|
31 ELR 10371 | Environmental Law Reporter | copyright © 2001 | All rights reserved Chevron, State Farm, and EPA in the Courts of Appeals During the 1990sChristopher H. Schroeder and Robert L. GlicksmanChristopher H. Schroeder is Professor of Law and Public Policy Studies, Duke University School of Law. Robert L. Glicksman is the Robert W. Wagstaff Professor of Law, University of Kansas School of Law. The authors thank Robert Drumm, University of Kansas School of Law, Class of 2002, and Charles Larsen, Nicole Wilson, Duke University School of Law, Class of 2001, and Ryan Vogt-Lowell and Greg Caplan, both of the Duke Class of 2002, for their valuable research assistance. The authors also thank the participants in the conference, "EPA at 30," held at Duke University School of Law, Dec. 7-8, 2000, for their helpful suggestions, with particular thanks to Jody Freeman and Mary Beth Ward, who were commenters on a version of this Article which we presented there. Professor Schroeder has presented parts of this work to the Environmental Policy Workshop at Duke University, to the "Sustainable Governance" conference at Duke, and to a General Counsel's lecture at the U.S. Environmental Protection Agency. Our thanks to attendees at each of these events. Thanks also to Jay Hamilton, Sidney Shapiro, and Richard Levy for valuable advice. Statistical analyses in this Article were done by Emily Schroeder, who merits much appreciation for her assistance, and who, along with the others who provided advice, bears no responsibility for any errors. Part of the contents of this Article will be adapted as a chapter in the forthcoming book, EPA at Thirty, to be published by RFF Press/Resources for the Future, 2001. I. Introduction Ten years ago, we analyzed how environmental policy, primarily as formulated by the U.S. Environmental Protection Agency (EPA), had fared in federal courts during the first two decades of the Environmental Era.1 Our primary interest then was to determine how emerging environmental values were being accommodated by federal courts when those courts reviewed administrative actions by EPA and other agencies charged with new environmental responsibilities. This Article updates, extends, and refines our earlier work using a different methodology and with related, but also different, objectives. In this Article we analyze how EPA and the federal courts interact under two broad aspects of judicial review. First, we study EPA's and the courts' interpretations of the Agency's statutory authority, with special emphasis on how the Agency has fared under the Chevron, U.S.A., Inc. v. Natural Resources Defense Council, Inc., (Chevron)2 doctrine. Second, we study how well EPA's rulemaking satisfies the requirements of "reasoned decisionmaking" as set forth in [31 ELR 10372] Citizens to Preserve Overton Park, Inc. v. Volpe (Overton Park),3 Motor Vehicle Manufacturers Ass'n of the United States v. State Farm Mutual Automobile Insurance Co. (State Farm)4 and other decisions. The Article thus focuses on how EPA has fared under judicial review, rather than on how environmental values have been treated by the federal courts. In subsequent work, we plan to use these same data to return to this latter question, which was the subject of our earlier work, an effort that will include examining the ways in which political preferences of the judges affect their decisions.5 II. Overall Characteristics of the Data Set We reviewed all of the cases in the federal appellate courts in which EPA or its Administrator appeared as a party, and which were decided between January 1, 1991 and August 1, 1999.6 Our initial Westlaw search produced 377 appellate decisions during the study period. Some of these decisions provide little or no information other than the parties and the disposition of the matter.7 Where indicated below, we retained these decisions when calculating disposition rates, although they necessarily drop out of the analysis when we turn to studying substantive or procedural decisions made by EPA in the course of conducting primary Agency business. Still other decisions resolved disputes that we viewed as entirely internal to the litigation process, such as decisions on motions to consolidate appeals, motions to stay mandates temporarily, motions for rehearing en banc, consensual motions to dismiss as moot and the like. Because we were interested in the interaction between EPA and the courts when some element of Agency decisionmaking on environmental issues was being challenged, we excluded these decisions from any further analysis. Excluding these litigation-related dispositions left us with 336 decisions. In addition to analyzing these cases from the 1990s, we also sampled cases from the 1980s by identifying all the cases in the federal appellate courts in which EPA or its Administrator appeared as a party, and which were decided between January 1, 1986 and January 1, 1988. This is not a random sample of 1980s cases, but it does capture the entirety of two years relatively soon after State Farm and Chevron had been decided. This sample provides us with one measure of whether the interaction between the EPA and the appellate courts under the Chevron and reasoned decision making doctrines has changed over time. Following the same procedures for our 1980s period as we did for the 1990s, our search produced a set of 80 decisions after winnowing. Table 1 presents the overall comparison.
Several characteristics of the entire data set of 1990s cases merit some consideration before we turn to more detailed analysis. First, the reversal rates in our cases from the 1990s and the 1980s show that EPA has fared somewhat better in the 1990s, and that these differences are statistically significant at the 95% confidence level. In our two-year sample of 1986-1987 cases, EPA prevailed 60% of the time, was reversed and remanded 35% of the time, and just remanded 5% of the time. In the 1990s, the comparable figures are 67%, 21%, and 12%. These figures suggest that the "Chevron effect"—an increase in affirmances due to more deference to agency determinations—which some thought might decline as the passage of time reduced the impact of the opinion8 has persisted into the 1990s and may even have strengthened. Second, the increased percentages of remand-only decisions are consistent with the observations of administrative law scholars that the D.C. Circuit in particular has increased its use of the technique of remanding without reversing, a procedure that can permit EPA to correct errors while much of the challenged rule remains in effect. If we disaggregate these figures, which include both adjudications and rulemakings, to focus exclusively on rulemaking challenges, they indicate that EPA is distinctive among federal administrative agencies. Of the 336 cases in the 1990s, 33% of the total (111) consisted of challenges to rulemaking. This is almost identical to the proportion of rules in our 1986-1987 sample, in which 34% (27/80), were rulemaking challenges. One source for comparing these figures with other agencies is the Schuck and Elliott study of the effect of the Chevron decision on the courts of appeals.9 In the course of their analysis, Schuck and Elliott estimated the overall distribution of challenges to rules and to adjudications in the periods they sampled. Summing across all administrative agencies, they found that rulemaking challenges comprised just 1% of the challenges to agency action in 1965, 12.3% of the total in 1975, and 9.4% of all challenges in 1984-1985.10 Thus, challenges to EPA rules in the 1980s and the 1990s were a much larger proportion of total [31 ELR 10373] appellate decisions involving the Agency than was the case with respect to agencies overall during any period that Schuck and Elliott examined. This difference is quite likely due to the relatively greater amount of rulemaking activity undertaken by EPA compared to other federal agencies. A check of all final rules issued by EPA in the 1990s shows that rulemaking activity by EPA increased in the 1990s compared to the 1980s.11 At first glance, this seems inconsistent with the so-called ossification effect, about which much has been written in recent years.12 The ossification effect is hypothesized as a consequence of aggressive judicial review, which prompts agencies to go through increasingly burdensome procedures in preparing final rules in order to guard against judicial reversal. As a result, scholars have hypothesized, agencies will tend to shift their emphasis from rulemaking to other forms of administrative action whenever those alternatives impose fewer costs on the agency and hold more promise for being sustained against legal challenges. The leading study of this effect is the Harfst and Mashaw work on the National Highway Transportation Safety Agency (NHTSA).13 Harfst and Mashaw concluded that the result of a series of reversals of NHTSA auto safety rules has been to move the agency away from rulemaking and toward adjudicative recall proceedings. EPA certainly does undertake a considerable amount of adjudication, in addition to issuing an appreciable number of documents that have not gone through the notice-and-comment rulemaking process and yet contain important policy guidance,14 but our summary statistics indicate that EPA continues to issue a substantial number of rules as well. The nature of the statutory responsibilities that have been imposed on EPA often mean that rulemaking is unavoidable, regardless of how burdensome it may be. In most of the statutes that EPA administers, Congress has established basic environmental policy with respect to air, water, solid waste, and other pollutants, sometimes in great detail, but almost always in ways that are incomplete until EPA has enacted rules and regulations translating that policy into administrable instructions. Congress, for instance, has declared that automobiles must meet certain emissions limitations before they can be marketed in the United States by means of a statute providing: "The Administrator shall by regulation prescribe … standards applicable to the emission of any air pollutant from … new motor vehicles."15 For the prohibitions on the sale of nonconforming motor vehicles under the Clean Air Act (CAA) to be effective, the EPA must first issue such regulations.16 In the CAA as in the other statutes EPA administers, Congress has laid out the framework for regulation, but regulation remains incomplete until EPA has supplied further details and implementing instructions. Overall, the statutes EPA administers are laced with responsibilities and actions that EPA cannot exercise except through issuing rules. Part III of this Article addresses issues regarding EPA's interpretations of its statutory authority, while Part IV addresses issues regarding the fact-finding, inference drawing, and policy decisions made in the course of implementing that authority. Interestingly, the U.S. Supreme Court frequently renders statutory interpretation decisions without expressly invoking the Chevron doctrine.17 Be that as it may, the early suggestion by Professor Merrill and others that the lower courts might also begin to disregard Chevron once they became cognizant of the high court's uneven use of the doctrine appears not to have been borne out over time.18 Instead, Judge Wald offered sound advice 10 years ago when she told administrative lawyers preparing briefs to her court to "think Chevron."19 Our study confirms that the courts of appeals view Chevron as the authoritative template for conducting judicial review of statutory interpretation. Explicit invocation of the Chevron framework was nearly universal in cases involving challenges to statutory interpretations. In 119 separate decisions raising such challenges, Chevron was cited in 102, or 86% of them. Of the 17 cases in our study not citing Chevron, only 3 were cases in which Chevron deference was clearly appropriate. As for the rest, one-half (eight) were decisions interpreting statutes governing the availability of judicial review.20 One involved a question of the retroactivity of a statute,21 another involved EPA's interpretation of the Comprehensive Environmental Response, Compensation, and Liability Act's (CERCLA's) civil liability standards,22 and in a third the court conducted a limited review of EPA's [31 ELR 10374] authority to act under the Leedom v. Kyne23 doctrine. These are all situations in which it is either settled that Chevron deference is not appropriate, or at least strongly arguable that it is not. In one case resolved on Step Two grounds, the court did not cite Chevron itself, but did cite a leading Supreme Court opinion elaborating on the content of Step Two, an opinion in which the Court itself explicitly employed the Chevron framework.24 Finally, two involved statutory intepretation in only a pickwickian sense, with the court doing little more than reciting statutory language to dispose of a challenge.25 The remaining three decisions are thus the only ones we found in which the Chevron methodology was clearly called for and yet not explicitly invoked—and each of these employed reasoning entirely consistent with Chevron.26 So Chevron provided the explicit template of analysis in 98% of the cases in our study in which it was clearly applicable. A. EPA's Overall Performance as an Interpreter of Statutes In recent years, EPA has suffered some notable setbacks in the courts of appeals. After engaging in time-consuming and elaborate rulemaking, EPA has lost cases in court involving its proposed national ambient air quality standards (NAAQS) for ozone and particulate matter (PM)27 its decision not to adjust the NAAQS for sulfur dioxide,28 its call for adjustments in the state implementation plans in 22 northeastern states which would have required adoption of California auto standards,29 its decision not to grant a waiver for the fuel additive methylcyclopentadienyl manganese tricarbonyl (MMT),30 its decision to allow highway projects from previously approved regional transportation plans to satisfy the CAA's "conforming plans" requirement,31 its standards for the discharge of sewage sludge under the Clean Water Act (CWA),32 and its decision permitting Indian tribes to implement a waste management permit program under the Resource Conservation and Recovery Act (RCRA).33 In these and other Agency setbacks, the courts of appeals have found that EPA erred in its interpretation of statutory authority. A recent study sharply criticizes EPA's low success rate in defending its interpretations of statutes in court.34 In an analysis of decisions involving EPA rules in the D.C. Circuit Court of Appeals for the years 1993-2000, this study found that of these 69 cases, EPA won only 33% of them.35 In addition, it found that in the 44 cases in which a Chevron question was litigated, EPA won just 41% of the time.36 These results led the study author to conclude that EPA acts "with little regard for the limits or obligations of its statutory authority."37 Our findings are more generous to EPA. Nonetheless, they still show EPA's rules being completely sustained in only 53% of the 111 rulemaking cases during our study period. Looking just at the 83 rulemaking challenges in which a Chevron question was raised, EPA is also sustained in 53% of the decisions in our data set (44/83). Whichever figures are chosen, it has been suggested that they show that EPA regularly ignores the laws that Congress has written, choosing instead to pursue its own "political goals and policy expediency."38 Figures such as these cannot support that conclusion. Studies of EPA's success in litigated cases cannot generate any reliable inferences regarding EPA's overall fidelity to statutory authority because they suffer from selection bias.39 The litigated cases are not the entire universe of instances in which EPA interprets environmental statutes, because many of the rules that EPA writes and other actions that it takes include statutory interpretations that are never challenged. From the entire universe of challenged and unchallenged instances of statutory interpretation, the challenged cases alone constitute a biased sample. One reason some interpretations are not challenged is that no potential litigant believes that EPA has erred in those interpretations. These are instances, then, of EPA's apparently staying faithful to its statutory authority, but they are excluded from any analysis that isolates on the litigated actions alone. On the other hand, the decision to litigate an EPA interpretation typically reflects a judgment that the interpretation is wrong, although the amount of confidence a petitioner has in that judgment varies. If litigants are economically rational actors, they will make litigation decisions on the basis of expected returns, so that a slight chance of achieving an extremely valuable victory could justify the effort. Likewise, a calculation that the value of the compliance cost savings for the period of time during which the challenged rule is suspended exceeds the costs of litigation could also do so, even if ultimate victory is quite unlikely. Litigants also might first decide to challenge a rule on the basis of other mistakes they believe EPA to have made, and then add a statutory interpretation challenge, which they would not otherwise have brought forward, because the marginal costs of the additional challenge are low. These [31 ELR 10375] complexities notwithstanding, it remains the case that statutory interpretations that are actually challenged in litigation are more likely to be wrong than are statutory interpretations that are never challenged in court. Therefore the set of reported decisions involving Chevron challenges to EPA rules presents a biased sample of all EPA statutory interpretations. In an effort to reduce the problem of selection bias and to gain a better understanding of EPA's fidelity to statutory instructions, we studied all of the rules that could have been challenged during our study period for which the Agency prepared a regulatory impact analysis (RIA). The vast majority of these RIAs were prepared because the rule was classified as major under the applicable executive order. There were 75 such rules, of which 25 resulted in 1 or more court challenges. In other words, 33% of all such rules were challenged. This is a much higher percentage than the percentage of EPA's total rules that were challenged.40 This set of rules with RIAs cannot be considered representative of the entire population of EPA rules because it is not a random sample. If anything, EPA's success in defending statutory interpretations in these high visibility rules should continue to overstate EPA's error rate, but less so than does a sample limited to the rules that have been challenged in court. After identifying these 75 rules, we studied the explanatory preamble in the Federal Register for each rule to determine the total number of statutory interpretations made by the Agency. Initially, this presented a specification problem that we needed to resolve. At one extreme, one could take the position that each and every time EPA promulgates any operationally distinct subpart of any rule, it has implicitly interpreted its statutory authority to authorize that subpart. If such implicit interpretations were counted, the number of statutory interpretations EPA has made in these 75 rules would be enormous. We concluded, however, that our attention could appropriately be narrowed by excluding instances in which EPA's authority was taken for granted, because the important issue is the fidelity of the deliberate choices EPA made in interpreting its law, and most implicit interpretations seem more compelled by a settled or noncontroversial construction of the relevant statute than the result of a deliberate choice. Therefore, we tried to isolate those parts of rules in which EPA deliberated about a question of statutory authority, whether because a question had been raised by a commenter or because someone within the Agency decisionmaking process had raised it. Operationally, we looked for passages in the preambles in which EPA explicitly referenced and discussed the issue of statutory authority. We have probably undercounted the number of distinct interpretations of statutes with this methodology, because it would miss any interpretations that were not self-identified by the Agency. We identified 132 different occasions of conscious statutory interpretation in the 75 rules. In the court decisions reviewing the 25 rules of the 75 that were subject to court challenge, the courts found 8 Chevron errors. This is an error rate of 8 out of 132, or 6%. EPA's interpretations of its authority either were not challenged or else were sustained 94% of the time. This is not equivalent to establishing that EPA was faithful to its statutory authority 94% of the time, because some of the unchallenged interpretations may have been mistaken. Most of these 75 rules, however, were ones with appreciable economic impact, and it seems probable that the vast majority of erroneous interpretations were flushed out through litigation, because the amounts at stake in these rules provide the kinds of incentives to challenge them mentioned earlier. In any event, this subset gives us a less biased estimate of EPA's fidelity than does the all-too-common focus on just litigated rules, and it shows a much smaller percentage of statutory interpretation errors than do studies limited to litigated cases. B. Chevron Review of EPA's Statutory Interpretations While not being a good basis for inferences about EPA's overall fidelity to statutes, the litigated cases are worthy of study for what they reveal about the interaction between EPA and the courts under the Chevron doctrine. In this section, in which we examine that interaction more closely, we continue to employ an individual statutory interpretation challenge as our unit of analysis, not an individual court decision. Simply totalling the cases in which EPA wins and loses Chevron challenges fails to illuminate EPA's overall performance as a statutory interpreter. If EPA had made eight or nine contestable interpretations of statutes and been sustained on all but one of them, an approach that totals case victories and defeats would misleadingly treat that case as a defeat for EPA. Engine Manufacturers Ass'n v. EPA41 illustrates the problem. The case involved at least six challenges to EPA's interpretation of provisions of the CAA giving it authority to regulate emissions from nonroad engines and vehicles, and also, according to EPA, impliedly preempting certain state regulations of those engines and vehicles. The court granted one of the petitions for review, but only as to a single element of EPA's rule, the scope of the implied preemption of state regulation. The effect of the judicial decision was to broaden the scope of preempted state regulations, leaving in place the ability of states other than California to choose whether to adopt California's standards or the federal standards, but not to adopt standards different from these two. It left the main structure of the federal regulatory scheme intact. The environmental effects of this decision are unpredictable, but probably negligible, as they depend upon whether states other than California would have adopted standards tougher than California's or the federal government's, which seems unlikely. The case ought to be counted as a substantial vindication of EPA's rule, despite the single issue reversal. Examining such cases issue by issue gives a much more respectable picture of EPA's performance. In order to conduct our examination, we first identified the cases in which Chevron challenges had been brought, and then separately analyzed each distinct challenge to an EPA intepretation of a statute. Because EPA engages in statutory interpretation in adjudications as well as rules, this analysis was not limited to judicial review of rules. In all we identified 102 cases in which 1 or more statutory interpretations of EPA were challenged, 83 of which were rulemaking proceedings.42 In these, we identified 202 separate Chevron challenges, or an average of 2 per decision. The aggregate results are in Figure 1. [SEE ILLUSTRATION IN ORIGINAL] As is well known, Step One of Chevron requires the court to decide independently whether or not Congress "has spoken to the precise question at issue."43 Whenever the court is unable to conclude that Congress has precisely spoken, it is to defer to any "permissible" or "reasonable" interpretation of the agency. From the day it was decided, one of the ongoing interpretational disputes over Chevron itself is whether or not the Supreme Court, in announcing this schematic approach to the judicial review of agency interpretations, intended to signal lower courts to be more deferential to agency interpretations.44 Since Chevron, some of the Court's subsequent formulations of Step Two have attempted to add further gloss on the question of how much deference is due. For instance, in a 1992 decision, the Court wrote: If the agency interpretation is not in conflict with the plain language of the statute, deference is due. In ascertaining whether the agency's interpretation is a permissible construction of the language, a court must look to the structure and language of the statute as a whole. If the text is ambiguous and so open to interpretation in some respects, a degree of deference is granted to the agency, though a reviewing court need not accept an interpretation which is unreasonable.45 However the Court has described the appropriateamount of deference, it has turned out that once an agency succeeds in convincing the court that the statute is ambiguous, Step Two has given the agency very considerable latitude. Indeed, Judge Stephen Williams of the D.C. Circuit once stated that Step Two asks whether the agency's interpretation could be presented with a straight face to a Kennedy School seminar.46 The truly contentious issue in applying Chevron has not been how much deference is due but rather when deference is due. The question, in other words, is what determines the boundary line between a statute that is clear and has a meaning discernible by the court without deference to the agency versus when it is ambiguous and results in the court according the agency a considerable degree of deference. To the untutored, it might seem that very few questions of statutory interpretation will be clear. Regardless of how detailed [31 ELR 10377] a statute Congress writes, implementation always raises issues and controversies that have not been squarely anticipated by the statutory text, and it is these that will give the agency and the parties the most difficulty.47 Chevron itself was an example, raising the question of whether the CAA's definition of "source" must be read only to mean each discrete piece of industrial equipment emitting the requisite volume of air pollutants, or whether it could also be read to include the facility taken as a whole. The distinction carried implications for a number of significant regulatory judgments, such as deciding whether through modifications a "source" had increased emissions sufficiently to require those modifications to obtain regulatory approval. If the entire facility were a source, plant operators could add new emitting equipment while closing down older equipment without triggering the approval requirement, whereas if each discrete emission point were a source, it could not offset additions with withdrawals in this way. Despite the significance of the issue and the detailed complexity of the CAA, the statute did not resolve the issue, and the Court deferred to EPA's reasonable decision to use the reading more generous to industry. Not only should few questions of statutory interpretation be clear, but when they are clear, one might expect that we will see very little litigation over them. A fundamental assumption of the ossification debate is that EPA has strong incentives to avoid reversal of its rules, an aversion which produces cautious behavior by the Agency in constructing a defensible record.48 If the statute is clear, EPA would know that it could be successfully challenged if it adopted an interpretation that belies the clear text, and thus would have strong reasons to base Agency action on a different interpretation, just as it has incentives to invest in extensive procedures above and beyond bare notice-and-comment requirements in order to avoid judicial reversal or, when possible, to eschew rulemaking altogether. Looking to the 1990s cases, our data on Chevron challenges to EPA confirm Judge Williams' opinion that Step Two review is very generous to agency interpretations. Even among the interpretations thought questionable enough to challenge in court, once EPA got beyond Step One, it was upheld 92% of the time. There has not been much movement in this result over time. In our sample of two years of cases from the 1980s, EPA was upheld 89% of the time in cases that reached Step Two. Our cases also show that success at Step One is quite a different matter. Clearly the courts are actively patrolling the boundary line between the two steps, because they are finding one-third of the challenges to EPA interpretations to fall on the Step One side of the line, most of them resulting in EPA losses. At Step One, EPA loses 59% of the time. According to the courts, in the litigated cases in which the statute is clear, EPA gets it wrong more often than right.49 Turning from the Agency's incentives to those of potential challengers may provide a partial explanation for these data. As already noted, some potential challengers place value on delaying the effective date of onerous regulations.50 In addition, petitioners may add statutory interpretation challenges to cases brought on other grounds because the marginal costs of bringing a statutory challenge are relatively small. Either impetus can produce statutory challenges even though EPA has interpreted the statute soundly. Were this a full description of EPA's and petitioners' behavior, we would expect to see some litigated statutory interpretations, as we do—202 litigated interpretations in the 1990s. In addition, however, we would still expect very few Step One challenges that EPA would lose, because the Agency's interpretation is something that it alone controls, and its incentive structure suggests that it would try to avoid making costly mistakes. So the data still require further explanation. Before proceeding further, we can supply a larger context by comparing EPA's performance to other agencies'. Comparison of our results with those of another study indicate that EPA's success rate at Step One is hardly unusual among regulatory agencies. A study of all cases in the courts of appeals in the years 1995-1996 found that in Chevron cases resolved at Step One, the agencies lost 58% of the time, a figure essentially identical to EPA's.51 Overall, the success under Chevron of all agencies during the same period was 73%, again nearly identical to EPA's 75.7% success rate during the 1990s.52 EPA, then, does not appear to be out of step with its brother and sister agencies. This raises the possibility that some of the conclusions we draw about EPA's success and failures under Chevron may be amenable to extrapolation to federal agencies more generally, although at this point we make no claims in that regard. Returning to EPA's Step One losses, one possible explanation for them is that our earlier assumptions about EPA's incentives were incomplete. EPA may have incentives to [31 ELR 10378] proceed with some interpretations of statutes even when it believes them to be erroneous. A particular interpretation may be necessary to satisfy an important constituency, for example. Adopting a rule utilizing a particular interpretation may enable the Agency to take credit for a measure; later on, the courts can take the blame for striking it down. Alternatively, EPA may have such strong policy preferences for its course of action that it is willing to run the litigation risk of reversal. (For this to be true, the expected gains from being able to pursue its chosen course—gains discounted by the probability of litigation success—must be larger than the expected gains from pursuing a policy less vulnerable to legal challenge.) Thus, although our aggregate statistics show that EPA does not deviate outside its statutory authority very often, in various ways the pursuit of "political goals and policy expediency"53 may still provide an explanation for some of the Step One reversals that EPA does suffer. Another distinct possibility for Step One reversals is that EPA and the courts systematically and in good faith disagree over the boundary between the two steps. This hypothesis has not been explored in the existing Chevron literature. That literature has noted that the determination of a clear meaning under Step One is hardly a mechanical task. The task often more closely resembles constructing a statutory meaning than it does finding one.54 The combination of the many different tools of construction available at Step One and the fluidity of how these tools are to be used together implies that different interpreters can reasonably reach different results.55 Some work has been done seeking to discover whether sources of systematic disagreement among interpreters can be found. For example, a number of commentators have explored whether the political ideology or the substantive environmental policy preferences of individual judges affect judicial votes in environmental cases.56 Little thought has been given to the possibility that the differences between agencies and courts as institutions might produce systematic and good-faith disagreements concerning statutory interpretation. Before examining the Step One cases, we develop the reasoning behind this possibility. The boundary line question, as Justice Scalia has posed it, is the question of "how clear is clear?"57 After posing it, he made a much quoted observation about the answer to it: In my experience, there is a fairly close correlation between the degree to which a person is (for want of a better word) a "strict constructionist" of statutes, and the degree to which that person favors Chevron and is willing to give it broad scope. The reason is obvious. One who finds more often (as I do) that the meaning of a statute is apparent from its text and from its relationship with other laws, thereby finds less often that the triggering requirement for Chevron deference exists. It is thus relatively rare that Chevron will require me to accept an interpretation which, though reasonable, I would not personally adopt. Contrariwise, one who abhors a "plain meaning" rule, and is willing to permit the apparent meaning of a statute to be impeached by the legislative history, will more frequently find agency-liberating ambiguity, and will discern a much broader range of "reasonable" inter-pretation that the agency may adopt and to which the courts must pay deference.58 This passage joins two of Justice Scalia's passions—his belief that he will be capable of deciding many of the Chevron cases brought before him at Step One and his rejection of legislative history as a source of statutory meaning—in a puzzling way.59 By rejecting legislative history as generative of "agency-liberating ambiguity," he in effect adopts Judge Leventhal's depiction of delving into legislative history as the equivalent of going to a cocktail party and scanning the crowd to pick out your friends.60 While Judge Leventhal more often picked out his friends in the course of reaching an independent judicial judgment of the statute's meaning, Justice Scalia sees that if legislative history contains numerous statements of varying import from which to choose, a judge under Chevron could just as easily use statements from the legislative history selectively to find an ambiguity in statutory meaning which would lead to deference to agency interpretations.61 Rejecting the use of legislative history to avoid that pit-fall, though, creates problems of its own in determining statutory meaning. Legislative history can resolve ambiguities as well as create them, so that excluding resort to legislative history risks lessening one's ability to find meaning.62 Statutes can say just so many things, regardless of how detailed they are, and interpretative disputes can in any event arise near the margins of what they do say. Chevron itself is illustrative. If, contrary to fact, the legislative history of the CAA had compellingly pointed to a definition, a strict constructionist would not be able to resolve the matter at Step One, whereas a judge using legislative history would. Thus, legislative history seems to have a meaning-generating capability as well as an ambiguity-generating one. It therefore seems odd that Justice Scalia could be so confident that a strict constructionist judge will be able to find meaning more often without its use.63 In advance of knowing what [31 ELR 10379] particular statutory interpretation problems one will face, and what the legislative history says, it is hard to see how a person could confidently be "one who finds more often … that the meaning of a statute is apparent from its text and from its relationship with other laws."64 To a considerable degree, this puzzle dissolves if the activity of statutory interpretation is an activity in which thinking a statute is clear (or ambiguous) helps make it so. We think a good deal of light can be shed on Justice Scalia's remarks concerning strict constructionism and on the low success rate for EPA at Step One, by adopting this hypothesis. EPA has an institutional interest in finding ambiguity, while Scalian judges have an incentive to discern clarity, and these differences produce differences in the arguments that each finds persuasive when confronting the task of statutory interpretation. The next section amplifies this hypothesis, beginning by examining statutory interpretation from the Agency's perspective. 2. Institutional Incentives and Motivated Reasoning When Congress creates an administrative agency such as EPA, it inevitably vests the agency with a mandate that extends well beyond any strict and narrow implementation of the regulatory components of the statutes it administers. The numerous statutes that contain those regulatory components also give EPA the responsibility to conduct ongoing research and inquiry into the vast array of environmental issues that the country faces. These inquiries are very often conducted in service of some regulatory objective. EPA analyzes studies of the impact of air pollution on human health and welfare, for example, in order to establish the scientific basis for ambient air quality criteria and standards under §§ 108 and 109 of the CAA. Much of the knowledge EPA gathers regarding environmental problems does not get immediately put to work in implementing regulatory obligations, however. EPA staff monitor emerging environmental research well before it might be put to such task-oriented purposes so that they will be somewhat up the learning curve when rulemaking begins. In addition, EPA studies problems not immediately addressed by a regulatory statute so that it can be in a position to recommend legislation to the President and the Congress, and also to assist in U.S. participation in multilateral or international discussions of transnational environmental problems prior to the existence of any domestic regulatory regime for addressing the issues raised in such fora. For example, EPA has studied global warming for years. These ongoing activities aimed at building general knowledge and expertise with respect to issues within or bordering on EPA's specific regulatory responsibilities are fundamental to one of the basic purposes for which administrative agencies have been created. In order better to promote the general public health and welfare, Congress and the President must have access to reliable expert knowledge on a continual basis. Even after the progressive administrative ideal of agencies' administering social policy on an expert apolitical basis has atrophied, acquiring and maintaining the knowledge necessary to develop, implement, and evaluate social policy still counts as a fundamental justification for such agencies as EPA. Another fundamental and obvious purpose of agencies such as EPA is to fill in the details in statutory frameworks, to implement these statutes through the issuance of rules and instructions more specific than the statutes themselves, as well as through the enforcement of those rules, to cope with unforeseen circumstances, and to resolve questions of detail that the statutes have not addressed. These functions demand their own brands of knowledge, including knowledge of how the targets of regulation operate and of the environments they face when asked to carry out responsibilities that laws and regulations may impose on them. They also inevitably entail knowledge of the art and science of administration, insofar as executing national regulatory programs requires a bureaucratic structure. To acquire the necessary intellectual resources, EPA hires professionals with the requisite training and skills to analyze incoming information, to define research problems and to formulate possible responses. These professionals, whether they be economists, lawyers, ecologists, toxicologists, biologists, administrators, or members of other professions, come to their work prepared to employ specialized orientations to problem definition, methodologies for analysis and norms for appropriate solutions. As a consequence of these circumstances and the responsibilities they impose on the Agency, individuals and divisions within EPA develop perspectives on the nature of environmental problems as well as the appropriateness of different solutions and implementation structures that stand in a semiautonomous relationship to the implementation of specific statutory provisions. We will refer to the ideas, recommendations, conclusions, and courses of action that internal Agency deliberation would generate independently of any congressional instructions as the autonomous views of the Agency. It is important to see that these autonomous views emerge as a legitimate consequence of the institutional structure, statutory responsibilities, and professional training of the Agency and its personnel. They are continually relied upon and legitimated by the very laws that Congress enacts, because insofar as Congress vests any administrative discretion and any implementation responsibility in the Agency, it does so with the expectation of drawing upon those very expert views. If the Agency did not have an independent perspective on better or worse ways to define problems, to approach problem-solving, and to implement programs, Congress would have a highly unattractive agent in whom to repose that discretion and those responsibilities. At the same time as these autonomous views are necessary, however, they set up a situation of potential tension between the instructions that Congress does give to the Agency with respect to problem definition, problem-solving, and implementation and with those views. Points of tension arise whenever the statutes that emerge from the political environment contain stipulations or requirements that are inconsistent with the judgments that the Agency would make if it were operating independently of those statutory strictures. Sometimes these tensions can be very large and long lasting. For example, EPA for years resisted the Delaney Clause applicable to food additives under the Food, Drug, and Cosmetic Act.65 Applying the clause to those substances [31 ELR 10380] would have had the effect of prohibiting the use of many pesticides on processed foods, because small residues of pesticides often remained after processing. EPA believed that de minimis quantities of such pesticides were acceptable from a public health perspective, that banning pesticides to comply with the Delaney Clause would impose substantial costs on farmers and consumers, and that the application of the Delaney Clause to processed foods was in any event irrational, in light of the absence of a Delaney Clause applicable to raw agricultural products.66 For many years, EPA struggled to find ways around the seemingly absolute instructions of the Delaney Clause, so that it could justify the de minimis exception that it employed in setting tolerances under the Federal Insecticide, Fungicide, and Rodenticide Act (FIFRA). When EPA's position was ultimately challenged in court, EPA lost. The court found the mandatory language in the Delaney Clause to be "clear and mandatory: … Congress intended to ban all carcinogenic food additives, regardless of amount or significance of risk, as the only safe alternative."67 The tension between EPA's autonomous views and statutory command can also be relatively obscure, apparent only to those mindful of the details of bureaucratic implementation. At issue in Environmental Defense Fund v. EPA,68 for instance, was EPA's attempt to comply with provisions of the CAA and the Urban Mass Transportation Act (UMTA) regulating any transportation project requiring the approval of a metropolitan planning organization (MPO), a regional body set up under the UMTA as a condition for receiving federal funding. Those provisions prohibit any MPO from approving a specific project unless it comes from a "conforming" plan, meaning a transportation plan that assists in the attainment or maintenance of ambient air quality standards.69 In their entirety, EPA found the conformity provisions added by the 1990 Amendments, including this one, to be "astonishingly confusing," and capable of "frustrating the process of state and federal cooperation and the integrated planning of transportation projects that [these provisions were] created to foster."70 EPA attempted to ameliorate some of the disruption by providing that for transportation projects not receiving federal funds, the statute was satisfied if the project came from a plan that had at one time been conforming, even if changes in the SIP requirements had subsequently made the plan nonconforming.71 In EPA's judgment, the "grandfathering" of nonfederally funded projects in this way (federally funded projects were not grandfathered at all) instilled a desirable flexibility, and hence made the integration of air policy and transportation planning more administrable, than effectively placing a moratorium on all transportation projects until new conforming plans had been written and approved. Over a dissent by Judge Stephen Williams, a panel of the D.C. Circuit found that the grandfather provision violated clear instructions of the Congress. The majority concluded that authorizing grandfathered projects could result in authorizing projects that might not be assured of not "causing or contributing to any new violation of any standard in the area," as required by 42 U.S.C. § 7506(c)(1)(A). Because the out-of-conformity old plan had not been approved for conformity to currently existing requirements, it might authorize projects inconsistent with new modeling or new ambient requirements, and thereby end up contributing to new violations.72 Such tensions will arise whenever the legislature creates an administrative bureaucracy to implement complex regulatory regimes requiring the acquisition of expertise and the exercise of discretionary judgment. The chances for such clashes have only increased in the modern era of American administrative government, in which the President and the Executive Branch have become recognized organs of national policy creation. In this environment, a President is expected to have independent positions on issues of national significance. Especially when environmental policy has been among the issues separating the presidential candidates, as it regularly has since President Carter ran against Governor Reagan in 1980, the victorious President feels entitled to seek to implement his positions, and the public expects him to do so. Even when those positions are at odds with existing statutes, Presidents legitimately seek to exercise what discretion the laws afford them in order to implement those positions, even while they attempt to change those laws. The Chevron decision provides the Supreme Court's current understanding of the way to resolve conflicts between the autonomous views of the Agency and the statutory commands of the Congress. Insofar as a statute clearly speaks to the precise question in dispute, the statute prevails. If the statute does not speak clearly, the Agency can legitimately and lawfully implement its autonomous views through the exercise of its administrative discretion, so long as its action is consistent with a reasonable construction of the statute. Chevron amounts to a lexical ordering: first satisfy Congress' clear instructions, then exercise reasonable discretion. Although written as a template for judicial review, Chevron speaks in the first instance to the Agency itself, because it is the Agency that must first determine whether there are clear instructions to be followed or, on the other hand, room for discretion to be exercised.73 If the Agency finds itself in the first category, the tension between its views and the Congress' is resolved by Congress' prevailing. If instead it finds itself in the second category, the tension is resolved by the Agency's views prevailing. Crucially, then, it is the Agency who in the first instance determines the boundary of its own discretion, not someone acting through processes external to the very institution whose structure, responsibilities, and professionalism produce the tension in the first place. Because the Agency has views about the appropriate ways to proceed with respect to the policy or administrative question at issue, the endogeneity of the Chevron determination means that the boundary is initially located by an actor with incentives to reduce the territory controlled by Congress and increase the territory that it controls. In Chevron terms, the Agency has a predisposition to find ambiguity. This predisposition does not operate exclusively of others, however. The Agency also has strong incentives to make decisions that are acceptable to Congress, and for several reasons. For one thing, congressional actions such as budgetary decisions and oversight pressures can affect the Agency's ability to accomplish its objectives. If it believes that Congress will monitor its statutory interpretations, this can provide a counterforce to pursuing its own preferences. In addition, the judiciary creates incentives as well—the incentives that contribute to the ossification effect—by signaling what it will take to survive judicial review. A final institutional element that functions as a counter-weight to Agency pursuit of its autonomous views is the professionalism of its staff attorneys, who by training are supposed to be committed to pursuit of the rule of law. While we do not doubt the general tendency of Agency lawyers to be respectful of Congress' instructions, these attorneys are still operating within the institution of the Agency itself, and so are themselves immersed in the distinctive norms of the institution.74 To the extent that these norms are internalized by the lawyers, they produce a desire to see the Agency's views effectuated. Even if such norms are not internalized, staff attorneys have self-interested incentives to find ways to enable their superiors to give effect to agency preferences, rather than to thwart them. So the shape of professional norms does not persist unaffected by the institutional setting—indeed, the concern of an attorney to advance rather than retard her client's interests is itself a significant aspect in the norms of the legal profession. In short, Agency lawyers are relatively poor candidates to follow unerringly in the tradition of Oliver Wendell Holmes' self-description. Holmes once wrote to Harold Laski about a case involving a federal statute. "I hope and believe," he wrote, "that I am not influenced by my opinion that it is a foolish law. I have little doubt that the country likes it and as I always say, as you know, that if my fellow citizens want to go to Hell I will help them. It's my job."75 While this disposition might be applicable to some Scalian judges, it is less so to most Agency attorneys. The existence of potentially countervailing preferences does not erase the Agency disposition to resolve the boundary question in its favor. Those countervailing preferences simply provide a check on it which may or may not be sufficient to blunt its effect. The Agency's autonomous views still motivate it to find statutory ambiguity. In other contexts, cognitive psychologists have found some evidence that motivations such as these can influence the very process of reasoning itself. They have developed a theory of motivated reasoning to explain these results, which hypothesizes that when someone has "a wish, desire, or preference that concerns the outcome of a given reasoning task," then this motivation "may affect reasoning through reliance on a biased set of cognitive processes: strategies for accessing, constructing, and evaluating beliefs."76 In other words, in reasoning through a problem, individuals will tend to be persuaded by arguments that take them to conclusions that they prefer. The model of the reasoning process lying behind the theory posits that when confronted with a reasoning problem we access an existing supply of beliefs, inference rules, evaluation techniques, and arguments. A considerable body of work shows that "people access different beliefs and rules on different occasions: They endorse different attitudes … express different self-concepts … make different social judgments … and use different statistical rules."77 The theory of motivated reasoning adds to this general result by positing that "directional goals" regarding the outcome of the reasoning process are among the conditions that can influence the beliefs and rules and the evaluations that people preferentially access, employ and find persuasive. For instance, in one study women who were also heavy caffeine consumers were less convinced by the evidence in an article claiming that caffeine posed risks for women than were women who were low consumers of caffeine. Each group evaluated aspects of the article in different ways, finding evidence or inferences stronger or weaker according to the result that accorded best with their personal habits. Men showed no such differential effects.78 Similarly, persons who endorsed the thesis that capital punishment deters crime were more likely to criticize a disconfirming study on the basis of such reasons as "insufficient sample size, nonrandom sample selection, or absence of control for important variables" than were those who already believed that capital punishment was not a deterrent.79 In short, "people are more likely to arrive at those conclusions that they want to arrive at."80 Motivated reasoning does not mean that anything goes so long as it leads to the desired conclusion. "People are not at liberty to believe anything they like; they are constrained by their prior beliefs about the acceptability of various procedures."81 The motivation to employ acceptable reasoning methods seems to be stimulated by the expectation that the results of the reasoning process will be evaluated by others. Research has confirmed that subjects who expected to be evaluated, or expected their judgments to be made public, reduced their tendency to rely upon cognitive biases thought [31 ELR 10382] to lead to less defensible judgments, such as the use of ethnic stereotypes, anchoring when making probability judgments, or excessively attributing causality to individuals when their choices were constrained by their environment (the fundamental attribution error).82 Unfortunately, we do not know of any empirical studies testing the theory of motivated reasoning directly on lawyers or judges. We believe, however, that any lawyer who has litigated a case has at some time had experiences consistent with the theory. Lawyers are trained to advocatevigorously for their clients, including presenting the best legal arguments to advance their client's interest, when called upon to do so. Finding the best argument for a client does not mean, however, that one has found the best argument for the outcome of the case. The best argument of the other side might be better. This is not, however, how lawyers many times experience the process of analyzing facts, researching arguments, testing theories, and developing their eventual legal position. More often than not, lawyers become convinced not only that they have found the best arguments for their side, but that their arguments are superior to the other side's. Having the strong motivation to find persuasive arguments to support their client's position, they will find persuasive those argument types, evaluative techniques, and other elements in the reasoning process so as to lead to the vindication of their client's position, just as the theory of motivated reasoning suggests.83 If the results of this research can be extrapolated to the context of statutory interpretation, it leads us to expect an agency like EPA would find ambiguity more often than would someone either disinterested or less interested in the outcome or someone motivated to prefer Congress' views. Thus, the average judge would seem less likely than EPA to find ambiguity in statutes. Adding the incentives of a Scalian judge would only enlarge this gap. By pre-committing himself to the position that texts quite often have clear meaning, a Scalian judge has further incentives to find clarity where others might find ambiguity. Under the theory of motivated reasoning, the differences in predispositions between the Agency and the judiciary operate within the process of reasoning itself by affecting the inferences, arguments, and conclusions that each finds persuasive. The theory of motivated reasoning thus provides one way to explain why a strict constructionist judge who believes that statutes usually have clear meanings will in fact and in good faith find them to have clear meanings more often than someone disposed to find ambiguity will in fact and in good faith find clarity in them. If this hypothesis is valid, we should find two features in examining statutory interpretation by courts and EPA. First, we should be able to find pairs of professionally acceptable argument-types that might appropriately be called upon at a particular point in a statutory analysis, one of which would aid the conclusion that the statute is ambiguous, the other of which would aid the conclusion that the statute is clear.84 Second, we ought to find EPA availing itself more often of the former, and the judiciary more often employing the latter. A significant number of EPA's Step One losses meet these two requirements. Those that do not illustrate further features of the process of statutory interpretation that are also consistent with the hypothesis. We begin with the group of cases exhibiting the two features just noted. Whenever a statutory challenge raises a question as to which the statute does not speak expressly, inferences from various sources of meaning will have to be employed to determine whether the statute nevertheless speaks clearly. According to the theory of motivated reasoning, Agency lawyers will find inference rules that tend to produce ambiguity more persuasive than Scalian judges do. Our set of Step One cases provides one clear confirmation of this effect: the government's reliance on, and the judiciary's skepticism toward, the principle that failure of a statute to prohibit EPA consideration of particular factors or considerations in developing regulations gives the Agency the discretion to consider them—in short, the principle that failure to exclude equals ambiguity and, hence, the discretion to include.85 In defending EPA rules, the government repeatedly relies upon a version of this argument—and the courts regularly reject it. In American Forest & Paper Ass'n v. EPA,86 for example, petitioners challenged an EPA decision to delegate discharge permit issuing authority under the CWA to Louisiana, but only on the condition that the state continue to consult with the U.S. Fish and Wildlife Service (FWS) and the National Marine Fisheries Service to ensure that endangered species would not be threatened by discharges contemplated by the permit. The CWA provides that EPA "shall" delegate permitting authority to a state program that meets nine specified requirements.87 None of these stated requirements could be construed to encompass the consultation requirement. EPA argued that it was authorized under the statute to add conditions above and beyond the nine, because the statute did not expressly say that the stated requirements were exclusive.88 The court rejected this interpretation, finding under Step One that the nine stipulated requirements were exclusive. American Petroleum Institute v. EPA89 involved a challenge to an EPA rule requiring standards for reformulated [31 ELR 10383] gasoline under the CAA mandating that 30% of the oxygen in reformulated gasoline be derived from renewable sources. The requirement had the effect of guaranteeinga market for ethanol, at that time the leading source of renewable oxygenate. The government relied upon EPA's statutory authorization to issue "requirements for reformulated gasoline …. Such regulations shall require the greatest reduction in emissions of ozone forming volatile organic compounds … achievable through the reformulation of conventional gasoline, taking into consideration the cost of achieving such emission reductions, any nonair-quality and other air-quality related health and environmental impacts and energy requirements."90 EPA asserted that the second sentence did not expressly limit the "requirements" that EPA could issue to those that achieved emissions reductions, and that the term was broad and undefined in the statute. This gave EPA the discretion, it claimed, to issue reformulated gasoline regulations designed to preserve the financial viability of renewables and to shift some portion of the additives used to comply with the content requirements of reformulated gasoline to renewables, which EPA viewed as environmentally advantageous.91 The court rejected this argument, finding that the second sentence of the provisions authorizing EPA's regulations stated their "sole purpose." Because EPA conceded that adding the renewables requirement was not expected to achieve greater emissions reductions, and might even lessen them, its rule was vacated as inconsistent with the statute, as the court interpreted it at Step One. Friends of the Crystal River v. EPA92 involved an objection the Agency had lodged against a dredge and fill permit application filed with Michigan's Department of Natural Resources. Acting under statutory authority and pursuant to regulations previously issued, EPA announced that it was transferring the further processing of the permit to the U.S. Army Corps of Engineers. Roughly one and one-half years later, EPA withdrew its objection and stated that it was appropriate for state processing to proceed. Petitioners argued that EPA lacked the authority to rescind the objection. EPA asserted that it did possess such authority, "because § 1344(j) [of the CWA] does not specifically preclude the EPA's withdrawal of its objections."93 The court agreed with the petitioners. The CWA provided a time limit within which the state must comply with EPA's objections, after which the Corps "may" issue the permit. Despite the use of the permissive qualifier, the court held that EPA lacked authority to pull its objection back after the state's time limit had elapsed. In Mead Corp. v. Browner,94 EPA had aggregated two noncontiguous sites under its Aggregation Policy, giving them a single Hazard Ranking System number, and placing the aggregated site on the national priority list (NPL) compiled under CERCLA, even though one site was more than a mile from the other, and would not have qualified if treated separately. As part of its defense of its Aggregation Policy, EPA noted that § 105 of CERCLA, which governs placing sites on the NPL, was silent on the issue of aggregation, meaning that EPA had discretion to aggregate in the way it had. In an analysis that blended the two steps of Chevron, the court disagreed. It analyzed CERCLA's text as a whole, as well as the purposes of its various provisions to find no basis for this kind of aggregation. Without squarely holding that EPA's interpretation violated Step One, the court also noted that even if EPA had discretion under § 105, permitting the aggregation of low-risk and high-risk sites for listing purposes was unreasonable. The conflict between EPA and the courts over the sounds of silence is the most evident pattern in the Step One reversals. In all we found a dozen cases in which EPA unsuccessfully advanced some version of the silence is ambiguity argument, only to be met by the court's counterthrust that the statutory conditions were exclusive.95 In Ethyl Corp. v. EPA,96 the D.C. Circuit supplied an extended analysis of the reasons for this disagreement. The case was brought by Ethyl Corporation after EPA had denied its application for a waiver from the CAA's ban on new fuel additives.97 The CAA provides that EPA "may waive" that ban if it "determines that the applicant has established that such … additive … will not cause or contribute to a failure of any emission control device or system … to achieve compliance [with the Act's emissions standards]."98 After multiple applications and litigation, EPA determined that Ethyl's MMT, an anti-knock additive, satisfied this condition. Nonetheless, EPA refused to issue the waiver, because it had concerns about the public health impact of MMT in the environment. It claimed that because the waiver provisions did not state that they were exclusive, the agency had discretion to consider "whether granting the waiver was consistent with the objectives of the [CAA]."99 Despite the presence of the permissive verb, "may," the court rejected EPA's position, finding at Step One that Congress had limited the factors that EPA may consider to those stated in the statute. Because Ethyl Corporation had satisfied these, the waiver must issue.100 In the court's judgment, "implicit in the EPA's argument [was] the notion that if Congress has not mentioned public health in section [7545(f)(4)]," then the statute was "'silent or ambiguous' as to that issue," thereby triggering Step Two instead of Step One review.101 Failure to state a condition or to declare expressly that the terms used were exclusive, however, was not going to be enough to satisfy the court that Congress had "explicitly or implicitly delegated discretionary authority" to the Agency. [31 ELR 10384] When Chevron was first decided, though, it was far from clear that EPA's argument would produce results "plainly out of keeping" with that decision, because the implications of the decision were hardly apparent on the face of the decision itself—the Court itself had not precisely spoken to the precedential import of Chevron.103 Ethyl Corp. reads a "strict constructionist" meaning into Chevron. In rejecting EPA's argument, Ethyl Corp. reins in the hegemony of the agencies by establishing the hegemony of the Scalian judge. Strict constructionist judges "do not start from the premise that the [statutory] language is imprecise. Instead, [they] assume that in drafting legislation, Congress said what it meant."104 Judges with such a disposition are going to find EPA's argument unpersuasive. Statutes are much more precise if, as a general rule, they are taken to exclude that which they do not expressly allow rather than, as EPA would have it, allowing what they do not expressly exclude. Silence can be heard many different places in federal statutes, though, and it would be highly unlikely that courts could adhere unequivocally to such a strict interpretive rule. As often as EPA fails to prevail in claiming an expansive interpretation of silence, it prevails in enough other circumstances so that it has legitimate reasons to continue to believe in and to press its approach notwithstanding the defeats it has suffered. For example, on several occasions the D.C. Circuit has permitted EPA to exercise discretion to weigh costs against benefits or to consider compliance costs in determining how to proceed under statutory language where such considerations are at best an awkward fit. One of these, for example, involved a rule issued by EPA to implement interstate pollution transport restrictions added to the CAA by the 1990 Amendments. Under these requirements, state implementation plans must prohibit any source from "contributing significantly" to nonattainment of the ambient air quality standards by any other state.105 In the challenged rule, EPA had issued a state implementation plan (SIP)-call requiring 22 states and the District of Columbia to require "highly cost-effective controls" on nitrogen oxide (NOx) emitters, defined as controls capable of removing NOx at a cost of $ 2,000 or less per ton.106 Within the overall structure of the CAA, using costs as a determinant of "significant" for purposes of attaining NAAQS in this way appears quite out of place. In constructing their SIPs, states are required under the CAA to assure compliance with NAAQS by statutorily established deadlines and, once in attainment, are to assure that NAAQS will be maintained, regardless of the cost.107 Under EPA's interpretation of the cross-boundary responsibilities of states, a downwind state could be forced to impose controls on intrastate NOx generators at control costs considerably in excess of $ 2,000 per ton while upwind sources continue to produce emissions that materially hamper downwind state compliance and which could be abated less expensively. Nonetheless, the court granted EPA the discretion to take costs into account in determining what amounts to a significant contribution to downwind air quality.108 In doing so, the court relied in part upon a rule followed in several prior D.C. Circuit decisions that only "clear congressional intent to preclude consideration of cost" will prevent an agency from doing so.109 It is beyond the purposes of this Article to debate the merits of the court's clear statement rule regarding costs.110 What is relevant is this: such a clear statement rule points in just the opposite direction from the court's rulings that reject EPA's arguments that failure to exclude amounts to permission. In its own sphere of application, the clear statement rule adopts a position indistinguishable from EPA's: failure to exclude the consideration of costs is to permit them. The clear statement cases are not flatly in conflict with the outcomes adverse to EPA which we have reviewed in this section, none of which addressed an Agency attempt to invoke cost considerations not provided for by the clear text of the statute. Nonetheless, rulings like these clear statement holdings do establish a jurisprudence of judicial review that can give Agency lawyers some justification for continuing to invoke the ambiguity-generating norm regarding silence, because they indicate that in some contexts the court will be more receptive to the argument than in others. Whether the court will accept or reject the ambiguity-generating norm thus turns out to be a matter of case-by-case decisionmaking, rather than a blanket rule. Because each statutory context can be distinguished from others, usually for a multitude of reasons, there will be plenty of contextual material to mine in search of potentially distinguishing characteristics that will take the next interpretive question out of the realm of the court's certainty-generating rule and into the realm in which it grants EPA the discretion to proceed. In Ethyl Corp., for example, the court held that a statute saying that EPA "may waive" a prohibition on new fuel additives for gasoline if EPA made certain determinations precluded EPA from considering the public health and welfare in deciding whether or not to exercise that authority.111 The use of the permissive phraselogy provided EPA a reasonable ground on which to distinguish that statutory context from others using mandatory phraselogy. It was not successful in that case, but ex ante it was a reasonable position to take. 4. Hard Cases A number of decisions illustrate that the project of finding a statute's plain meaning can be a very complex task, whether or not one looks to legislative history as a legitimate interpretive tool. Even the most widely accepted tools of construction can be deployed in complex ways. One layer of complexity stems from arguments based on the "text and structure of the statute taken as a whole" as an aid in that process, as well as arguments that avail themselves of other provisions of the U.S. Code. In the case of other provisions of the statute itself, judges employ the principle that "a court must … interpret the statute as a symmetrical and coherent regulatory scheme," and "fit, if possible, all parts into an harmonious whole."112 While the specific uses of the "text and structure" arguments aimed at achieving statutory harmony vary, they tend to take on several general forms. Beginning with the statutory provision or provisions that are the textual focus of the interpretive dispute, the court tentatively inserts one of the candidate interpretations and then traces out the implications of that interpretation for the operation of other parts of the statute, or sometimes for other statutes. Alternatively, judges can invoke other statutory provisions that define a term similar or identical to a term in controversy in other contexts. Both techniques were on display in the recent Food & Drug Administration v. Brown & Williamson Tobacco Corp.113 decision holding that the Food and Drug Administration (FDA) lacked statutory authority to regulate tobacco. The Supreme Court analyzed the portions of the Food, Drug & Cosmetic Act (FDCA) instructing FDA as to what regulatory options it possessed once it asserted jurisdiction over a drug or a drug delivery device. The majority concluded that the remedial parts of the FDCA would require banning tobacco products, because they could not be shown to satisfy the statutory criterion of safety. This result, it further concluded, was contrary to congressional pronouncements in other statutes, which the Court read to preclude prohibiting the sale of tobacco products. Use of text and structure and comparisons with other statutes can require tracing the implications of a contested provision through multiple steps, especially in fields as statutorily dense as environmental, health and safety legislation. Each step introduces the possibility of disagreement over the interpretation of other statutory provisions. As significant as any other factor capable of producing good-faith disagreement, what counts as a "coherent" and "harmonious" reading of the statute admits of no unique answer. No better known example of the possibility of conflicting harmonizations of a statute exists than in the litigation recently viewed by millions of Americans as it was being presented to the Florida Supreme Court. Florida election law has long provided that the Secretary of State "shall" certify the votes received from each of Florida's 67 counties, 7 days after an election.114 Returns from counties not certifying votes to the Secretary by that time "shall" be ignored. The state's election statutes also provide a procedure to conduct a manual recount of votes originally tabulated by machine,115 and in large counties this is not a task that can be finished in seven days, especially if a candidate waits until near the end of her statutorily provided window after the election for making such a request.116 A recent addition to the Florida election law provides that returns from counties filed after the seven-day deadlines "may" be ignored and the election "may" be certified.117 Finally, Florida election law provides that "write-in, absentee and manually counted results shall constitute the official return of the election."118 Secretary of State Katherine Harris "harmonized" these statutory provisions by concluding that the mandatory "shall" language imposed an obligation on her which the permissive "may" language did not negate, meaning that she was required by law to ignore manual recounts filed after the seventh day. In her view, manual recounts were harmoniously accommodated under the certification process to the extent they could be completed within seven days. The statutory procedures existing to contest an election after certification were adequate to accommodate justifiable recounts after that. Alternatively, the Palm Beach County Canvassing Board "harmonized" these statutory provisions by concluding that the more recently enacted permissive "may" language took precedence over the mandatory "shall," that the importance of counting every Floridian's vote argued for the statutorily authorized manual recounts to count in the final certification and that this reading was most consistent with the language including manually recounted votes in the "official return" for the election. Without invoking outside-the-statute values, such as the paramount importance of counting every Floridian's vote, each of these harmonizations does a sensible job in fitting together the pieces of Florida's election laws.119 Uses of the tools of "text and structure" and additional statutes often seem to be most persuasive when they are cumulative, as when one can show that the opposition's suggested interpretation creates anomalies or inconsistencies in several different parts of the statute or when compared to usages in the provisions of other statutes. In other words, the more statutory material that can affirmatively be harmonized, or the more statutory glitches or rough edges that can be revealed, the more persuasive the argument for, or [31 ELR 10386] against, a particular reading. In the American Trucking Ass'n v. Browner120 case, for instance, the government's brief before the Supreme Court traces the implications of their opponents' proferred interpretation of § 110 of the CAA through seven other distinct provisions of the Act to show that the interpretation proferred by the industry petitioners creates tensions or inconsistencies with these other provisions and that the government's reading best harmonizes the statute. The other tools of statutory construction also produce strands of argument that contribute cumulatively to the court's ultimate conclusion. Appeals to statutory purposes and goals, to punctuation, to dictionaries, and to the drafting history of enactments all can play roles, depending upon the statutory provision at issue and the larger statutory context in which that provision is found. Furthermore, there does not seem to be any complete hierarchical ranking of these tools. Rather, each type of argument seems to gain strength, and thus be harder to negate with countervailing arguments, to the extent it is clear and decisive, measured against arguments ofits type. A dictionary-based argument in which all consulted dictionaries agree is stronger than one in which there is variance among them, for example. No case addresses how to compare a strong dictionary argument to a competing purposes and goals argument, though. Overall, then, statutory interpretation often seems to involve the building of a cable of interwoven strands, rather than the construction of a continuous chain of deductive reasoning.121 This way of reasoning, familiar to pragmatists,122 can entail numerous junctures at which presumptions, biases or other intellectual tools that assist in drawing inferences must be employed. It also means that, at the end of the day, the overall strength of the cable must be assessed, because disproving or rejecting any single strand does not itself invalidate the ultimate conclusion reached. Each of these junctures presents occasions at which persons motivated to find ambiguity and persons motivated to find settled meaning can diverge. We found a group of decisions in which a dense statutory thicket necessitated complex reasoning of this sort, and in which the courts ended up finding plain meaning where the Agency had found ambiguity. After a careful analysis of the competing arguments, it is hard for us to fault EPA for asserting the interpretation that it did. In Coalition for Clean Air v. EPA,123 EPA tried unsuccessfully to avoid producing a federal implementation plan (FIP) for the Los Angeles air basin, as it had been ordered to do in a prior ruling after years of Agency foot dragging. Los Angeles has some of the most intractable air quality problems in the United States, and EPA has been extremely reluctant over the years to exercise effective federal authority to impose a FIP. Its statutory argument for avoiding the effect of the earlier court order was based on the 1990 Amendments to the CAA, passed shortly after the original order mandating a FIP. EPA effectively argued that by virtue of the new SIP standards and requirements contained in the 1990 Amendments, the state of California had by statute been given the opportunity to produce a compliant SIP before EPA had authority to impose a FIP. The structure of the CAA has always given primacy to state implementation, and giving California the first opportunity to meet the new 1990 standards was consistent with that structure.124 Either the FIP would have to be written against the old, pre-1990 requirements, or it would have to be written against the new 1990 requirements. It would seem counterproductive to proceed under the first assumption, because doing so would put in place a FIP that was outmoded and subject to supersedence even before it was finished; and the second approach would deny California the first opportunity to address the 1990 requirements. The statutory triggering mechanism for FIPs contained in the 1990 Amendments states that EPA "shall promulgate a [FIP] within 2 years after the Administrator … disapproves a State implementation plan,"125 language suggesting an event that could occur only after the 1990 Amendments were enacted (as compared to, for example, "has disapproved"). The 1990 Amendments also make available some compliance flexibility not granted under the prior law which would not be available to EPA in drafting a FIP, and it seems inconsistent with the statutory scheme to deny California access to that flexibility.126 Thus, EPA sought to build an interpretive cable with multiple strands of argument, based on (1) the words used in a specific provision of the Act; (2) the text and structure of the entire Act; (3) the effect of an interpretation of the triggering provision on other specific sections of the Act; and (4) the purposes of the Act. It was not strong enough to persuade the court. Interestingly, the court found none of EPA's arguments persuasive within its own argument-category, and ended up drafting an opinion in which it used arguments of the same general kind as those used by EPA, except that in its cable of argument the strands reinforced the opposite conclusion. Coalition for Clean Air illustrates a pattern. Notwithstanding the widely varying statutory contexts in which they arise, and regardless of how complex the set of arguments made, all of the individual argument strands in opinions resolving questions of statutory interpretation almost always line up as supportive of the same conclusion. Very seldom does the court acknowledge that the relevant considerations are somewhat in conflict, but that the weight of the evidence supports its interpretation. Given that statutes and pieces of statutes can be written at different times, by different committees, this pattern of unanimity is counterintuitive. The style of reasoning in the cases we examined can also be captured by the pragmatist's "web of beliefs" metaphor. Pragmatists use this metaphor, which is related but not identical to the cable-chain metaphor, to depict reasoning as a [31 ELR 10387] process in which we compare with one another different values and propositions related to our reasoning project, going back and forth among them until we reach an overall judgment that has the strongest web of support.127 Within this construct, through the effects of motivated reasoning, the separate argument strands tend eventually to line up in the same direction. As we begin to reach a judgment, which may be based on one or several particularly strong argument-types, we begin to privilege or find persuasive arguments in the other categories that reinforce that judgment. In the end, for example, arguments based on text and structure that support our conclusion become more persuasive than arguments setting up a conflict with that conclusion; this reinforces the persuasiveness of arguments based on purpose and goals that support our conclusion; and so on.128 Coalition for Clean Air seems to be both a complex case and a close one, despite its seriatim dismissal of each and every one of EPA's arguments. Although the court never says so, the conclusion of the majority may be influenced by the judges' frustration in EPA's continual refusal to carry out direct orders of the court in a timely fashion, such that it would have required something close to explicit language in the 1990 Amendments to be convinced that EPA was going to be able to slip out of those responsibilities via a legislative bail out. EPA, of course, had equally strong policy preferences for reading the statute as it did. It had been trying to avoid writing a FIP for the Los Angeles basin for nearly two decades, due to the intractable nature of the air quality problems in that region of the country. 5. Easy Cases, Strong Policy Preferences We found a small group of cases that seem most readily explained as instances in which EPA had strong policy preferences that it chose to pursue even though the arguments that its authorizing statute could be interpreted to allow that pursuit were extremely weak. Because under the theory of motivated reasoning "people are not at liberty to believe anything they like [and] are constrained by their prior beliefs about the acceptability of various procedures,"129 these cases do not square with the hypothesis that EPA lost the case because of a good-faith disagreement with the courts over statutory interpretation. Instead, it seems likely that EPA's lawyers knew they were advancing weak arguments. In Backcountry Against Dumps v. EPA,130 petitioners challenged a decision by EPA to approve a solid waste management program authored by the Campo Band of Mission Indians under § 4005(c) of RCRA. The approval gave the Campos authority to regulate sanitary landfills within its jurisdiction—and specifically the landfill that it wanted to operate itself—with more flexibility than permitted had their program not received § 4005(c) approval. This section authorized "states" to seek approval for such plans. The definitional sections of RCRA defined a "state" as "any of the Several States, the District of Columbia, the Commonwealth of Puerto Rico, the Virgin Islands, Guam, American Samoa, and the Commonwealth of the Northern Mariana Islands."131 A separate section defined "municipality" to include "an Indian tribe or authorized tribal organization."132 The court agreed with petitioners that EPA lacked authority to grant program approval to the Campos, because that section applied to states, and RCRA defined tribes to be municipalities, not states. While it probably would have found the plain language of the statute enough to justify its decision, the court also thought it significant that "when Congress wants to treat Indian tribes as states, it does so in clear and precise language," as in several other environmental statutes.133 The Agency had plausible policy preferences for bringing Indian tribes within the category of governmental entities eligible for the more flexible permitting authority, some fairly weak ones specific to RCRA itself, and a strong one of more general applicability. Its RCRA-specific objective was to avoid the "regulatory gap" that would exist if the government entity with jurisdiction over the use of tribal land was unable to operate a permitting program for its sanitary landfills that contained the full statutory authorities provided to states. When pushed at oral argument, however, the Agency admitted that any such gap was minor in light of the conceded ability of the Campos to seek approval of a site-specific regulation for its landfill that would contain all the flexibility granted under § 4005(c).134 In all likelihood, the reason pushing EPA to take the position that it had taken was left unstated: One hallmark of the Clinton Administration was to accord American Indians heightened respect. The very fact that Indian tribes were treated as states under some environmental statutes but not RCRA was undoubtedly galling to Indian organizations, as well as unappealing to the Administration. It seems likely that EPA decided to treat Indian tribes as having the same status as states under RCRA because of a strong general policy toward that issue—even though the particular statute under which the question was raised in this instance seemed quite clearly to prohibit doing so. Easy cases do not have to be simple ones. Davis County Solid Waste Management & Energy Recovery Special Service District v. EPA135 involved complex statutory analysis. After EPA had issued proposed emission control standards for municipal waste combustors (MWCs) pursuant to its § 111 CAA authority, Congress enacted the 1990 Amendments to that Act. The 1990 Amendments added a new § 129 dealing specifically with the regulation of MWCs. EPA [31 ELR 10388] therefore was required to incorporate the new § 129 provisions into its approach to MWC regulation. Under its original approach, EPA had proposed to regulate MWCs on the basis, inter alia, of the total per day processing capacity of all the MWCs at a single facility. One consequence of this was that a relatively small single unit might be required to meet stringent standards along with much larger units, because each was part of a large facility. Section 129 did not say in so many words what were permissible categories of units for regulatory purposes, but § 129(a)(1) did provide a timetable for issuing emission standards for different categories of MWCs. Two of these categories were defined by the size of the MWC—those with capacities greater and those lesser than 250 tons/year that were combusting municipal solid waste (MSW). Two other categories were defined by the type of waste being processed—those combusting hospital or medical waste, and those combusting commercial or industrial waste. Section 129(a)(2) also authorized EPA to "distinguish among classes, types … and sizes of units within a category in establishing such standards." After EPA proceeded with a regulatory approach that continued to divide MWCs into categories according to the capacity of the entire facility, it was challenged by petitioners claiming that § 129 required that all MWCs that combusted MSW and had less than 250 tons/year capacity be put in the same category. The issue was consequential for petitioners because the stringency of the standards applicable to existing MWCs depended upon the category in which they were classified. EPA believed that Congress did not intend radically to revise the approach the Agency had taken to MWC regulation in its already commenced rulemaking, as would be the result if the petitioners' challenge were correct. On its face, the statute reflected Congress' awareness of the already in progress rulemaking proceeding, for it contained an express provision stating that nothing in § 129 altered schedules for issuing standards established pursuant to settlement or consent decrees, as EPA's schedules had been. At the same time as Congress was aware of the ongoing proceedings, nothing in the text of the statute or in the legislative history expressed disapproval of the basic structure of the approach EPA was taking; the sole informative references to EPA's ongoing rulemaking in the legislative history suggested that the statutory approach was "building upon" EPA's, and that the statutory language was intended to tighten emissions controls compared to EPA's proposals. As an agency that had deliberated prior to issuing its notice of proposed rulemaking regarding the best way to proceed, EPA had every reason to want to continue with that basic approach unless persuaded otherwise by adverse comments received during the notice-and-comment period, or unless instructed otherwise by the Congress. EPA concluded that it had not been instructed otherwise. EPA's problem was that the statutory scheme enacted by Congress proved very difficult to square with the Agency's original approach. Section 129(a)(1)(A)-(D) identifies the four groups of incinerators named earlier; (E) gives EPA authority to designate "other categories," indicating quite plainly that Congress considered MWCs of less than 250 tons/year, MWCs of greater than 250 tons/year capacity, medical waste incinerators, and commercial incinerators each to be a separate "category" of MWC. Section 129(a)(2) then establishes a methodology for determining the minimum control level for existing units in each category that is dependent upon the membership of each category, by being set in the statute equal to "the average emissions limitation achieved by the best performing 12 percent of units in the category."136 The effect of following Congress' apparent scheme is to group all existing MWCs of less than 250 tons/year capacity together, while the effect of following EPA's original rule groups these smaller MWCs together with larger ones, if each is found within a facility that has a capacity greater than 2,200 tons/year. The court's analysis of the text and structure of the Act invoked 10 different sections or subsections of the Act, through which it constructed a compelling case in support of petitioners' challenge. Ultimately, EPA was left to arguing that Congress in effect misspoke, or miswrote, and that its references to unit capacities of various types in § 129 actually were references to total facility capacities of various types.137 Were legislative history given heavy weight by the court, EPA's arguments might have moved the court, because along with the legislative history discussed above, EPA did adduce some additional legislative historical evidence suggesting a drafting mistake. However, whatever weight such considerations might have were negated in the court's view by what may have proved to be the most damaging statutory subsection of all: § 129(g)(1) explicitly defines an MWC "unit" to be "a distinct operating unit of any facility." That definition is impossible to square with EPA's contention that when Congress wrote "unit" elsewhere in the statute that it really meant "facility." The court was unwilling to contradict the statute's own definitional section. In Davis County EPA ended up having very little going for it in terms of arguments based on the text and structure of the 1990 Amendments. Its only arguments emanated from legislative history. Whatever else may be in dispute about the manner in which courts analyze statutes to determine their meaning, this much stands out: when the textual and structural strands of argument support a particular meaning, arguments based on legislative history are not going to change that result. Davis County thus stands as a counter-Holy Trinity Church v. United States138 for the post-Chevron era. Whereas the Court in Holy Trinity was willing to work around plain statutory language in order to achieve a result suggested by the legislative history, the Davis County court declined to do so.139 6. Conclusions Regarding Step One Reversals Many of the Step One cases in our data set are consistent with the hypothesis that EPA personnel and the judiciary have systematic and good-faith disagreements about the [31 ELR 10389] question of when deference is due—the how clear is clear question. The only group that cannot easily be reconciled with this hypothesis is the group of easy cases but strong policy preferences.140 Admittedly, it is contestable whether the cases we find easy should be classified that way, but if that classification be granted, then these cases represent situations in which EPA's defense of its interpretation was beyond the bounds of plausibility. Because motivated reasoning posits that individuals are checked by norms of acceptability in constructing arguments, motivated reasoning cannot explain these outcomes. A much more common approach in the growing literature on the relationship between agencies and the courts explains disagreements between the two as some combination of conscious decisionmaking that trades off policy preferences and rule of law or other values. Both sets of individuals are modeled as people who maximize utility, making trade offs when preferences conflict. Judges may choose between "craft" and "policy," for instance, and agency personnel may choose between maximizing budgets (by keeping relations good with Congress) and policy.141 Whatever the trade offs involved, individuals are seen as making conscious choices between obeying the law or pursuing policy goals, so that the legal arguments they assert are regularly taken to be strategic, not the product of conviction. Except for this last group of decisions, however, the good-faith disagreement hypothesis we have sketched here leads to similar predictions about behavior as the strategic approach does. The mechanisms through which policy influences choice under our approach, however, are internal to the legal reasoning process itself. Arguments and analyses are experienced as more persuasive because they produce decisions that individuals otherwise are motivated to reach. Under this approach, when individuals make choices that favor their policy preferences, they may not be trading off compliance with the rule of law versus those preferences. Their processes of legal reasoning may have inclined them to conclude that the rule of law is consistent with those policy preferences.142 There is no reason why the internal process and the strategic process cannot each contribute to human decisionmaking. Indeed, we think that some combination of these two mechanisms might well provide a better description of many legal decisionmaking processes than either one in isolation. 7. Step Two and the Reluctance of the Courts to Resolve Questions of Degree The Step Two cases raise some distinct issues. So far, we have argued that agencies and judges may have systematically different motivations toward finding certainty in statutory interpretation, and that this motivation helps explain why judges find statutes to have clear meanings more often than the agency does. This only partially helps us understand the 67% of Chevron decisions in which the court agrees with the agency that the statute is ambiguous, but not why EPA is so overwhelmingly successful at this stage of Chevron analysis (over 92% success rate in the 1990s). The number of times courts agree with the agency that a statute is ambiguous suggests that norms of the interpretive judicial craft do indeed impose some constraints on the courts, as Shapiro and Levy, along with others, have postulated.143 Within those constraints, not every statute can be read to have a plain meaning. It is a separate question, though, why EPA is so overwhelmingly successful at Step Two. One possible explanation for the Agency's high success rate emerges from a deeper inquiry into what count as acceptable craft norms. Step Two evaluations, by definition, are made once the court has verified that the statute is susceptible to being given more than one meaning. The reasons that an agency has for choosing between the legitimate meanings emanate from the same set of autonomous views that motivated it to find ambiguity in the first place. Those autonomous views reflect agency resolution of conflicting choices based on its own all-things-considered policy considerations, mixed undoubtedly with considerations of political expediency as well. At Step Two, however, the agency's pursuit of "policy goals and political expediency" ceases to be anathema, as it was at Step One. Now this is precisely what our system of government expects and anticipates the agency will and should do. "The power of an administrative agency to administer a congressionally created … program necessarily requires the formulation of policy and the making of rules to fill any gap left, implicitly or explicitly, by Congress."144 The rationale of Chevron itself depends heavily on the idea that agencies have been appropriately tasked by Congress to reconcile conflicting policy goals and interests. Not only have the agencies been given this responsibility, the courts have emphatically not been given that responsibility. In the key theoretical passage in Chevron, theCourt wrote: Judges are not experts in the field, and are not part of either political branch of the Government. Courts must, in some cases, reconcile competing political interests, but not on the basis of the judges' personal policy preferences. In contrast, an agency to which Congress has delegated policymaking responsibilities may, within the limits of that delegation, properly rely upon the incumbent administration's views of wise policy to inform its judgments. While agencies are not directly accountable to the people, the Chief Executive is, and it is entirely appropriate for this political branch of the Government to make such policy choices—resolving the competing interests which Congress itself either inadvertently did not [31 ELR 10390] resolve, or intentionally left to be resolved by the agency charged with the administration of the statute in light of everyday realities.145 This is the language upon which were based many of the predictions that Chevron meant great deference. The agency's choice among the different options that an ambiguous statute makes available will typically involve a decision that one meaning advances a mix of competing policy interests, to a better degree than another—the very kind of choice that courts are ill-equipped to make.146 If appellate courts are faithful to this aspect of Chevron, they will adopt a very deferential attitude toward the agency's choice at Step Two. In contemporary times, the judiciary's aversion to conducting policy analysis that has initially been assigned to federal agencies constitutes a continuing reaction to Lochner v. New York.147 Courts do not wish to be charged with imposing their own policy views on the elected branches of government. At a minimum, some argument anchoring such action in the text of the Constitution is required before the courts will take that step, and there is no such anchor available to judges assessing the validity of agency decisions that by the Court's own characterization only involve the weighing of competing policy interests under constitutionally valid delegations by Congress. After Lochner, no ordinary constitutional anchor will suffice, either. Lochner, after all, was itself crafted as a decision based on the Constitution. The only conceivable constitutional anchor for second-guessing an agency balance appears to be the Due Process Clause's protection against irrational government action, but this has been given such an undemanding interpretation that it could not easily support intrusive review. Lochner-phobia fairly compels courts to give agencies a wide berth at Step Two. In contrast to Step Two, Step One finds courts on much firmer ground. Agencies have no comparative or democratic advantage in determining what Congress said. Here, courts may even have pride of place, because by superintending the action of agencies they ensure that the laws that Congress writes will not be lost in the halls of the bureaucracies. So their democratic pedigree is much more solid at Step One. So, too, is the style of reasoning to be employed. If Step Two were to be a serious check on agency decisions, courts would have to weigh competing policies themselves. The third noticeable dimension of the jurisprudence of Justice Scalia that can be seen within the judiciary as a whole is an antipathy toward judicial analysis that requires weighing or balancing of values.148 Vigorous Step Two review would run square up against the argument that weighing or balancing of competing policy interests is highly inappropriate judicial activity. Instead of willingly engaging in the weighing of competing policies that the agency has been delegated authority to administer, Justice Scalia advocates what Thomas Grey has aptly named a "new formalism."149 In fact, in Professor Grey's typology, textualism is subsumed as a part of formalist thinking.150 The new formalism denies any role in legal reasoning for the kind of consequentialist policies that agencies routinely consider, which Grey terms "acceptability."151 The acceptability of the real-world consequences of judicial reasoning comes into play only as a tie-breaker in rare cases, or in extreme cases in which formal reasoning leads to outrageously unacceptable results, or in a justification of the formalist system as a whole. This, however, does not mean that the new formalism is a stiff-legged mechanical jurisprudence. To the contrary, Grey argues that it turns out to be "moderate and pragmatic" formalism.152 This suggests that the new formalism can comfortably subscribe to the cable versus chain understanding of the reasoning process, so long as the ultimate object of that thought process is to ascertain determinate meaning, and so long as one of the strands in the cable does not involve determining the preferred mix of policies assigned to agency implementation. The new formalism, in a word, is very much at home with Step One as we have described it, and very much averse to infusing Step Two with much content. Our inquiry into the judicial treatment of EPA's use of science when challenged in litigation reflects, we believe, another manifestation of the judicial tendency toward formalism and categorical distinctions and aversion to answering questions of more or less. There the contrast shows up in a striking reluctance to second-guess EPA when the challenge requires an evaluation of degree, such as whether EPA's selection of data upon which to rely was better or worse than another possible data set, compared to a greater willingness to overrule EPA when the nature of the challenge is categorical, such as whether a conclusion drawn by EPA was without any factual support, or whether EPA had completely failed to explain a link in its chain of reasoning.153 We examine the science cases in detail in the next section. Finally, we report on our statistical findings concerning the tools that the courts in the 1990s employed to assist them in statutory interpretation. Our results show that the tools associated with textualist interpretive approaches predominated. Figure 2 presents the percentages of statutory interpretation decisions in which the court made reference to various common tools of intepretation. The most clearly text-based arguments are heavily dominant, with arguments centering on the meaning of discrete provisions being made in 88% of the cases, and harmonization arguments being made in 68% of them.154 [SEE ILLUSTRATION IN ORIGINAL] There is also good evidence that nontext-based arguments are not heavily relied upon in cases involving EPA. The rejection of legislative history has been part and parcel of the textualist movement in statutory interpretation, and we find strong evidence that legislative history is not playing a significant role in determining the outcomes of EPA cases. In our data collection, we first coded "legislative history cited" any time legislative history was identified in the opinion of the court in connection with an argument regarding the meaning of the statute. This excludes references to legislative material to establish some procedural fact about the legislation at issue. This coding protocol is consistent with some previous studies of legislative history used in the Supreme Court.155 We found legislative history cited in this fashion in 33% of our Chevron cases. This figure is midway between the 18% rate Professor Merrill found in the statutory interpretation cases of the Supreme Court during the 1992 Term and the 49% rate Professor Schacter found in such cases decided during the 1996 Term. We think this 33% figure may overstate the role that legislative history is playing in the courts of appeals. Admittedly, in our data set opinions are often written so that it is difficult to tell the precise role that legislative history plays in the court's overall opinion. One consequence of the "web of belief" or "strands of the cable" style of argument referred to earlier,156 is that the opinion does not indicate one way or the other whether the result would have been different if legislative history had been strongly contrary to the meaning suggested by text-based tools alone. A judge could respond to a legislative history argument because it had been vigorously argued by a party, not because the judge viewed the argument as dispositive. On the other hand, a judge for whom legislative history was material to her conclusion could write an opinion without relying upon it in order to bring along a colleague who was skeptical about its use. Legislative history also functions differently at Step One than it does at Step Two. At Step Two, even judges who eschew legislative history at Step One may refer to legislative history for information about what policies and objectives seem relevant to the interpretation of the statute, so that they can pass judgment on whether or not the agency interpretation constitutes a reasonable accommodation of conflicting policy interests. Because policies do not have to have the status of law, such sources are more legitimate for such information than they are to a textualist trying to answer the Step One question. Notwithstanding the interpretive difficulties, we did code legislative history arguments separately when the language of the opinion itself suggested that such history was doing more than providing optional affirmation of a decision reached on other grounds. In coding for the "legislative history relied upon" variable, we excluded cases in which the decision takes the legislative history to be itself ambiguous or otherwise unhelpful, we excluded cases in which a party's arguments based on legislative history are rejected,157 and we excluded cases in which the structure of the opinion strongly suggests that a conclusion would have been reached even if the legislative history cited did not exist. Those limitations reduce to 8% the percentage of cases in which legislative history was relied upon, in the sense just defined, rather than simply cited.158 IV. Judicial Review of EPA Scientific Determinations A. EPA's Overall Performance in Making Scientific Judgments The manner in which the courts review statutory interpretations of agencies such as EPA has attracted considerable attention in the academic literature, particularly since the Supreme Court's 1984 decision in Chevron. It is not only EPA's interpretations of its enabling statutes that have been the frequent focus of judicial review, however. Litigants also routinely challenge the manner in which EPA implements thesestatutes, i.e., the manner in which EPA applies those statutes to the facts before it in a given situation. This aspect of the relationship between EPA and the courts has attracted relatively less academic attention, despite its obvious importance to the fate of the Agency's efforts to fulfill its statutory mandates. In this section we examine the relationship between the courts and EPA with respect to this aspect of EPA's activities, with special emphasis on challenges that have been brought against EPA's use of science. During the 1990s, litigants with pro-industrial or pro-development agendas were responsible for initiating 79% of the challenges to the validity of EPA's scientific determinations, which was up from 70% in the 1980s. During the period 1986-1987, EPA lost 15% of challenges brought against its scientific determinations (we analyzed 105 cases in the 1990s, and 33 during the 1986-1987 period). During the 1990s, that figure rose to 22%. Pro-environmental litigants succeeded on 32% of their challenges to EPA's science in the 1990s, compared to 40% in 1986-1987. The figures for pro-industry litigants moved in the opposite direction. During the period 1986-1987, pro-industry litigants prevailed in only 4% of their challenges to EPA's scientific determinations. In the next decade, they prevailed 19% of the time. The nearly five-fold increase in the pro-industry success rate is statistically significant at the 90% level.159 These figures raise some intriguing questions. First, why were more than three out of every four challenges during the 1990s industry-based? One possibility, of course, is that EPA sides more frequently with pro-environment interests. If pro-environmental interests are satisfied more frequently with the outcome of EPA's scientific determinations than are affected industries, they will have less impetus to challenge those decisions than will disgruntled industrial concerns. A tilt by EPA toward environmentally protective scientific determinations is not the only possible explanation, however, for the predominance of industry-based challenges to EPA science. Environmental groups may be more selective in raising challenges in court to EPA's scientific determinations, even if the Agency reaches results unacceptable to those groups just as often as it reaches results unacceptable to industry. Science challenges are expensive to pursue. A challenger typically must hire experts to assess the validity of the process by which EPA reached the disputed conclusions, and in some instances may want those experts to perform their own assessments to compare the results with EPA's. If environmental groups lack the financial resources available to regulated industries, they may simply be unable to initiate as many science-based challenges as does industry. In a related vein, environmental groups may prefer to allocate the bulk of their resources to law-based as opposed to fact-based challenges because of a perception that law-based challenges are likely to have a wider impact than will a challenge to resolution of a particular scientific issue.160 If so, environmental groups may prefer to confine their scientific challenges to high impact proceedings such as broad-based rulemaking proceedings in which EPA's resolution of scientific issues yielded regulations deemed by these groups to be insufficiently protective. Another possible related explanation is that industrial interests may pursue challenges that have lower prospects for success than environmental interests do. If a judicial challenge results in a stay of a new regulation pending court disposition, industry can benefit from that delay, whether or not it ultimately prevails on the merits.161 Environmental groups are less likely to face situations in which they benefit from delay. They could so benefit if the Agency adopts a less stringent regulation in place of a more stringent one, and they are able to convince a court to stay the new rule pending the outcome of the challenge. They will not benefit, however, from the stay of a regulation that affects an industry or a problem for the first time or that tightens a regulation already in place. Thus, in deciding whether to challenge the regulation based on the validity of the science upon which it is based, environmental groups face different strategic options and as a result may have a higher threshold for anticipated success on the merits than do potential industrial challengers. A second interesting question relates to EPA's overall success rate in these challenges. While we cannot rule out the possibility that the increase from 15% losses to 22% losses was due to chance, there are several reasons to think that this increase might be due to systemic features of the EPA-court interaction. One is that the courts might have been somewhat more aggressive and less deferential in their review of EPA's scientific determinations in the 1990s. Commentators have noted a possible tension between the D.C. Circuit, on which conservative judges predominate, and the Clinton Administration, which might manifest itself in this way. Another possibility is that the tenor of judicial review remained constant but that EPA was willing to regulate on the basis of evidence that it would not have been willing to rely upon at an earlier time. Some have suggested that EPA has relied less on science—and used less credible science—in the Clinton Administration.162 Even if the threshold [31 ELR 10393] of judicial acceptability remained the same, EPA may not have been able to convince the courts to approve of its actions on the basis of evidence that appeared relatively weak. Alternatively, it may be that the problems EPA tackled in the 1990s were more difficult and intransigent than the ones that the courts reviewed in the earlier period. If the problems were less susceptible to easy resolution, it is perhaps not surprising that EPA reached the "wrong" result, from the courts' perspective, or at least was not able to convince the courts that it reached a defensible result, in a greater percentage of cases than during the earlier period. This explanation is consistent with the conventional wisdom that EPA was able to "cherry pick" the easiest problems during the early years of its existence and that the solutions to problems subsequently became less obvious. It is also consistent with the perception that some of the problems EPA has tackled in the last decade are more controversial than were earlier subjects of the Agency's attention. There could be little dispute that the Cuyahoga River, when it caught fire, was polluted and in need of cleanup. It was not as obvious during the late 1980s and early 1990s that acid rain was a phenomenon in need of a regulatory fix or that global warming was attributable to pollution from human sources. Finally, why has the success rate for industrial challenges but not environmental group challengers risen during the last decade? One possibility is that any tilt EPA has experienced in favor of more pro-environment scientific decisions has resulted in an increased percentage of the decisions that are harder for EPA to defend. Another explanation may be that the courts have not only become more aggressive in their willingness to review EPA's scientific determinations, but that they have also become more skeptical of the value of pro-environment decisions and are therefore more inclined to reverse them (and less likely to reverse EPA's pro-industry determinations) than they were during the 1980s. This thesis presents the possibility that the judges have become more pro-industry at precisely the time that EPA has inclined more heavily toward pro-environmental scientific determinations. Such a combination, not surprisingly, would yield a higher percentage of reversals in scientific challenges filed by industry and a lower percentage of reversals in challenges advanced by environmental groups. It is difficult to assess some of the factors that are relevant to answering these questions. Direct evidence that EPA has slanted its scientific determinations increasingly toward a pro-environment agenda would be difficult to come by, for example. Likewise, the judges on the courts of appeals are not likely to admit to being more skeptical of, no less antagonistic toward, environmental protection endeavors than judges in days of yore may have been. It is possible, however, to scrutinize closely the cases in which litigants have challenged EPA's scientific determinations to ascertain whether the cases present any patterns, such that one can identify the kinds of issues on which EPA was likely to lose during the 1990s and those on which it was likely to win. Those patterns, in turn, may shed some light on the approach that the courts have taken in recent years to the judicial review of the manner in which EPA implements its statutory mandates and, in particular, the manner in which the Agency resolves scientific disputes before it. B. Judicial Review of EPA's Scientific Determinations 1. The Analytical Framework for Judicial Review of EPA's Scientific Determinations As we have indicated above, the Supreme Court's decision in the Chevron case has become the nearly universal template for judicial review of the validity of EPA's statutory interpretations by the courts of appeals. In determining whether those courts had applied a similarly broad-based analytical framework for the review of EPA's scientific determinations, we identified two obvious candidates: the Overton Park163 and State Farm164 cases.165 These are the seminal Supreme Court decisions setting forth the "reasoned decisionmaking" requirement for judicial review of questions of fact and policy raised against agency rulemaking.
Courts of Appeals Reliance on State Farm and Overton Park Given the small size of the database involved, it is impossible to draw definitive conclusions from the data in Table 2 about the impact of a court's decision to cite one or the other or both of these cases on the outcome of the case. The numbers are nonetheless interesting. In the five cases in which the courts cited Overton Park but not State Farm, EPA prevailed in every one (three of which involved challenges by pro-industry forces, one of which involved a challenge by a pro-environmental group, and one of which involved challenges by both sides). In the 20 cases in which the courts cited State Farm but not Overton Park, EPA prevailed at least in part in 12 and lost on at least one issue in another 12.184 In cases in which the courts cited both Supreme Court opinions, challengers lost in more than twice as many (seven) as they won (three). In the 33 cases in which the courts cited neither precedent, EPA prevailed at least in part in 22 cases and lost at least in part in 14 others. Thus, EPA's record appears to have been better when the courts cited neither precedent than when they cited State Farm alone, but not as good as when the courts cited only Overton Park or both Overton Park and State Farm.185
The D.C. Circuit's Reliance on State Farm and Overton Park As Table 3 indicates, our figures also reveal that the D.C. Circuit was more inclined than the courts of appeals as a whole were to cite State Farm than to cite the Overton Park precedent. In the 40 cases we reviewed, the D.C. Circuit never cited Overton Park alone, whereas it cited State Farm alone in 12 cases involving 17 sets of issues. As we have indicated above, the Court in State Farm quoted from and relied on its previous decision in Overton Park in describing the content of the arbitrary and capricious standard as applied to judicial review of the manner in which an agency applies its substantive enabling authority. Moreover, the versions of the standard enunciated in both cases have at least one factor in common—whether the agency based its decision on a consideration of the factors made relevant by the statute.186 It appears that the courts of appeals engage in the same kind of inquiry, and consider the same factors, whether or not they invoke State Farm or Overton Park, even if their review tends to be more deferential under the Overton Park rubric. Table 4 breaks down the percentage of the cases involving EPA science challenges in which the attacks on EPA's determinations reflected each of the factors set forth in the two cases.
Basis for and Outcome of EPA Science Challenges Table 4 reveals that pro-industry litigants brought more challenges in five of the seven identified categories than did their pro-environment counterparts and an equal number of challenges in the remaining two categories (considered irrelevant factors and failed to consider alternatives). For both categories of challengers, the least successful attack involved the charge that the Agency considered irrelevant factors. The Agency never lost on that ground. The claim that the agency failed to consider alternatives was little more successful. Pro-environment litigants never won on that ground, and only one pro-industry litigant prevailed on that ground. Both categories of claimants had considerable success in arguing that EPA provided inadequate notice. Pro-environment groups prevailed in 5 of the 7 cases in which they leveled that charge, while industry litigants won in 7 of the 17 cases in which they grounded their attack on inadequate notice. Both categories of litigants were notably less successful on the other purely procedural ground—that EPA failed to respond to comments. Environmental groups never won on that ground, while industry challengers won in 3 of 14 cases in which the issue was raised.187 Of the more substantive factors, the one demonstrating the greatest discrepancy in pro-environment and pro-industry success rates was failure to consider relevant factors. Environmental groups never prevailed on this ground, while industry challengers won 20% of the time. This difference may indicate that while EPA is rarely if ever susceptible to the charge that it completely ignored its principal mandate by failing to consider the extent to which its contemplated action would protect health and the environment, it was more susceptible to charges that it ignored factors likely to be the focus of industrial concern, such as cost and technological feasibility. At least two other points jump out from the numbers in Table 4. First, both pro-environment and pro-industry litigants had remarkable success arguing that EPA provided an inadequate explanation for its actions. The environmental groups won in 7 of the 8 cases in which the courts addressed that concern, while pro-industry groups won in 13 of 14 cases. This was the argument that had by far the highest success rates of the arguments on which the courts of appeals ruled in the science cases. Perhaps this success rate is attributable to the catch-all nature of the category. Courts unsatisfied with the Agency's reasoning process may have been more inclined to strike down action on the ground of inadequate explanation, regardless of the nature of the specific defects. Second, pro-industry groups were more successful (11 out of 22 cases) than were pro-environment groups (3 out of 10 cases) in arguing that EPA provided an inadequate factual basis for its decision. The figures in Table 4 seem to raise more questions than they answer. For one thing, as noted above, the inadequate explanation factor is so amorphous that it is difficult to determine what kinds of defects in agency explanations were the most likely to prove fatal to the validity of the action being challenged. The insufficient factual basis factor raises a similar problem. The figures in Table 4 do not make it possible to identify what kinds of factual gaps most troubled the courts or why pro-industry litigants were relatively more successful at convincing the courts that such gaps existed. The raw numbers also fail to provide much enlightenment on the question of whether the cases verify or refute the notion that EPA has engaged in "bad science" in a significant percentage of the cases in which its scientific reasoning has been challenged. The next section explores the science cases in more detail with the aim of digging deeper into the resolution of science challenges by the courts of appeals to determine if any more useful patterns emerge. The discussion focuses first on the cases in which the courts of appeals upheld EPA's decisions in the face of attacks on the Agency's science either by pro-environment or pro-industry groups and then proceeds to analyze the cases in which the courts struck down EPA's decisions based on the ground that it implemented its statutory mandate improperly. Within each group of cases, we have created two categories of decisions: those in which the attacks on EPA's science focused upon flaws in the Agency's data collection and interpretation and those in which litigants attacked the persuasiveness or consistency of the Agency's reasoning process. In other words, the first category focuses on the adequacy of the Agency's factual basis (the fourth factor in Table 4), while the second hones in on the adequacy of its explanation (the fifth factor).188 We determined in our review of the statutory interpretation cases that the courts of appeals during our survey period tended to be leery of second-guessing the political branches on questions that required them to evaluate whether one policy judgment was better or worse than another, because deciding such questions of degree, or weighing conflicting policy interests, is peculiarly the province of the agency, while they were more comfortable making the formalistic or linguistic judgments required to determine whether the statute was clear. The divergence manifested itself in a greater rate of reversal at Step One than at Step Two of the Chevron process. We were interested in determining whether a similar pattern would emerge from a study of the cases involving judicial review of EPA's scientific determinations, and if so, why. We also tested in Part III above the hypothesis that the mindset of what we called a "Scalian judge" could well explain a systematic disagreement between agency decisionmakers and judges on the clarity of statutory text. We wondered whether a similar mindset might explain some of the results in the statutory implementation cases. One possible starting point in the search for an answer to these questions is in Justice Scalia's dissenting opinion in Mistretta v. United States,189 in which the majority upheld the constitutionality of the U.S. Sentencing Commission and the Sentencing Guidelines the Commission issued. Although Justice Scalia dissented from the Court's conclusion that the Guidelines did not contravene the separation of powers, he agreed with its holding that Congress did not engage in an excessive delegation of legislative power to the Commission. Justice Scalia found judicial enforcement of the nondelegation doctrine to be problematic: Once it is conceded, as it must be, that no statute can be entirely precise, and that some judgments, even some judgments involving policy considerations, must be left to the officers executing the law and to the judges applying [31 ELR 10398] it, the debate over unconstitutional delegation becomes a debate not over a point of principle but over a question of degree.190 Because Congress is "better equipped to inform itself of the 'necessities' of government" and since the factors bearing upon those necessities are multifarious and highly political, "it is small wonder that we have almost never felt qualified to second-guess Congress regarding the permissible degree of policy judgment that can be left to those executing or applying the law."191 By contrast, in Justice Scalia's view, the courts are perfectly well equipped to enforce the structural provisions of the Constitution by characterizing, for example, the nature of a power executed by an administrative entity. In that case, Justice Scalia concluded that the Commission was improperly exercising legislative as opposed to executive or judicial functions.192 It is possible that a Scalian judge would draw a similar distinction judging attacks on EPA scientific determinations. Such a judge would be reluctant to become embroiled in assessments of questions of degree, and would be more willing to assess whether EPA had committed some more categorical failure to discharge statutory or administrative responsibilities at all. Such either-or determinations include assessing whether the Agency had reached a conclusion without any factual support, failed completely to explain a link in its reasoning process, failed completely to take into account a relevant consideration, or engaged in inconsistent reasoning.193 None of these bases for reversal turns on a question of degree and none entails review of questions over which Agency policymakers possess an obvious advantage over judges in institutional competence. As the analysis below indicates, our review of the cases that involve attacks on EPA's scientific reasoning indicates that the case results are consistent with such a judicial inclination to avoid immersing the judiciary in the adjudication of questions of degree. 2. Cases in Which the Courts Upheld the Validity of EPA's Scientific Reasoning a. Attacks on EPA's Data Collection and Interpretation Attacks based on the nature of the evidence upon which EPA relied in making its scientific determinations were uniformly unsuccessful during the survey period. We could not identify any cases in which EPA lost because of the data selection choices it made. Instead, the courts always deferred to these choices. In one case, for example, the D.C. Circuit deferred to EPA's decision to rely on 60-day instead of 4-day studies of the bioaccumulation of toxic chemicals in trout in deciding to list a substance as a hazardous waste under RCRA.194 In another case, the same court concluded that EPA reasonably decided to include data from commercial Subtitle C waste management facilities' wastes in defining the scope of an exemption from RCRA regulation for hazardous wastes. It also held that the Agency acted reasonably in applying the low hazard criterion used to determine eligibility for the exemption to data from only two lead processing facilities because those facilities represented 40% of the plants in the industry.195 In other cases, the courts rejected contentions that the data EPA relied on were too old.196
2. 467 U.S. 837, 14 ELR 20507 (1984). 3. 401 U.S. 402, 1 ELR 20110 (1971). 4. 463 U.S. 29, 13 ELR 20672 (1983). 5. See infra note 56 for some citations to this growing literature. 6. The beginning date was the cutoff date of our earlier study. The end date was simply the time at which we had to cease collecting cases in order to begin analyzing them. While the statistical measures in this Article are all drawn from this data set, in discussing different issues we occasionally will refer to significant decisions outside of these time boundaries, such as Michigan v. EPA, 213 F.3d 663, 30 ELR 20407 (D.C. Cir. 2000). 7. Many of these are issued as unpublished Table Decisions. During our study period, the appellate courts decided 77 of the cases involving EPA via Table Decisions. 8. Schuck and Elliott raised this as a possibility in 1990. See Peter H. Schuck & E. Donald Elliott, To the Chevron Station: An Empirical Study of Administrative Law, 1990 DUKE L.J. 984. Cf. Sidney A. Shapiro & Richard E. Levy, Judicial Incentives and Indeterminacy in Substantive Review of Administrative Decisions, 44 DUKE L.J. 1051, 1069-70 (1995) (citing figures showing declining rates of deference in U.S. Supreme Court decisions and in the courts of appeals). 9. Schuck & Elliott, supra note 8. 10. Unfortunately, the Schuck and Elliott data do not break out EPA separately, so no direct comparison of our EPA-specific findings with their data has been possible. 11. Estimates of the gross number of rules issued by EPA also bear out the hypothesis that EPA has not responded to the ossification problem by appreciably shifting away from rulemaking. A simple Westlaw search in the Federal Register database for all final rules issued by EPA in the 1980s and 1990s returned a total of 2,761 rules in the 1980s and 3,553 in the 1990s. 12. E.g., Thomas O. McGarity, Some Thoughts on "Deossifying" the Rulemaking Process, 41 DUKE L.J. 1385 (1992); Richard J. Pierce Jr., Two Problems in Administrative Law: Political Polarity on the District of Columbia Circuit and Judicial Deterrence of Agency Rulemaking, 1988 DUKE L.J. 300. 13. JERRY L. MASHAW & DAVID L. HARFST, THE STRUGGLE FOR AUTO SAFETY (1990). 14. See James T. Hamilton & Christopher H. Schroeder, Strategic Regulators and the Choice of Rulemaking Procedures: The Selection of Formal vs. Informal Rules in Regulating Hazardous Waste, 57 LAW & CONTEMP. PROBS. 111 (1994). 15. 42 U.S.C. § 7521(a), ELR STAT. CAA § 202(a). 16. Id. § 7522(a), ELR STAT. CAA § 203(a). 17. Thomas W. Merrill, Judicial Deference to Executive Precedent, 101 YALE L.J. 969 (1992). 18. Id. at 984; accord Paul Caron, Tax Myopia, or Mamas Don't Let Your Babies Grow Up to Be Tax Lawyers, 13 VA. TAX REV. 517, 562 (1994). 19. PETER L. STRAUSS ET AL., GELLHORN & BYSE'S ADMINISTRATIVE LAW: CASES AND COMMENTS 620 (9th ed. 1995) (quoting Judge Wald's advice in a 1994 speech to practitioners: "Now for you agency case lawyers. Chevron is the password. In every case involving statutory interpretation, think Chevron."). 20. Boise Cascade Corp. v. EPA, 942 F.2d 1427, 22 ELR 20007 (9th Cir. 1991); Clinton County Comm'rs v. EPA, 116 F.3d 1018, 27 ELR 21397 (3d Cir. 1997); Laguna Gatuna, Inc. v. Browner, 58 F.3d 564, 25 ELR 21192 (10th Cir. 1995); Montrose Chem. Corp. of Cal. v. EPA, 132 F.3d 90, 28 ELR 20495 (D.C. Cir. 1998); North Shore Gas Co. v. EPA, 930 F.2d 1239, 21 ELR 21038 (7th Cir. 1991); Razore v. Tulalip Tribes of Wash., 66 F.3d 236, 26 ELR 20063 (9th Cir. 1995); South Holland Metal Finishing Co. v. Browner, 97 F.3d 932, 27 ELR 20325 (7th Cir. 1996); PEACH v. Army Corps of Eng'rs, 87 F.3d 1242 (11th Cir. 1996). 21. United States v. Olin Corp., 107 F.3d 1506, 27 ELR 20778 (11th Cir. 1997). 22. In re Bell Petroleum Servs., Inc., 3 F.3d 889, 23 ELR 21474 (5th Cir. 1993). 23. West Virginia Coal Ass'n v. Reilly, 932 F.2d 964, 22 ELR 20092 (4th Cir. 1991). 24. ALM Corp. v. EPA, 974 F.2d 380, 23 ELR 20126 (3d Cir. 1992) (citing National R.R. Passenger Corp. v. Boston & Maine Corp., 505 U.S. 407 (1992)). 25. Hobbs v. United States, 947 F.2d 941, 22 ELR 20331 (4th Cir. 1991); Pennsylvania Dep't of Envtl. Resources v. EPA, 932 F.2d 269 (3d Cir. 1991). 26. James City County v. EPA, 12 F.3d 1330, 24 ELR 20182 (4th Cir. 1993); Navistar Int'l Transp. Corp. v. EPA, 941 F.2d 1339, 21 ELR 21363 (6th Cir. 1991); Systech Envtl. Corp. v. EPA, 55 F.3d 1466, 25 ELR 21247 (9th Cir. 1995). 27. American Trucking Ass'n v. EPA, 175 F.3d 1027, 29 ELR 21071 (D.C. Cir.), reh'g granted in part, denied in part, 195 F.3d 4, 30 ELR 20119 (D.C. Cir. 1999), aff'd in part, rev'd in part, and remanded, 2001 WL 182549 (U.S. Feb. 27, 2001). 28. American Lung Ass'n v. EPA, 134 F.3d 388, 28 ELR 20481 (D.C. Cir. 1998). 29. Virginia v. EPA, 108 F.3d 1397, 27 ELR 20718 (D.C. Cir.), modified on reh'g, 116 F.3d 499, 27 ELR 21380 (D.C. Cir. 1997). 30. Ethyl Corp. v. EPA, 51 F.3d 1053, 25 ELR 20817 (D.C. Cir. 1995). 31. Environmental Defense Fund v. EPA, 167 F.3d 641, 29 ELR 20631 (D.C. Cir. 1999). 32. Leather Indus. of Am., Inc. v. EPA, 40 F.3d 392, 25 ELR 20158 (D.C. Cir. 1994). 33. Backcountry Against Dumps v. EPA, 100 F.3d 147, 27 ELR 20471 (D.C. Cir. 1996). 34. JONATHAN ADLER, ENVIRONMENTAL PERFORMANCE AT THE BENCH: THE EPA'S RECORD IN FEDERAL COURT 11 (2000). 35. Id. at 8, fig. 1. 36. Id. at 9, fig. 2. 37. Id. at 11. 38. Id. at 16. 39. Selection bias involves "choosing observations in a manner that systematically distorts the population from which they were drawn." GARY KING ET AL., DESIGNING SOCIAL INQUIRY 28 (1994). 40. Of 3,553 total rules issued in the 1990s, 111 of them, or 3%, resulted in decisions in the courts of appeals. 41. 88 F.3d 1075, 26 ELR 21477 (D.C. Cir. 1996). 42. In all, we found Chevron cited in 109 separate decisions. Two of these were cases in which EPA was interpreting its own regulations, and five others were decided on procedural grounds without reaching the Chevron questions. 43. Chevron, 467 U.S. at 842, 14 ELR at 20508. 44. See, e.g, Shapiro & Levy, supra note 8, at 1069: The language in Chevron appeared to order courts to go to step two if there was any ambiguity concerning Congress' intent about the meaning of a statutory term. Since few statutes are absolutely clear, this approach appeared to create a presumption of statutory ambiguity that would be difficult to rebut. Schuck & Elliott, supra note 8, at 1023 ("Most commentators agreed that in [Chevron and Chemical Manufacturers] the Court strongly signaled its intention that reviewing courts defer more to agency interpretations of their statutory authority."); Ronald Levin, The Anatomy of Chevron: Step Two Reconsidered, 72 CHI.-KENT L. REV. 1253, 1256-60 (1997). Some of the early commentary interpreting Chevron as a signal to lower courts to accord agencies more deference, while disagreeing over the desirability of that signal, include Richard J. Pierce, Chevron and Its Aftermath: Judicial Review of Agency Interpretations of Statutory Provisions, 41 VAND. L. REV. 301 (1988) (favorable to expanded deference); Peter L. Strauss, One Hundred Fifty Cases Per Year: Some Implications of the Supreme Court's Limited Resources for Judicial Review of Agency Action, 87 COLUM. L. REV. 1093 (1987); Kenneth Starr, Judicial Review in the Post-Chevron Era, 3 YALE J. ON REG. 283, 284 (1986); Cynthia R. Farina, Statutory Interpretation and the Balance of Power in the Administrative State, 89 COLUM. L. REV. 452 (1989) (critical of expanded deference). 45. National R.R. Passenger Corp. v. Boston & Maine Corp., 112 S. Ct. 1394, 1401-02 (1992) (internal citations omitted). 46. Panel Discussion, Developments in Judicial Review With Emphasis on the Concepts of Standing and Deference to the Agency, 4 ADMIN. L.J. 113, 124 (1990) (comments of Judge Stephen Williams). 47. E.g., Merrill, supra note 17, at 991: If the inquiry at step one is formulated in terms of whether the statutory text discloses that Congress has spoken to the precise question at issue, this results in even greater deference to agency views, because Congress has undoubtedly "spoken to" fewer issues in text than it has through some combination of textual and nontextual sources. 48. E.g., DANIEL FARBER, ECO-PRAGMATISM 189 (1999) ("Because agencies are never sure just how 'hard' the hard look will be, they strive to protect themselves against the most unsympathetic judges."). 49. In addition to analyzing overall performance in the 1990s, we analyzed three different two-year periods in the 1990s—1991-1992, 1994-1995, 1997-1998—to see if there were any significant variations with the decade. We did find some variations. EPA did less well in the middle period (63% overall success rate) than in the other two (76% and 79%, respectively), but these results are not statistically significant. Thus the shift from litigating Bush-era rules in the 1991-1992 period to Clinton-era rules in the rest of the decade did not have a significant impact on EPA's success rate. 50. See supra note 39 and accompanying text. 51. Orin S. Kerr, Shedding Light on Chevron: An Empirical Study of the Chevron Doctrine in the U.S. Courts of Appeals, 15 YALE J. ON REG. 1, 31 (1998). As with us, Kerr used a single Chevron interpretation or application, not a single judicial decision, as his unit of analysis. 52. Id. Our data on Step Two are harder to compare with Kerr's. He found that a considerable number of cases citing Chevron ultimately collapsed Step One and Step Two into a single "reasonableness" analysis, and he tabulated these separately from Step Two cases. Id. In our review of decisions involving EPA, we found almost no such cases. We cannot explain the divergence between Kerr's results and ours. If, however, his "reasonableness" cases plus his Step Two cases are treated as equivalent to our Step Two cases, then he finds the Agency prevailed 84.7% of the time, compared to our figure for EPA of 92.6%. 53. ADLER, supra note 34, at 16. 54. Merrill, supra note 17, at 972. 55. See Shapiro & Levy, supra note 8. 56. E.g., Richard J. Pierce Jr., Is Standing Law or Politics, 77 N.C. L. Rev. 1741 (1999); Emerson H. Tiller & Frank B. Cross, A Modest Proposal for Improving American Justice, 99 COLUM. L. REV. 215 (1999); Gregory C. Sisk et al., Charting the Influenceson the Judicial Mind: An Empirical Study of Judicial Reasoning, 73 N.Y.U.L. REV. 1377 (1999); Harry T. Edwards, Collegiality and Decisionmaking on the D.C. Circuit, 84 VA. L. REV. 1335 (1998); Richard L. Revesz, Environmental Regulation, Ideology, and the D.C. Circuit, 83 VA. L. REV. 1717 (1997). 57. Antonin Scalia, Judicial Deference to Administrative Interpretations of Law, 1989 DUKE L.J. 511, 521. 58. Id. at 521. 59. See especially the discussion in Thomas W. Merrill, Textualism and the Future of the Chevron Doctrine, 72 WASH. U.L.Q. 351, 366-73 (1994). 60. Patricia M. Wald, Some Observations on the Use of Legislative History in the 1981 Supreme Court Term, 68 IOWA L. REV. 195, 214 (1983). 61. Justice Scalia has argued that the use of legislative history "has facilitated rather than deterred decisions that are based upon the courts' policy preferences, rather than neutral principles of law." ANTONIN SCALIA, A MATTER OF INTERPRETATION 35 (1997). 62. See Merrill, supra note 59, at 366-67 (defenders of the use of legislative history "point out that statutory language is often vague or ambiguous and that it is often necessary to determine the purpose of the statute in order to particularize the meaning …"). 63. Justice Scalia has separation-of-powers reasons to limit the use of legislative history as well as the functional reasoning that it is ambiguity-creating. Here, we are only interested in the implications of legislative history on the question of whether more disputes could be resolved at Step One with or without its use, not whether its use is proper, all things considered. 64. See Merrill, supra note 59, at 370 ("Justice Scalia's explanation for why the rise of textualism would produce a decline in Chevron deference … is at best unproven."). See also id. at 368-70, for suggestions on how the question of whether legislative history increases ambiguity or increases certainty might be tested. 65. 21 U.S.C. § 348(c)(3)(A). 66. See ROBERT PERCIVAL ET AL., ENVIRONMENTAL REGULATION: LAW, SCIENCE, AND POLICY 494-505 (3d ed. 2000). 67. Les v. Reilly, 968 F.2d 985, 988-89, 22 ELR 21303, 21305 (9th Cir. 1992). The Food and Drug Administration (FDA) had likewise tried to circumvent the Delaney Clause for years. See Richard Merrill, FDA's Implementation of the Delaney Clause: Repudiation of Congressional Choice or Reasoned Adaptation toScientific Progress?, 5 YALE J. ON REG. 1 (1988). Its efforts also proved unsuccessful. Public Citizen v. Young, 831 F.2d 1108, 18 ELR 20173 (D.C. Cir. 1987). 68. 167 F.3d 641, 29 ELR 20631 (D.C. Cir. 1999). 69. Id. at 643-44, 29 ELR at 20631. 70. Environmental Defense Fund v. EPA, 82 F.3d 451, 468, 26 ELR 20968, 20976 (D.C. Cir. 1996). 71. 167 F.3d at 646, 29 ELR at 20632-33. 72. Id. at 646-47, 29 ELR at 20632-33. 73. The interpretive responsibilities of the administrative agency are closely analyzed in Peter Strauss, When the Judge Is Not the Primary Official With Responsibility to Read: Agency Interpretation and the Problem of Legislative History, 66 CHL-KENT L. REV. 321 (1990). 74. By focusing on lawyer's professionalism, we should not be taken to suggest that nonlawyer policymakers are easily prone to ignore the law. In our experience, the vast majority of appointed officials take the rule of law very seriously, too. As the text indicates, we think both lawyers and nonlawyers within the Agency experience cross-cutting desires. A lawyer's training may, we believe, lead her to experience the rule-of-law desires more intensely than a nonlawyer, but that is just a central tendency, not an inevitability. 75. Letter from Oliver W. Holmes to Harold J. Laski (Mar. 4, 1920), in 1 HOLMES-LASKI LETTERS 1916-1935 248-49 (Mark D. Howe ed., 1953). 76. Ziva Kunda, The Case for Motivated Reasoning, 108 PSYCHOL. BULL. 480, 480 (1990). 77. Id. at 483. 78. Ziva Kunda, Motivation and Inference: Self-Serving Generation and Evaluation of Evidence, 53 J. PERSONALITY & SOC. PSYCHOL. 636 (1987). 79. Kunda, supra note 76, at 490 (citing C.G. Lord et al., Biased Assimilation and Attitude Polarization: The Effects of Prior Theories on Subsequently Considered Evidence, 37 J. PERSONALITY & SOC. PSYCHOL. 2098 (1979)). 80. Id. at 495. 81. Id. at 490. 82. Id. at 481 (citing studies). 83. This hypothesis could in theory be tested. It ought to be possible to construct a sample of litigated cases in which the ex ante probabilities of success were obtained from both sides, and both sides were asked whether they had the better arguments for deciding the outcome. If the motivated reasoning hypothesis is correct, the combined probabilities for individual cases should systematically sum to greater than 100%, and in a significant number of cases the losing side should have said before hand that they had the better arguments. 84. Of course, the overall enterprise of statutory interpretation is notorious for exhibiting such paired argument-types. Karl N. Llewellyn, Remarks on the Theory of Appellate Decision and the Rules or Canons About How Statutes Are to Be Construed, 3 VAND. L. REV. 395, 401-06 (1950) (listing canons and counter canons). 85. We refer to the government's use of this argument in order to highlight the fact that once EPA rules end up in court, litigation decisions, including what arguments to advance, are controlled by the U.S. Department of Justice. 86. 137 F.3d 291, 28 ELR 21122 (5th Cir. 1998). 87. 33 U.S.C. § 1342(b), ELR STAT. FWPCA § 402(b). 88. In part, it relied upon another section of the CWA instructing EPA to promulgate "guidelines for the purpose of establishing uniform application forms and other minimum requirements for the acquisition of information," as well as "the minimum procedural and other requirements of any State program." 33 U.S.C. § 1314(i), ELR STAT. FWPCA § 304(i). 89. 52 F.3d 1113, 25 ELR 20824 (D.C. Cir. 1995). 90. 42 U.S.C. § 7545(k)(1), ELR STAT. CAA § 211(k)(1). 91. 52 F.3d at 1118, 25 ELR at 20826. 92. 35 F.3d 1073, 24 ELR 21490 (6th Cir. 1994). 93. Id. at 1080, 24 ELR at 21493. 94. 100 F.3d 152, 27 ELR 20446 (D.C. Cir. 1996). 95. In addition to the cases reviewed in the text, see Backcountry Against Dumps v. EPA, 100 F.3d 147, 150-51, 27 ELR 20471, 20472 (D.C. Cir. 1996); Davis County Solid Waste Management & Energy Recovery Special Serv. Dist. v. EPA, 101 F.3d 1395, 1404-05, 27 ELR 20476, 20478 (D.C. Cir. 1996); Engine Mfrs. Ass'n v. EPA, 88 F.3d 1075, 1088, 26 ELR 21477, 26 ELR 21481-82 (D.C. Cir. 1996); Fertilizer Inst. v. EPA, 935 F.2d 1303, 1310, 21 ELR 21122, 21125 (D.C. Cir. 1991); Legal Envtl. Assistance Found. v. EPA, 118 F.3d 1467, 1474, 27 ELR 21385, 21388-89 (11th Cir. 1997). 96. 51 F.3d 1053, 25 ELR 20817 (D.C. Cir. 1995). 97. 42 U.S.C. § 7545(f)(1), ELR STAT. CAA § 211(f)(1). 98. Id. § 7545(f)(4), ELR STAT. CAA § 211(f)(4). 99. 51 F.3d at 1054, 25 ELR at 20817. 100. Id. at 1058, 25 ELR at 20820. 101. Id. at 1060, 25 ELR at 20821. 102. Id. (internal quotations and citations omitted). 103. See, e.g., Levin, supra note 44. 104. United States v. LaBonte, 520 U.S. 751 (1997) (Thomas, J.). 105. 42 U.S.C. § 7410(a)(2)(D)(i)(1), ELR STAT. CAA § 110(a)(2)(D)(i)(I). 106. Michigan v. EPA, 213 F.3d 663, 669, 30 ELR 20407, 20408 (D.C. Cir. 2000). 107. Union Elec. Co. v. EPA, 427 U.S. 246, 6 ELR 20570 (1976). 108. Michigan, 213 F.3d at 677-78, 30 ELR at 20411-12. 109. Id. See also George E. Warren Corp. v. EPA, 159 F.3d 616, 622-24 (D.C. Cir 1998), reh'g granted, 164 F.3d 676, 29 ELR 20252 (D.C. Cir. 1999); Grand Canyon Air Tour Coalition v. Federal Aviation Admin., 154 F.3d 455, 475, 29 ELR 20075, 20082-83 (D.C. Cir. 1998), cert. denied, 526 U.S. 1158 (1999). 110. This clear statement rule has been invoked by industry cross-petitioners and their amici in the American Trucking Ass'n v. Browner case currently pending in the Supreme Court. The eventual Supreme Court decision in the matter could give further guidance on its use. 111. See supra notes 96-100 and accompanying text. 112. Food & Drug Admin. v. Brown & Williamson Tobacco Corp., 120 S. Ct. 1291, 1301 (2000) (citations omitted). 113. Id. 114. 2000 Fla. Laws ch. 102.111(1) (2000). 115. Id. ch. 102.166(5)(c). 116. Candidates have until the county certifies its votes, or until 72 hours after the election, whichever comes later, to file a request. Id. ch. 102.166(4)(b). 117. Id. ch. 102.112. 118. Id. ch. 101.5614(8). 119. Relying on this value, which it derived from the state Constitution, the Florida Supreme Court sided with Palm Beach County's interpretation. Palm Beach Cty. Canvassing Bd. v. Harris, Nos. SC00-2346 et al. (Fla. Nov. 21, 2000), vacated, Bush v. Palm Beach County Canvassing Bd., 121 S. Ct. 471 (2000). Food & Drug Admin. v. Brown & Williamson Tobacco Corp. also involved dueling statutory harmonizations. Contrary to the majority, the dissent concluded that the remedial sections of the FDCA provided the FDA with options other than prohibiting the sale of tobacco products, so that there was no conflict between the FDA asserting jurisdiction and permitting the continued sale of tobacco products, conditioned on compliance with FDA regulations. 120 S. Ct. 1291, 1322-25 (2000) (Breyer, J., dissenting). Even if it were the case that jurisdiction over tobacco implied prohibition of tobacco sales, though, this hardly put the two portions of the FDCA into disharmony, according to the dissent. The dissent, who did not read the ancillary statutes invoked by the majority to preclude a ban on sales, found it just as harmonious to read the statute so as to ban tobacco as it was to read it so as to deny the FDA jurisdiction. Id. at 1322. 120. Brief for Federal Respondents at 32-34, American Trucking Ass'n v. Browner, 120 S. Ct. 2193 (2000) (No. 99-1426). The government also argues that other statutes, such as the Unfunded Mandates Reform Act, are compatible with the government's reading, thus negating the affirmative use to which their opposition seeks to put those statutes. 121. Deductive reasoning can play a role within each strand, of course. 122. The cable-chain metaphor is Charles Peirce's, 5 CHARLES S. PEIRCE, COLLECTED PAPERS P264 (Charles Hartshome & Paul Weiss eds., 1960). 123. 971 F.2d 219, 22 ELR 21274 (9th Cir. 1992). 124. This concept is codified by 42 U.S.C. § 7401(a)(3), ELR STAT. CAA § 101(a)(3), which states that the control of air pollution is "the primary responsibility of States and local governments." 125. Id. § 7410(c)(1), ELR STAT. CAA § 110(c)(1). 126. Id. § 7511a(e)(5), ELR STAT. CAA § 182(e)(5); see Coalition for Clean Air, 971 F.2d at 226, 22 ELR at 21277. 127. See, e.g., W.V. QUINE & J.S. ULLIAN, THE WEB OF BELIEF (1970); STEVEN J. BURTON, AN INTRODUCTION TO LAW AND LEGAL REASONING 132-36 (1985); William N. Eskridge Jr. & Philip P. Frickey, StatutoryInterpretation as Practical Reasoning, 42 STAN. L. REV. 321, 348 (1990). 128. Others with whom we are basically in agreement regarding the pragmatic nature of reasoning in statutory interpretation cases have highlighted cases in which the most convincing arguments in different categories seem to point in opposite directions, but we did not find any Chevron Step One opinions in our data set that fit that description. See, e.g., Eskridge & Frickey, supra note 127, at 352 (describing the Bob Jones and Weber decisions as ones in which "an apparently clear text" yielded to strong arguments in other categories). But see id. at 355-56 (arguing that in Weber several different interpretations were "textually justifiable," and that interpreters who initially believe that a statute is clear, but who then find other arguments "strongly cut against the apparent meaning," "might reevaluate her belief that the text is really all that clear"). 129. Id. at 490. 130. 100 F.3d 147, 27 ELR 20471 (D.C. Cir. 1996). 131. 42 U.S.C. § 6903(31), ELR STAT. RCRA § 1004(31). 132. Id. § 6903(13), ELR STAT. RCRA § 1004(13). 133. 100 F.3d at 150, 27 ELR at 20473. 134. Id. at 152, 27 ELR at 20473. 135. 101 F.3d 1395, 27 ELR 20476 (D.C. Cir. 1996). 136. 42 U.S.C. § 7429(a)(2), ELR STAT. CAA § 129(a)(2). 137. 101 F.3d at 1409, 27 ELR at 20482. 138. 143 U.S. 457 (1892). The case raised the question of whether an act making it unlawful to assist the immigration of aliens under contract to perform "labor or services of any kind" applied to the Church, which had contracted with an English cleric to become its rector. Concluding that the legislative history indicated the statute was not meant to apply to "brain toilers," the Court held in favor of the church. 139. For a recent argument that the Court in Holy Trinity actually failed to give effect to Congress' intentions as expressed in the best reading of the legislative history, see Adrian Vermeule, Legislative History and the Limits of Judicial Competence: The Untold Story of Holy Trinity Church, 50 STAN. L. REV. 1833 (1998). 140. See supra notes 129-39 and accompanying text. 141. See Shapiro & Levy, supra note 8, for a model of judges as trading offcraft and policy. Jonathan Adler treats EPA as if it consciously ignores what it knows to be the law in preference for policy goals and political expediency. See ADLER, supra note 34. The examples in the text are illustrative only; we are not ourselves proposing a complete model of judicial or agency behavior. 142. Thus, while we agree with a great deal of the perceptive analysis of Thomas Merrill, we part company insofar as he seems to attribute Justice Scalia's efforts to find meaning via textualist approaches as entirely the product of a conscious decision on the Justice's part. See, e.g., Merrill, supra note 59, at 371 (Justice Scalia has lectured widely that textualism does provide more certainty while the use of legislative history creates more ambiguity, thus "it would hardly be surprising that, whateverthe validity of the assumptions underlying these positions, Justice Scalia would be loath to expose his positions too often to a contrary demonstration from a committed intentionalist"). Our suggestion is that Justice Scalia's precommitments make the assumption that textualism produces certainty true for him, because in disputes over meaning, they influence what arguments Justice Scalia will find persuasive. 143. Shapiro & Levy, supra note 8. 144. Morton v. Ruiz, 415 U.S. 199, 231 (1974). 145. Chevron, 467 U.S. at 836-37, 14 ELR at 20514. 146. See also infra text at notes 189-92, discussing Justice Scalia's views on questions of degree as expressed in the Mistretta v. United States, 488 U.S. 361 (1989), decision. 147. 198 U.S. 45 (1905). 148. For further discussion, see infra notes 189-92 and accompanying text. 149. THOMAS C. GREY, THE NEW FORMALISM (Stanford Public Law and Legal Theory, Working Paper No. 4, 1999) (on file with authors). 150. Id. at 2. 151. Id. at 26. 152. "At its theoretical core, the new formalism is just the old legal pragmatism." Id. at 29. 153. See infra Part IVB. 154. Arguments based on statutory objectives may also be considered text-based, insofar as they rely upon objectives articulated in the statutory text. Unfortunately, we failed to code so as to distinguish such text-based objectives arguments from nontext-based ones. We suspect that most of the objectives-based arguments fall into the first category, but we would need to go back through the data set to confirm this, and as yet we have not. 155. Jane Schacter, The Confounding Common Law Originalism in Recent Supreme Court Statutory Interpretation: Implications for the Legislative History Debate and Beyond, 51 STAN. L. REV. 1, 12-13 (1998). 156. See supra notes 122-23, 127-28 and accompanying text. 157. If the court rejected one party's legislative history arguments but elsewhere used legislative history arguments affirmatively, we counted the case. 158. The use of legislative history in the 1990s cases involving review of EPA scientific determinations was even rarer, even though part of the task of a court engaging in review under the arbitrary and capricious test is determining "whether the [agency] acted within the scope of its authority," Citizens to Preserve Overton Park v. Volpe, 401 U.S. 402, 1 ELR 20110 (1971), and whether the Agency's decision was based on consideration of factors made relevant by the statute. Motor Vehicle Mfrs. Ass'n of the United States v. State Farm Mut. Auto. Ins. Co., 463 U.S. 29, 13 ELR 20672 (1983). Of the scientific review cases we discuss in the next part, the courts relied on legislative history as part of their substantive analysis in only three cases (BP Exploration & Oil, Inc. v. EPA, 66 F.3d 784, 26 ELR 20037 (6th Cir. 1995); Natural Resources Defense Council v. EPA, 966 F.2d 1292, 22 ELR 20950 (9th Cir. 1992); Solite Corp. v. EPA, 952 F.2d 473, 22 ELR 20376 (D.C. Cir. 1991)), and in one of those cases (BP Exploration, 66 F.3d at 796, 26 ELR at 20042), the court used the legislative history only to confirm its interpretation of the statutory text. 159. P-value = 0.0673, using Fisher's Exact test. This is the only change between EPA's 1980s track record and the 1990s identified in this paragraph that is statistically significant at the 90% level. 160. By comparison, industry and environmental Chevron challenges are more evenly balanced. Industry brought 59% of the Chevron challenges in the 1990s; environmental organizations and individuals representing environmental positions brought 41%. 161. See supra note 39 and accompanying text. 162. See E. Donald Elliott et al., Science, Agencies, and the Courts: Is Three a Crowd?, 31 ELR 10125 (Jan. 2001). Similar remarks were made by several commentators at the "EPA at 30" conference, Duke University School of Law, Dec. 7-8, 2000. 163. Citizens to Preserve Overton Park v. Volpe, 401 U.S. 402, 1 ELR 20110 (1971). 164. Motor Vehicle Mfrs. Ass'n of the United States v. State Farm Mut. Auto. Ins. Co., 463 U.S. 29, 13 ELR 20672 (1983). 165. Another possibility was Baltimore Gas & Elec. Co. v. Natural Resources Defense Council, 462 U.S. 87, 13 ELR 20544 (1983). Only five cases in our sample cited that case, and among those five, three also cited State Farm (Southwestern Pa. Growth Alliance v. Browner, 121 F.3d 106, 111, 27 ELR 21540, 21542 (3d Cir. 1997); American Iron & Steel Inst. v. EPA, 115 F.3d 979, 992, 27 ELR 21241, 21246 (D.C. Cir. 1997); and Central Ariz. Water Conservation Dist. v. EPA, 990 F.2d 1531, 1540 n.8, 23 ELR 20678, 20682 n.8 (9th Cir. 1993)) and another also cited Overton Park (BP Exploration & Oil, Inc. v. EPA, 66 F.3d 784, 792, 26 ELR 20037, 20040 (6th Cir. 1995)). EPA prevailed in all of those cases (except on one issue in American Iron & Steel Inst.), regardless of whether the challenge was brought by pro-industry or pro-environmental groups. The only case in which a court cited Baltimore Gas & Elec. without also citing State Farm or Overton Park, was an unpublished decision, Alaska Miners Ass'n v. EPA, 931 F.2d 896
185. We were initially surprised that the cases citing Overton Park may reflect more deferential review than the cases citing State Farm. After all, the Agency lost in Overton Park as well as State Farm. It is true that Overton Park characterizes the APA's arbitrary and capricious test as "a narrow one," and warns reviewing courts not to "substitute [their] judgment for that of the agency." Overton Park, 401 U.S. at 416, 1 ELR at 20113. The case also mandates a "searching and careful" review, however. Id. It is possible that courts seeking a deferential analytical framework have latched onto the initial characterization and warning. But the State Farm Court repeated precisely that language. State Farm, 463 U.S. at 43, 13 ELR at 20676. Thus, a court seeking a deferential framework could just as easily cite State Farm as Overton Park. Perhaps theexplanation lies elsewhere. Two of our colleagues have posited that a key difference between the two cases is that Overton Park invites judges not only "to establish their own standards of what an agency should consider in order to produce an adequately reasoned decision," but also "to consider the result reached by the agency, rather than its reasons." Sidney A. Shapiro & Richard E. Levy, Heightened Scrutiny of the Fourth Branch: Separation of Powers and the Requirement of Adequate Reasons for Agency Decisions, 1987 DUKE L.J. 387, 437. State Farm, on the other hand, "makes clear that the proper focus is not the result reached by an agency, but rather the reasons given to support that result." Id. If, as we argue below, judges during our survey period appeared to be more comfortable identifying gaps in agency reasoning processes than second-guessing agencies on policy-based results and on agency resolution of questions of degree, then State Farm might afford a more inviting precedent to cite when a court decides to reverse the agency in a case involving an attack on EPA scientific determinations. 186. Analysis of whether the Agency considered relevant factors may involve inquiries into two separate but related factors listed in State Farm—whether the Agency failed to consider relevant factors and whether it considered irrelevant factors. 187. Industry litigants raised a failure to respond argument in a much higher percentage of cases than did pro-environment litigants. 188. An attack on the adequacy of the Agency's explanation could conceivably entail the charge that the Agency failed to consider relevant factors, considered irrelevant factors, or failed to consider alternatives. Thus, there is the potential for overlap between the fifth factor and each of the first three. 189. 488 U.S. 361 (1989). 190. Id. at 415-16 (Scalia, J., dissenting). 191. Id. at 416. 192. Id. at 420. 193. See Mark Seidenfeld, Demystifying Deossification: Rethinking Recent Proposals to Modify Judicial Review of Notice and Comment Rulemaking, 75 TEX. L. REV. 483, 510 (1997) ("Lawyers thrive on applying critical reasoning skills to cases and judges, who usually come from the ranks of successful lawyers, are therefore well qualified to spot logical weaknesses in arguments and gaps in reasoning."). 194. Dithiocarbamate Task Force v. EPA, 98 F.3d 1394, 27 ELR 20224 (D.C. Cir. 1996). 195. Solite Corp. v. EPA, 952 F.2d 473, 487, 498, 22 ELR 20376, 20383-84, 20390 (D.C. Cir. 1991). See also Appalachian Power Co. v. EPA, 135 F.3d 791, 814, 28 ELR 20521, 20529 (D.C. Cir. 1998) (rejecting assertion that EPA artificially inflated the costs of emission control by failing to exclude from its database a few extraordinarily high cost plants). 196. Sierra Club v. EPA, 107 F.3d 923
|