, see Tolson v. James, 315 F. Supp. 2d 110, 116 (D.D.C. 2004), and certainly do not represent concrete evidence establishing an unacceptable risk of error in the section 106 UAO process.

The remaining four sites do offer some support for GE's case. GE has demonstrated that EPA's initial remedy selection at two sites was erroneous. At the Vega Alta site, GE took issue with an EPA-proposed remedy to clean up the site. Supp. McAlister Decl. ¶¶ 18-19. GE proposed its own remedy, and EPA eventually withdrew its previous remedy and adopted GE's approach. Id. ¶¶ 22-23. Similarly, at the Solvent Savers site, EPA replaced its initial remedy with a PRP-selected one after the PRPs demonstrated that EPA's initial remedy was inefficient. Id. ¶¶ 30-36. At two other sites, GE has shown that EPA erroneously issued a UAO. At the Rome site, GE was issued a UAO even though GE had negotiated a clean-up plan with state environmental officials under the Resource Conservation and Recovery Act ("RCRA"), 42 U.S.C. § 6901 et seq. McAlister Decl. ¶ 134. GE was able to seek review through RCRA, and EPA withdrew the UAO before the court could act. Id. ¶¶ 146, 148. At the Palmerton Zinc Pile site, Horsehead Industries,

48

another PRP, was issued a UAO to clean up a site that Horsehead had never owned or operated. Krablin Decl. ¶¶ 18-21. Horsehead was able to obtain review because of a parallel cost recovery action. Id. ¶ 29. The court eventually issued a consent order preliminarily enjoining enforcement of the UAO against Horsehead. Id. ¶ 31.

GE's evidence from reimbursement proceedings is revealing but does not advance GE's position. Under section 106(b)(2), PRPs may seek reimbursement upon completion of the clean- up if they believe that they have been issued a UAO in error. If EPA refuses to reimburse the PRP, then the PRP may file suit in a federal district court, where the PRP can challenge both liability and remedy selection. See 42 U.S.C. §§ 9606(b)(2)(C)-(D). Given that thousands of PRPs have complied with UAOs, reimbursement proceedings would appear to be the most fertile ground for finding EPA error. But GE concedes that "there has only been one Section 106(b) case that has resulted in a final judgment adverse to EPA requiring reimbursement from the government." GE SUF at ¶ 154. At the summary judgment stage, the paucity of evidence of error from reimbursement proceedings is telling. See Celotex, 477 U.S. at 322.

In sum, when all four categories of evidence proffered by GE are carefully scrutinized, GE has pointed to just five instances of error -- four examples from specific sites described in GE's declarations and one example of a complying PRP successfully obtaining reimbursement from the government. EPA has issued 1,705 UAOs to more than 5,400 PRPs since 1982. See Rouhani Report at ¶¶ 4.3, 6.5.2. If GE were attempting to present an exhaustive list of EPA's errors in issuing UAOs, this would represent a minuscule error rate. But even if GE is not presenting an exhaustive list of EPA's errors for all PRPs, the Court can presume that GE has at least thoroughly surveyed all 68 of the UAOs that GE itself has been issued. In the evidence presented, GE has

49

demonstrated that EPA has made only three errors22 in issuing those 68 UAOs -- an error rate of 4.4 percent. Although not insignificant, 4.4 percent is an "acceptable rate of error" for EPA in issuing UAOs under section 106 of CERCLA. See Shands, 602 F.2d at 1160 (holding that a four percent error rate constitutes "substantial compliance" with a statute); see also Humphries v. County of Los Angeles, Civ.A.No. 05-56467, 2008 WL 4791868 at *20 (9th Cir. Nov. 5, 2008) (holding that a 50 percent error rate would constitute a substantial risk of erroneous deprivation); Dupuy v. Samuels, 397 F.3d 493, 505, 512 (7th Cir. 2005) (holding that a 74.6 percent error rate is unacceptably high). And given the size of the record after several years of exhaustive discovery, the paucity of evidence demonstrating a higher rate of error for all PRPs casts doubt on assuming an error rate even as high as four percent.

4. Balancing Having identified the private interests, government interests and risk of error, the final step in this due process assessment is to balance these factors against one another. See Mathews, 424 U.S. at 334. The balancing process is dynamic -- for example, while greater process could reduce the risk of error, it also adds cost and delay, thereby burdening both government and private interests. Here, GE has stated a preference for a full judicial hearing. See Tr. at 9:8-10:1. But courts seldom import the full judicial model into the administrative decision-making process. In lieu of a full judicial hearing, GE has stated that the key ingredient for the additional process it thinks necessary is a neutral decision-maker. See id. As discussed above, courts have endorsed both ALJs, see Rosen, 735 F.2d at 570, and detached "presiding officers" within an agency, see

22

Those are alleged errors at the Vega Alta site, the Solvent Savers site, and the Rome site.

50

Chemical Waste, 873 F.2d at 1485, as adequate neutral decision-makers. Therefore, the Court will undertake the Mathews balancing to determine whether an agency presiding officer or an ALJ is constitutionally necessary.23 The precise nature of the interests at stake is worth repeating. The private interest depends on whether or not a PRP complies. If the PRP complies, then the average costs of compliance are $4 million and the deprivation lasts for an average of three years. If the PRP does not comply, then the average size and length of the deprivation are substantial but unclear. And whether or not a PRP complies, the deprivations are primarily financial, although for some PRPs the financial deprivations are sufficiently large to have collateral effects on operations. As for the government interest, EPA does not issue UAOs in emergency situations. But EPA nonetheless has a substantial financial and administrative interest. Because UAOs are issued frequently, even minimal additional process would tax EPA's resources. Finally, the risk of error -- as represented through the actual rate of error over many years in this pattern and practice case -- is low. GE has provided concrete evidence to establish that EPA has made errors in issuing five UAOs, including three UAOs issued to GE itself. Assuming that GE has plumbed for error all 68 of the UAOs that it has received, and assuming that GE's experience is representative, EPA makes some sort of error in roughly four percent of its UAOs.

The Court will first consider whether a pre-issuance hearing before a presiding officer within EPA would so reduce the risk of error, without unduly burdening the government interest, as to be constitutionally required. At first glance, the burden on the government interest appears

23 If neither a presiding agency officer nor an ALJ as a neutral decision-maker is constitutionally required for the section 106 UAO pre-issuance process, then the substantially more burdensome full judicial hearing would not be necessary either.

51

minimal. For any given UAO, a hearing before a presiding officer would add only weeks or a few months to an issuance process that usually takes years. See Rouhani Report at ¶ 6.1.3. Moreover, the costs of a single hearing before a presiding officer are minimal, especially considering the size of the private interests at stake. But the Court may not examine the burden on the government at such a granular level. See Gray Panthers v. Califano, 466 F. Supp. 1317, 1324 (D.D.C. 1979), rev'd on other grounds sub nom. Gray Panthers v. Schweiker, 652 F.2d 146 (D.C. Cir. 1980). Rather, the Court must consider the cumulative effect of the additional process. Id. As previously discussed, EPA issued 1,705 UAOs to more than 5,400 PRPs in the 285 months from August 16, 1982 to May 25, 2006. See Rouhani Report at ¶¶ 4.3, 6.5.2. That is a rate of six UAOs issued to nineteen PRPs every month. Because UAOs are normally only issued at the conclusion of an extended negotiation period, many (if not all) PRPs would likely invoke the opportunity for a hearing before a presiding officer. In the aggregate, then, the requirement of a presiding officer would certainly constitute a substantial financial and administrative cost to the government.

On the other hand, a pre-decision hearing by an agency presiding officer would not significantly reduce the risk of error. As demonstrated in this pattern and practice case by the low rate of error, the risk of error is already very small. The cost of additional process must indeed be minimal to justify adding to a process that already results in an acceptably low risk of error. See Tillman v. Lebanon Cty. Correctional Facility, 221 F.3d 410, 422 (3d Cir. 2000). However, as discussed above, the cost of a presiding officer is not as minimal as it might initially appear.

Finally, the size and nature of the private interests of the PRP are not so great as to justify increased government costs with only a marginal improvement in the rate of error. Although the privates interest are significant, GE has not demonstrated that the primarily financial interests at

52

stake here routinely have such collateral effects as to constitute a "brutal need" requiring greater pre-deprivation process. See Arnett v. Kennedy, 416 U.S. 134, 169 (1974). Moreover, any benefit to those private financial interests would be offset somewhat by the additional costs of the enhanced pre-issuance process -- full adversarial hearings with legal representation, even if just before an agency official, are expensive. Hence, because the benefit of additional process is low and the cost of that process is fairly high, the Court finds that a hearing before a presiding officer is not constitutionally necessary.

This result does not change when the pre-issuance hearing is before an ALJ rather than a presiding officer. The Court's weighing of the private interests remains unchanged. The type of hearing does not impact the size and nature of the deprivation. Nor would the impact of a hearing before an ALJ change the Court's rate of error analysis. It is hard to reduce the rate of error when it is very low to begin with. Only the analysis of the government interest changes when considering a hearing before an ALJ instead of a presiding officer. But a pre-issuance hearing before an ALJ would increase the burden on the government interest. Hearings before an ALJ are more formal and have more of the trappings of the judicial process than informal hearings before a presiding officer. Hearings before an ALJ would undoubtedly be lengthier and more time- consuming. Because these hearings frequently would be invoked by PRPs, requiring an ALJ would constitute an even greater financial and administrative cost to the government than a hearing before a presiding officer. Based on the balancing of these factors, then, the Court concludes that the Constitution does not require a pre-issuance hearing before an ALJ either.

In summary, although GE has presented evidence of isolated errors by EPA, such infrequent errors -- constituting a reasonably low overall risk of error -- do not warrant the

53

sweeping changes GE requests. Errors should be addressed by a PRP when they occur -- either by not complying with a UAO and defending a subsequent enforcement proceeding or by complying with a UAO and seeking post-completion reimbursement. Those avenues remain available to PRPs under CERCLA as a more effective means to address the occasional errors revealed by the record before this Court. To the extent that GE continues to believe that EPA generally overuses or abuses UAOs, thereby overstepping its mandate, any broader remedy should be sought from Congress, not the courts.

CONCLUSION

In their motions for summary judgment, GE and EPA focus on two due process arguments: unconstitutional coercion under Ex parte Young and unconstitutional deprivation of property or liberty under Mathews v. Eldridge. For the reasons explained above, and based on the extensive record developed through years of discovery, the Court agrees with EPA on both points and concludes that GE has not shown that EPA's pattern and practice of administering section 106 of CERCLA violates due process. Therefore, GE's motion for summary judgment is denied and EPA's motion for summary judgment is granted. A separate order has been issued on this date.

/s/ John D. Bates

JOHN D. BATES

United States District Judge

Dated: January 27, 2009