United States Court of Appeals
          FOR THE DISTRICT OF COLUMBIA CIRCUIT



Argued September 18, 2020             Decided October 9, 2020

                         No. 19-5284

    THE AMERICAN COUNCIL OF THE BLIND AND PATRICK
                     SHEEHAN,
                    APPELLANTS

                       OTIS STEPHENS,
                         APPELLEE

                               v.

    STEVEN T. MNUCHIN, SECRETARY OF THE TREASURY,
                      APPELLEE


         Appeal from the United States District Court
                 for the District of Columbia
                     (No. 1:02-cv-00864)


    Jeffrey A. Lovitky argued the cause and filed the briefs for
appellants.

    Daniel Winik, Attorney, U.S. Department of Justice,
argued the cause for appellee. With him on the brief were
Ethan P. Davis, Acting Assistant Attorney General, and
Charles W. Scarborough, Attorney.

    Before: HENDERSON and WALKER, Circuit Judges, and
SILBERMAN , Senior Circuit Judge.
                                2
    Opinion for the Court filed by Circuit Judge HENDERSON.

     KAREN LE CRAFT HENDERSON, Circuit Judge: Under the
terms of a 2008 injunction, the Secretary of the United States
Department of the Treasury (Treasury) must make the various
Federal Reserve Notes distinguishable to the visually impaired
no later than the next scheduled redesign of each denomination.
These redesigns have been significantly delayed. In 2016, the
American Council of the Blind (Council) moved to impose a
firm deadline on the Secretary. The district court denied the
motion, we vacated its order and on remand the district court
denied a similar motion. The Council challenges this most
recent denial on two grounds. First, it argues that the district
court relied on Treasury’s financial burden without obtaining
“concrete estimates” of that burden, thereby violating our
mandate in American Council of the Blind v. Mnuchin, 

(D.C. Cir. 2017) (ACB II). Second, the Council
contends that the district court abused its discretion because its
decision lacked evidentiary support. We disagree on both
points and affirm the district court.

                      I.   BACKGROUND

    Federal Reserve Notes—that is, U.S. paper currency—are
identical in size and texture and nearly identical in color. These
characteristics make using paper currency difficult for
individuals who are blind or nearly blind. In 2002, the Council
sued the Treasury Secretary, alleging that the design of U.S.
paper currency violates the Rehabilitation Act, 29 U.S.C.
§ 794, by denying a reasonable accommodation to the visually
impaired. So began a nearly two decades-long litigation
odyssey.

    In 2006, the Council won partial summary judgment, see
Am. Council of the Blind v. Paulson, 

(D.D.C. 2006), which we affirmed, see 

(D.C.
                               3
Cir. 2008) (ACB I). On remand, the district court disclaimed
both the “expertise” and the “power” to “choose among the
feasible alternatives, approve any specific design change, or
otherwise to dictate to the Secretary of the Treasury how he can
come into compliance with the law.” 

, 2
(D.D.C. 2008) (quoting 

). Thus,
the district court’s injunction—currently in force—required the
Secretary to take:

       such steps as may be required to provide
       meaningful access to United States currency for
       blind and other visually impaired persons,
       which steps shall be completed, in connection
       with each denomination of currency, not later
       than the date when a redesign of that
       denomination is next approved by the Secretary
       of the Treasury.

     In response, Treasury developed a three-pronged plan for
providing meaningful access to currency. See Meaningful
Access to United States Currency for Blind and Visually
Impaired Persons, 75 Fed. Reg. 28,331 (proposed May 20,
2010). First, the Bureau of Engraving and Printing (Bureau)
would produce new banknotes with a raised tactile feature
(RTF). Second, the Bureau would continue its project, begun
in 1997, of incorporating large, high-contrast numerals on new
currency. Third, the Bureau would distribute handheld
currency readers to the visually impaired.

    Although the Bureau has made some progress on the
second and third approaches, it has had little success
developing an RTF. The Secretary set four goals for selecting
an RTF: accuracy (the RTF must effectively allow the visually
impaired    to     distinguish    among      denominations),
manufacturability (it must be able to be produced in massive
                                4
quantities), durability (it must not degrade prematurely) and
processability (it must be capable of being processed by
currency counting machines, vending machines, ATMs, etc.).
The Bureau contends that balancing these factors is daunting;
for example, larger, thicker RTFs may increase accuracy and
durability    while     sacrificing   manufacturability   and
processability.

     In addition to the RTF’s technical challenges, advances in
counterfeiting have also pushed back the expected timeline for
the accessibility redesign.1 Under the 2008 injunction, the
accessibility redesign is coupled with the security redesign. In
2012, the Bureau director attested that the expected
accessibility timeline would not be met because of unexpected
delays redesigning the $100 note to incorporate new
anticounterfeiting technologies. Later, in a 2016 status report,
Treasury said that it had “recently learned of significant
developments in counterfeiting technology that bear upon the
long-term effectiveness of the security features” it had planned
on integrating into new notes. This setback necessitated the
design and development of new security features.

     The 2008 injunction was premised on the expectation that
each denomination would be redesigned every seven to ten
years. Treasury’s most recent timeline contemplates that the
required accessibility redesigns may not be completed until the
2030s—some three decades after the district court first found
that Treasury was in violation of federal law. Understandably

    1
       The original timeline for the $5 note redesign was 2015–2018
but—as of 2017—the new timeline is 2028. Other notes are facing
similar or greater delays. For the $10 note, 2013–2016 has become
2026. For the $20 note, 2010–2013 has become 2030. For the $50
note, 2011– 2014 has become 2032–2035. And for the $100 note,
which was not included in the original injunction, the expected
timeline is 2034–2038.
                               5
frustrated with the delays, the Council moved under Rule 60(b)
to modify the injunction to a “deadline of December 31, 2020
by which date the Secretary must provide meaningful access to
the $10 bill, which is the next bill scheduled to be redesigned.”
The Council further requested that the modified injunction
require the Secretary “to make the remaining denominations of
currency accessible to the blind and visually impaired not later
than December 31, 2026.” In sum, whereas the existing
injunction couples the security and accessibility redesigns—
relying on Treasury’s emphasis on the former to also produce
the latter—the Council’s proposal decouples the redesigns and
forces the accessibility redesign by a specific date.

     The district court denied the Rule 60(b) motion,
concluding that decoupling the redesigns “could create
unnecessarily duplicative work and potentially increase costs
for both the government and the private sector.” Am. Council
of the Blind v. Lew, No. 02-CV-00864, 

, at
*2 (D.D.C. Jan. 6, 2017). In ACB II, we reversed. Because the
district court relied on the financial burden imposed by the
requested relief without obtaining a concrete estimate of that
burden, we concluded that it had abused its discretion. 

. Our holding was straightforward:

       If the district court is to properly conclude that
       withholding meaningful access to paper
       currency from millions of visually impaired
       individuals for eight to twenty years longer than
       expected . . . remains equitable because of the
       potential financial burden resulting from
       granting the plaintiffs’ modification, the district
       court needs more concrete estimates of the costs
       that matter.

     On remand, the Council again moved to modify the
injunction to require Treasury to provide meaningful access to
the $10 note by the end of 2020 and to the other denominations
by the end of 2026. Treasury opposed the motion. In addition,
the Bureau expressed doubt about the efficacy of the three-
pronged approach it had begun to develop in 2010. For
example, a MITRE Corporation study showed significant
percentages of experienced Braille readers were unable to
correctly identify currency with an RTF. And because Braille
readers were over-represented in the study compared to the
overall population of visually impaired individuals, there was,
in the Bureau’s opinion, reason to believe that real-world
results would be even worse. This was the factual background
for the Bureau’s December 2018 status report expressing
“concerns regarding its ability to create a tactile feature for U.S.
currency with an accuracy rate that will make it suitable for
reliable use in commerce.”

     In any event, the district court again denied the Council’s
motion but for ostensibly different reasons from those we
rejected in ACB II. See Am. Council of the Blind v. Mnuchin,

(D.D.C. 2019). First, it found that
decoupling the accessibility and security redesign “would
divert Treasury resources and attention from pressing anti-
counterfeiting measures.”

Second, the court opined
that the Council’s proposed modification would require
Treasury to begin immediately incorporating RTFs into U.S.
currency, notwithstanding no suitable RTF is currently
available.

The Council now appeals that
decision.

                         II. ANALYSIS

    We review the district court’s denial of a Rule 60(b)
motion for abuse of discretion. See Pigford v. Johanns, 416
                                 

, 16 (D.C. Cir. 2005). Under this standard, we reverse
“if the district court applied the wrong legal standard or relied
on clearly erroneous findings of fact.” In re Vitamins Antitrust
Class Actions, 

, 1209 (D.C. Cir. 2003). The
district court also abuses its discretion if it violates the mandate
of the court of appeals. See United States v. Kpodi, 

, 491 (D.C. Cir. 2018).

     In denying the Council’s motion to modify the injunction,
the district court was persuaded by two considerations: security
and feasibility.      Although those considerations can be
indirectly affected by additional resources, they are
distinguishable from the direct financial burden the district
court considered in ACB II. The question, then, is whether they
should be distinguished or whether our holding should be read
more broadly. Both the language we used in ACB II and the
purpose of the mandate rule counsel against concluding that the
district court violated our mandate. Of course, we do not affirm
merely because the court below did not violate the law of the
case; its rationale must also be supported by record evidence.
We conclude that it is.

                                     A.

     An “inferior court has no power or authority to deviate
from the mandate issued by an appellate court.” Briggs v. Pa.
R.R. Co., 

, 306 (1948). The so-called “mandate
rule” is a “more powerful version of the law-of-the-case
doctrine, which prevents courts from reconsidering issues that
have already been decided in the same case.” Indep. Petroleum
Ass’n of Am. v. Babbitt, 

, 597 (D.C. Cir. 2001)
(internal quotation marks omitted). In other words, the district
court cannot “fashion[] a remedy that is ‘inconsistent with
either the spirit or express terms’” of our holding. Ass’n of Am.
R.Rs. v. Dep’t of Transp., 

, 554 (D.C. Cir. 2018)
                                 8
(Tatel, J., dissenting) (quoting Quern v. Jordan, 

,
347 n.18 (1979)). The obligation to follow past mandates is
especially compelling in long-running litigation. See Morley v.
C.I.A., 

, 401 (D.C. Cir. 2018) (per curiam)
(Henderson, J., dissenting).

    The district court violated neither the letter nor spirit of our
mandate in ACB II. Its first rationale in denying the Council’s
motion is that the Council’s proposed modification would
“divert Treasury resources and attention” from its anti-
counterfeiting responsibilities, thereby “posing risks to the
global security of U.S. currency with concomitant serious
ramifications for consumers, businesses, and the health of the
global 

–90. The Council
argues that this concern necessarily implicated the financial
burden of its proposal, which means the district court should
have required concrete estimates of the costs involved in
adopting its proposal.

     There is a certain logic to the Council’s argument. Almost
any concern about financial burden could be reframed in terms
of a resource constraint. “I can’t afford the gas for the trip,”
for example, can be reframed as “I don’t have enough gas to
make the trip.” Unless it is physically impossible to refill the
gas tank, the two statements are different ways of phrasing the
same thought. It would violate the spirit of ACB II’s mandate
if the district court, possessing no new cost estimates, merely
rephrased its earlier decision but left its logic unaltered.

    Whether this accurately describes the district court’s
decision depends on whether we adopt a narrow or broad
reading of ACB II. On the narrow reading, ACB II’s holding
prevented the district court from basing its decision on direct
financial costs without having concrete estimates of the direct
costs involved. Adopting this view, the Secretary points out
                                9
that the district court did not say “conducting separate security
and accessibility redesigns would be costlier than conducting a
single redesign” but instead that “the Bureau can do only so
many things at once.” On the broad understanding, however,
our holding extended to any rationale open to cost analysis.

     Looking at both the language we used in ACB II and the
underlying logic behind the mandate rule, we adopt the narrow
understanding.       The phrase “financial burden” means
something specific to most speakers and audiences. Although
any obstacle short of pure impossibility can be framed as a
financial obstacle, a financial burden usually refers to costs
within a fixed budgetary framework rather than quality or
timeline obstacles that might theoretically be solved with an
unlimited budget. The district court found that a short-term
prioritization of the accessibility redesign by itself could impair
the Secretary’s ability to execute a timely security redesign.

–91. Because the Congress, not the
Secretary, determines the resources at the Secretary’s disposal,
this point is logically different from the justification we
rejected in ACB II. The district court’s security rationale is a
management consideration, not a budgetary one.

     The distinction between management, i.e., policy, and
budgetary considerations may be somewhat artificial but it is
recognized in other areas of the law. In Federal Tort Claims
Act cases, for example, the “discretionary function exception”
to liability is implicated if the government’s action is
“susceptible to policy analysis.” United States v. Gaubert, 

, 325 (1991). In a “reverse” context, sister circuits
have held that a financial consideration is not necessarily a
policy consideration. As the Ninth Circuit noted, “[b]udgetary
constraints underlie virtually all governmental activity” so the
fact that government employees are required to “work within a
budget” does not turn every government action into a policy
                                10
decision. ARA Leisure Servs. v. United States, 

,
195–96 (9th Cir. 1987). Similarly, the Eleventh Circuit
recognized that “financial considerations alone may not make
a decision one involving policy.” Hughes v. United States, 

, 769 (11th Cir. 1997). Here, we make the corollary
point: a policy consideration is not synonymous with a
financial one, even if there is substantial overlap.

     The purpose of the mandate rule also supports a narrow
reading.     The mandate rule is a doctrine of judicial
administration; its goal is to “achieve finality,” making it
possible for appellate courts to do their job. Greater Bos.
Television Corp. v. F.C.C., 

, 279 (D.C. Cir. 1971).
The law of the case doctrine—of which the mandate rule is a
species—“does not seek to sweep under its coverage all
possible issues arising out of the facts of the case.” U.S. ex rel.
Dep’t of Labor v. Ins. Co. of N. Am., 

, 1041 (D.C.
Cir. 1997). Although there may be good reason to compel the
district court to quantify its security rationale in dollar-
denominated terms, ACB II does not require it.

      The district court’s feasibility rationale also comports with
ACB II’s mandate. The court was persuaded that the Council’s
proposed modification would require circulating an RTF even
though no RTF has met the Bureau’s circulation benchmarks.

–98. The Council replies that this
all translates into financial burden (meaning the district court
needed a concrete estimate of that burden). Its logic works
something like this: because the effectiveness of an RTF is
“largely a question of size,” the Bureau could comply with the
proposed deadline simply by making the RTF larger. By
effectiveness, we understand the Council to mean accuracy,
which is only one of the Bureau’s four benchmarks. In the
Council’s view, however, the other benchmarks—
manufacturability, durability and processability—are solely
                               11
economic concerns. Lack of durability could be resolved
through more frequent currency replacement and lack of
processability could be resolved by retrofitting existing cash-
processing machines. But, as we noted earlier, ACB II should
not be read to encompass any policy tradeoff that could be
theoretically improved with infinite resources. The Secretary’s
decision to pursue an RTF that meets all four of its benchmarks
is nothing if not a policy decision. We therefore reject the
Council’s suggestion that the district court’s feasibility
rationale can be boiled down to financial burden.

    That the district court met ACB II’s mandate does not
decide whether the district court abused its discretion for some
other reason—a question we take up next.

                                 B.

     Rule 60(b) provides that, upon motion, the court may
relieve a party of an injunction if “applying it prospectively is
no longer equitable.” Fed. R. Civ. P. 60(b)(5). The district
court should grant a Rule 60(b) motion if “a significant change
either in factual conditions or in law renders continued
enforcement detrimental to the public interest.” Horne v.
Flores, 

, 447 (2009) (internal quotation marks
omitted). The Secretary does not dispute that circumstances
have changed. In ACB II, after all, we recognized that a “much
greater than planned delay in providing meaningful access to
visually impaired individuals is unquestionably a change in
factual 

–67. The issue is whether
this change renders enforcement of the 2008 injunction
“detrimental to the public interest.” The Council has the
burden of proving that it does. See 

.

     The district court’s rationales for denying the Council’s
motion are sufficiently supported by the record. In setting out
its security rationale, the district court cited the Secretary’s
                                12
estimate that adding the RTF to the $10 note by the end of 2020
would likely push back the security redesign of each
denomination by at least two years—possibly more. 

n.42. The Council’s only direct rebuttal is
that “neither the Secretary nor the district court explains the
counter-intuitive conclusion that imposing an obligation to
complete a task by the end of 2020 would somehow necessitate
a delay to an event that is not scheduled to be completed” until
2026.

     But the Secretary did explain his conclusion. In response
to an interrogatory from the district court, the Secretary noted
that “[t]he incorporation of an RTF would necessarily affect the
overall design of the bills, and thus would affect the design of
each denomination, including potentially the effectiveness of
security features.” The complexity of this interaction means
that “[t]here is no way to predict the results of this testing, and,
if notes containing any feature failed the testing, there is no way
to predict how much additional development and testing would
be required.” In other words, each new feature—accessibility-
or security-related—requires extensive testing and may
interact with existing features in unpredictable ways; this
requires an irreducible amount of testing, no matter how distant
the deadline.

     The Secretary’s conclusion appears to be the result of a
hard-learned lesson. The most recently released note, the $100
note, was delayed for over three years after the Bureau
determined that the new security ribbons were producing
creases that made the notes unusable. Worse, this defect was
only discovered after large-scale production was underway. A
later review concluded that the Bureau had engaged in
“insufficient preproduction testing.” The district court cited
the Secretary’s statement in support of its conclusion that the
harm of a later than expected accessibility redesign does not
                               13
justify the risks of “delaying an already postponed security
redesign even 

;

not an abuse of discretion for the district court to credit the
Secretary’s estimate and weigh the tradeoffs as it did.

     Granted, we do not know much about how the Secretary
derived his estimate of a two-year delay in the security
redesign. The estimate is “concrete” but it may fail to fully live
up to the “show your work” spirit that animated ACB II.
Nonetheless, the Secretary’s estimate is a predictive judgment
that “involves deductions based on the expert knowledge of the
agency” and therefore does not—if, indeed, it can—require
“complete factual support in the record.” Rural Cellular Ass’n
v. F.C.C., 

, 1105 (D.C. Cir. 2009) (quoting
Melcher v. F.C.C., 

, 1151 (D.C. Cir. 1998)).
Additionally, the estimate does not stand to benefit from
quantitative support in the same way as a cost-based rationale.
Budgetary calculations, after all, are uniquely susceptible to
line-by-line disaggregation. And our caselaw rejects the notion
that the only acceptable agency analyses are quantitative. See,
e.g., Am. Farm Bureau Fed’n v. E.P.A., 

, 535
(D.C. Cir. 2009) (“[T]he fact that the EPA’s analysis is
qualitative rather than quantitative does not undermine its
validity.”); Ctr. for Sustainable Econ. v. Jewell, 

,
611 (D.C. Cir. 2015) (“Our decisions afford greater leeway to
Interior to evaluate qualitatively costs that are difficult to
quantify.”).

     The district court’s feasibility rationale is also well-
supported by the record. The Council does not contest that no
RTF feature is currently available. Instead, its primary rebuttal
is that today’s RTF difficulties do not suggest that one cannot
be ready for circulation by December 31, 2026. In other words,
December 2026 is a long time away so how can the court be
certain that the Council’s proposed deadline is infeasible? But
                               14
this flips the burden: it is the responsibility of the Council, as
the moving party, to justify the modification. And in any event,
it was not an abuse of discretion for the district court not to
compel the Secretary to take an action that might be possible.

     The Council’s attack on the feasibility rationale is wrong
for an additional reason—one that strikes at the heart of our
institutional role. When the Council insists that the Secretary
could approve a short-term solution, such as notches or paper
roughening, neither of which has been approved by the
Bureau’s experts, or when it suggests functionally eliminating
three of the Bureau’s four requirements for a successful RTF,
it seeks not merely to compel the Secretary to comply with the
law but to exercise control over the manner in which he must
do so.

     Like the district court in 2008, we have serious doubts
about the authority of a court to compel the Secretary to
implement any specific design feature. See Paulson, 581 F.
Supp. 2d at 2. The Congress charged the Secretary with
“furnish[ing] suitable notes for circulation . . . in the best
manner to guard against counterfeits and fraudulent
alterations.” 12 U.S.C. § 418. The Bureau carries out this
responsibility under the direction of the Secretary, who has the
authority to direct the “form and tenor” of each note.

law gives us the authority to require that
U.S. currency take a particular form. “Courts have no license
to usurp from agencies the discretion that Congress has granted
them.” 33 Charles A. Wright, Charles H. Koch, Jr., and
Richard Murphy, Federal Practice and Procedure § 8384
(2018); cf. Norton v. S. Utah Wilderness All., 

, 63–
64 (2004) (holding, in APA context, federal court can compel
agency to take only discrete actions it is required to take). And
even if we had the authority, it is a task for which we lack
expertise.
                               15
     It would be odd if this court, while eschewing the intention
to require any specific design, imposed a timetable that
necessarily compelled a particular design or forced the Bureau
to abandon plans for its preferred design. Such an action would
violate “the principle that courts should not impose remedies
that interfere with an agency’s proper discretion.” Wright,
Koch & Murphy, supra, § 8384. Thus, we are not swayed by
the Council’s suggestion that “other viable alternatives” to RTF
“have never been considered by the Secretary.” Instead, we
think the district court was wise to abide by the principle that
an “agency’s own timetable for performing its duties in the
absence of a statutory deadline is due ‘considerable
deference.’” Cobell v. Norton, 

, 1096 (D.C. Cir.
2001) (quoting Sierra Club v. Gorsuch, 

, 658
(D.C. Cir. 1983)); see also Mexichem Specialty Resins, Inc. v.
E.P.A., 

, 555 (D.C. Cir. 2015) (same); Nat. Res.
Def. Council, Inc. v. Train, 

, 712 (D.C. Cir. 1974)
(“The courts cannot responsibly mandate flat guideline
deadlines when the Administrator demonstrates that additional
time is necessary.”).

     We sympathize with the Council’s frustration on behalf of
those it has so faithfully represented. As the Council said in its
initial complaint over 18 years ago, “individuals with visual
disabilities suffer needless impediments in purchasing
groceries, transportation, and a multitude of other goods and
services” and are at “heightened risk of fraud and deceit.”
Because Treasury has yet to comply with federal law, “millions
of visually impaired individuals” remain largely “dependent on
the kindness of others” to handle those everyday transactions
that individuals with sight take for granted. ACB 

. Nevertheless, the same equitable power that permitted
the district court to craft the 2008 injunction gave it
considerable discretion in determining whether to modify it.
                              16
The district court did not exceed that discretion by denying the
Council’s Rule 60(b) motion.

     For the foregoing reasons, the district court’s judgment is
affirmed.

    So ordered.