No Safe Harbor in Delay: Fifth Circuit Confirms EPA May Disapprove Overdue Good Neighbor SIPs Using Updated Modeling and Requires Independent “Maintenance” Analysis Beyond Formal Nonattainment Designations

Introduction

In State of Texas v. EPA, No. 16-60670 (5th Cir. Sept. 22, 2025), the Fifth Circuit denied Texas’s petition for review challenging the Environmental Protection Agency’s 2016 disapproval of Texas’s State Implementation Plan (SIP) addressing the Clean Air Act’s (CAA) Good Neighbor Provision for the 2008 ozone National Ambient Air Quality Standards (NAAQS). Writing for the court, Judge Higginson held that EPA (1) retained authority to disapprove a SIP notwithstanding EPA’s failure to meet the 12‑month statutory deadline; (2) provided adequate process under the Administrative Procedure Act (APA); (3) permissibly relied on updated interstate transport modeling developed after Texas’s 2012 SIP submission; and (4) reasonably concluded that Texas’s SIP was substantively deficient because it failed to independently analyze “interfere with maintenance” and ignored downwind areas not yet formally designated nonattainment. Judge Smith dissented, urging vacatur based on EPA’s delay and its reliance on post-deadline data.

The decision sits at the intersection of cooperative federalism, interstate ozone transport, and post–Loper Bright administrative review. It clarifies core aspects of Good Neighbor SIP content and the scope of EPA’s review, while rejecting the notion that agency delay strips EPA of power to act or freezes the evidentiary baseline at the time of state submission.

Summary of the Opinion

• Administrative Record: The court denied Texas’s and intervenors’ motion to strike documents related to EPA’s concurrent CSAPR Update rulemaking, holding APA review proceeds on the “whole record” and EPA properly acknowledged contemporaneous technical information without making the FIP dispositive of the SIP disapproval.
• Procedural Rulings:
  • Missed 12-month deadline does not void EPA’s authority. The court, relying on Supreme Court guidance, held that when a statute imposes a timing requirement without specifying consequences for noncompliance, the remedy is not automatic nullification of agency action. Alternative remedies (e.g., citizen suits to compel action) exist, and Texas showed no prejudicial error.
  • Adequate notice and process. Under Vermont Yankee, EPA met its APA obligations. The agency was not required to pre-supply specific metrics or modeling choices to states, and Texas ignored multiple pre-rulemaking opportunities to supplement its SIP with updated data that EPA had circulated.
• Substantive Rulings:
  • EPA may independently assess Good Neighbor compliance. Section 110(k)(3) obliges EPA to approve SIPs only if they meet “all applicable requirements.” EPA can disapprove SIPs that fail to substantively satisfy the statute; state deference does not extend to ignoring federal standards.
  • Maintenance clause has independent force. The Good Neighbor Provision’s “interfere with maintenance” language is distinct from “contribute significantly to nonattainment.” A state must analyze maintenance threats at downwind receptors that currently meet the NAAQS but may struggle to maintain attainment.
  • “Nonattainment” refers to failing to meet the NAAQS, not just formally designated nonattainment areas. SIPs must consider exceedances even in areas not yet formally designated as nonattainment.
  • EPA’s use of 2017 modeling and updated data was reasonable. Selecting 2017 aligned with downwind attainment schedules. EPA could rely on newer, more accurate modeling and acknowledge concurrent rulemaking analyses to corroborate SIP deficiencies.
  • No predetermination or bad faith. EPA did not improperly demand SIP conformity to a preordained FIP. Proposing a FIP before final SIP disapproval is statutorily permissible and did not taint the SIP action.
• Dissent: Judge Smith would vacate under APA § 706(2)(D), reasoning that EPA’s overdue action and reliance on post-deadline data were arbitrary and capricious, and that the court should adopt a bright-line rule against such use of updated data after statutory deadlines.

Analysis

Precedents Cited and Their Influence

• Loper Bright Enterprises v. Raimondo, 603 U.S. 369 (2024): The court applied Loper Bright’s post-Chevron framework: courts independently determine statutory meaning and the scope of any agency discretion, then police the boundaries. Within those bounds, agencies get APA deference for discretionary choices and “most deferential” review for complex scientific judgments. This shaped the opinion’s structure: text-first analysis of § 110(k)(3), followed by deferential review of EPA’s technical modeling and record-based reasoning.
• EPA v. EME Homer City Generation, L.P., 572 U.S. 489 (2014): Validated EPA’s multi-state modeling approach to allocating upwind responsibilities and underscored that EPA need not provide specific pre-submission metrics to states. Also recognized EPA’s FIP timing authority; the Fifth Circuit relied on EME Homer to reject Texas’s “metrics” and predetermination arguments and to contextualize EPA’s 2017 modeling year choice.
• North Carolina v. EPA, 531 F.3d 896 (D.C. Cir. 2008): Required independent significance for the “interfere with maintenance” clause; informed the court’s insistence that Texas’s SIP address maintenance at clean-but-vulnerable receptors, not just designated nonattainment areas.
• Wisconsin v. EPA, 938 F.3d 303 (D.C. Cir. 2019) and Midwest Ozone Group v. EPA, 61 F.4th 187 (D.C. Cir. 2023): Reinforced the principle of aligning Good Neighbor obligations with downwind attainment timeframes and validated updated modeling; the Fifth Circuit echoed this alignment in upholding EPA’s 2017 target year.
• Brock v. Pierce County, 476 U.S. 253 (1986) & Barnhart v. Peabody Coal Co., 537 U.S. 149 (2003): Deadlines lacking specified consequences generally do not strip agencies of power to act; courts avoid imposing forfeitures where Congress provided other remedies. These cases undergirded the rejection of Texas’s “deadline-as-jurisdictional-bar” theory.
• General Motors Corp. v. United States, 496 U.S. 530 (1990): In the CAA context, delay in acting on SIP revisions does not bar enforcement; citizen suits and equitable considerations are the remedies. The court drew on this CAA-specific guidance to conclude EPA retained authority to disapprove despite delay.
• BCCA Appeal Group v. EPA, 355 F.3d 817 (5th Cir. 2003); Texas v. EPA (Texas I), 690 F.3d 670 (5th Cir. 2012); and Texas v. EPA (Texas II), 132 F.4th 808 (5th Cir. 2025): These Fifth Circuit decisions emphasize cooperative federalism’s limits: states choose control strategies, but EPA ensures SIPs meet the CAA’s substantive requirements. Texas II, in particular, rejected the argument that EPA lacks discretion to judge Good Neighbor SIP adequacy—an argument Texas renewed here.
• Alaska Department of Environmental Conservation v. EPA, 540 U.S. 461 (2004): Even within programs granting states significant discretion (PSD), EPA may oversee state decisions to ensure compliance. The Fifth Circuit applied that logic a fortiori to § 110(k)(3) compliance determinations.
• Vermont Yankee Nuclear Power Corp. v. NRDC, 435 U.S. 519 (1978): Courts cannot impose extra-procedural requirements beyond the APA. The court used this to reject Texas’s demand for tailored pre-submission guidance or special notice beyond statutory rulemaking procedures.
• State Farm, Prometheus, and data-update cases: The court relied on State Farm’s “relevant data” requirement and a line of cases holding agencies may (and sometimes must) consider new, better data to avoid arbitrary action. This supported EPA’s use of updated modeling and awareness of concurrent CSAPR Update analysis.

Legal Reasoning

1) Standard of Review after Loper Bright

The court independently construed the CAA’s relevant provisions, then reviewed EPA’s discretionary judgments for arbitrariness. On scientific matters—like interstate ozone transport modeling—the court applied especially deferential review given EPA’s technical expertise.

2) EPA’s Authority to Disapprove a SIP Despite Delay

Section 110(k)(2) sets a 12‑month deadline to act on a technically complete SIP, but contains no explicit consequence for missing it. Relying on Brock, Peabody Coal, and General Motors, the court held EPA’s tardiness did not void its power to act. Congress provided less drastic remedies (e.g., citizen suits to compel action), and under the APA’s “rule of prejudicial error,” Texas showed no cognizable prejudice: the delay left Texas operating under precisely the no‑additional‑control posture it had proposed. The court rejected converting the deadline into a jurisdictional bar.

3) Cooperative Federalism: States’ Discretion vs. EPA’s Gatekeeping Role

Texas argued EPA had to confine its review to Texas’s own justification and defer to the state’s methodological choices. The court disagreed. Section 110(k)(3) directs EPA to approve a SIP only if it “meets all applicable requirements.” That statutory command necessarily empowers EPA to determine whether a SIP complies, and to disapprove if it does not—even where the state has chosen its own control strategy. The court emphasized it would be “nonsense” to force EPA to approve a substantially inadequate SIP only to issue a SIP call later for the same deficiencies.

4) Independent Meaning of “Interfere with Maintenance”

Echoing North Carolina, the court held the Good Neighbor Provision’s maintenance clause carries independent force. A SIP must analyze whether upwind emissions will interfere with downwind maintenance at receptors that currently measure attainment but are projected to have future maintenance concerns. Texas’s SIP, which limited its analysis to a narrow set of designated nonattainment areas and offered no quantitative analysis tying Texas emissions to maintenance risks elsewhere, failed this requirement.

5) “Nonattainment” Means a Condition, Not Just a Formal Designation

The court rejected Texas’s view that “nonattainment” in § 110(a)(2)(D)(i)(I) refers only to formally designated nonattainment areas. Instead, “nonattainment … with respect to [a] NAAQS” signifies the condition of not achieving the standard. That condition can occur at any receptor regardless of formal designation, and the Good Neighbor obligation is forward-looking: SIPs must ensure emissions “will” not contribute significantly to future downwind nonattainment. Reading “nonattainment” to mean only designated areas would undercut the statute’s prophylactic function and timeline alignment with downwind attainment obligations.

6) Use of Updated Modeling and 2017 Target Year

EPA used 2017 as the modeling year to align with downwind attainment deadlines and relied on updated transport modeling developed during the CSAPR Update rulemaking. The court held that EPA could consider “the most up-to-date information” and was not limited to data available in 2012 when Texas submitted its SIP. Under State Farm and allied cases, ignoring better, newer data can be arbitrary; EPA’s explanation for using the 2017 modeling and acknowledging concurrent rulemaking analysis was reasonable.

7) Administrative Record and Predetermination

The court denied the motion to strike CSAPR Update materials from the SIP docket. EPA properly included them as relevant context; the SIP disapproval did not hinge on the FIP. Nor was there evidence EPA acted in bad faith or predetermined the disapproval. Proposing a FIP before final SIP disapproval is contemplated by the CAA’s timing provisions, and the record showed EPA independently assessed Texas’s SIP on its own terms.

Impact

For States

• Forward-looking, nationwide analysis is mandatory. Good Neighbor SIPs must:
  • Analyze and give independent effect to “interfere with maintenance,” including receptors that currently attain but face future risks.
  • Assess contributions to nonattainment as a condition—not just formally designated areas—and explain geographic scope choices.
  • Quantify upwind contributions and the impact of proposed controls; generic trend charts and wind roses, without linkages to downwind concentrations, are insufficient.
• Do not rely on obsolete programs or stale data. SIPs that lean on superseded trading programs (e.g., CAIR) or dated inventories risk disapproval. If EPA circulates updated modeling, states ignore it at their peril.
• Delay is not a defense. An agency’s lateness will not immunize an inadequate SIP. The remedy for EPA delay is to compel action, not to deem a deficient SIP federal law by default.

For EPA

• Post–Loper Bright: stay textual, explain delegations. Courts will police the boundaries of statutory discretion; inside those bounds, EPA’s scientific judgments get deference when reasoned and adequately supported.
• Use current science and align with downwind schedules. The Fifth Circuit endorsed EPA’s reliance on updated modeling and alignment with attainment deadlines; documenting why a particular modeling year is chosen is key.
• Concurrent rulemakings may inform SIP actions. EPA can acknowledge and leverage technical analyses from related proceedings when they corroborate deficiencies, provided the SIP action stands on its own statutory footing.

For Litigation and Policy

• Deadline-based vacatur arguments weakened in the Fifth Circuit. Absent explicit statutory consequences, missed § 110(k)(2) deadlines will not by themselves void EPA’s authority to act; prejudice must be shown.
• Uniform approach to Good Neighbor analysis reinforced. The decision harmonizes with D.C. Circuit jurisprudence (North Carolina, Wisconsin, Midwest Ozone Group) on maintenance, future projections, and time-aligned modeling, helping ensure multi-state coordination.
• Dissent signals potential for further review. Judge Smith’s bright-line rule against post-deadline data could fuel petitions elsewhere or for en banc/Supreme Court review, but the majority’s approach now governs in the Fifth Circuit.

Complex Concepts Simplified

• Good Neighbor Provision: Requires each state’s plan to prevent emissions that (a) significantly contribute to another state’s failure to meet a NAAQS (nonattainment), or (b) interfere with another state’s ability to stay in attainment (maintenance).
• SIP vs. FIP: A SIP is a state’s plan to meet federal air standards; EPA must approve it if it meets all requirements. If a SIP is missing or inadequate, EPA must issue a Federal Implementation Plan (FIP) to fill the gap.
• NAAQS and Design Value: NAAQS are national limits on pollutant concentrations. The “design value” is a statistic from monitoring data used to determine attainment status.
• Nonattainment vs. Maintenance: Nonattainment means a location exceeds the NAAQS; maintenance refers to staying in attainment going forward. The Good Neighbor clause protects both.
• Receptors: Downwind air monitoring sites or modeled receptors where pollutant concentrations are assessed for attainment/maintenance.
• Modeling Year: A future year chosen for emissions and air quality modeling that typically aligns with regulatory attainment deadlines.
• Post–Loper Bright Review: Courts no longer defer under Chevron to agency interpretations by default. They first interpret the statute themselves and then review agency discretion for reasonableness under the APA, giving special deference to technical expertise.
• Arbitrary and Capricious: An agency must consider relevant factors and articulate a rational connection between the facts found and the choice made; ignoring important aspects or better data can render action arbitrary.
• Rule of Prejudicial Error: Even if procedure is imperfect, a court will not set aside agency action without showing the error caused meaningful prejudice.
• Vermont Yankee Principle: Courts cannot require agencies to adopt procedural steps beyond those specified by statute and the APA.

Key Takeaways

• EPA’s failure to meet the 12‑month SIP action deadline in § 110(k)(2) does not void subsequent action absent a statutory consequence; alternative remedies exist, and prejudice must be shown.
• Good Neighbor SIPs must separately analyze “interfere with maintenance” and consider future nonattainment at receptors not limited to formally designated nonattainment areas.
• EPA may rely on updated, post-submission modeling and data when acting on a SIP, especially to align with downwind attainment deadlines, provided it reasonably explains its choices.
• States’ discretion to choose control measures does not insulate SIPs from EPA disapproval where the plan fails to meet “all applicable requirements.”
• Proposing a FIP before final SIP disapproval is permissible and does not, without more, evidence predetermination.
• Under Loper Bright, courts will police statutory boundaries while deferring to EPA’s technical expertise on complex scientific judgments supported by the record.

Conclusion

State of Texas v. EPA cements several important principles in Good Neighbor litigation after Loper Bright. First, agency delay does not nullify EPA’s authority to act on SIPs; the statute’s timing provision is not a jurisdictional bar. Second, Good Neighbor SIPs must be forward-looking, addressing interference with maintenance and nonattainment as factual conditions wherever they may manifest—not only in already-designated areas. Third, EPA can and often should rely on current, robust modeling aligned with statutory attainment schedules, even where that modeling post-dates the SIP submission. Finally, cooperative federalism is not an invitation to substitute state judgment for statutory compliance: § 110(k)(3) charges EPA with ensuring SIPs meet “all applicable requirements.”

The decision aligns the Fifth Circuit’s approach with D.C. Circuit transport jurisprudence, provides practical clarity for states crafting Good Neighbor SIPs, and signals that, in complex scientific arenas like interstate ozone transport, updated evidence and reasoned explanations will carry the day.