Before SMITH, WIENER, and ELROD, Circuit Judges. JERRY E. SMITH, Circuit Judge: Texas sued the Equal Employment Opportunity Commission (“EEOC”) and the Attorney General (“Defendants”), challenging EEOC’s guidance on employers’ use of criminal records in hiring. Texas alleged that guidance constituted an unlawfully promulgated substantive rule and sought to enjoin its enforcement. The state also asked for a declaration per the Declaratory Judgment Act (“DJA”) that it could lawfully exclude felons from state employment. The district court dismissed for want of jurisdiction, but a different panel of this court reversed. Texas v. EEOC (Texas I), 827 F.3d 372 (5th Cir. 2016). That panel, however, withdrew its opinion and remanded so the district court could apply United States Army Corps of Engineers v. Hawkes Co., 136 S. Ct. 1807 (2016), in the first instance. Texas v. EEOC (Texas II), 838 F.3d 511 (5th Cir. 2016) (per curiam). On remand, considering cross-motions for summary judgment, the district court dismissed Texas’s DJA claim but enjoined defendants from enforcing EEOC’s guidance against Texas until EEOC complies with the notice-and- comment rulemaking requirements of the Administrative Procedure Act (“APA”), 5 U.S.C. § 553. We modify the injunction and affirm it as modified. I. A. In April 2012, EEOC issued “Enforcement Guidance on the Consideration of Arrest and Conviction Records in Employment Decisions Under Title VII” (“the Guidance”).[1] Citing data suggesting that blanket bans on hiring individuals with criminal records disproportionately impact minorities, the Guidance declares, With respect to criminal records, there is Title VII disparate impact liability where the evidence shows that a covered employer’s criminal record screening policy or practice disproportionately screens out a Title VII-protected group and the employer does not demonstrate that the policy or practice is job related for the positions in question and consistent with business necessity. Guidance at 9. The Guidance also describes how EEOC will assess whether considering criminal records in hiring decisions has a disparate impact on protected groups. Notably, the Guidance concludes that “[n]ational data . . . supports a finding that criminal record exclusions have a disparate impact based on race and national origin [and] provides a basis for the Commission to further investigate such Title VII disparate impact charges.” Id. at 10. The Guidance details how an employer may show that its policy is job- related and consistent with business necessity and thus may defend against a charge that its criminal-record policy gives rise to disparate impact liability under Title VII. The Guidance provides that “[a]n employer’s evidence of a racially balanced workforce will not be enough to disprove disparate impact.” Id. Instead, an employer “needs to show that [its criminal-record hiring] policy operates to effectively link specific criminal conduct, and its dangers, with the risks inherent in the duties of a particular position.” Id. at 14. The Guidance presents “[t]wo circumstances in which [EEOC] believes employers will consistently meet the ‘job related and consistent with business necessity’ defense”: (1) by establishing a validated, multi-factor screening system per the Uniform Guidelines on Employee Selection Procedures standards or (2) by “develop[ing] a targeted screen . . . and then provid[ing] an opportunity for an individualized assessment for people excluded by the screen.” Id. (emphasis added). Significantly, the Guidance condemns any “policy or practice requiring an automatic, across-the-board exclusion from all employment opportunities” because it “does not focus on the dangers of particular crimes and the risks in particular positions.” Id. at 16.[2] “All entities covered by Title VII are subject to [the Guidance's] analysis,” id. at 27 n.2—including “state[] and local governments,” id. at 3. The Guidance was “issued as a part of the Commission’s efforts to eliminate unlawful discrimination in employment screening, for hiring or retention, by entities covered by Title VII, including private employers as well as federal, state, and local governments.” Id. (emphasis added).[3] And the Guidance stresses that it should be followed by “employers considering the use of criminal records in their selection and retention processes”; “individuals who suspect that they have been denied jobs or promotions, or have been discharged because of their criminal records”; and “EEOC staff who are investigating discrimination charges involving the use of criminal records in employment decisions.” Id.[4] Although the scope of the Guidance is purportedly broad, EEOC has limited rulemaking and enforcement power with respect to Title VII. It may issue only “procedural regulations” implementing Title VII and may not promulgate substantive rules. See 42 U.S.C. § 2000e-12(a); see also EEOC v. Arabian Am. Oil Co., 499 U.S. 244, 257 (1991). And despite that EEOC can bring civil enforcement proceedings against private employers for violating Title VII, it may only investigate state employers.[5] The Attorney General is the only federal government entity or person who may directly sue state employers to enforce Title VII. Both EEOC and the Attorney General, however, may issue aggrieved individuals “right-to-sue” letters, allowing those persons to sue a state employer for violating Title VII. See 42 U.S.C. §§ 2000e-5(f), 2000e-6. B. By state law and the policies of individual state agencies, Texas has long excluded persons with felony convictions—or at least those convicted of specified categories of felonies—from many public jobs. The Texas Department of Public Safety and the Texas Department of Aging and Disability Services, for instance, categorically exclude all convicted felons from employment. Texas schools and the Texas Juvenile Justice Department, moreover, categorically exclude applicants convicted of specified felonies. Soon after EEOC issued the Guidance, Texas received notice that an individual who had been rejected for a Department of Public Safety job had filed a complaint with EEOC, challenging Texas’s no-felon hiring policy as having a disparate impact in violation of Title VII. Texas, in turn, sued EEOC and the Attorney General, contending that “EEOC’s rule purports to limit the prerogative of employers, including Texas, to exclude convicted felons from employment.” Texas averred that it had to choose either to “violate state and local laws that prohibit the ‘individualized assessments’ that EEOC requires and consider convicted felons for hire as Troopers, jailers, and school teachers—or [to] . . . ignore the EEOC’s rule and risk an enforcement action.” Texas brought one claim under the DJA, 28 U.S.C. §§ 2201-2202 (“Count One”), and another under the APA, 5 U.S.C. § 702 (“Count Two”). In Count One, Texas asked for “a declaration of its right to maintain and enforce its laws and policies that absolutely bar convicted felons (or particular categories of convicted felons)” from specified jobs. Texas also asked for an injunction that EEOC and the Attorney General “cannot enforce the interpretation of Title VII that appears in its Felon-Hiring Rule, nor . . . issue right-to-sue letters pursuant to that rule.” In Count Two, Texas urged the court to set aside the Guidance, contending that the Guidance exceeded EEOC’s power under Title VII, was promulgated without notice and comment in violation of the APA, and was substantively unreasonable. The district court dismissed for want of subject matter jurisdiction, but a divided panel reversed. Texas I, 827 F.3d 372. The panel majority held that Texas had Article III standing to challenge the Guidance, id. at 380, and that the Guidance was a final agency action eligible for judicial review under the APA, id. at 387-88. Judge Higginbotham dissented, contending that the controversy did not satisfy “Article III’s demand of ripeness, injury, and adversarial engagement.” Id. at 388. Judge Higginbotham also maintained that the Guidance was not a final agency action. Id. The panel withdrew its opinion, vacated the judgment, and remanded. Though the Supreme Court had announced Hawkes before the panel issued Texas I—and the panel applied Hawkes in Texas I—the district court had not had a chance to apply Hawkes in the first instance. Texas II, 838 F.3d 511. Noting that this court had remanded another case “relat[ing] closely to the issue that the Supreme Court decided in Hawkes,” the panel did the same. Id.[6] On remand, the district court denied Defendants’ renewed motion to dismiss for lack of jurisdiction, largely adopting the reasoning in Texas I. Following cross-motions for summary judgment, the court dismissed Texas’s DJA claim, “declin[ing] to declare that Texas has a right to maintain and enforce its laws and policies that absolutely bar convicted felons (or certain categories of convicted felons) from serving in any job the State and its Legislature deems appropriate.” The court also “decline[d] to enjoin the EEOC from issuing right- to-sue letters.” Regarding the APA claim, the district court granted Texas’s motion for summary judgment in part and denied Defendants’ motion. The court held “that the Guidance . . . is a substantive rule issued without notice and the opportunity for comment.” The court thus enjoined EEOC and the Attorney General “from enforcing the EEOC’s interpretation of the Guidance against the State of Texas until the EEOC has complied with the notice and comment requirements under the APA for promulgating an enforceable substantive rule.” The court did not reach the questions whether EEOC has the power to promulgate a substantive rule interpreting Title VII or whether the Guidance was substantively unreasonable.[7] EEOC appealed, and Texas cross-appealed. II. First we must decide two jurisdictional issues: Whether the Guidance is a final agency action that we may review[8] and whether Texas has standing to challenge the Guidance. We review de novo whether the district court had jurisdiction.[9] We begin with whether the Guidance is a reviewable final agency action because that analysis contextualizes the standing inquiry. The APA allows judicial review only of a “final agency action,”[10] meaning an action that (1) “mark[s] the consummation of the agency’s decisionmaking process” and (2) “by which rights or obligations have been determined, or from which legal consequences will flow.”[11] The Supreme Court has “long taken” a “‘pragmatic’ approach . . . to finality,”[12] viewing “the APA’s finality requirement as ‘flexible.’”[13] Defendants do not dispute that the Guidance is “the consummation of [EEOC's] decisionmaking process.” Bennett, 520 U.S. at 178 (internal quotation marks and citation omitted). Reviewability vel non of the Guidance thus turns on the second Bennett prong—whether “rights or obligations have been determined” by it, or whether “legal consequences will flow” from it. Id. (citation omitted). A. Courts consistently hold that an agency’s guidance documents binding it and its staff to a legal position produce legal consequences or determine rights and obligations, thus meeting the second prong of Bennett. “The primary distinction between a substantive rule”—which is, by definition, a final agency action[14]—”and a general statement of policy . . . turns on whether an agency intends to bind itself to a particular legal position.”[15] Whether an action binds the agency is evident “if it either appears on its face to be binding[] or is applied by the agency in a way that indicates it is binding.”[16] Courts have looked for mandatory language to determine whether an agency’s action binds it and accordingly gives rise to legal consequences. In some cases, “the mandatory language of a document alone can be sufficient to render it binding.”[17] Similarly, actions that retract an agency’s discretion to adopt a different view of the law are binding. “[I]f a statement denies the [agency] discretion in the area of its coverage[,] then the statement is binding, and creates rights or obligations.” Texas, 809 F.3d at 171 (alterations omitted) (quoting Gen. Elec., 290 F.3d at 382). Put differently, “where agency action withdraws an entity’s previously-held discretion, that action alters the legal regime, binds the entity, and thus qualifies as final agency action.”[18] That withdrawal of discretion distinguishes a policy statement—which leaves the agency “the discretion and the authority to change its position . . . in any specific case” and “does not seek to impose or elaborate or interpret a legal norm”—from a final agency action. Synchor, 127 F.3d at 94. Another indication that an agency’s action binds it and thus has legal consequences or determines rights and obligations is whether the document creates safe harbors protecting private parties from adverse action. “When the language of the [agency] document is such that private parties can rely on it as a norm or safe harbor by which to shape their actions, it can be binding as a practical matter.” Cohen, 578 F.3d at 9 (internal quotation marks and citation omitted). Hawkes is illustrative. There, the Court held that jurisdictional determinations (“JDs”) made by the U.S. Army Corps of Engineers regarding the applicability of the Clean Water Act (“CWA”) to different tracts of land were final agency actions. The Court noted that when the Corps issued a “negative” JD, the landowner was assured that he would be free from CWA enforcement actions for five years. Hawkes, 136 S. Ct. at 1814. But if the Corps found that a tract was subject to the CWA—and thereby issued an “affirmative JD”—the landowner would be denied the benefits of a negative JD. In other words, the issuance of JDs produced “legal consequences,” giving plaintiffs a safe harbor or not. That the agency’s action binds its staff or creates safe harbors demonstrates that legal consequences flow from it, even when the agency lacks authority to promulgate substantive regulations implementing the statute it administers. What matters is whether the document “ha[s] practical binding effect” such that “affected private parties are reasonably led to believe that failure to conform will bring adverse consequences.” Gen. Elec., 290 F.3d at 383 (internal quotation marks and citation omitted).[19] B. Defendants do not dispute that the Guidance binds EEOC, and for good reason. The Guidance indicates that it binds EEOC staff to an analytical method in conducting Title VII investigations and directs their decisions about which employers to refer for enforcement actions. It also limits discretion respecting the use of certain evidence, mandating that evidence of a racially-balanced workforce cannot overcome a showing of disparate impact. And by broadly condemning “[a] policy or practice requiring an automatic, across-the- board exclusion from all employment opportunities,” Guidance at 16, the Guidance leaves no room for EEOC staff not to issue referrals to the Attorney General when an employer uses a categorical felon-hiring ban. Those characteristics are comparable to the ones in Natural Resources Defense Council v. EPA, 643 F.3d 311 (D.C. Cir. 2011). There, the court held that an EPA guidance was a final agency action because it withdrew agency employees’ discretion to refuse to find state emission-control plans non-compliant with EPA air quality standards on particular grounds. Id. at 319-20. The Guidance here similarly tells EEOC staff that across-the-board limitations are unlawful and forbids staff from considering certain evidence—that of a balanced workforce—when deciding whether an employer has satisfied Title Vll’s requirements. The Guidance also prescribes a multi-factor framework for employers to use in designing “targeted exclusion” policies that agency personnel must presumptively follow in determining whether an employer’s felon-hiring policy violates Title VII. Guidance at 14, 17-18. That corresponds to Appalachian Power Co. v. EPA, 208 F.3d 1015, 1022 (D.C. Cir. 2000), holding that a guidance document requiring agency staff to use a multi-factor analysis in deciding whether a regulated entity’s activity complied with governing law was a final agency action. The Guidance, moreover, tells employers how to avoid Title VII disparate-impact liability. It describes “[t]wo circumstances in which the Commission believes employers will consistently meet the ‘job related and consistent with business necessity’ defense.” Guidance at 14. Those “safe harbors” are, admittedly, not so definite as is the effect of JDs in Hawkes, nor are the consequences of failing to abide by the Guidance as severe as the penalties threatened in Hawkes. But viewing the finality of the Guidance flexibly and pragmatically, as we are bound to do, the Guidance affects “the field of potential plaintiffs” in a way that carries legal consequences and dictates employers’ rights and obligations. Hawkes, 136 S. Ct. at 1814. The Guidance is “binding as a practical matter” because “private parties can rely on it as a norm or safe harbor by which to shape their actions.” Cohen, 578 F.3d at 9 (citation omitted). In fact, the Guidance explicitly declares that it is intended to be a play- book for employers to use to avoid liability, and it describes “best practices” for employers. Guidance at 3; see id. at 25 (listing “Employer Best Practices”). Further, it is supposed to be used by “individuals who suspect that they have been denied jobs or promotions, or have been discharged because of their criminal records,” id. at 3, thus opening the “field of potential plaintiffs,” Hawkes, 136 S. Ct. at 1814. C. Conceding that the Guidance binds EEOC, Defendants, for three reasons, insist that legal consequences do not flow from it. First, they posit that because EEOC has no power to bring a Title VII enforcement action against Texas, its Guidance has no legal consequences for the state. EEOC emphasizes that the Guidance “applies solely to how the EEOC conducts a preliminary, non-final step in the administrative process,” i.e., how it investigates a charge of discrimination and decides whether to issue a right-to-sue letter. EEOC staff have no “authority to impose penalties [on Texas] for non-compliance with its provisions or applicable provisions of Title VII.” That theory makes the Guidance’s finality dependent on plaintiff’s identity. EEOC’s position allows the Guidance to constitute final agency action if the plaintiff is a private employer against which EEOC can bring an enforcement action, but non-final if the plaintiff is a public employer. Finality, however, cannot vary depending on who sued the agency; it depends on the rule itself.[20] That argument, furthermore, is hard to reconcile with Hawkes, which teaches that legal consequences may flow from an agency action even if “no administrative or criminal proceeding can be brought for failure to conform” to the action. Hawkes, 136 S. Ct. at 1815. The decision in Frozen Food Express v. United States, 351 U.S. 40 (1956), likewise undermines Defendants’ contention that, because EEOC lacks enforcement authority over Texas, the Guidance is not final agency action. In Frozen Food Express, the Interstate Commerce Commission issued an order that listed commodities it considered statutorily exempt from regulation. Id. at 42. “Although the order ‘had no authority except to give notice of how the Commission interpreted’ the relevant statute, and ‘would have effect only if and when a particular action was brought against a particular carrier,’ [the Supreme Court] held that the order was . . . immediately reviewable.” Hawkes, 136 S. Ct. at 1815 (internal citation omitted) (first quoting Abbott, 387 U.S. at 150, then citing Frozen Food Express, 351 U.S. at 44-45). The order, moreover, warned carriers that they risked civil and criminal penalties by transporting the non-exempt commodities, thus indicating that the order was a reviewable agency action. Frozen Food Express, 251 U.S. at 44.[21] Invoking Luminant, 757 F.3d at 442, Defendants secondly object that any legal consequences flow from Title VII, not the Guidance, because the Guidance’s interpretation of Title VII disparate impact liability has force of law only if a court presiding over an enforcement action agrees with the Guidance. Defendants reason that whatever consequences flow from the Guidance itself—namely, that Texas state agencies might have to change their felon-hiring policies to avoid liability—are practical, not legal, consequences.[22] But as we have explained, whether the agency action binds the agency indicates whether legal consequences flow from that action. Moreover, Luminant is distinguishable. There we held that the Clean Water Act—not the EPA’s notices of violation to the plaintiff energy companies—set forth the plaintiffs’ rights and obligations. Luminant, 757 F.3d at 442. Put differently, the EPA notices “merely expressed the agency’s opinion about the legality of the plaintiff’s conduct; it did not . . . commit the administrative agency to a specific course of action should the plaintiff fail to comply with the agency’s view.” Texas I, 827 F.3d at 384 (discussing Luminant).[23] Here, the Guidance dictates how EEOC must assess claims of Title VII disparate-impact liability targeting employers with felon-hiring policies. The Guidance does not merely comment on a single employer’s practices; it tells EEOC staff and all employers what sort of policy is unlawful. Defendants thirdly insist that the Guidance’s alleged safe harbors are not safe at all, observing that the Guidance equivocally “states only that the Commission ‘believes’ that certain practices will ‘consistently’ meet statutory requirements.” Defendants maintain, moreover, that the Guidance does not— and could not—”create a safe harbor guaranteeing that the Attorney General will not sue an employer.” That is true, but, as discussed above, whether the Guidance is final agency action does not hinge on whether a private or public employer challenges it. Such an approach would flout the Supreme Court’s repeated instruction to approach finality flexibly and pragmatically. See Abbott Labs., 387 U.S. at 149; Qureshi, 663 F.3d at 781. The Guidance purports to bind EEOC staff when conducting investigations. It instructs staff that if employers use one of the two stipulated methods for assessing applicants, they will have demonstrated that their exclusion is job-related and consistent with business necessity. The Guidance accordingly establishes safe harbors, thereby producing legal consequences and determining rights and obligations of regulated parties. In sum, the Guidance “has the effect of committing the agency itself to a view of the law that, in turn, forces the plaintiff either to alter its conduct, or expose itself to potential liability.” Texas I, 827 F.3d at 383. Legal consequences thus flow from the Guidance, and it determines rights and obligations. The Guidance accordingly satisfies the second prong of Bennett and is a final agency action that we have jurisdiction to review. III. We also must decide whether Texas has standing to sue EEOC and the Attorney General to challenge the legality of the Guidance. As the party invoking federal jurisdiction, Texas must establish Article III standing by showing that it has suffered “an injury that is ‘concrete, particularized, and actual or imminent; fairly traceable to the challenged action; and redressable by a favorable ruling.’” Texas, 809 F.3d at 150 (quoting Clapper v. Amnesty Int’l USA, 568 U.S. 398, 409 (2013)). “[E]ach element [of standing] must be supported in the same way as any other matter on which the plaintiff bears the burden of proof, i.e., with the manner and degree of evidence required at the successive stages of the litigation.” Lewis v. Casey, 518 U.S. 343, 358 (1996) (first alteration added) (quoting Lujan v. Defs. of Wildlife, 504 U.S. 555, 561 (1992)). “In response to a summary judgment motion,” therefore, the plaintiff “must set forth by affidavit or other evidence specific facts, which for purposes of the . . . motion will be taken to be true.” Id. (quoting Defs. of Wildlife, 504 U.S. at 561). We “review[] questions of standing de novo.” McKinley v. Abbott, 643 F.3d 403, 406 (5th Cir. 2011) (citation omitted). A. 1. Because it is the object of the Guidance and has suffered multiple injuries as a result, Texas has constitutional standing. If, in a suit “challenging the legality of government action,” “the plaintiff is himself an object of the action . . . there is ordinarily little question that the action or inaction has caused him injury, and that a judgment preventing or requiring the action will redress it.”[24] “Whether someone is in fact an object of a regulation is a flexible inquiry rooted in common sense.” Contender Farms, 779 F.3d at 265. That common-sense inquiry is easy here. The Guidance explicitly states that it applies to state employers and “[wa]s issued as part of [EEOC's] efforts to eliminate unlawful discrimination in employment screening . . . by entities covered by Title VII,” including state employers. Guidance at 3. Thus, by its own terms, the Guidance covers Texas. Texas suffers at least two injuries as an object of the Guidance. “An increased regulatory burden typically satisfies the injury in fact requirement.” Contender Farms, 779 F.3d at 266. Additionally, “being pressured to change state law constitutes an injury,” because “states have a sovereign interest in the power to create and enforce a legal code.” Texas v. United States, 787 F.3d 733, 749 (5th Cir. 2015) (cleaned up).[25] The Guidance does both. It deems unlawful the hiring practices of multiple Texas agencies by rejecting across-the-board felon hiring screens. Guidance at 16. And it warns the state that the state will “consistently” be able to show that its current policies for hiring police officers, game wardens, school teachers, and youth corrections offers are lawful under Title VII only if it abandons those policies and adopts one of two procedures authorized by the Guidance. See id. at 14. Because the Guidance tells EEOC staff to treat it as binding in its investigations, id. at 3, Texas faces the possibility of investigation by EEOC and referral to the Attorney General for enforcement proceedings if it fails to align its laws and policies with the Guidance.[26] In sum, Texas has opted to express certain values by excluding felons from many positions of public employment, and the Guidance imposes a regulatory burden on Texas to comply with the Guidance to avoid enforcement actions and, consequently, pressures it to abandon its laws and policies. 2. Beyond those injuries, Texas has adequately established that it suffered a procedural injury jeopardizing its concrete interests. A plaintiff can show a cognizable injury if it has been deprived of “a procedural right to protect [its] concrete interests.” Summers v. Earth Island Inst., 555 U.S. 488, 496 (2009) (emphasis omitted). A violation of the APA’s notice-and-comment requirements is one example of a deprivation of a procedural right. See Sierra Club v. EPA, 699 F.3d 530, 533 (D.C. Cir. 2012). The redressability requirement is lighter when the plaintiff asserts deprivation of a procedural right. “When a litigant is vested with a procedural right, that litigant has standing if there is some possibility that the requested relief will prompt the injury-causing party to reconsider the decision that allegedly harmed the litigant.” Texas, 809 F.3d at 150-51 (quoting Massachusetts v. EPA, 549 U.S. 497, 518 (2007)). We assume, for purposes of the standing analysis, that Texas is correct on the merits of its claim that the Guidance was promulgated in violation of the APA. See Sierra Club, 699 F.3d at 533. And that violation undercuts Texas’s concrete interest, as a sovereign state, in maintaining compliance with its laws. Texas, 787 F.3d at 749. The Guidance warns employers that they may not defend a felon-exclusion policy on the ground that it was adopted to comply with a state or local law or regulation.[27] The Guidance consequently encourages employers, to avoid liability, to deviate from state law when it conflicts with the Guidance. 3. Defendants respond on two fronts. First, they revert to their objection that the Guidance is not a final agency action, so Texas cannot show that it is an object of such an action. We already have rejected that contention, which, in any event, erroneously conflates the finality analysis with standing.[28] Second, Defendants maintain that Texas cannot show an immediate injury because “the Guidance does not compel Texas to do anything.” But it would strain credulity to find that an agency action targeting current “unlawful” discrimination among state employers—and declaring presumptively unlawful the very hiring practices employed by state agencies—does not require action immediately enough to constitute an injury-in-fact.[29] B. Texas’s injuries are fairly traceable to EEOC’s promulgation of the Guidance. The Guidance, not Title VII, condemns Texas’s felon-hiring policies, and it, not Title VII, pressures Texas to change its laws and policies or risk referral to the Attorney General by EEOC.[30] Texas’s injuries are likewise traceable to the Attorney General. On remand, Defendants began distinguishing between Texas’s alleged lack of standing to sue the Attorney General, and any standing to sue EEOC. Defendants now stress that the Attorney General is not bound to enforce the Guidance, and, moreover, DOJ currently eschews EEOC’s approach to Title VII disparate-impact claims “in numerous respects.” Defendants also allege that current DOJ policy bars it from enforcing the Guidance. They conclude that “there is no material probability that the [Attorney General] would seek to enforce the Guidance at all” and that Texas, accordingly, has failed to show an injury-in-fact fairly traceable to the Attorney General. We disagree. Defendants miss the mark by focusing on DOJ’s current litigation position. “In identifying an injury that confers standing, courts look exclusively to the time of filing.”[31] Thus, that DOJ recently changed its position and no longer shares the Guidance’s views on disparate-impact liability and criminal-record hiring policies does not impact our standing analysis. We focus, instead, “on whether the party invoking jurisdiction had the requisite stake in the outcome when the suit was filed.” Davis v. FEC, 554 U.S. 724, 734 (2008). With the proper time frame in view, we have no trouble concluding that Texas’s injuries are fairly traceable to the Attorney General. The pressure on Texas to change its laws exists, in part, because the Attorney General has prosecutorial power to bring enforcement actions against Texas based on EEOC referrals or a pattern-or-practice claim. That was true when Texas filed the suit, and it remains so now. That the Attorney General has not attempted to enforce the Guidance against Texas does not deprive it of standing.[32] Finally, Texas’s injuries would be redressed by a favorable ruling. An injunction forbidding EEOC and the Attorney General from enforcing the Guidance would safeguard Texas’s sovereign interests. C. Although standing is assessed as of the date on which suit was filed, courts may not decide cases that since have become moot because there is no longer a live case or controversy. See, e.g., U.S. Parole Comm’n v. Geraghty, 445 U.S. 388, 397 (1980). Thus, even though Defendants do not raise the possibility that Texas’s claims against the Attorney General are moot considering DOJ’s new position, we must ponder that circumstance. A case does not necessarily become moot when a defendant voluntarily ceases the objectionable conduct. United States v. W. T. Grant Co., 345 U.S. 629, 632 (1953) (“[V]oluntary cessation of allegedly illegal conduct does not deprive the tribunal of power to hear and determine the case, i.e., does not make the case moot.”). That is because “[t]he defendant is free to return to his old ways.” Id. But a “case may nevertheless be moot if the defendant can demonstrate that there is no reasonable expectation that the wrong will be repeated.” Id. at 633 (internal quotation marks and citation omitted). “The burden is a heavy one.” Id. The defendant must show that “subsequent events made it absolutely clear that the allegedly wrongful behavior could not reasonably be expected to recur.”[33] Defendants have not carried that heavy burden. There is no direct evidence that the Attorney General has committed not to honor referrals from EEOC based on the Guidance or that Texas otherwise cannot reasonably expect its injury to recur. The only indication that the Attorney General will not follow the Guidance is a general memorandum from the Associate Attorney General to the Heads of Civil Litigating Components and United States Attorneys dated January 25, 2018, which instructs that “effective immediately for [Affirmative Civil Enforcement] cases, the Department may not use its enforcement authority to effectively convert agency guidance documents into binding rules.”[34] That broad memorandum does not “ma[k]e it absolutely clear that the allegedly wrongful behavior [can] not reasonably be expected to recur.” McKinley, 643 F.3d at 406 (citation omitted). And even if it did, Defendants— the Attorney General included—endorsed parts of the Guidance on remand, acknowledging that, to avoid liability, employers must meet one of the two safe harbors in the Guidance and that “the Guidance is fairly encompassed by Title VII.”[35] To summarize, Texas has suffered cognizable injuries that are fairly traceable to EEOC and the Attorney General. Defendants’ mutual authority to enforce the Guidance are two sides of the same coin. While the Guidance is in place, EEOC pressures Texas to comply with the threat of referral to the Attorney General for further legal action. The Attorney General’s statutory authority to sue Texas is another source of leverage. Both sources would be redressed by a judgment in Texas’s favor. The Attorney General’s recent change of position has not mooted the suit. IV. Because we have jurisdiction, we turn to the challenges to the scope and phrasing of the injunction. Texas contends that EEOC lacked power to promulgate the Guidance at all. It thus asserts that, instead of barring enforcement until the Guidance goes through notice-and-comment rulemaking, the district court should have enjoined Defendants from treating the Guidance as binding. Defendants, for their part, aver that the injunction is impermissibly vague and overbroad because it fails to specify whether it bars EEOC and the Attorney General from enforcing the Guidance as such or from enforcing the interpretation of Title VII embodied in the Guidance. We review de novo the scope of an injunction. See FDIC v. Faulkner, 991 F.2d 262, 267 (5th Cir. 1993). A. The district court’s injunction states that Defendants EEOC and the Attorney General of the United States (in any enforcement action against the State of Texas) are ENJOINED from enforcing the EEOC’s interpretation of the Guidance against the State of Texas until the EEOC has complied with the notice and comment requirements under the APA for promulgating an enforceable substantive rule. Emphasizing that the Guidance is effectively a substantive rule and that EEOC is not authorized to promulgate substantive rules to implement Title VII,[36] Texas urges us to find that EEOC lacked power to promulgate the Guidance at all. It therefore asks us to modify the injunction so that it will unconditionally enjoin treatment of the Guidance as binding.[37] We agree that the Guidance is a substantive rule subject to the APA’s notice-and-comment requirement and that EEOC thus overstepped its statutory authority in issuing the Guidance. That conclusion follows naturally from our holding that the Guidance is a final agency action. The district court held that the Guidance “is a substantive rule,” but it did not deem it “necessary to the adjudication of the claims” to reach the issue whether EEOC lacked authority to promulgate a substantive rule. The injunction, however, implies the answer, given that it would permit the Guidance to stand if it went through notice-and-comment rulemaking—a process used to promulgate substantive rules. Because the Guidance is a substantive rule, and the text of Title VII and precedent confirm that EEOC lacks authority to promulgate substantive rules implementing Title VII, we modify the injunction by striking the clause “until the EEOC has complied with the notice and comment requirements under the APA for promulgating an enforceable substantive rule.” B. Under Federal Rule of Civil Procedure 65(d)(1)(B) and (C), every injunction must “state its terms specifically; and . . . describe in reasonable detail—and not by referring to the complaint or other document—the act or acts restrained or required.” “The specificity requirement is not unwieldy. An injunction must simply be framed so that those enjoined will know what conduct the court has prohibited.” Meyer v. Brown & Root Constr. Co., 661 F.3d 369, 373 (5th Cir. Nov. 1981); accord Scott v. Schedler, 826 F.3d 207, 213 (5th Cir. 2016) (per curiam). Defendants contend that the injunction equivocates whether “the EEOC and Attorney General are barred from enforcing the Guidance as such[] or are barred from enforcing an interpretation of Title VII that is embodied in the Guidance.” But Defendants also suggest that the meaning is evident from the context of the suit, noting that the only interpretation of the injunction’s scope consistent with the district court’s ruling against EEOC on procedural grounds is that the injunction bars EEOC and the Attorney General from treating the Guidance as binding. Because “[a]n injunction must simply be framed so that those enjoined will know what conduct the court has prohibited,” and Defendants appear to admit that the injunction bears just one construction, the injunction meets the specificity requirement of Rule 65. See id. But to avoid any confusion, we modify the injunction to clarify that EEOC and the Attorney General may not treat the Guidance as binding in any respect. V. Finally, we consider whether to reach the merits of Texas’s DJA claim. The DJA “confer[s] on federal courts unique and substantial discretion in deciding whether to declare the rights of litigants.” Wilton v. Seven Falls Co., 515 U.S. 277, 286 (1995). Ample precedent establishes that we should not exercise our discretion to extend declaratory relief when a challenged law or policy no longer affects the plaintiff.[38] Because we affirm the injunction, we decline to consider Texas’s DJA claim. We thus vacate the district court’s ruling on the merits of that claim. The injunction is AFFIRMED as modified, the district court’s ruling on the merits of Texas’s DJA claim is VACATED, and the DJA claim is DISMISSED. No remand is needed.
