Since at least 1980, there has been a documented increase in regulatory activity at the end of presidential terms, especially in the post-election period when the outgoing President’s successor is from the other party. This phenomenon has come to be known as “midnight regulation,” and the products of end-of-term legislative rulemaking are referred to as “midnight rules.” While a study I conducted for the Administrative Conference of the United States revealed that most midnight rules are routine, some are not and are designed to project the agenda of the outgoing administration into the future and force the incoming administration to expend precious time and political capital on unwinding the last minute regulatory frenzy.
Incoming administrations have developed strategies to deal with the midnight regulation problem and more generally take control of the administrative state. Typical steps have included freezing the issuance of new rules until they can be reviewed by an appointee of the new President, postponing the effective dates of rules that had been published but not yet gone into effect, and withdrawing rules that had been sent to the Federal Register but not yet published. While it is clear that the administration has authority to pause the issuance of new rules and that revocation of already-published rules requires a new round of notice and comment, two aspects of this strategy have always been plagued by procedural uncertainty. First, does a delay in the published effective date of a rule recently promulgated require advance notice and comment or at least notice and comment as soon as possible after the delay goes into effect? Second, and more relevant to this essay, what is the legal status of a rule that has been submitted to the Federal Register but not yet published? The last five administrations, dating back to President Bill Clinton, have ordered the withdrawal of all unpublished rules from the Federal Register and have treated them as if they had never been promulgated, even though the Secretary of the relevant Department would have already signed the rules and possibly posted them on the agency’s website before sending them to the Federal Register for publication.
In at least two memoranda on midnight rulemaking, the Congressional Research Service has concluded, without much analysis, that rules may be freely withdrawn until they are published in the Federal Register. Recently, a panel of the U.S. Court of Appeals for the D.C. Circuit, in an opinion by Democratic appointee Judge David Tatel, joined by Democratic appointee Judge Patricia Millet, cast serious doubt on this conclusion, holding that once a rule has been sent to the Federal Register and “placed on public inspection,” which is the last step before actual publication, the rule has become legally effective and may be withdrawn only pursuant to a new round of notice and comment. Humane Society of the United States v. USDA, 41 F.4th 54 (2022).
At first I was skeptical of the court’s conclusion, but on further consideration, especially after reviewing the text of the relevant provisions of the APA, I have come to agree with the D.C. Circuit that once a rule has been finalized within the agency, it has become legally effective and the incoming administration cannot simply treat it as if it never existed.
The Department of Agriculture’s rule at issue in Humane Society would have outlawed a procedure in which show horses’ legs are “sored” to create their distinctive gait. As the court explained, “soring” involves cutting, burning or otherwise inflicting pain on a horse’s legs to alter the horse’s gait to that of a trained show horse, saving the time and expense of a lengthy training regimen. On January 11, 2017, just three days before the end of President Obama’s second term in office, the Department of Agriculture posted the signed rule on its website and sent it to the Office of the Federal Register (OFR) for publication. On January 19, the OFR “scheduled the rule for publication and made it available for public inspection.” On January 20, Donald Trump became President and, pursuant to general instructions issued by his Chief of Staff Reince Priebus, the rule was withdrawn from the Federal Register and never heard from again. The Humane Society and some of its members sued to challenge the rule’s withdrawal, arguing that it became final when OFR made it available for public inspection or even earlier when the Department of Agriculture posted the signed rule it on its website.
In light of the fact that several administrations have withdrawn rules that have been submitted to the Federal Register and treated them as if they had not been issued, it might be surprising to learn that the text of the Administrative Procedure Act (APA), properly understood, strongly supports the D.C. Circuit’s conclusion that publication in the Federal Register is not required before a rule becomes legally effective. Clearly, the APA requires that any “substantive rules of general applicability” be published in the Federal Register after they are issued. 5 U.S.C. §552(a)(1)(D). This is where the role of publication becomes clear: The penalty for failure to publish a rule is not ineffectiveness, but rather that, “Except to the extent that a person has actual and timely notice of the terms thereof, a person may not in any manner be required to resort to, or be adversely affected by, a matter required to be published in the Federal Register and not so published.” 5 U.S.C. § 552(a)(1). By providing that a person with “actual and timely notice” of an unpublished rule may “be required to resort” to it, the APA declares that issued, but unpublished rules, are legally in effect. Publication, in other words, provides notice of the preexisting fact that the rule has gone into effect.
Interestingly, this is not the analysis by which Judge Tatel’s opinion reached the panel’s conclusion. Rather, it relied primarily on the Federal Register Act (FRA), which on this matter tracks the APA:
Making a document available for public inspection “is sufficient to give notice of the contents of the document to a person subject to or affected by it.” 44 U.S.C. § 1507. A document “is not valid as against a person who has not had actual knowledge of it until . . . [it is] made available for public inspection.” Id. Federal Register publication then “creates a rebuttable presumption” that the document was “duly issued, prescribed, or promulgated” and that it was properly “made available for public inspection at the day and hour stated in the printed notation.” Id.
Far from bolstering the government’s position, the Federal Register Act forecloses its argument that an agency prescribes a rule only once the rule is published in the Federal Register. The statute repeatedly distinguishes between the publication of a document and its issuance, prescription, or promulgation.
Humane Society, 41 F.4th at 569.
In fact, as the court noted, when “[c]onfronted with this language at oral argument, government counsel conceded that ‘a rule can be issued, prescribed, or promulgated without publication in the Federal Register or prior to publication in the Federal Register.’” Id. The court then resorted to the APA only to explain that the “APA requires notice and comment before ‘repealing a rule.’ 5 U.S.C. §§ 551(5), 553. ” Id. at 569-70. Judge Tatel may be correct that the FRA demonstrates that rules are effective once they have been placed on public inspection, but because the APA provides the procedure for rulemaking, in my view it would have been preferable for the opinion to rely primarily on the APA and use the Federal Register Act only as support.
The dissent, discussed at length below, raised the specter of agencies being unable to recall rules with errors or other infirmities that were not noticed until the rule became public. In response, the court noted, an agency can invoke the APA’s “good cause” exception to attempt to dispense with notice and comment before repealing or delaying the effectiveness of a midnight rule. I have sympathy for this view: as I have described in previous work, midnight rulemaking presents an incoming administration with challenges that can threaten to overwhelm agencies when they ought to be free to focus on advancing the new administration’s regulatory agenda. In my view, new administrations ought to be relatively free to renounce recently promulgated rules that have not yet gone into effect. But the courts, including the D.C. Circuit, have not been receptive to this strategy, holding that once a rule has been finalized by a prior administration, a new administration must employ APA procedures to delay, amend or repeal it. Air Alliance Houston v. EPA, 906 F.3d 1049 (D.C. Cir. 2018); Natural Resources Defense Council v. National Highway Traffic Safety Administration, 894 F.3d 95 (2d Cir. 2018); see also Environmental. Defense Fund v. Gorsuch, 713 F.2d 802 (D.C. Cir. 1983); Natural Resources Defense Council v. Abraham, 355 F.3d 179 (2d Cir. 2004).
There is one complication that might affect the applicability of the Humane Society rule against treating unpublished rules as if they had never been promulgated: agencies often specify that a rule’s effective date is the date of publication in the Federal Register. The OFR then fills in the effective date upon publication. The Second Circuit confronted this issue in a case involving claims for asylum based on China’s “one child” policy. Zhang v. Slattery, 55 F.3d 732 (2d Cir. 1995). Attorney General William Barr had signed a rule in the final days of the George W. Bush administration, but the rule was never published and the effective date specified in the text of the rule was as follows: “EFFECTIVE DATE: [Insert date of publication in the FEDERAL REGISTER.]” Because the effective date had not been filled in, the court held that “the January 1993 ‘final rule’ never became effective in the first place. Until the ‘EFFECTIVE DATE’ was reached—by publication—there was no rule to repeal.” Id. at 749. To prevent easy reversal by an incoming administration, outgoing administrations must be careful to specify an effective date rather than leave it open for OFR to specify.
Republican appointee Judge Naomi Rao dissented, citing the decades-long tradition of incoming administrations treating unpublished rules as a nullity. In her view, “publication determines the adoption, finality, and effectiveness of a substantive rule.” Humane Society, 41 F.4th at 576 (Rao, J., dissenting). She relied primarily on two arguments, first D.C. Circuit precedent, namely, Kennecott Utah Copper Corporation v. Department of the Interior, 88 F.3d 1191 (D.C. Cir. 1996), and second that APA § 553(d) specifies that rules are effective only after publication in the Federal Register and that this section should be interpreted as amending the FRA. However, although I agree with her as a matter of policy, neither of her legal arguments withstand scrutiny.
Her first argument is that the D.C. Circuit had already decided, in Kennecott Utah Copper Corporation v. Department of the Interior, 88 F.3d 1191 (D.C. Cir. 1996), that agency rules like the one at issue in Humane Society become legally effective upon publication and not before. Kennecott Copper involved regulations that were sent to the OFR by the Department of the Interior on January 19, 1993, the final full day of George W. Bush’s administration. On January 21, 1993, “before the OFR filed the document for public inspection” Interior withdrew the document, and these regulations were treated by the incoming administration as if they had never existed. Id. at 1200-01 (emphasis supplied). The agency then re-opened the comment period and, pursuant to the original rulemaking record and the new one, promulgated a new set of regulations that would have superseded the 1993 regulations even if they had been treated as legally operative. The problem with Judge Rao’s characterization of the Kennecott Copper case is that the rules in that case were withdrawn before being placed on public inspection during what OFR’s regulations characterized as a “three-day confidential processing period” during which OFR regulations provided for withdrawal of rules. The court then stated that OFR’s rules, “which allow agencies to withdraw documents during the confidential processing period” reasonably interpret the FRA. 88 F.3d at 1207. Kennecott Copper said nothing at all about withdrawal of rules after they have been made available for public inspection. Nor could it create binding precedent on that subject since the rules in that case had been withdrawn before they reached that stage.
Judge Rao’s second argument, that the APA requires publication before a rule can be considered legally effective, is, in my view, not the best interpretation of APA § 553(d). That subsection provides that “[t]he required publication or service of a substantive rule shall be made not less than 30 days before its effective date.” She interprets this provision to mean that the effective date of all rules is measured from either service or publication. She relies upon the Attorney General’s Manual on the APA for support, but the Manual provides stronger support for an interpretation that does not implicate a rule’s legal status, but only that a person cannot be adversely affected by a legally valid rule less than 30 days after the person has actual or constructive notice of it. The Manual notes that “[t]he purpose of the time lag required by section 4 (e) is to ‘afford persons affected a reasonable time to prepare for the effective date of a rule or rules or to take any other action which the issuance of rules may prompt’. Sen. Rep. p. 15; H.R. Rep. p. 25.” Attorney General’s Manual at 36. Because the FRA and Kennecott Copper equate filing for public inspection with actual publication, there is no inconsistency between the APA and the FRA on this matter.
Judge Rao correctly notes that the Manual observes that “the delayed effective date provision now codified at 5 U.S.C. § 553(d), may be interpreted as amending Section 7 of the Federal Register Act.” 41 F.4th at 581 (Rao, J. dissenting). But Judge Rao turns the “may” in this statement to a “must,” arguing “[t]hat this consequence followed from the plain meaning of the APA was understood at the time of its enactment.” Id. This is an overstatement at best—the Manual was expressing the possibility that courts would arrive at such an interpretation and was certainly not stating an opinion as to § 553(d)’s definitive meaning or presenting a contemporaneous understanding of that meaning. Rather, given the Manual’s characterization of the legislative history, the more likely import of the provision was to require thirty days’ notice before adverse consequences may flow from an otherwise valid rule. Otherwise, on Judge Rao’s view, rules could be freely withdrawn even 29 days after publication, a conclusion no one suggests.
As we are learning quite often these days, even decades of consistent practice is often insufficient to create binding administrative law. Judge Rao’s view was clearly supported by decades of practice by incoming Presidents and agencies, who treated rules that were withdrawn before publication as a nullity. Judge Rao’s view is also supported by policy considerations surrounding the midnight rulemaking phenomenon. Incoming administrations are often left with hundreds of pages of newly-minted rules to sift through. Some of them may have been rushed, while others may fly directly in the face of the policies upon which the incoming President was elected. Reviewing all of them occupies precious time and effort on the part of the incoming administration. Allowing liberal withdrawal of midnight rules would enable the incoming administration to better pursue the agenda that led to its election and would generally allow for a smoother transition. It is only the law that seems to be an impediment to the desire of incoming administrations to treat unpublished rules as if they had never even existed.
Jack M. Beermann is Professor of Law and Political Science and Harry Elwood Warren Scholar, Boston University.
 Congressional Research Service, R42612, Midnight Rulemaking: Background and Options for Congress (updated Oct. 4, 2016) ; Congressional Research Service, LSB10566, Responses to Midnight Rulemaking: Legal Issues (Jan. 21, 2021).
 See Jack M. Beermann, Presidential Power in Transitions, 83 B.U. L. Rev. 947 (2003); Jack M. Beermann & William P. Marshall, The Constitutional Law of Presidential Transitions, 84 N.C. L. Rev. 1253 (2006).