National Mining Association v. Bruce Babbitt, Secretary, United States Department of Interior

Opinion for the Court filed by Circuit Judge SILBERMAN. SILBERMAN, Circuit Judge: National Mining Association challenges four regulations promulgated by the Secretary of the Interior (Office of Surface Mining Reclamation and Enforcement) as part of a package of regulations governing damage to land, structures, and certain water supplies caused by mining subsidence. The district court rejected appellant’s claims. We, however, agree with appellant that two of the agency’s regulations are arbitrary and capricious. I. The Surface Mining Control and Reclamation Act, 30 U.S.C. §§ 1201 et seq. (1994), sets forth permit requirements and performance standards for coal mining operations. An important aspect of this statutory scheme is its regulation of subsidence caused by underground mining. Subsidence, as used in the Act, apparently refers to the kind of earth movement that occurs “when a patch of land over an underground [coal] mine sinks, shifts, or otherwise changes its configuration.” National Wildlife Fed’n v. Hodel, 839 F.2d 694, 739 (D.C.Cir.1988); see also Keystone Bituminous Coal Ass’n v. DeBenedictis, 480 U.S. 470, 474-75, 107 S.Ct. 1232, 94 L.Ed.2d 472 (1987) (discussing coal mine subsidence and its effects). Although the word subsidence literally means lowering, tending downward, or “flattening] out so as to form a depression,” Webster’s Third New Int’l Dictionary 2279 (1971), and *909 there is no definition of the term in the statute or the regulations, the parties agree it is used only to describe the kind of subsidence caused by underground coal mining. For purposes of this case, we accept that definition. The central- statutory provision governing “subsidence” provides that an underground mining permit issued by an approved State or Federal program must require the operator to “adopt measures consistent with known technology in order to prevent subsidence causing material damage to the extent technologically and economically feasible, maximize mine stability, and maintain the value and reasonably foreseeable use of such surface lands....” 30 U.S.C. § 1266(b)(1) (1994). Subsidence regulation under the Mining Act has had various incarnations and has generated a fair amount of litigation before us. See, e.g., National Wildlife Fed’n v. Lujan, 928 F.2d 453, 455-60 (D.C.Cir.1991); National Wildlife Fed’n v. Hodel, 839 F.2d at 739-41; In re Permanent Surface Mining Regulation Litig., 653 F.2d 514 (D.C.Cir.1981) (en banc). In the aftermath, Congress added a new § 720 to the Mining Act, see Energy Policy Act of 1992, Pub.L. No. 102-486, see. 2504(a)(1), § 720, 106 Stat. 2776, 3104 (1992), which provides: (a) Requirements. Underground coal mining operations conducted after October 24, 1992, shall comply with each of the following requirements: (1) Promptly repair, or compensate for, material damage resulting from subsidence caused to any occupied residential dwelling and structures related thereto, or non-commercial building due to underground coal mining operations. Repair of damage shall include rehabilitation, restoration, or replacement of the damaged occupied residential dwelling and structures related thereto, or noncommercial building. Compensation shall be provided to the owner of the damaged occupied residential dwelling and structures related thereto or noncommercial building and shall be in the full amount of the diminution in value resulting from the subsidence. Compensation may be accomplished by the purchase, prior to mining, of a noncan-cellable premium-prepaid insurance policy- (2) Promptly replace any drinking, domestic, or residential water supply from a well or spring in existence prior to the application for a surface coal mining and reclamation permit, which has been affected by contamination, diminution, or interruption resulting from underground coal mining operations. Nothing in this section shall be construed to prohibit or interrupt underground coal mining operations. (b) Regulations. Within one year after October 24, 1992, the Secretary shall, after providing notice and opportunity for public comment, promulgate final regulations to implement subsection (a) of this section. 30 U.S.C. § 1309a (1994). In order to implement this new statutory provision, the Secretary in 1993 proposed subsidence regulations revising the subsidence regulations previously promulgated under the Mining Act. See 58 Fed.Reg. 50,174 (1993). After a notice and comment period, the Secretary modified the proposed regulations and issued them in final form in 1995. See 60 Fed.Reg. 16,722 (1995). II. Appellant National Mining Association brought this action in the district court challenging 10 parts of the new regulations as arbitrary and capricious, see 30 U.S.C. § 1276(a)(1); 5 U.S.C. § 706(2)(A), and moved for summary judgment. The Secretary, along with intervenors National Wildlife Federation and Kentucky Resources Council, Inc., filed cross-motions for summary judgment, which the district court granted. The Association limits its appeal to four of the district court’s rulings. We consider them in turn. A. The Angle of Draw Presumption The Association’s most vigorous challenge is to the regulation establishing a rebuttable presumption of causation: *910 If damage to any non-commercial building or occupied residential dwelling or structure related thereto occurs as a result of earth movement within an area determined by projecting a specified an-' gle of draw from the outermost boundary of any underground mine workings to the surface of the land, a rebuttable presumption exists that the permittee caused the damage. The presumption will normally apply to a 30-degree angle of draw. 30 C.F.R. § 817.121(c)(4)(i) (1998). As the agency explained, the angle of draw “is the angle of inclination between the vertical at the edge of the underground mine workings and the point of zero vertical displacement at the edge of a subsidence trough.” 60 Fed.Reg. at 16,738. It “is one way to define the outer boundary of subsidence displacement that may occur at the surface.” Id. Once the presumption is triggered, the burden shifts to the mining company to offer evidence that the damage is attributable to another cause. The regulation suggests some examples: that the “damage predated the mining in question; the damage was proximately caused by some other factor or factors and was not proximately caused by subsidence; or the damage occurred outside the surface area within which subsidence was actually caused by the mining in question.” 30 C.F.R. § 817.121 (c)(4)(iv). Appellant claims that this presumption actually shifts the burden of proof to mining companies to show a negative — that their operations did not cause subsidence — and that such a shift runs afoul of the Administrative Procedure Act given the agency’s posture in such situations as proponents of an order. See 5 U.S.C. § 556(d) (1994); Director, Office of Workers’ Compensation Programs, Dep’t of Labor v. Greenwich Collieries, 512 U.S. 267, 281, 114 S.Ct. 2251, 129 L.Ed.2d 221 (1994). The government responds that the agency’s regulation merely shifts the burden of production not the burden of persuasion, and it is only the latter that the APA forbids. See id. at 279-80, 114 S.Ct. 2251. Although we recognize that at a certain point along an evidentiary continuum a shift in the burden of production can become de facto a shift in the burden of persuasion, we do not think it is necessary in this case to draw the line. For a factual presumption that causes a shift in the burden of production must be reasonable (as we explain below, this means essentially that the circumstances giving rise to the presumption must make it more likely than not that the presumed fact exists, see Secretary of Labor v. Keystone Coal Mining Corp., 151 F.3d 1096, 1100-01 (D.C.Cir.1998)). For two reasons, the agency’s presumption fails that test. The first is that the nature of subsidence evidence that triggers the presumption has become hopelessly confused in this litigation, and the second is that the geographical boundary in which the presumption obtains — the angle of the draw — is irrationally broad. The regulation states that the presumption is employed wherever damage to a structure covered by the Energy Policy Act “occurs as a result of earth movement” within the angle of the draw, 30 C.F.R. § 817.121(c)(4), which gives rise to the question: What is the conceptual relationship between “subsidence” (which, it will be recalled, the parties agree refers to mining subsidence) and “earth movement”? On appeal, contrary to its position in the district court, the government explains that “earth movement” does not mean any earth movement but only “earth movement consistent with subsidence” and that any resulting damage to covered structures (also necessary to trigger the presumption) is also consistent with the kind of damage caused by subsidence. Appellant cries foul. It claims the government cannot shift the meaning of the regulation during litigation. Before the district court the government appeared to argue that proof of subsidence in fact (that is, coal mining subsidence) was necessary *911 to trigger the presumption. It is not uncommon, however, when an agency regulation is challenged in a facial attack, as Congress permits parties to do, that the meaning of disputed provisions does not appear clearly until the case is before the courts of appeals. We typically accept the agency’s construction — which often eliminates or narrows the dispute — because we recognize the agency is entitled to deference as to the meaning of its own regulation. Auer v. Robbins, 519 U.S. 452, 462-63, 117 S.Ct. 905, 137 L.Ed.2d 79 (1997); Serono Laboratories, Inc. v. Shalala, 158 F.3d 1313, 1325 (D.C.Cir.1998); Association of Bituminous Contractors, Inc. v. Apfel, 156 F.3d 1246, 1251-52 (D.C.Cir.1998); Paralyzed Veterans of Am. v. D.C. Arena L.P., 117 F.3d 579, 584 (D.C.Cir.1997), ce rt. denied sub nom. Pollin v. Paralyzed Veterans of Am., - U.S. -, 118 S.Ct. 1184, 140 L.Ed.2d 315 (1998). Yet, we are entitled to expect that at least by the time of oral argument the agency will have settled on an interpretation that reflects its “fair and considered judgment,” Auer, 519 U.S. at 462, 117 S.Ct. 905 — and that such an interpretation is understandable. The government assures us that, despite some vacillation, it has come to rest, but, unfortunately, if it has, we do not understand its conceptual resting point. Pressed at oral argument to define just what “earth movement consistent with subsidence” means, the government, with manifest circularity, explained that the term refers to earth movement not caused by non-subsidence (i.e., earth movement not caused by earthquakes, floods, etc.). Establishing the cause of the damage, however — whether subsidence or something else — is the whole purpose of the evidentiary presumption. Indeed, as appellant correctly observes, one of the ways in which a mining company rebuts the presumption is to produce evidence that the damage was proximately caused by some factor other than mining. See 30 C.F.R. § 817.121(c)(4)(iv). It thus does not help to define the trigger for the presumption in terms of the presumption’s intended result. In short, we have no clue what “earth movement consistent with subsidence” means and the government’s efforts to enlighten us created instead hopeless confusion. We, of course, cannot approve a regulatory presumption that the agency’s lawyers cannot interpret in an intelligible fashion. But even if we did understand the relationship between subsidence and earth movement, as those terms are used in the regulation, it could not stand because the geographical boundary of the area in which the presumption operates— the angle of the draw — is both arbitrary and capricious. That is to say, the presumption simply does not serve as a reasonable proxy to explain subsidence damage to structures. Appellant contends that the record reveals that the kind of subsidence that occurs within the angle of draw (pure vertical subsidence) is not the kind of subsidence (differential vertical and horizontal displacement) that ordinarily results in damage to surface structures. The very authors of the scientific studies on which the agency relied in its proposed rulemak-ing, the Association points out, have specifically rejected the predictive value of the angle of draw. And, even this evidence does not support the agency’s decision to base a natiomoide presumption of causation on the angle of draw. The nationwide presumption is further deficient, it is argued, because it overlooks the key distinction between high extraction and partial extraction mining (damage to structures from the latter method is said to be unlikely); it ignores the critical variable of mining depth (the deeper the mine, the greater surface area is covered by the 30-degree angle of draw, yet the less likely is actual structural damage at the surface); and it ignores other non-subsidence causes of earth movement that are just as *912 likely to cause damage to structures within the angle of draw. 1 The government’s response is anemic. It emphasizes that the regulation permits a state regulatory authority to petition the Department for a different angle if it can demonstrate that its proposed angle is more reasonable, see 30 C.F.R. § 817.121(c)(4)(i), and also permits a mining company to request a different site-specific angle if it too could demonstrate a more reasonable calculation, see id. § 817.121(c)(4)(ii). But the government does not claim—nor could it—that these safety valve provisions could save the regulatory presumption if we thought it unreasonable. As we have said repeatedly, an evidentiary presumption is “only permissible if there is a sound and rational connection between the proved and inferred facts, and when proof of one fact renders the existence of another fact so probable that it is sensible and timesaving to assume the truth of [the inferred] fact ... until the adversary disproves it.” Keystone Coal Mining, 151 F.3d at 1100-01 (quoting Chemical Mfrs. Ass’n v. Department of Transp., 105 F.3d 702, 705 (D.C.Cir.1997) (quoting NLRB v. Curtin Matheson Scientific, Inc., 494 U.S. 775, 788-89, 110 S.Ct. 1542, 108 L.Ed.2d 801 (1990))) (emphasis added) (internal citation and internal quotation marks omitted) (alterations in original). “If there is an alternate explanation for the evidence that is also reasonably likely, then the presumption is irrational.” Id. We think the government has failed to justify its presumption. It has not offered any support, scientific or otherwise, that even begins to establish that the angle of draw delimits the surface area -within which it is logical or reasonable to employ an evidentiary presumption of causation. Indeed, the government apparently eon-cedes that it is not supported by available science. This is not surprising since, as appellant argues, science seems more supportive of the view that the angle of draw has nothing whatever to do with identifying subsidence-caused damage to structures. A leading textbook in the field, which the government paradoxically listed as support for the presumption, see 60 Fed. Reg. at 16,738, defines the angle of draw in such a way as to seriously undermine (pun intended) the government’s position. See Syd S. Peng, Coal Mine GROUND Control 422-23 (2d ed.1986) (defining angle of draw, noting that it varies from 15 to 45 degrees, and then stating: “The angle of draw is more or less of academic interest. Because the subsidence profile levels off and subsidence becomes very small far before it reaches the edges of the subsidence basinf,] from [a] structural damages point of view, it is practically meaningless.”) (emphasis added). Nonetheless, the government thinks it sufficient to explain that the angle of draw is used merely “to define the boundaries of the area within which earth movement resulting from subsidence, if any, will most probably occur.” But the real question in using the angle of draw to set the applicable boundaries is whether subsidence-caused damage to structures within the angle is more likely than not to occur. After all, the mining company’s potential liability is for causing damage to protected structures, not for causing “earth movement resulting from subsidence.” Unless the government can establish a likely connection between the former and the angle of draw, the presumption cannot stand. The agency rejected the suggestion that the presumption be limited to the so-called “angle of critical deformation”—a smaller *913 angle, within the angle of draw, that measures the inclination from the edge of the underground mining area to the surface point exhibiting the “maximum tensile strain” or stretching. See 60 Fed.Reg. at 16,738. (Dr. Peng, a leading expert in the field, estimates that the angle of critical deformation is on average 10 degrees smaller than the angle of draw. See Peng at 423.) Although the government concedes that subsidence-caused damage to structures within the angle of critical deformation portion of the angle of draw is more likely to occur than in the portion of the angle of draw beyond the angle of critical deformation, it contends that its decision to use the entire angle of draw is justified on “possibility” grounds — the larger angle defines the “outer boundary of subsidence displacement that may occur at the surface.” (Neither in its brief nor in the preamble to the final rule, see 60 Fed.Reg. at 16,738-39, we should note, has the government offered any scientific support for even the possibility of subsidence-caused damage within the angle of draw.) To impose a presumption of causation of damage on a party based merely on the possibility that the party caused the damage is to convert a factual presumption into a counterfactual presumption. Although the government has never wavered from its insistence that the presumption is a factual one, we do not see how a counter-factual procedural device could be justified even as a matter of policy, see Allentown Mack Sales & Serv., Inc. v. NLRB, 522 U.S. 359, -, 118 S.Ct. 818, 828, 139 L.Ed.2d 797 (1998), since the statute imposes liability only for causation. Moreover, the government simply has no response at all to the devastating objection that the angle of draw is an inherently illogical measure since, as a matter of geometry, the deeper the mine the wider the angle. Yet, it seems undisputed that the likelihood of any structural damage on the surface decreases with the depth of the mine. Since the agency recognized, in its preamble to the final rule, see 60 Fed.Reg. at 16,740, that the depth and location of the mine (and the amount of coal extracted) are all factors that bear on the likelihood of subsidence-caused damage, the Department never even adequately explains why it wishes to employ any sort of nationwide presumption. Essentially the government argues that its presumption is justified by efficiency; it is easier to establish a mine’s liability. There are limits to that justification, otherwise the government could dispense with enforcement proceedings altogether. To be sure, we would be obliged to defer to a reasonable agency determination of probabilities — including predictions based on its own expertise and policies. But we see nothing of the sort here. We have no difficulty concluding that this regulation is both arbitrary and capricious because it is irrationally overbroad, and we therefore vacate it. B. Pre-Subsidence Survey Appellant challenges the regulation requiring all applicants for a mining permit to conduct, inter alia, a pre-subsidence “survey of the condition of all non-commercial buildings or occupied residential dwellings and structures related thereto, that may be materially damaged or for which the reasonably foreseeable use may be diminished by subsidence, within the area encompassed by the applicable angle of draw.” 30 C.F.R. § 784.20(a)(3) (1998). Appellant’s principal objection is that, when promulgating this regulation in 1995, the agency failed adequately to explain its deviation from its prior policy not to require pre-subsidence surveys. The government responds with two internally inconsistent arguments: that the change in policy was justified (if not mandated) by the Energy Policy Act of 1992, and that the requirement of a presubsidence survey of the condition of structures is not a change of position at all because the regulation has always implicitly required such a survey. We take this latter argument as an alternative one (though it would have *914 been helpful if counsel had designated it as such). Even so, it is plainly wrong, and we are surprised that the government would choose to advance it. It cannot seriously be maintained that prior to 1995 the regulation required a pre-subsidence survey. The government admits as much in describing the preEnergy Policy Act regulatory regime: beginning in 1983, "[i]nstead of conducting a pre-subsidence survey, the applicant was now only required to identify any lands or structures that could be materially damaged by subsidence." And in the preamble to the proposed version of the current regulation, the Secretary explained that "the survey itself [required by the prior regulation] is proposed to be changed from a mere inventory of structures, and renewable resource lands, to a survey of the condition of structures, facilities, and surface features." 58 Fed. Reg. at 50,179 (emphasis added). The government's argument that a pre-subsidence survey of the condition of structures has always been implicitly required is palpably in conflict with its own account of the regulatory history. Much the better argument is that the change in policy was justified by Congress' explicit instruction to the Secretary in the Energy Policy Act to promulgate regulations to implement the new Act, which conferred greater protection to structures and land from subsidence-caused damage. Once again, however, the government overplays its hand and suggests that Congress' instruction to promulgate new regulations completely absolves the agency of the requirement to supply a reasoned explanation for its change in policy. The government relies for this proposition on our decision in City of Las Vegas v. Lujan, 891 F.2d 927 (D.C.Cir.1989), in which we stated that "[a] change in congressional instructions, of course, absolves the Secretary from explaining why his policy has changed, if indeed it has," id. at 934. But Lvjan involved a specific congressional instruction to take a precise agency action-issuing emergency regulations-more readily than the Secretary previously had seen fit to take. See id. The Energy Policy Act does not refer to pre-subsidence surveys at all, and so we do not see how Congress' general instruction to implement statutory protection of structures through a repair or compensate obligation absolves the government of explaining its decision to require a pre-subsidence survey that it previously had opted against. Be that as it may, the agency did actually say in the preamble to the final rule that the new policy was necessary "to effectively implement the requirements of the Energy Policy Act," 60 Fed.Reg. at 16,730, and more specifically that the information gathered was "essential to establish a baseline against which the effects of subsidence may be measured and to ensure full implementation [of the Energy Policy Act]," id. at 16,729. Consistent with the notion that the Energy Policy Act's mandate justified the Secretary's decision to increase mining companies' information-gathering responsibilities, the agency limited the survey requirement to structures and water supplies protected by the Energy Policy Act. See id. at 16,730. Greater protection for structures (albeit only through a repair or compensate obligation) was the main impetus of the Energy Policy Act, and we think the agency's reliance on that impetus is a satisfactory explanation for its change of policy. Appellant also contends that the agency failed to respond adequately to comments that the regulation is overly burdensome, and that it irrationally requires information about the condition of structures at the time of the permit application, rather than at the time (often years later) when mining operations actually commence. In our view, the agency has said enough and its regulations are not unreasonable. The Secretary specifically responded to the comments complaining about the cost of the survey by limiting the structures to which the requirement applied, and by modifying some of the mapping requirements of the survey. See id. *915 We do not agree with appellant, moreover, that the timing of the survey (pre-application as opposed to pre-mining) is without purpose. As the government points out, the results of the presubsidence survey are used to determine whether the applicant is also required to submit a subsidence control plan with the application. See 30 C.F.R. § 784.20(b). The agency concluded that, based on its experience in this area, “the proposed format for the survey information is the minimum needed to adequately assess the need for a subsidence control plan.” 60 Fed.Reg. at 16,730. The government also explains that the possible effects of subsidence is a relevant factor in its determination of whether to grant a permit in the first place. We see no reason why the agency should be precluded from requiring all of this information at the time of the application just because conditions might change before mining begins. As the Secretary explained, the mining company can always supplement outdated information later in the process. See 60 Fed.Reg. at 16,730. Nevertheless, the regulation as currently written must be vacated along with the first one discussed because it defines the area within which the pre-subsidence survey is required by reference to the angle of draw. The government did not argue that the survey requirement could be sustained independent of the angle of draw — and we do not see how it could. C. Planned Subsidence Minimization The Association next challenges two provisions of the new regulations that impose an obligation on mining permittees who use a “planned subsidence” mining technique to minimize subsidence damage. If a permittee employs mining technology that provides for planned subsidence in a predictable and controlled manner, the permittee must take necessary and prudent measures, consistent with the mining method employed, to minimize material damage to the extent technologically and economically feasible to non-commercial buildings and occupied residential dwellings and structures related thereto.... 30 C.F.R. § 817.121(a)(2); see also 30 C.F.R. § 784.20(b)(7) (providing that mining companies required to submit a subsidence control plan and projecting to use planned subsidence must, with certain exceptions, include “a description of methods to be employed to minimize damage from planned subsidence to non-commercial buildings and occupied residential dwellings and structures related thereto”). Planned subsidence refers to mining methods that make it possible to predict the time and manner of the resulting subsidence (one such method is “longwall mining”). Appellant disputes the regulation’s damage minimization requirement for planned subsidence on the ground that such a requirement is contrary to § 516(b)(1) of the Mining Act, which provides that any permit issued must require the operator to adopt measures consistent with known technology in order to prevent subsidence causing material damage to the extent technologically and economically feasible, maximize mine stability, and maintain the value and reasonably foreseeable use of such surface lands, except in those instances where the mining technology used requires planned subsidence in a predictable and controlled manner.... 30 U.S.C. § 1266(b)(1) (1994) (emphasis added). Since Congress explicitly approved long-wall mining, notwithstanding that, by definition, it causes subsidence, appellant argues that the regulation would frustrate Congress’ intent. The government’s minimization of damage requirement, we are told, is merely a paraphrase of the prevention of damage requirement from which longwall mining is exempted. The government argues that it is entitled to Chevron deference in interpreting the statutory lan *916 guage for the phrase “predictable and controlled manner” can be interpreted as a manner that restricts collateral damage. Appellant, by contrast, would read predictable and controlled manner as not imposing any separate obligation but as words that simply describe longwall mining. If we were interpreting the statute de novo, we might well agree that appellant has the better argument. But we are not. And although the government’s reading is a bit of a stretch, we think it passes the Chevron test. See Chevron U.S.A. Inc. v. Natural Resources Defense Council, Inc., 467 U.S. 837, 842-43, 104 S.Ct. 2778, 81 L.Ed.2d 694 (1984). D. Waivers The Association’s last challenge is to the regulation that implements the Energy Policy Act’s “repair or compensate” obligation. See 30 C.F.R. § 817.121(c)(2) (“The permittee must promptly repair, or compensate the owner for, material damage resulting from subsidence caused to any noncommercial building or occupied residential dwelling or structure related thereto that existed at the time of mining.”). The Association asserts that the regulation is unreasonable “to the extent it purports to nullify prior agreements between owners of eligible structures and underground mine operators.” In appellant’s view, nothing in the Energy Policy Act suggests an intent to override such waiver agreements that would otherwise be binding under state common law. Moreover, it is urged, if the statute were read to authorize abrogation of waiver agreements, it would confer a windfall to the landowner, and consequently might constitute an unconstitutional taking of the mining company’s contract rights. The thrust of appellant’s argument focuses on the unfairness (and asserted conflict with the statutory language) caused by a “double recovery” regulatory regime, under which a landowner is compensated pre-subsidence damage (and possibly pre-Energy Policy Act) by selling a waiver of rights to the mining company, and then post-subsidence, postdamage through an enforcement action under the Energy Policy Act. The Association contends that Congress could not possibly have intended such a result. At oral argument, however, the government stated unequivocally (which it did not do in its brief or in the preamble to the regulation) that any compensation owed to a landowner under the Act will always be reduced by at least the amount previously paid in a contractual waiver of subsidence rights-both pre- damage, pre-Act waivers, and pre-damage, post-Act waivers. 2 In other words, the value of the landowner’s rights under the federal regime may well exceed (indeed typically would) the present value of what- ever compensation the landowner received under the previous legal climate. But the landowner would not be entitled to a dou- ble recovery. The government’s oral clari- fication seems to strip appellant’s chal- lenge of its force and appellant waived rebuttal. Nonetheless, the Association’s brief appeared to argue that such waivers must be honored fully, and that the set-off approach also is inconsistent with the Act If appellantmeans to deny any obli- gation to compensate beyond the amount a mining company originally paid the land- owner for the waiver, no matter how it would compare to the landowner’s legal rights post passage of the federal statute, we reject its position. We previously up- held the government’s limitation of the occupied residential *917 obligation to repair or compensate for damage to structures only to the extent required by state law, see National Wildlife Fed’n v. Lujan, 928 F.2d at 457-59, in part because the Mining Act at the time did not explicitly impose an obligation to compensate for such damage, see id. at 458 n. 3. The Energy Policy Act imposes just such an obligation on its face. See 30 U.S.C. § 1309a(a)(l) (“Compensation shall be provided to the owner of the damaged occupied residential dwelling and structures related thereto or non-commercial building and shall be in the full amount of the diminution in value resulting from the subsidence.”) (emphasis added). It is therefore wholly consistent with the statute — indeed it might even be mandated— for the Secretary to require the mining companies further to compensate landowners for damages to which the new federal law entitled them. That is not to say, of course, that a landowner and mining company would be barred from entering into a post-Act fair contract based on anticipated damages that would extinguish the landowner’s claim if the damages turned out to be more than anticipated. But we do not understand the agency to deny that. As for appellant’s takings challenge, the argument is insufficiently developed to warrant much attention. Appellant seems to assume that interference with contract rights is a per se taking, despite the well-settled rule that “legislation [that] disregards or destroys existing contractual rights does not always transform the regulation into an illegal taking.” Connally v. Pension Benefit Guar. Corp., 475 U.S. 211, 224, 106 S.Ct. 1018, 89 L.Ed.2d 166 (1986). To develop a real takings argument, appellant would be compelled to demonstrate why “(1) the economic impact of the regulation on the claimant; (2) the extent to which the regulation has interfered with investment-backed expectations; and (3) the character of the governmental action,” id. at 225, 106 S.Ct. 1018 (quoting Penn Cent. Transp. Co. v. City of New York, 438 U.S. 104, 124, 98 S.Ct. 2646, 57 L.Ed.2d 631 (1978)) (internal quotation marks omitted), warrant the conclusion that an unconstitutional taking necessarily would result. See also Eastern Enters. v. Apfel, 524 U.S. 498, -, 118 S.Ct. 2131, 2146-49, 141 L.Ed.2d 451 (1998) (reviewing takings precedents and concluding that “Congress has considerable leeway to fashion economic legislation, including the power to affect contractual commitments between private parties”). As we understand appellant’s argument, based solely on the possibility that the government’s alteration of a mining company’s previously settled contract rights “could expose” the government to liability for an unlawful taking, we should construe the Energy Policy Act to mandate an exemption in the Secretary’s regulations for private waiver agreements. But the avoidance canon is not applicable when the statute or regulation would effect a taking, if at all, only in certain situations. See United States v. Riverside Bayview Homes, Inc., 474 U.S. 121, 127-28, 106 S.Ct. 455, 88 L.Ed.2d 419 (1985); Bell Atlantic Tel. Cos. v. FCC, 24 F.3d 1441, 1445 (D.C.Cir.1994); Railway Labor Executives’ Ass’n v. United States, 987 F.2d 806, 816 (D.C.Cir.1993). We will not “frustrate[] permissible applications of a statute or regulation,” Riverside Bayview Homes, 474 U.S. at 128, 106 S.Ct. 455, based on the specter — rather implausible from what we can tell now — of a future unconstitutional taking. For the foregoing reasons, we reverse the district court in part and affirm- in part. Footnotes [1] Appellant also claimed that the Department unlawfully relied on studies not mentioned in the proposed rule. But informal rulemaking does not contemplate a closed record; the government is entitled to rely on information not exposed to comment so long as it is supplementary. See generally Air Transport Ass'n of America v. FAA, 169 F.3d 1, 6-7 (D.C.Cir.1999). The real problem with the studies is that they are inadequate support for the presumption. [2] In the preamble, to be sure, the Secretary stated that ''[t]he use of pre- and post-subsidence agreements would be an acceptable means of fulfilling the requirement so long as the terms meet the requirement under paragraph 817.121(c)(2) that the permittee repair or compensate any subsidence-related material damage to any non-commercial building or dwelling or related structure.” 60 Fed. Reg. at 16,735 (emphasis added). Although the Secretary did not make clear that this policy applies to both pre-Act and post-Act waivers, the government's posi- tion at oral argument was that it does. We accept the agency's interpretation of its own regulation. own regulation.