FOR PUBLICATION FILED
UNITED STATES COURT OF APPEALS SEP 30 2020
MOLLY C. DWYER, CLERK
U.S. COURT OF APPEALS
FOR THE NINTH CIRCUIT
NATIONAL URBAN LEAGUE; LEAGUE No. 20-16868
OF WOMEN VOTERS; BLACK
ALLIANCE FOR JUST IMMIGRATION; D.C. No. 5:20-cv-05799-LHK
HARRIS COUNTY, Texas; KING Northern District of California,
COUNTY, Washington; CITY OF LOS San Jose
ANGELES, California; CITY OF
SALINAS, California; CITY OF SAN ORDER
JOSE, California; RODNEY ELLIS;
ADRIAN GARCIA; NAVAJO NATION;
NATIONAL ASSOCIATION FOR THE
ADVANCEMENT OF COLORED
PEOPLE; CITY OF CHICAGO, Illinois;
COUNTY OF LOS ANGELES, California;
GILA RIVER INDIAN COMMUNITY,
Plaintiffs-Appellees,
v.
WILBUR L. ROSS, in his official capacity
as Secretary of Commerce; UNITED
STATES DEPARTMENT OF
COMMERCE; STEVEN DILLINGHAM, in
his official capacity as Director of the U.S.
Census Bureau; UNITED STATES
CENSUS BUREAU,
Defendants-Appellants,
and
STATE OF LOUISIANA; STATE OF
MISSISSIPPI,
Intervenor-Defendants.
Before: RAWLINSON, CHRISTEN, and BUMATAY, Circuit Judges.
Order by Judges RAWLINSON and CHRISTEN, Dissent by Judge BUMATAY
On August 3, 2020, the United States Census Bureau (Bureau) adopted a
census plan (Replan) that dramatically advanced critical deadlines for conducting
the 2020 census. Appellees challenged this action pursuant to the Enumeration
Clause of the United States Constitution and the Administrative Procedure Act
(APA). On September 24, 2020, the district court entered a preliminary injunction
staying the Replan’s schedule for completion of census field operations and for
reporting the census results to the President and enjoining the government from
implementing these deadlines. The government has filed an emergency motion to
stay the preliminary injunction pending appeal, and a request for an immediate
administrative stay pending resolution of the stay motion. In this order, we
consider only the request for an administrative stay.
The decennial census is an enormous and complex nationwide operation. It
requires nearly a decade of planning and hundreds of thousands of dedicated
workers to accomplish. In 2018, after years of planning and testing, the Bureau
adopted a plan to complete the 2020 census. The plan called for an extraordinary
2
effort on the part of the government including hiring 340,000–500,000 field staff.
For reasons stated in the record, the district court found that due to significant
challenges encountered in the wake of COVID-19, the Bureau suspended field
operations in March 2020. When operations resumed, the Bureau was unable to
recruit sufficient numbers of field staff. In July 2020, the Bureau estimated that it
only retained 38% of the field staff required to complete an accurate and timely
census.
As a result of these serious challenges, the district court found that as early
as April 2020, the Bureau, the Department of Commerce, and even the President
had all publicly acknowledged that the December 31 deadline was no longer
attainable. The Bureau adopted a new census plan in April to accommodate the
delays caused by COVID-19 (“COVID-19 Plan”). The COVID-19 plan extended
the deadline for each step in the process and contemplated that the Bureau would
ask Congress for a 120-day extension of the December 31, 2020 delivery deadline
for the completed census report. The Bureau’s work proceeded according to the
COVID-19 Plan until August 2020.
In early August, a “senior Department [of Commerce] official” directed the
Bureau to change course and prepare a new plan for completing the census by the
December 31, 2020 statutory deadline. Senior Bureau staff were given just four to
five days to develop this “Replan.” On August 3, 2020, the Bureau announced its
3
adoption of the Replan, and its central feature: accelerating the COVID-19 Plan’s
deadline for the completion of field work and data collection from October 31 to
September 30. On September 24, the district court entered a preliminary
injunction preventing the Bureau from implementing the September 30 deadline to
stop field work and data collection. The government requests an immediate
administrative stay of the district court’s injunction.
I
The government has filed a single emergency motion seeking a stay pending
appeal, and also seeking an administrative stay pending resolution of the motion
for stay pending appeal. We recently established that an administrative stay “is
only intended to preserve the status quo until the substantive motion for a stay
pending appeal can be considered on the merits, and does not constitute in any way
a decision as to the merits of the motion for stay pending appeal.” Doe v. Trump,
944 F.3d 1222, 1223 (9th Cir. 2019). Based on our preliminary review of the
record, we conclude that the status quo would be seriously disrupted by an
immediate stay of the district court’s order.
As explained above, until August of this year, the Bureau had been operating
for several months under the COVID-19 plan. That plan represented a revised
schedule to account for the challenges caused by the COVID-19 pandemic. It
included extended deadlines based on the understanding that the Bureau would
4
need additional time to complete the necessary field work and data processing to
produce an accurate census report. The district court’s September 5 temporary
restraining order and September 24 preliminary injunction preserve the status quo
because they maintain the Bureau’s data-collection apparatus pending resolution of
the appeal. By the time the district court entered its order, the Bureau had already
begun winding down its field operations and terminating census field workers in
anticipation of the Replan’s accelerated September 30 deadline. The process of
disbanding thousands of census workers will resume if an administrative stay is put
in place, eliminating the Bureau’s ability to conduct field work. Accordingly, on
the facts of this case, staying the preliminary injunction would upend the status
quo, not preserve it.
We are mindful of the potential harms faced by both parties. Here, not only
would the status quo be upended by an administrative stay, the Bureau’s ability to
resume field operations would be left in serious doubt. Thousands of census
workers currently performing field work will be terminated, and restarting these
field operations and data collection efforts, which took years of planning and hiring
efforts to put in place, would be difficult if not impossible to accomplish in a
timely and effective manner. Granting the administrative stay thus risks rendering
the plaintiff’s challenge to the Replan effectively moot.
We also recognize that missing the December 31 statutory deadline risks
5
serious harm to the government. However, the record does not demonstrate that
the Bureau’s ability to meet that deadline is affected by the district court’s
injunction. Rather, the evidence in the administrative record uniformly showed
that no matter when field operations end—whether September 30 under the Replan
or October 31 under the COVID-19 Plan—the Bureau will be unable to deliver an
accurate census by December 31, 2020. The President, senior Bureau officials,
senior Department of Commerce officials, the Office of Inspector General, the
Census Scientific Advisory Committee, and the Government Accountability Office
have all stated that delivering a census by December 31 without compromising
accuracy is practically impossible, and has been for some time. As the district
court recognized, after the Bureau realized the pandemic would prevent it from
adhering to its original schedule, the Bureau made two requests to Congress: first,
it requested the December 31 deadline be extended to April 2021. When no final
congressional action had been taken on that request in July, the Bureau requested
$443 million to cover the additional cost to complete the census by year’s end.
Contrary to the dissent’s repeated assertion, the only undisputed fact in this
sequence was that Congress has not given the Bureau the extension or the
additional funding it needs to meet the statutory deadline.
The government did not counter the Appellees’ showing on this point.
Citing the chorus of statements made by the Bureau and other officials, the district
6
court found that the Bureau could not meet the December 31 deadline. Indeed,
despite the government’s persistent argument in the district court and before our
court that the September 30 deadline for terminating field operations is essential to
meeting its December 31 statutory deadline, the administrative record
compellingly supports the district court’s conclusion that moving the October 31
deadline to September 30 will not allow the Bureau to complete the census on
time.
Finally, we note that notwithstanding the pendency of the government’s
emergency request for an immediate administrative stay to allow the Replan’s
September 30 deadline to take effect, on September 28 the government again
changed the deadline for completing field work. The government informed us in a
September 28, 2020 letter, without explanation, that it now intends to end field
operations on October 5, 2020. This abrupt change contradicts the government’s
argument that the September 30 date is vitally important to the Bureau’s ability to
meet its statutory reporting deadline. Our dissenting colleague cites a September
28 estimate suggesting that the census is 98% complete. This is still below the
enumeration rate required by the Bureau’s internal standards for generating an
accurate census report. Further, the district court ruled on September 24 and
found, as of that date, the Bureau had met its standard in only four states.
Given the extraordinary importance of the census, it is imperative that the
7
Bureau conduct the census in a manner that is most likely to produce a workable
report in which the public can have confidence. The Bureau must account for its
competing constitutional and statutory obligation to produce a fair and accurate
census report. The hasty and unexplained changes to the Bureau’s operations
contained in the Replan, created in just 4 to 5 days, risks undermining the Bureau’s
mission.
Our dissenting colleague makes four errors. First, the dissent applies the
wrong standard for a preliminary administrative stay. In Doe #1 v. Trump, our
circuit definitively resolved which standard applies to administrative stay motions.
We are not free to depart from that standard. Miller v. Gammie, 335 F.3d 889, 899
(9th Cir. 2003) (en banc) (holding that a three-judge panel may not overrule a prior
decision of the court). Citing the dissent from Doe #1 v. Trump, our colleague
applies the factors used when we consider a motion for stay pending appeal. This
analysis erroneously collapses the distinct legal analyses for an administrative stay
and a motion for stay pending appeal. When considering the request for an
administrative stay, our touchstone is the need to preserve the status quo. We defer
weighing the Nken 1 factors until the motion for stay pending appeal is considered.
See Doe #1, 944 F.3d at 1223.
Second, as a consequence of its threshold error, the dissent does not grapple
1
Nken v. Holder, 556 U.S. 418, 426 (2009).
8
with the factor that drives the outcome of the government’s motion: the Bureau’s
apparatus for conducting field work will be dismantled before the motion for stay
pending appeal can be decided. The dissent does not dispute that issuing an
administrative stay in this case would return the Bureau to the process of
dismantling its data-collection infrastructure and terminating its field staff.
Third, although we need not wade into the underlying merits of the issues on
appeal, we would be remiss if we did not note that the dissent hinges on the
unsupported premise that the Bureau can meet the December 31 deadline if an
administrative stay is issued. The dissent’s assumption that the agency can still
meet its deadline relies entirely upon one conclusory statement that was not in the
administrative record but was instead prepared for litigation. Dep’t of Homeland
Sec. v. Regents of the Univ. of California, 140 S. Ct. 1891, 1908 (2020) (explaining
that an agency’s post hoc rationalizations “must be viewed critically”); Arrington
v. Daniels, 516 F.3d 1106, 1113 (9th Cir. 2008) (rejecting a justification for agency
action that “is entirely absent from the administrative record”). Given the
consistent picture painted by the administrative record, it is not surprising the
district court was unpersuaded by this sole conclusory statement.
Fourth, the dissent addresses several issues that are not properly before us at
the administrative stay stage. The government’s emergency motion does not
contest the district court’s conclusion that Appellees have standing to bring their
9
claims. Nor does the emergency motion challenge the district court’s conclusion
that the Bureau’s decision to adopt the Replan is an unreviewable political
question. Thus, those issues are not properly before us and we do not reach them.
Because the status quo would be upended, rather than preserved, if an
administrative stay is issued, the government’s request for an immediate
administrative stay set forth in Docket Entry No. 4 is denied.
Appellees’ response to the emergency motion is due October 2, 2020.
Appellants’ optional reply is due by October 3, 2020.
10
FILED
National Urban League v. Ross, No. 20-16868 SEP 30 2020
Bumatay, J., dissenting
MOLLY C. DWYER, CLERK
U.S. COURT OF APPEALS
COVID-19 has wreaked an undeniable toll on the Nation. The virus has
already stolen too many American lives. Even more have been hospitalized or
fallen ill. And nearly every American’s plans this year have been roiled by the
virus. But it cannot roil the law. Contorting the Administrative Procedure Act,
and liberating itself from any semblance of judicial restraint, the district court
injected itself into a sensitive and politically fraught arena: the 2020 census. After
the Department of Commerce adopted a plan to address census delays from the
COVID-19 pandemic, plaintiffs brought suit under the APA. Upon reviewing the
internal deliberative emails of the agencies, the district court decided that it knows
better than the Secretary of Commerce. Based on internal discussions about the
agency’s ability to complete the census in a timely and accurate fashion, the
district court essentially overruled the Secretary’s decision to adopt the revised
plan. But it is undisputed that this new plan was the only way to meet the statutory
obligation to report the census results to the President by December 31, 2020. No
matter for an adventurous district court: it simply cast aside the statutory deadline
as part of its injunction.
Because the district court was without authority to issue its injunction, the
defendants are likely to succeed on the merits, and they will be irreparably harmed
1
without relief, I would have granted the request for an administrative stay.
Accordingly, I respectfully dissent.
I.
A census is required by our Constitution, which provides that the “actual
Enumeration” of the population shall be conducted “in such Manner as [Congress]
shall by Law direct.” U.S. Const. Art. I, § 2, cl. 3. As should be evident from this
text, besides requiring that such an enumeration shall occur, the Constitution
otherwise vests “virtually unlimited discretion” with Congress. Wisconsin v. City
of New York, 517 U.S. 1, 19 (1996); see also Baldrige v. Shapiro, 455 U.S. 345,
361 (1982) (recognizing Congress’s broad discretion over the census). Congress,
in turn, has vested substantial discretion with the Secretary of Commerce to
determine how to conduct the decennial census. See 13 U.S.C. § 141(a);
Wisconsin, 517 U.S. at 19 (“Through the Census Act, Congress has delegated its
broad authority over the census to the Secretary.”). But there’s one aspect that
Congress did not delegate: the date for completion of apportionment counts. 13
U.S.C. § 141(b). That deadline is etched in stone: December 31, 2020. 1 And
1
Congress has provided for other deadlines as well. For example, the Census
Bureau must “take a decennial census of the population” starting on April 1, 2020,
and report the results to the President by December 31, 2020 (the deadline
primarily at issue in this case). See 13 U.S.C. § 141(a)-(b). After receiving this
report, the President must calculate “the number of Representatives to which each
State would be entitled” and transmit that information to Congress by January 10,
2
there’s one branch Congress has not delegated any census decisions to: the
judiciary.
Cognizant of its statutory deadlines—but unaware of the looming health
crisis—the Census Bureau adopted a final operational plan for the 2020 Census in
December 2018. This plan has two major phases: a data-collection phase and a data-
processing phase. During the data-collection phase, field employees follow up at
non-responding addresses and collect other crucial information. Only after this
phase is complete can the Bureau begin processing the collected data to report to the
President by the December 31 deadline.
But even the best laid plans can go awry. Just as the data collection phase was
set to begin, the COVID-19 pandemic struck, forcing the Bureau to suspend its field
operations for four weeks. To resume those operations, the Bureau adopted the
COVID-19 Plan on April 13, 2020, which set new deadlines for the data collection
and dating processing phases, on the assumption that Congress would extend the
statutory deadlines by 120 days. Congress did not act, however, so the Bureau
adopted the “Replan” schedule, which outlined expedited deadlines designed “to
accelerate the completion of data collection [] by our statutory deadline of December
31, 2020, as required by law[.]” According to the Bureau, it was able to meet this
2021. See 2 U.S.C. § 2a(a). Finally, the Bureau must report a tabulation of
population for redistricting to the states by March 31, 2021. See 13 U.S.C. §
141(c).
3
compressed timeframe by (1) offering financial incentives to increase the number of
hours each enumerator worked and achieve the “same work hours as would have
been done under the original time frame”; and (2) taking advantage of updated
software and processing capabilities not available during the 2010 Census in order
to maximize enumerator effectiveness. An Associate Director at the Bureau attests
that the agency “is confident that it can achieve a complete and accurate census and
report apportionment counts by the statutory deadline following the Replan
Schedule.” (emphasis added).2 Under this plan, field operations would conclude by
September 30, and data processing would begin on October 1. The Bureau asserts
that it must complete the data collection phase by September 30 and turn to the data
processing phase by October 1 to meet its December 31, 2020 deadline. See Motion
at 1. On September 28, 2020, the Bureau extended its internal deadline slightly:
setting October 5, 2020 as the target date for concluding field operations.3 As of
September 28, 2020, the Bureau reports over 98% enumeration nationwide. 4
II.
2
Inexplicably, the majority’s decision simply ignores this attestation when
claiming that even under the Replan, “the Bureau will be unable to deliver an
accurate census by December 31, 2020.” Majority Op. at 5.
3
United States Census Bureau, 2020 Census Update,
https://www.census.gov/newsroom/press-releases/2020/2020-census-update.html
4
United States Census 2020, Total Response Rates by State,
https://2020census.gov/en/response-rates/nrfu.html
4
Whether to grant a request for a stay is governed by the familiar four-factor
test: “(1) whether the stay applicant has made a strong showing that he is likely to
succeed on the merits; (2) whether the applicant will be irreparably injured absent a
stay; (3) whether issuance of the stay will substantially injure the other parties
interested in the proceeding; and (4) where the public interest lies.” Doe #1 v.
Trump, 944 F.3d 1222, 1225 (9th Cir. 2019) (Bress, J., dissenting) (simplified).5
We should have granted an administrative stay here because defendants are
likely to succeed on the merits. The Secretary’s decision to adopt the Replan—
rather than simply ignore a statutory deadline—was not arbitrary and capricious.
At bottom, the district court’s APA analysis seems to turn on the court’s apparent
disagreement with whether the census will be sufficiently accurate under the
Replan. But the accuracy of the census is likely a nonjusticiable political question;
a properly deferential review would find the Replan satisfies statutory and
constitutional requirements; and the plaintiffs here do not appear to have standing
5
The majority suggests that I apply the “wrong standard for a preliminary
administrative stay.” Majority at 7. But as Judge Bress has already persuasively
explained: “the instant request for a temporary stay is part of the request for a stay
pending appeal, and the Court cites no authority for why the usual stay factors—
including likelihood of success on the merits—would not apply.” Doe #1, 944
F.3d at 1226 (Bress, J., dissenting). We can’t simply ignore the fact that the
government is likely to prevail on the merits here. That’s particularly true where,
like here, the parties have addressed the merits in the request for a stay and the
opposition thereto. See id.
5
because their alleged injuries are not redressable. I discuss each flaw with the
district court’s injunction in turn.
A.
Putting aside momentarily the fact that the crux of this case is not justiciable,
see, infra, § II-B and II-C, and assuming that the APA applies here and that the
Replan can be considered a “final agency action,” cf. NAACP v. Bureau of the
Census, 945 F.3d 183, 189 (4th Cir. 2019) (challenges to 2020 census “design
choices” were not final agency actions under the APA), the Replan does not violate
the APA.
Under the APA, agencies must engage in “reasoned decisionmaking.”
Michigan v. EPA, 576 U.S. 743, 750 (2015). Where census decisions are
concerned, this only requires the Secretary to “examine the relevant data and
articulate a satisfactory explanation for his decision.” Department of Commerce v.
New York, 139 S.Ct. 2551, 2569 (2019). “We may not substitute our judgment for
that of the Secretary.” Id. Nor may we “subordinat[e] the Secretary’s
policymaking discretion to the Bureau’s technocratic expertise.” Id. at 2571
(Bureau staff’s conclusions are not “touchstones of substantive reasonableness.”);
accord Wisconsin, 517 U.S. at 23 (Because it is the Secretary “to whom Congress
has delegated its constitutional authority over the census,” “the mere fact that the
6
Secretary’s decision overruled the views of some of his subordinates is by itself of
no moment in any judicial review of his decision.”).
To make reasoned decisions, agencies must consider “significant
alternatives.” Mt. Diablo Hosp. v. Shalala, 3 F.3d 1226, 1232 (9th Cir. 1993). The
defendants did so. As a Bureau Associate Director explained, the Bureau
“considered a variety of options and evaluated risks” in crafting the Replan,
ultimately “select[ing] those that we believed presented the best combination of
changes to allow us to meet the statutory deadline without compromising quality to
an undue degree.” Although the Replan compressed several steps, which might
“increase the risk” of errors, the Associate Director explained that efficiencies new
to the 2020 Census nevertheless allowed the Replan to “achieve a complete and
accurate census.”
The core of the district court’s reasoning is that the Secretary erred in
considering the deadline fixed and then trying to maximize accuracy within that
constraint. The court thought the Secretary should have been more flexible and
considered other alternatives. But all of the alternatives would require the Bureau
to consciously blow a statutory deadline. For example, the district court suggests
the defendants could have considered “not adopting the Replan while striving in
good faith to meet statutory deadlines.” Or, as the plaintiffs put it, “Defendants
could have continued to operate under the COVID-19 Plan while striving to meet
7
statutory deadlines.” But the COVID-19 Plan was premised on Congress
extending the statutory deadlines. By adhering to that plan despite Congress’s
inaction, the defendants would necessarily not be striving in good faith to meet the
deadline; they would be consciously abandoning it.6 “An agency is under no
obligation to consider every possible alternative to a proposed action, nor must it
consider alternatives that are unlikely to be implemented or those inconsistent with
its basic policy objectives.” Seattle Audubon Soc’y v. Moseley, 80 F.3d 1401, 1404
(9th Cir. 1996). Thus, a fortriori, an agency need not consider alternatives that
violate the law. The Bureau cannot be liable for failing to consider an alternative
that would undisputedly violate the clear deadline set by Congress to obtain
marginal improvements (of some unknown degree) to the census.
The district court also erred in determining that the Secretary’s reason for
adopting the Replan ran contrary to the facts. The district court noted that some
Bureau employees thought it would be impossible to accurately complete the
census by December 31, given the COVID-19 delays.7 But each statement relied
6
This same core defect infects the other proposed alternatives, such as making
“good faith efforts to meet the deadline” short of adopting the Replan and
“balanc[ing]” accuracy and timeliness concerns.
7
The court also suggests that the Commerce Department pressured the Bureau to
cease seeking an extension of the deadline, though nothing in the record before this
panel suggests this is so, and the district court’s citations show only that the Bureau
did not affirmatively request an extension in certain instances. Even if that were
true though, it cannot undermine the Bureau’s stated reason that it adopted the
8
on was made before the Replan, which the Bureau’s Associate Director has
attested will reach sufficient levels of accuracy. In any event, “there is nothing
even unusual” about a Cabinet secretary “disagreeing with staff, or cutting through
red tape.” New York, 139 S. Ct. at 2580 (Thomas J., concurring in part and
dissenting in part). The Secretary is owed “wide discretion” in this arena because
“it is he to whom Congress has delegated its constitutional authority over the
census.” Wisconsin, 517 U.S. at 22; see 13 U.S.C. § 141(a). Dissent from inferior
employees at the Bureau cannot constitute “facts” that the Secretary’s decision
runs “contrary” to. See Wisconsin, 517 U.S. at 23 (“[T]he mere fact that the
Secretary’s decision overruled the views of some of his subordinates is by itself of
no moment in any judicial review of his decision.”). To hold otherwise would
impermissibly “subordinat[e] the Secretary’s policymaking discretion to the
Bureau’s technocratic expertise.” New York, 139 S. Ct. at 2571.
Finally, the district court concluded that the defendants “failed to sufficiently
consider” their obligations to produce an accurate census because “the Replan will
decrease the census’s accuracy and undercount historically undercounted
Replan because it realized Congress would not extend the deadline. See New York,
139 S. Ct. at 2576 (2019) (Thomas J., concurring in part and dissenting in part)
(courts defer to executive agency and it is entitled to a presumption of regularity in
part because crediting accusations of pretext, which can be easily lodged by
“political opponents of executive actions to generate controversy,” could “lead
judicial review of administrative proceedings to devolve into an endless morass of
discovery and policy disputes”).
9
individuals.” But the need to consider accuracy does not give courts license to act
as a super Census Bureau. The Secretary is “required to consider the evidence and
give reasons for his chosen course of action,” but “[i]t is not for us to ask whether
[the] decision was ‘the best one possible’ or even whether it was ‘better than the
alternatives.’” New York, 139 S. Ct. at 2571 (citation omitted). The Bureau
fulfilled the deliberative requirement by considering the Replan’s impact on
accuracy. See Providence v. Yakima Med. Ctr. v. Sebelius, 611 F.3d 1181, 1190
(9th Cir. 2010) (Agency action is arbitrary and capricious where the agency
“entirely failed” to consider an important aspect of the problem.) (emphasis
added).
B.
Although the district court ostensibly conducted APA review of the
procedures the Secretary used to adopt the Replan, the crux of the court’s decision
is its view that the Replan would not produce an accurate census. But the
“accuracy” requirement is a general duty arising from the Census Act, not a
specific statutory or constitutional mandate. See New York, 139 S. Ct. at 2568–69
(“[B]y mandating a population count that will be used to apportion representatives,
see [13 U.S.C.] § 141(b), 2 U. S. C. § 2a, the Act imposes a duty to conduct a
census that is accurate and that fairly accounts for the crucial representational
rights that depend on the census and the apportionment.”) (simplified). And it is
10
for the Secretary, under the authority Congress delegated to him, to balance the
need for accuracy against the statute’s hard deadline.
Although justiciability arguments are only raised briefly on the pending
motion for a stay, “federal courts have an independent obligation to ensure that
they do not exceed the scope of their jurisdiction, and therefore they must raise and
decide jurisdictional questions that the parties either overlook or elect not to press.”
Henderson ex rel. Henderson v. Shinseki, 562 U.S. 428, 434 (2011). Deciding
whether the census meets a free-floating concept of “accuracy” is exactly the type
of political question that courts are powerless to adjudicate. Virtually all of the
factors announced in Baker v. Carr, 369 U.S. 186 (1962), support a finding of this
being a nonjusticiable political question.8 Principally, the district court’s
“accuracy” requirement is not amenable to “judicially discoverable and
manageable standards.” See id. at 217. How accurate is accurate enough? See,
e.g., Department of Commerce v. United States House of Representatives, 525 U.S.
316, 322 (1999) (“[T]he Bureau has always failed to reach—and has thus failed to
8
These factors include: a textual commitment of the issue to a coordinate political
branch, a lack of judicially discoverable and manageable standards for resolving it,
the impossibility of deciding without an initial policy determination of the kind
clearly for nonjudicial discretion, or the impossibility of a court’s undertaking
independent resolution of the question without expressing a lack of respect due
coordinate branches of government. Baker, 369 U.S. at 217.
11
count—a portion of the population.”). 9 And what standard are courts to use when
evaluating accuracy anyway? Neither the district court nor this panel offer any
answers.
But the answer is actually quite simple: it would be impossible for us to
decide this case “without an initial policy determination of the kind clearly for
nonjudicial discretion.” See Carr, 369 U.S. at 217. Even under ordinary
circumstances, the Secretary and Bureau must juggle many important
considerations when designing the census plan. For example, in choosing the date
for when to end its data-collection phase and begin its data-processing phase, the
defendants must consider the trade-offs between terminating field operations (even
though not everyone has been counted) against the time needed to process the data
into the Secretary’s report to the President and the States. See 2 U.S.C. § 2a(a); 13
U.S.C. § 141(c); see also NAACP, 945 F.3d at 191 (“‘Setting aside’ one or more of
these ‘choices’ necessarily would impact the efficacy of the others, and inevitably
would lead to court involvement in ‘hands-on’ management of the Census
9
See also Wisconsin, 517 U.S. at 6 (“[Various] errors have resulted in a net
‘undercount’ of the actual American population in every decennial census.”);
Karcher v. Daggett, 462 U.S. 725, 732 (1983) (recognizing that “census data are
not perfect,” and that “population counts for particular localities are outdated long
before they are completed”); Gaffney v. Cummings, 412 U.S. 735, 745 (1973)
(remarking that census data “are inherently less than absolutely accurate”); accord
C. Wright, History and Growth of the United States Census 16-17 (1900) (noting
that the accuracy of our first census in 1790 was seriously questioned by the man
who oversaw its implementation as Secretary of State, Thomas Jefferson).
12
Bureau’s operations.”). With each decision, the Bureau must consider (and choose
among) the various tradeoffs each option presents. By requiring the Bureau to
prioritize an elusive standard of accuracy over and above the interest in completing
the census in a timely manner, as prescribed by Congress, the court substitutes its
own policy determination for those set by Congress and delegated to the Secretary.
Analogous cases have held similar claims to be nonjusticiable political
questions. Just last year the Court held that trying to decide among “different
visions of fairness” for districting maps is an “unmoored determination of the sort
characteristic of a political question beyond the competence of the federal courts.”
Rucho v. Common Cause, 139 S. Ct. 2484, 2499–2500 (2019) (internal quotations
omitted); accord Nixon v. United States, 506 U.S. 224 (1993) (constitutional
provision granting the “the sole Power to try all Impeachments” does not “provide
an identifiable textual limit on the authority which is committed to the Senate”).
So too here: determining the “accuracy” of the census is no more of a judicial
question than determining the “fairness” of districting maps.10
10
Nor does the fact that plaintiffs brought their claims under the APA change the
political question analysis. See 5 U.S.C. § 702 (“Nothing herein . . . affects other
limitations on judicial review or the power or duty of the court to dismiss any
action or deny relief on any other appropriate legal or equitable ground[.]”); Int’l
Refugee Assistance Project v. Trump, 883 F.3d 233, 366 (4th Cir. 2018)
(Niemeyer, J., dissenting) (“§ 702(1)’s recognition of ‘other limitations’ on the
scope of APA review reflects Congress’s intent to maintain longstanding
prudential limits confining the judiciary to its proper role in our constitutional
13
To be sure, courts may entertain some challenges to census-related
decisions. But cases treating such challenges as justiciable involved narrow and
deferential review—not a freewheeling inquisition into the “accuracy” of the
census. In Department of Commerce v. New York, for example, the Court
considered whether the Secretary could add a citizenship question to the census
consistent with the Enumeration Clause and Census Act. 139 S. Ct. at 2566, 2569.
On the constitutional challenge, the Court reviewed only for whether the addition
of the challenged question bore a “reasonable relationship to the accomplishment
of an actual enumeration.” Id. at 2566. On the statutory question, the Court
deferentially considered “whether the Secretary examined the relevant data and
articulated a satisfactory explanation for his decision.” Id. at 2569. The Court’s
other census cases likewise involved this type of narrow and deferential review.
See Wisconsin, 517 U.S. at 19–20 (“[S]o long as the Secretary’s conduct of the
census is consistent with the constitutional language and the constitutional goal of
equal representation, it is within the limits of the Constitution.”) (simplified);
Franklin v. Massachusetts, 505 U.S. 788, 801 (1992) (similar); see also U.S. Dep’t
of Commerce v. Montana, 503 U.S. 442, 458–59 (1992) (“The polestar of equal
system, such as the political question doctrine.”); Mobarez v. Kerry, 187 F. Supp.
3d 85, 92 (D.D.C. 2016) (holding that political question doctrine precluded review
of APA claims).
14
representation does not provide sufficient guidance to allow us to discern a single
constitutionally permissible course” among multiple options.).
When our review morphs beyond these precedents into an interrogation of
“accuracy,” of the type underlying the district court’s APA analysis here, we are
beyond our proper role as judges. Some legal questions—even ones arising under
the same constitutional provision as previously justiciable questions—might prove
to be nonjusticiable. See New York v. United States, 505 U.S. 144, 185 (1992)
(“[T]he Court has suggested that perhaps not all claims under the Guarantee Clause
present nonjusticiable political questions” even if most do). Thus, while the court
might be competent to decide whether a particular decision bears a “reasonable
relationship” to the goal of an “actual enumeration,” the same cannot be said of
evaluating the “accuracy” of a census. Indeed, the Court has rejected the claim
that its prior cases require “a census that was as accurate as possible” and has
recognized that “[t]he Constitution itself provides no real instruction” on how to
measure the “accuracy” of a census. Wisconsin, 517 U.S. at 18; see also Tucker v.
U.S. Dep’t of Commerce, 958 F.2d 1411, 1417 (7th Cir. 1992) (Posner, J.) (“It
might be different if the apportionment clause, the census statutes, or the
Administrative Procedure Act contained guidelines for an accurate decennial
census, for that would be some evidence that the framers of these various
enactments had been trying to create a judicially administrable standard.”).
15
We cannot mechanically apply the political question doctrine, which must be
considered in light of the important separation of powers function it performs. A
court’s authority to act depends on a threshold question of the “appropriate role for
the Federal Judiciary”: whether the claims brought “are claims of legal right,
resolvable according to legal principles, or political questions that must find their
resolution elsewhere.” Rucho, 139 S. Ct. at 2494 (emphasis in original). Here,
these background principles weigh in favor of not adjudicating this dispute. No
census has been, or can be, fully accurate, according to the Court. See Wisconsin,
517 U.S. at 6 (“Although each [census] was designed with the goal of
accomplishing an ‘actual Enumeration’ of the population, no census is recognized
as having been wholly successful in achieving that goal.”). Determining what
level of accuracy is sufficient is simply not something that the judicial branch is
equipped to do. 11 Indeed, “[i]t would be difficult to think of a clearer example of
the type of governmental action that was intended by the Constitution to be left to
the political branches directly responsible—as the Judicial Branch is not—to the
11
The district court and plaintiffs seem to think that the district court’s injunction
does not require judicial supervision over the accuracy of the census. Instead, they
frame the injunction as merely preventing the Secretary from adopting the Replan
because it failed to follow the requisite procedures for doing so. But the crux of
the district court’s injunction is its disagreement with the Secretary’s resolution of
how to balance accuracy of the census against the statutory deadline. See, infra,
§ II-A. And in ordering relief, the district court has inserted itself at the top of the
Executive branch’s census operation. See Motion at 17 (describing ongoing
supervision of the district court under the preliminary injunction).
16
electoral process. Moreover, it is difficult to conceive of an area of governmental
activity in which the courts have less competence.” Gilligan v. Morgan, 413 U.S.
1, 10 (1973). By allowing census-accuracy supervision under the guise of APA
review, we have “given the green light for future political battles to be fought in
this Court rather than where they rightfully belong—the political branches.” DHS
v. Regents of the Univ. of California, 140 S. Ct. 1891, 1919 (2020) (Thomas, J.,
concurring in part and dissenting in part).
C.
Plaintiffs also likely fail to establish Article III standing, given that they
have not shown that their alleged injury is redressable by the courts, even assuming
the other standing requirements are met. An injury is necessarily not redressable if
the court has no authority to authorize the relief requested. See Gonzales v.
Gorsuch, 688 F.2d 1263, 1267 (9th Cir. 1982) (Kennedy, J.) (“Redressability
requires an analysis of whether the court has the power to right or to prevent the
claimed injury.”); Republic of Marshall Islands v. United States, 865 F.3d 1187,
1199 (9th Cir. 2017) (holding that a lawsuit seeking to enforce a treaty right was
not redressable because “the federal courts have no power to right or to prevent . . .
violat[ions of] a non-self-executing treaty provision”).
Clearly, a district court has no authority to order an Executive agency to
disobey a Congressional statute. Neither the district court nor plaintiffs have cited
17
any authority for this unprecedented expansion of the judicial power to decide
cases and controversies. See U.S. Const. Art. III, § 2. Congress makes laws, the
Executive enforces them, and we interpret them in the course of adjudicating
disputes. Absent the metaphorical “striking down” of an unconstitutional statute,
we are impotent to set aside congressionally enacted laws. See United States v.
Booker, 543 U.S. 220, 283 (2005) (Stevens, J., dissenting in part) (“[T]he Court
simply has no authority to invalidate legislation absent a showing that it is
unconstitutional. To paraphrase Chief Justice Marshall, an ‘act of the legislature’
must be ‘repugnant to the constitution’ in order to be void.’” (quoting Marbury v.
Madison, 1 Cranch 137, 177 (1803))). Here, no one challenges the
constitutionality of the statute establishing the Secretary’s deadline. Accordingly,
the district court had no authority to ignore it—let alone order an Executive agency
to do so.
All of the cases relied on by the district court to enjoin operation of the
statute, despite not finding any constitutional infirmity, are wholly inapposite.
None suggest that a court can require an agency to disobey a statute; they merely
confirm that an agency is not necessarily precluded from acting, even if it is doing
so after a statutory deadline. See, e.g., Barnhart v. Peabody Coal Co., 537 U.S.
149, 158 (2003) (holding that despite statute’s mandatory deadline, post-deadline
action taken by the agency was not void because there was no Congressional intent
18
that agency would be deprived of statutory authority to act if it did so beyond the
deadline); Linemaster Switch Corp. v. EPA, 938 F.2d 1299, 1304 (D.C. Cir. 1991)
(similar). The fact that an agency can—depending on the text, structure, and
history of the statue at issue—continue to act beyond its statutory deadline, says
nothing about a court’s authority to require an agency to do so.
D.
An agency’s decision on how to respond to a once-in-a-century pandemic, in
order to meet its statutory deadline, is quintessentially the type of decision we
should give substantial deference to. Throughout this pandemic, we’ve deferred to
the elected branches to determine how to best respond, even when shuttering our
churches and businesses. See, e.g., S. Bay United Pentecostal Church v. Newsom,
959 F.3d 938, 939 (9th Cir. 2020) (denying Free Exercise Clause challenge to
application of California’s stay-at-home order to in-person religious services based
on deference to elected branches during pandemic). We’ve done so despite our
role in protecting individuals’ constitutional rights. See On Fire Christian Ctr.,
Inc. v. Fischer, 2020 WL 1820249, at *6 (W.D. Ky. 2020) (although “a state may
implement emergency measures that curtail constitutional rights” during a
pandemic, it cannot enact measures that are “beyond all question, a plain, palpable
invasion of rights secured by the fundamental law”). If deference is appropriate
there, surely it is doubly appropriate here, where courts are already required to
19
show deference to the agencies. See New York, 139 S. Ct. at 2578 & n.3 (Thomas,
J., concurring in part and dissenting in part) (explaining highly deferential review
of an “agency’s discretionary choices and reasoning under the arbitrary-and-
capricious standard”); Wisconsin, 517 U.S. at 19–20 (explaining narrow and
deferential review of Secretary’s census decision). Simply put, there’s no basis to
anoint ourselves supervisors of this sensitive process at the eleventh hour.
III.
At a minimum, we should have granted an administrative stay while we
further considered the underlying motion to stay the injunction pending appeal.
The government faces irreparable harm from our refusal to do so. It’s undisputed
that if the government cannot finalize the data collection phase of the census and
move into the data processing phase in a timely fashion, it will likely miss its
statutory deadline.
Thus, even if the court ultimately rules for the defendants on the merits, it
might not matter much: the plaintiffs will have effectively secured the relief they
seek on the merits (e.g., a delay of moving into the data processing phase). In
contrast, the defendants have said only that it would be “difficult” to rehire and
redeploy workers once terminated, if they are allowed to do so, but not that it
would be impossible to revamp these workers if needed. Accordingly, although an
administrative stay would be inefficient if ultimately reversed later, the damage
20
would not be irreparable. At most it would present a bureaucratic hassle for the
agencies. The same cannot be said for the majority’s decision to deny the
administrative stay. Similarly, the district court, and now the majority, fail to
consider the harms that irreparably flow to other States. See Amicus Brief at 8
(“The effect of the TRO was to run up the census tally in Plaintiffs’ jurisdictions at
the expense of lagging jurisdictions like Louisiana and Mississippi.”); id. at 8–9
(noting “disruption of redistricting and reapportionment in 24 states that have
constitutional or statutory deadlines” tied to census).
Finally, the status quo here, to the extent that’s relevant, is the legal
landscape that would have existed prior to the district court’s judicial
misadventure. See Doe #1, 944 F.3d at 1229 (Bress, J., dissenting) (explaining that
preserving the status quo is not an enumerated factor, but in any event, an
administrative “stay simply suspends judicial alteration of the status quo, while the
injunctive relief granted below constitutes judicial intervention upending it”)
(simplified). Accordingly, we should have granted the request for an
administrative stay to restore the parties to the positions they were in prior to the
district court’s decision.
IV.
Despite its errors, the district court deserves some credit. It seems to have
been motivated by a valiant attempt to balance two competing priorities: accuracy
21
of the census versus timeliness under the statutory deadline. But the elected
branches have already done this balancing. The Secretary of Commerce was
briefed on all of the Bureau employee concerns the district judge found persuasive.
The Secretary considered those concerns, and then, in exercising the role that the
President appointed him to perform, made the decision to proceed with the Replan.
“By second-guessing the Secretary’s weighing of risks and benefits and penalizing
him for departing from the Bureau’s” views about the Replan, the district court,
and now the majority, “substitute[] [their] judgment for that of the agency.” New
York, 139 S.Ct. at 2571. Likewise, Congress was aware of the potential problem
and did not extend the deadline. The House of Representatives held committee
hearings and ultimately voted on a bill to extend the deadline. The Senate received
the bill, held committee hearings on it, but then took no further action—and hasn’t
since July 2020.12 Plaintiffs suggest that the Senate might act on the bill soon. 13
There is no basis for the judiciary to inject itself into this sensitive political
controversy and seize for itself the decision to reevaluate the competing concerns
between accuracy and speed, after the elected branches have apparently done so
12
https://www.congress.gov/bill/116th-congress/house-bill/6800/all-
actions?overview=closed
13
Plaintiffs’ Opposition at 14 n.1 (citing Hansi Lo Wang, Bipartisan Senate Push
to Extend Census Begins Weeks Before Count Is Set to End, NPR (Sept. 15,
2020), https://www.npr.org/2020/ 09/15/913163016/bipartisan-senate-push-to-
extend-census-begins-weeks-before- count-is-set-to-end)
22
already—or are actively doing so now. See Clinton v. City of New York, 524 U.S.
417, 449 (1998) (Kennedy, J., concurring) (“Failure of political will does not
justify unconstitutional remedies.”). Plus, had we ruled for the defendants, nothing
would have prevented the elected branches from revisiting this dispute at a later
date. A belated fix might entail additional cost and delay that the district court’s
injunction avoids. But in our constitutional design, courts are not empowered to
swoop in and rescue the elected branches from themselves. If additional cost and
delay is the consequence of Congress’s inaction, or the Secretary’s decision to
adopt the Replan, then so be it. The recourse for such problems lies with the
People themselves at the ballot box—not with unelected and unaccountable judges
in chambers.
I respectfully dissent.
23