American Federation of Government Employees v. FLRA

United States Court of Appeals
FOR THE DISTRICT OF COLUMBIA CIRCUIT

Argued October 7, 2021 Decided February 1, 2022

No. 20-1396

AMERICAN FEDERATION OF GOVERNMENT EMPLOYEES,
AFL-CIO,
PETITIONER

v.

FEDERAL LABOR RELATIONS AUTHORITY,
RESPONDENT

Consolidated with 20-1397, 20-1404

On Petitions for Review of an Order
of the Federal Labor Relations Authority

Matthew W. Milledge argued the cause for petitioners.
With him on the briefs were David A. Borer, Andres M.
Grajales, Judith E. Rivlin, Teague P. Paterson, Gregory
O=Duden, Julie M. Wilson, and Allison C. Giles.

Noah Peters, Solicitor, Federal Labor Relations Authority,
argued the cause for respondent. With him on the brief were
Rebecca J. Osborne, Deputy Solicitor, and Sarah C.
Blackadar, Attorney.

Before: TATEL, PILLARD and JACKSON, Circuit Judges.
Opinion for the Court filed by Circuit Judge JACKSON.
 2

JACKSON, Circuit Judge: By statute, certain federal
employers are required to engage in collective bargaining with
their employees’ representatives whenever there is a
management-initiated change to the “conditions of
employment affecting such employees.” Federal Service
Labor-Management Relations Statute, 5 U.S.C. § 7103(a)(12);
see also id. §§ 7102(2), 7103(a)(14). Congress has defined
“conditions of employment” to include “personnel policies,
practices, and matters . . . affecting working conditions,” with
certain enumerated exceptions. Id. § 7103(a)(14). And from
the mid-1980s until the policy statement challenged here, the
Federal Labor Relations Authority (“FLRA”) interpreted these
statutory provisions to require collective bargaining over any
workplace changes that have more than a de minimis effect on
such working conditions.
In September of 2020, the FLRA adopted a new threshold
for when collective bargaining is required. Under the agency’s
new standard, the duty to bargain is triggered only if a
workplace change has “a substantial impact on a condition of
employment.” U.S. Dep’t of Educ., 71 F.L.R.A. 968, 971
(2020). The petitioners are public-sector labor unions that
challenge the FLRA’s decision to alter the bargaining
threshold; they maintain that the FLRA’s new standard is both
inconsistent with the governing statute and insufficiently
explained, and is therefore arbitrary, capricious, and contrary
to law.
In the opinion that follows, we hold that the FLRA’s
decision to abandon its de minimis exception in favor of a
substantial-impact threshold was not sufficiently reasoned, and
thus is arbitrary and capricious in violation of section 706 of
the Administrative Procedure Act (“APA”), 5 U.S.C.
§ 706(2)(A). We therefore grant the unions’ petitions for
review and vacate the FLRA’s policy statement.
BACKGROUND
Before the fall of 2020, it was the longstanding view of the
FLRA that, despite certain federal employers’ clear statutory
 3
duty to engage in collective bargaining over “conditions of
employment,” 5 U.S.C. § 7103(a)(12), some public-sector
management decisions were not subject to bargaining if they
had only a “de minimis impact” on such conditions. Dep’t of
Health & Human Servs. Soc. Sec. Admin., 24 F.L.R.A. 403, 407
(1986); see also Dep’t of Health & Human Servs. Soc. Sec.
Admin. Reg. V, 19 F.L.R.A. 827, 834 (1985) (McGinnis,
concurring); Dep’t of Health & Human Servs. Soc. Sec. Admin.
Chi. Region, 15 F.L.R.A. 922, 924 (1984). The de minimis
impact exception, the FLRA explained, “derive[s] from the
Latin phrase ‘De minimis non curat lex,’ which . . . mean[s]
that the law does not care for, or take notice of, very small or
trifling matters[.]” Dep’t of Health & Human Servs. Soc. Sec.
Admin, 24 F.L.R.A. at 407 & n.2 (quoting De Minimis Non
Curat Lex, BLACK’S LAW DICTIONARY (5th ed. 1979)). In
other words, the FLRA read 5 U.S.C. § 7103(a)(12) and (14) to
imply a narrow exception to the statute’s collective bargaining
requirement; one that pertained to management decisions that
had only a trivial effect on conditions of employment.
In October of 2019, the United States Departments of
Education and Agriculture jointly requested that the FLRA
issue a general statement of policy changing the impact
threshold at which collective bargaining becomes mandatory.
See 5 C.F.R. § 2427.2 (providing for general statements of
policy or guidance). The Departments alleged several
problems with the longstanding de minimis standard. For
instance, they asserted that the de minimis policy was
insufficiently concrete to permit consistent application, causing
unnecessary litigation. They also asserted that “effective and
efficient government would be promoted by the establishment”
of a clearer standard, and they maintained that the ideal
standard would be one under which only a “substantial
change”—or, rather, a change having a “substantial impact” on
conditions of employment—triggers the duty to bargain.
Request for General Statement of Policy or Guidance at 3–4,
J.A. 3–4.
On September 30, 2020, the FLRA adopted the
Departments’ proposed standard over a dissent and without
 4
soliciting public comment. In a four-page policy statement, the
FLRA announced that “an agency will not be required to
bargain over a change to a condition of employment unless the
change is determined to have a substantial impact on a
condition of employment.” Dep’t of Educ., 71 F.L.R.A. at 971.
The FLRA stated that the “more than de minimis” test for
determining when the duty to bargain is triggered is “not the
appropriate standard,” id., in part because that test had resulted
in bargaining “whenever management has made any decision,
no matter how small or trivial” and is also “unpredictable,” id.
at 969. The FLRA further faulted the initial agency decision to
adopt the de minimis standard on the grounds that it was
insufficiently explained. Id. at 970. With respect to the
substantial-impact test, the FLRA suggested that this new
standard would create “a line that [was] meaningful and
determinative[,]” and noted that substantial impact is the
governing bargaining threshold in the private sector. Id.
The FLRA’s dissenting member argued that the agency’s
adoption of a new bargaining threshold was an unjustifiable
departure from past precedent and that there was a clear legal
basis for the agency’s prior adoption of the de minimis
standard. See id. at 972 (DuBester, dissenting). The dissent
also maintained that the substantial-impact test was contrary to
the governing statute, id. at 973, and that the FLRA’s policy
change was not sufficiently reasoned or explained, id. at 973–
76.
The petitioners in these consolidated cases—the American
Federation of Government Employees, AFL-CIO, the National
Treasury Employees Union, and the American Federation of
State, County and Municipal Employees, AFL-CIO—are labor
unions that represent employees of government agencies that
the FLRA’s policy change covers. Each filed a timely petition
for review of the FLRA’s policy statement, and we
consolidated the petitions. The unions have standing as
employee representatives whose “bargaining position” would
be “fundamentally diminished” under the FLRA’s new
interpretation, see Nat’l Treasury Emps. Union v. Chertoff, 452
F.3d 839, 853 (D.C. Cir. 2006), and we have jurisdiction over
 5
their petitions for review, see 5 U.S.C. § 7123(a).
The unions challenge the FLRA’s policy statement on two
grounds, both of which implicate APA standards. First, they
argue that the new substantial-impact threshold rests on an
impermissible reading of 5 U.S.C. §§ 7103(a)(12) and (a)(14);
they contend, in particular, that removing management-
initiated changes whose effects are more than de minimis but
less than substantial from the scope of collective bargaining is
contrary to the plain and unambiguous language of these
statutory provisions. Second, the unions assail the FLRA’s
policy statement adopting the substantial-impact standard as
arbitrary and capricious. See 5 U.S.C. § 706(2)(A). In this
regard, according to the unions, the FLRA’s explanation for
dispensing with its old policy and adopting the new one was
insufficient to support the policy change.
DISCUSSION
Our analysis begins with the petitioners’ arbitrary and
capricious challenge. The petitioners also challenge the
substantial-impact exception as contrary to the statute and
unsupported by any canon that would justify its treatment as
“inherent in most statutory schemes by implication.” See Ass’n
of Admin. Law Judges v. FLRA (AALJ), 397 F.3d 957, 962
(D.C. Cir. 2005) (quoting Env’t Def. Fund, Inc. v. EPA, 82 F.3d
451, 466 (D.C. Cir. 1996)). Because we hold that the policy is
unreasoned, we need not and do not reach the statutory claim.
Cf. Shays v. FEC, 414 F.3d 76, 97 (D.C. Cir. 2005) (declining
to assess a claim about the meaning of a statutory provision
where the failure of the challenged action under the arbitrary
and capricious standard offered a sufficient basis for decision).
Under the arbitrary and capricious standard, “we must
ensure that the [FLRA] ‘examine[d] the relevant data and
articulate[d] a satisfactory explanation for its action including
a rational connection between the facts found and the choice
made.’” AFGE v. FLRA (AFGE 2020), 961 F.3d 452, 456
(D.C. Cir. 2020) (alterations in original) (quoting Fred Meyer
Stores, Inc. v. NLRB, 865 F.3d 630, 638 (D.C. Cir. 2017)). The
key question is, at its core, whether the FLRA “engage[d] in
 6
reasoned decisionmaking.” Fred Meyer, 865 F.3d at 638
(internal citations omitted); see also FCC v. Prometheus Radio
Project, 141 S. Ct. 1150, 1158 (2021). And while the FLRA
certainly “may depart from its precedent,” in so doing, it “must
supply a reasoned analysis indicating that prior policies and
standards are being deliberately changed.” AFGE 2020, 961
F.3d at 457 (quoting Nat’l Fed’n of Fed. Emps. v. FLRA, 369
F.3d 548, 553 (D.C. Cir. 2004)). The agency must also show
that “the new policy is permissible under the statute, that there
are good reasons for it, and that the agency believes it to be
better” than the previous policy. FCC v. Fox Television
Stations, Inc., 556 U.S. 502, 515 (2009) (emphasis omitted).
With these standards in mind and for the reasons explained
below, we conclude that the FLRA’s decision to abandon its
longstanding precedents and adopt the substantial-impact
standard was not sufficiently reasoned in several critical
respects.
A
The first problem with the FLRA’s reasoning is that its
policy statement falls short on explaining the purported flaws
of the de minimis standard.
To start, the policy statement’s description of the problem
it seeks to solve is inconsistent. At the beginning of its
substantive discussion, the FLRA laments that “[it] has
effectively extended the bargaining obligation under the de
minimis test to . . . trigger[] an agency’s duty to bargain
whenever management has made any decision, no matter how
small or trivial.” Dep’t of Educ., 71 F.L.R.A. at 969. The
policy statement explains that this purportedly ubiquitous
misapplication of the de minimis standard to require bargaining
in any and all circumstances has “drained” that standard “of
any determinative meaning[.]” Id. In the preceding
paragraphs, however, the policy statement’s background
section highlights its concern that the de minimis standard is
“unpredictable” and “has created uncertainty,” citing several
examples of cases in which the FLRA concluded that a
workplace change was de minimis and thus beyond the scope
 7
of bargaining. Id. at 969 & n.14; see also id. at 970 n.18
(explaining that the FLRA forewent solicitation of public
comments “because we are well aware of the confusion sown
by nearly thirty-five years of our caselaw”).
It is not at all clear how the de minimis standard could both
lead inexorably to the conclusion that all management
decisions “no matter how small or trivial” must be subject to
bargaining and at the same time yield unpredictable results,
including, by the FLRA’s own telling, many instances in which
the duty to bargain was not triggered. See id. at 969 & n.14.
Yet that is how the FLRA’s policy statement reads: the existing
standard both purportedly subjects every minor decision to
review and is unworkable because it is impossible to predict.
Even looking past this apparent contradiction, the FLRA’s
claim that the de minimis standard’s unpredictability has
“created uncertainty that has negatively impacted labor-
management relations” is unconvincing on its own terms. Id.
at 969 (internal quotation marks omitted). As evidence of the
“vast differences of opinion among arbitrators, judges, and the
[FLRA itself] as to what matters affect conditions of
employment sufficiently to require bargaining,” the FLRA
cites four pairs of decisions that involve purportedly similar
facts in which the FLRA found certain management-initiated
changes to be more than de minimis (and thus subject to
bargaining) and others to be de minimis (and thus not). Id. at
969 & n.14. But closer inspection reveals that these divergent
results are readily explained by distinguishable contexts. See
id. at 973 (DuBester, dissenting) (“[T]he cases cited by the
majority simply reflect the inherently fact-dependent nature of
the de minimis exception.”).
Take, for instance, the policy statement’s citation to two
decisions about office and seating arrangements that the agency
now claims were contradictory. In Social Security
Administration, Baltimore, Maryland, the FLRA concluded
that “changes in seating assignments . . . including the
movement of four employees (one-fourth of all [bargaining]
unit employees) and one employee[’]s[] loss of access to a
 8
window” when considered together were “sufficient to support
the . . . conclusion that the changes in seating arrangements
were more than de minimis.” U.S. Dep’t of Health & Human
Servs., Soc. Sec. Admin., Balt., Md., 36 F.L.R.A. 655, 668, 688
(1990)). In GSA, Region 9, on the other hand, the FLRA
determined that an employer’s decision to temporarily move a
single employee to another building—after her union
suggested that such a temporary move would benefit the
employee—had, in light of “equitable considerations,” only a
de minimis impact on her working conditions. 52 F.L.R.A.
1107, 1108–09, 1111–12 (1997).
To describe these decisions is to distinguish them. And
the latter decision, GSA, Region 9, indeed expressly
distinguished the former as being about a “permanent
relocation[] of employees that w[as] instituted by agency
management for operational reasons,” rather than a temporary
move at the employee’s union’s suggestion. See id. at 1112
(citing U.S. Dep’t of Health & Human Servs., Soc. Sec. Admin.,
Balt., Md., 36 F.L.R.A. at 655). The FLRA’s summary
assertion that these decisions were irreconcilable and thus
evidence of the de minimis exception’s unpredictability thus
lacks merit.
The FLRA’s other examples are more of the same. For
instance, the agency speciously summarizes one of its
decisions as holding that “[r]equiring [an] employee to give up
a ‘second’ office while keeping [his] primary office” was
“more than de minimis” and another as determining that
“moving an employee permanently to a vacant office
[was] held not to be more than de minimis.” Dep’t of Educ.,
71 F.L.R.A. 969 n.14 (first citing U.S. Dep’t of the Air Force,
Air Force Materiel Command, Space & Missile Sys. Ctr.,
Detachment 12, Kirtland Air Force Base, N.M., 64 F.L.R.A.
166, 173–74 (2009); and then citing Nat’l Treasury Emps.’
Union, Chapter 26, 66 F.L.R.A. 650, 653 (2012)).
But these decisions too are readily distinguishable on their
facts. U.S. Department of the Airforce concerned an
employer’s decision to downsize the workspace of a trainer
 9
who was a member of the relevant collective bargaining unit.
64 F.L.R.A. at 173–74. In that case, the FLRA held that the
employer’s decision to downsize the trainer’s primary office at
the same time it took away a second workspace that he had used
to conduct certain face-to-face training and store equipment—
changes made on short notice and without relocation
assistance—when viewed alongside problems with the new
workspace, constituted a more-than-de minimis change. See
id. In stark contrast, National Treasury Employees’ Union,
Chapter 26, was not about the reassignment of a union
employee at all; instead, it concerned a union’s challenge to an
employer’s decision to assign an employee outside the
bargaining unit to a vacant workstation in the same office as
employees in the bargaining unit. 66 F.L.R.A. at 652–53.
Unsurprisingly, the FLRA held that the mere presence of an
employee outside the bargaining unit in the same office as unit
members had only a de minimis effect on the union employees’
conditions of employment. Id. at 653. Once again, the FLRA’s
evidence of an irretrievably broken standard falls flat.
As this court observed once before, “there is little
indication that the de minimis exception” has created the sort
of “confusion” the FLRA now claims. AALJ, 397 F.3d at 963.
Rather than demonstrate such confusion, the examples that the
FLRA cites to frame the problem it seeks to address seem to us
to demonstrate rigorous application by the FLRA of a fact-
intensive standard to varying factual contexts. Put another
way, far from demonstrating the de minimis standard is
unworkable, the FLRA’s policy statement simply appears to
demonstrate how it works.
B
The FLRA’s condemnation of the de minimis test also fails
to grapple with the agency’s own past policy choices and this
court’s decisions upholding them.
1
In the policy statement under review, the FLRA insists that
the agency’s initial adoption of the de minimis threshold back
in 1985 was insufficiently explained and reasons that this initial
 10
failure supports the 2020 policy change. In this regard, the
FLRA explains that the agency had used a substantial-impact
standard for a few years before it adopted the de minimis
threshold in 1985. And it calls the decision to discard the
substantial-impact standard in favor of the de minimis
threshold “specious” for failing to “provide any rationale as to
why the substantial impact standard was incorrect.” Dep’t of
Educ., 71 F.L.R.A. at 970 n.24. The FLRA further suggests
that if a rigorous statutory analysis had been done in 1985, the
agency would have discovered that “the de minimis standard is
inconsistent with the purposes of the Statute.” Id. at 971. Thus,
the FLRA seeks to cast its policy change as merely correcting
thirty-five-year-old procedural and interpretive errors.
That rationale cannot withstand scrutiny. It is true that
before it adopted the de minimis exception in 1985 the FLRA
had applied a substantial-impact standard like the one at issue
here, apparently as a carryover from the executive-order
regime that governed public-sector labor relations before the
enactment of the Labor-Management Relations Statute in 1978.
But the mere fact that the FLRA briefly used a substantial-
impact standard soon after the agency’s creation does not
provide inherent support for the present decision to discard
thirty-five years of intervening precedent and return to that past
policy. Indeed, Executive Order 11491, which governed
before the Labor-Management Relations Statute, did not
require bargaining at all. Instead, it directed agencies to
“meet . . . and confer in good faith with respect to personnel
policies and practices and matters affecting working
conditions, so far as may be appropriate,” Exec. Order No.
11,491 § 11(a), 3 C.F.R. (1966-1970), which is in marked
contrast to the Labor-Management Relations Statute’s
requirement that agencies bargain over “any condition of
employment,” 5 U.S.C. § 7114(b)(2); see also Nat’l Fed’n of
Fed. Emps. v. FLRA, 369 F.3d 548, 554 (D.C. Cir. 2004)
(“Congress passed the Federal Labor Relations Act to
encourage collective bargaining between federal employees
and their employers.”). And the FLRA has itself recognized
that the Labor-Management Relations Statute is more
 11
protective of collective bargaining than was the Executive
Order and practice thereunder. See Dep’t of Educ., 71 F.L.R.A.
at 974 & nn. 42–44 (DuBester, dissenting).
Furthermore, the FLRA’s characterization of its earlier
decisions as lacking explanation for its adoption of the de
minimis exception is misleading. In fact, as the dissenting
Commissioner points out, id. at 974–75 (DuBester, dissenting),
those earlier decisions expressly considered the question, and
in light of Congress’s clear purpose of expanding public-sector
bargaining rights, the mid-1980s FLRA specifically “rejected
the ‘substantial impact’ test” in favor of the de minimis
exception, Dep’t of Health & Human Servs. Soc. Sec. Admin.
Chi. Region, 15 F.L.R.A. at 924. What is more, the agency
explained back then that it was rejecting the substantial-impact
threshold both because the de minimis test struck the correct
balance between government efficiency interests and labor
rights, see Dep’t of Health & Human Servs. Soc. Sec. Admin.
Reg. V, 19 F.L.R.A. at 834 (McGinnis, concurring), and
because “[t]he limited scope of Federal sector bargaining
caused by external laws, rules, and regulations also demands
that the [FLRA] not impose further limitations unless they are
based on clear statutory authority and are buttressed by sound
policy considerations,” Dep’t of Health & Human Servs. Soc.
Sec. Admin., 24 F.L.R.A. at 406–07. And the FLRA has since
reiterated the view that the de minimis standard is “the
appropriate threshold” under the relevant statutory provisions.
Soc. Sec. Admin. Off. of Hearings & Appeals, 59 F.L.R.A. 646,
653 (2004), aff’d, AALJ, 397 F.3d at 964.
To be sure, the FLRA’s reasons for replacing the
substantial-impact standard with the de minimis exception
back in the mid-1980s were not tidily arranged in a single
decision. But the FLRA’s present assertion that the agency had
offered no “explanation or rationale to support the change”
from the substantial-impact test to the de minimis standard,
Dep’t of Educ., 71 F.L.R.A. at 970, is simply incorrect.
Notably, even if the FLRA had failed to provide an
adequate explanation for its adoption of the de minimis
 12
standard more than thirty-five years ago, any such failure
would not absolve the agency of its present-day responsibility
to explain its decision to jettison the precedents that apply the
de minimis threshold. The FLRA cannot point to a latent and
unchallenged purported defect in the original adoption of its
prior policy and offer that as an independent basis for adopting
a new one. Rather, the decision to adopt a new policy must be
sufficiently explained on its own terms, because “[i]n
administrative law, as elsewhere, two wrongs do not make a
right.” Am. Wild Horse Pres. Campaign v. Perdue, 873 F.3d
914, 928 (D.C. Cir. 2017). We have long held that “we cannot
condone the ‘correction’ of one error by the commitment of
another,” id. (quoting Gray v. Mississippi, 481 U.S. 648, 663
(1987)), and we decline to do so here.
2
Finally, to the extent that the FLRA now asserts that the de
minimis standard must go because it is categorically
“inconsistent with the purposes of the [Labor-Management
Relations] Statute,” Dep’t of Educ., 71 F.L.R.A. at 971, that
contention is plainly contrary to this court’s past interpretation
of the statute.
In AALJ, for instance, a union challenged the FLRA’s
application of the de minimis exception to a new category of
management decisions. The agency argued the exception was
inherent in the statute, and thus urged us to hold that collective
bargaining was not required for a de minimis change to
conditions of employment. We agreed, concluding that an
exception for de minimis effects on conditions of employment
is “inherent in” the statute and thus “neither contrary to the text
nor unreasonable in light of” the Labor-Management Relations
Statute’s purpose. AALJ, 397 F.3d at 959, 962.
Our holding in AALJ is a clear recognition of the
appropriateness of a de minimis exception to the duty to
bargain, as a matter of law. Thus, when the FLRA nevertheless
reached the conclusion that the de minimis test is not
“appropriate” for determining whether the duty to bargain is
triggered, Dep’t of Educ., 71 F.L.R.A. at 971, it not only
 13
ignored its own decisions outlining the reasons for its adoption
of the de minimis exception in the first place, but also “departed
from precedent” that had expressly decided that the de minimis
exception is consistent with the Labor-Management Relations
Statute, AFGE 2020, 961 F.3d at 459. For this reason, too, we
conclude the need for the policy change “is not sensibly
explained.” Id. (internal quotation marks omitted).
C
Up to this point, we have focused on the deficiencies in the
FLRA’s reasons for its conclusion that the de minimis standard
was so problematic that it needed to be changed. Beyond the
various ill-defined reasons that the FLRA offers for departing
from the de minimis threshold, the policy statement under
review also undertakes to explain why the FLRA now prefers
the substantial-impact standard. The FLRA points to two
principal reasons for this decision. And, once again, neither is
sufficiently explained.
1
The FLRA first heralds the substantial-impact standard’s
bona fides by suggesting that it draws “a line that is meaningful
and determinative” for agencies and their employees seeking
to navigate labor negotiations. Dep’t of Educ., 71 F.L.R.A. at
970. And in its brief the FLRA further contends that its finding
that the new standard will be more predictable and
administrable is “precisely the sort of predictive judgment” to
which this court must defer. FLRA Br. 63.
But neither the FLRA’s challenged decision nor its brief
on review analyzes the relative administrability of the
substantial-impact and de minimis standards. This is a critical
“gap in [the FLRA’s] reasoning[.]” AFGE 2020, 961 F.3d at
459. And the FLRA simply ignores the guidance and
elaboration provided in its own opinions. See, e.g., Dep’t of
Health & Human Servs. Soc. Sec. Admin. Region V, 19
F.L.R.A. at 834–35 (McGinnis, concurring) (identifying five
factors informing application of the de minimis standard). It is
especially striking that the FLRA does not provide a
comparative analysis of the two standards or in its policy
 14
statement even argue that the substantial-impact standard used
by the National Labor Relations Board (NLRB) in the private-
sector context has in fact led to more predictable results, since,
as noted above, the FLRA frames inconsistent application as
the principal problem the new standard is designed to fix.
There is no obvious reason to expect that labor unions and
employers will disagree less frequently about whether any
given management decision has a “substantial impact” on
conditions of employment than they previously did over
whether such a decision had a more than de minimis effect. But
see Dep’t of Educ., 71 F.L.R.A. at 971. Indeed, these two
standards share many characteristics that might lead one to
expect just as much disagreement—and, for that matter, just as
many “differences of opinion among arbitrators, judges, and
the [FLRA] as to what matters affect conditions of employment
sufficiently to require bargaining[,]” id. at 969 (internal
quotation marks omitted)—when the substantial-impact
threshold is applied. See also id. at 973–74 (DuBester,
dissenting) (“The majority fails to explain how adoption of its
new standard will produce decisions that are any less fact-
dependent than those applying the current standard.”). Thus,
the new standard that the FLRA has adopted is not the sort of
“common sense” measure for “advanc[ing the decision’s
stated] goals”—here, predictability and streamlined
administration—that we have found adequate to withstand
arbitrary and capricious review. Free Access & Broad.
Telemedia, LLC v. FCC, 865 F.3d 615, 618 (D.C. Cir. 2017).
To address this obvious shortcoming, the FLRA argues for
the first time in its brief that the new substantial-impact
standard will have more predictable results because the FLRA
will be able to draw on fifty years of NLRB decisions applying
the substantial-impact test. The FLRA does not explain in its
brief or otherwise why the NLRB’s substantial-impact
decisions would be easier to apply than the FLRA’s own thirty-
five years of precedents using the de minimis test. But,
regardless, this argument is forfeit because it was not raised
“where it counts”—i.e., in the challenged decision itself.
AFGE 2020, 961 F.3d at 457; see Motor Vehicle Mfrs. Ass’n of
 15
U.S., Inc. v. State Farm Mut. Auto. Ins. Co., 463 U.S. 29, 50
(1983) (“[C]ourts may not accept appellate counsel’s post hoc
rationalizations for agency action.”).
What remains is the FLRA’s bald assertion that its own
“determination that the substantial impact test would draw a
line that is [more] meaningful and determinative is precisely
the sort of predictive judgment to which this [c]ourt accords
heightened deference.” FLRA Br. 63 (internal quotation marks
omitted). We fully recognize that we must defer to expert
agencies’ “reasoned predictions about technical issues[.]”
BNSF Ry. Co. v. Surface Transp. Bd., 526 F.3d 770, 781 (D.C.
Cir. 2008). But there is nothing technical about the
predictability assessment that the FLRA makes here, and we
are not bound by the FLRA’s conclusory and counterintuitive
assertions about the consistency with which its new standard is
likely to be applied in subsequent adjudications, especially
when the record contains no factual basis for making such a
forecast.
With respect to its consistency concern, the FLRA must at
least explain why and how it has concluded that the substantial-
impact threshold is “better” than the standard it was
relinquishing. Fox Television, 556 U.S. at 515. It has failed to
do so.
2
The FLRA’s final tack is to invoke the NLRB’s
longstanding use of a substantial-impact standard in the
private-sector context, as an independent reason for the
FLRA’s adoption of the same collective bargaining threshold.
The FLRA’s policy statement asserts that, because “collective
bargaining in the public sector” under the Labor-Management
Relations Statute “must be narrower” than in the private sector,
“[i]t is incongruous . . . [to apply] a standard more lenient than
the test applied by the [NLRB] . . . to determine whether a
change requires bargaining.” Dep’t of Educ., 71 F.L.R.A. at
970. In its brief, the FLRA adds that it is not just permitted but
indeed required to consider NLRB precedents when it
administers parallel provisions of its statute; in this regard, it
 16
points to our instruction that “when the [FLRA] departs from a
familiar principle rooted in private sector precedent, it should
either identify ‘practical distinctions between private and
governmental needs’ that justify the departure, or offer some
evidence in the language, history, or structure of the statute
suggesting that Congress intended a different result.” AFGE v.
FLRA (AFGE 1988), 853 F.2d 986, 992 (D.C. Cir. 1988)
(citation omitted).
But NLRB precedent cannot save the FLRA’s unreasoned
and unreasonable determination to import the substantial-
impact standard. To start, the FLRA flips the controlling
question on its head. Unlike in AFGE 1988, we are not being
called upon to review a decision to part ways with NLRB
precedent in the first instance. Instead, we are evaluating the
FLRA’s recent departure from its own longstanding precedents
that since the 1980s have struck a balance that is different from
the NLRB’s decisions. In other words, the baseline for our
review is the FLRA’s longstanding and repeatedly reaffirmed
decision to diverge from NLRB policy, which makes AFGE
1988’s explanation of what the FLRA must do to rationalize a
new departure from NLRB precedent inapposite.
The FLRA’s decision to adopt the NLRB’s substantial-
impact test also fails to account for the agency’s own past
assessments of how the differences between the public-sector
and private-sector bargaining contexts inform the appropriate
bargaining threshold. We have previously cautioned that the
FLRA must “be careful to appreciate fully those distinctions
between the private and public sectors that might necessitate a
different legal analysis and conclusion” with respect to
collective bargaining, since “the bargaining status of any given
subject is determined by different statutory provisions and by
different policy considerations.” Library of Cong. v. FLRA,
699 F.2d 1280, 1287 (D.C. Cir. 1983).
The FLRA apparently took those distinctions into account
in its 1986 decision in Department of Health & Human
Services Social Security Administration, when it acknowledged
the NLRB’s substantial-impact standard was “similar[]” to the
 17
FLRA’s de minimis exception, 24 F.L.R.A. at 406 n.1, but
nevertheless opted to apply the de minimis standard. Critically,
that decision highlighted “[t]he limited scope of Federal sector
bargaining caused by external laws, rules, and regulations[,]”
and observed that this context “demands that the [FLRA] not
impose further limitations unless they are based on clear
statutory authority and are buttressed by sound policy
considerations.” Id. at 406–07 (emphasis added); cf. Library
of Cong., 699 F.2d at 1287 & n.33 (explaining that “[t]he scope
of collective bargaining is far broader in the private sector,” in
terms of the categories of conditions subject to bargaining,
since the Labor-Management Relations Statute “exclu[des]
from the scope of federal sector bargaining [] matters provided
for by federal statute, such as the pay rate or hours of
employment”).
Thus, the FLRA has held in the past that the relative
substantive narrowness of the public-sector bargaining
mandated under the Labor-Management Relations Statute in
fact supports the de minimis standard notwithstanding the
NLRB’s more stringent bargaining threshold. See Dep’t of
Health & Human Servs. Soc. Sec. Admin, 24 F.L.R.A. at 407.
That conclusion is precisely the opposite of the one that the
FLRA reached here. The agency now ignores its earlier
balancing of the factors unique to public-sector bargaining and
fails to address the reasons that it previously found persuasive
when it decided to select a test that differs from the one that
pertains to private-sector bargaining.
We conclude that, whatever the virtues of the FLRA’s
present analysis concerning the statutory scheme for public-
sector bargaining relative to its private-sector counterpart
(which we do not here decide), the FLRA must acknowledge
that in the past it reached the opposite conclusion about the
need for congruity between the public- and private-sector
bargaining thresholds. Thus, “[i]t is not enough” for the agency
to now rest on abstract invocations of the “narrowness” of
public-sector bargaining or “rely vaguely on [it]s general duty
to interpret the statute with government efficiency in mind.”
AFGE 1988, 853 F.2d at 993. Rather, the FLRA’s adoption of
 18
the substantial-impact threshold after it previously and
specifically rejected that standard must be built on a more
“solid foundation,” id., including an explanation of the
agency’s view of why the new approach better comports with
all of the provisions that Congress enacted to govern collective
bargaining in the public sector. See, e.g., 5 U.S.C.
§ 7101(a)(1)(B) (finding that collective bargaining
“contributes to the effective conduct of public business”).
The FLRA’s failure to address its previous balancing of
the Labor-Management Relations Statute’s priorities—a
balancing that led it to adopt a different policy than that of the
NLRB—is yet another indication that the FLRA has not, in
fact, “engage[d] in reasoned decisionmaking.” Fred Meyer,
865 F.3d at 638; see 5 U.S.C. § 706(2)(A).
CONCLUSION
The cursory policy statement that the FLRA issued to
justify its choice to abandon thirty-five years of precedent
promoting and applying the de minimis standard and to adopt
the previously rejected substantial-impact test is arbitrary and
capricious for the reasons explained above. Consequently, the
unions’ petitions for review are granted and the FLRA’s
September 30, 2020 general statement of policy is vacated.
So ordered.