United States v. Fokker Services B.V.

 United States Court of Appeals
         FOR THE DISTRICT OF COLUMBIA CIRCUIT



Argued September 11, 2015             Decided April 5, 2016

                       No. 15-3016

               UNITED STATES OF AMERICA,
                       APPELLEE

                            v.

                 FOKKER SERVICES B.V.,
                     APPELLANT


                Consolidated with 15-3017


       Appeals from the United States District Court
               for the District of Columbia
         and Petition for the Writ of Mandamus
                  (No. 1:14-cr-00121-1)


    Edward C. O’Callaghan argued the cause for appellant.
With him on the briefs was David D. DiBari.

    Aditya Bamzai, Attorney, U.S. Department of Justice,
argued the cause for appellee. With him on the briefs were
Vincent H. Cohen, Jr., Acting U.S. Attorney, Mary B.
McCord, Principal Deputy Assistant Attorney General, U.S.
Department of Justice, Steven M. Dunne, Chief, Appellate
Unit, and Elizabeth Trosman and Elizabeth H. Danello,
Assistant U.S. Attorneys.
                              2
    Adam G. Unikowsky argued the cause for court-appointed
amicus curiae. With him on the brief was David W. DeBruin,
appointed by the court.

   Before: SRINIVASAN, Circuit Judge, and SILBERMAN and
SENTELLE, Senior Circuit Judges.

    Opinion for the Court filed by Circuit Judge SRINIVASAN.

    SRINIVASAN, Circuit Judge: The Constitution allocates
primacy in criminal charging decisions to the Executive
Branch.    The Executive’s charging authority embraces
decisions about whether to initiate charges, whom to
prosecute, which charges to bring, and whether to dismiss
charges once brought. It has long been settled that the
Judiciary generally lacks authority to second-guess those
Executive determinations, much less to impose its own
charging preferences.       The courts instead take the
prosecution’s charging decisions largely as a given, and
assume a more active role in administering adjudication of a
defendant’s guilt and determining the appropriate sentence.

     In certain situations, rather than choose between the
opposing poles of pursuing a criminal conviction or forgoing
any criminal charges altogether, the Executive may conclude
that the public interest warrants the intermediate option of a
deferred prosecution agreement (DPA). Under a DPA, the
government formally initiates prosecution but agrees to
dismiss all charges if the defendant abides by negotiated
conditions over a prescribed period of time. Adherence to the
conditions enables the defendant to demonstrate compliance
with the law. If the defendant fails to satisfy the conditions,
the government can then pursue the charges based on facts
admitted in the agreement.
                               3
     This case arises from the interplay between the operation
of a DPA and the running of time limitations under the
Speedy Trial Act. Because a DPA involves the formal
initiation of criminal charges, the agreement triggers the
Speedy Trial Act’s time limits for the commencement of a
criminal trial. In order to enable the government to assess the
defendant’s satisfaction of the DPA’s conditions over the time
period of the agreement—with an eye towards potential
dismissal of the charges—the Speedy Trial Act specifically
allows for a court to suspend the running of the time within
which to commence a trial for any period during which the
government defers prosecution under a DPA.

     In this case, appellant Fokker Services voluntarily
disclosed its potential violation of federal sanctions and
export control laws.     After extensive negotiations, the
company and the government entered into an 18-month DPA,
during which Fokker would continue cooperation with federal
authorities and implementation of a substantial compliance
program. In accordance with the DPA, the government filed
criminal charges against the company, together with a joint
motion to suspend the running of time under the Speedy Trial
Act pending assessment of the company’s adherence to the
agreement’s conditions. The district court denied the motion
because, in the court’s view, the prosecution had been too
lenient in agreeing to, and structuring, the DPA. Among
other objections, the court disagreed with prosecutors’
decision to forgo bringing any criminal charges against
individual company officers.

    We vacate the district court’s denial of the joint motion to
exclude time under the Speedy Trial Act. We hold that the
Act confers no authority in a court to withhold exclusion of
time pursuant to a DPA based on concerns that the
government should bring different charges or should charge
                              4
different defendants. Congress, in providing for courts to
approve the exclusion of time pursuant to a DPA, acted
against the backdrop of long-settled understandings about the
independence of the Executive with regard to charging
decisions. Nothing in the statute’s terms or structure suggests
any intention to subvert those constitutionally rooted
principles so as to enable the Judiciary to second-guess the
Executive’s exercise of discretion over the initiation and
dismissal of criminal charges.
     In vacating the district court order, we have no occasion
to disagree (or agree) with that court’s concerns about the
government’s charging decisions in this case. Rather, the
fundamental point is that those determinations are for the
Executive—not the courts—to make. We therefore grant the
government’s petition for a writ of mandamus and remand for
further proceedings consistent with this opinion.
                              I.
                              A.
     The Speedy Trial Act establishes time limits for the
completion of various stages of a criminal prosecution. See 18
U.S.C. §§ 3161-3174. For instance, the Act requires the
commencement of trial within seventy days of the filing of an
information or indictment by the government.                Id.
§ 3161(c)(1). The Act also excludes various pretrial periods
from the running of that seventy-day time clock. Of
particular relevance, the Act excludes “[a]ny period of delay
during which prosecution is deferred by the attorney for the
Government pursuant to written agreement with the
defendant, with the approval of the court, for the purpose of
allowing the defendant to demonstrate his good conduct.” Id.
§ 3161(h)(2).
                               5
     That exemption exists to enable prosecutors to resolve
cases through DPAs. DPAs, along with their out-of-court
analogues, non-prosecution agreements (NPAs), afford a
middle-ground option to the prosecution when, for example, it
believes that a criminal conviction may be difficult to obtain
or may result in unwanted collateral consequences for a
defendant or third parties, but also believes that the defendant
should not evade accountability altogether. Both DPAs and
NPAs generally include an admitted statement of facts,
require adherence to “conditions designed . . . to promote
compliance with applicable law and to prevent recidivism,”
and remain in effect for a period of one to three years. U.S.
Attorney’s Manual § 9-28.1000 (2015). During that period, if
the defendant fails to abide by the terms of the agreement, the
government can prosecute based on the admitted facts. While
prosecutors at one time seldom relied on NPAs and DPAs,
their use has grown significantly in recent years.
     DPAs differ from NPAs primarily with regard to the
filing of criminal charges. With an NPA, “formal charges are
not filed and the agreement is maintained by the parties rather
than being filed with a court.” Craig S. Morford, Selection
and Use of Monitors in Deferred Prosecution Agreements and
Non-Prosecution Agreements with Corporations, at 1 n.2
(Mar. 7, 2008). A DPA, by contrast, “is typically predicated
upon the filing of a formal charging document by the
government.” Id.
     For that reason, a DPA’s viability depends on the
specific exclusion of time for such agreements set forth in the
Speedy Trial Act, 18 U.S.C. § 3161(h)(2). The filing of an
information or indictment would ordinarily trigger the Act’s
seventy-day clock within which trial must commence. See id.
§ 3161(c)(1). But in the case of a DPA, if the defendant were
to fulfill the agreement’s conditions, the prosecution would
move to dismiss all charges with prejudice at the end of the
                              6
specified time period, ordinarily one to three years. Without
the statutory exclusion of time for DPAs provided in
§ 3161(h)(2), the government would relinquish its ability to
prosecute based on the conceded facts if the defendant were to
violate the agreement after seventy days. That would largely
eliminate the leverage that engenders the defendant’s
compliance with a DPA’s conditions. The statutory exclusion
of time for DPAs therefore is essential to the agreements’
effective operation.
                              B.
     Fokker Services, a Dutch aerospace services company,
provides technical and logistical support to owners of aircraft
manufactured by its predecessor company. In 2010, Fokker
voluntarily disclosed to the United States Departments of
Treasury and Commerce that it had potentially violated
federal sanctions and export control laws concerning Iran,
Sudan, and Burma. At the time Fokker came forward, no
government agency had initiated any investigation focused on
the company.
     Over the course of the next four years, Fokker cooperated
in the wide-ranging investigation conducted by federal
authorities. The company facilitated interviews of relevant
witnesses, expedited the government’s requests to Dutch
authorities for documents under the Mutual Legal Assistance
Treaty, and initiated its own internal investigation. Fokker’s
internal investigation revealed that, from 2005 to 2010, the
company had participated in 1,147 illicit transactions through
which it earned some $21 million in gross revenue. The
company instituted remedial measures to improve its
sanctions compliance program, adopting a set of procedures
to track parts and bolstering its employee training
requirements. It also fired its president and demoted or
reassigned other employees who had been involved in the
                               7
violations. The company’s compliance efforts have been
described by government officials as “a model to be followed
by other corporations.” Gov’t Supp. Mem. in Support of
DPA Reached with Fokker Services, B.V., at 15.
      In light of Fokker’s cooperation, remediation efforts, and
other mitigating factors, federal agencies negotiated a global
settlement with the company. The settlement included, as an
integral component, an 18-month DPA. During the DPA’s
18-month period, Fokker was to: continue full cooperation
with the government, implement its new compliance policy,
and pay fines and penalties totaling $21 million (a sum
equaling the gross revenues gained by the company from the
illicit transactions). Fokker also accepted responsibility for
the acts described in the stipulated factual statement
accompanying the DPA.
     On June 5, 2014, pursuant to the agreement, the
government filed with the district court a one-count
information against Fokker, together with the DPA. The
information charged Fokker with conspiracy to violate the
International Emergency Economic Powers Act. See 18
U.S.C. § 371; 50 U.S.C. § 1705. The same day, the
government and Fokker filed a joint motion for the exclusion
of time under the Speedy Trial Act, in order to “allow [the
company] to demonstrate its good conduct and implement
certain remedial measures.” Joint Consent Motion for
Exclusion of Time Under the Speedy Trial Act, at 1.
     The district court then held a series of status conferences,
during which it repeatedly emphasized its concerns about the
absence of any criminal prosecution of individual company
officers. Tr. of Status Conference (June 25, 2014), at 4; Tr. of
Status Conference (July 9, 2014), at 5. The court requested
several additional written submissions from the government.
The government was asked to explain why the interests of
                               8
justice supported the court’s approval of the deal embodied by
the DPA, and also to address whether Fokker’s initial
disclosures to the government had in fact been voluntary. See
Tr. of Status Conference (June 25, 2014), at 3-4; Tr. of Status
Conference (July 9, 2014), at 5-6.           In response, the
government described why the “proposed resolution with
Fokker Services is fair and is an appropriate exercise of the
government’s discretion,” Gov’t Mem. in Support of DPA
Reached with Fokker Services, B.V., at 2, and affirmed the
absence of any indication “that Fokker Services was
motivated to make its disclosures out of fear about a
nonexistent U.S. government investigation,” Gov’t Status
Report, at 15. The district court later expressed that it might
still reject the DPA because it was “too good a deal for the
defendant.” Tr. of Status Conference (Oct. 29, 2014), at 4.
     On February 5, 2015, the district court denied the joint
motion for the exclusion of time. In explaining the reasons
for its decision, the court criticized the government for failing
to prosecute any “individuals . . . for their conduct.” United
States v. Fokker Services, B.V., 79 F. Supp. 3d 160, 166
(D.D.C. 2015). According to the court, approval of an
agreement in which the defendant had been “prosecuted so
anemically for engaging in such egregious conduct for such a
sustained period of time and for the benefit of one of our
country’s worst enemies” would “promote disrespect for the
law.” Id. at 167. The court further noted that certain
employees had been permitted to remain with the company;
that the DPA contained no requirement for an independent
monitor; and that the amount of the fine failed to exceed the
revenues Fokker gained from the illegal transactions. Id. at
166. Based on those considerations, the court rejected the
DPA as an “[in]appropriate exercise of prosecutorial
discretion.” Id. at 167.
                               9
     The district court’s order marks the first time any federal
court has denied a joint request by the parties to exclude time
pursuant to a DPA. Both parties filed a timely notice of
appeal. Because both parties seek to overturn the district
court’s denial of their joint motion to exclude time, we
appointed an amicus curiae to present arguments defending
the district court’s action.
                              II.
     Although we face a threshold question concerning our
jurisdiction to review the district court’s interlocutory order,
our assessment of the jurisdictional issue is substantially
informed by our consideration of the merits of the parties’
challenge to the district court’s action. Consequently, in
accordance with our approach in parallel circumstances, see
In re Kellogg Brown & Root, Inc., 756 F.3d 754 (D.C. Cir.
2014), we first consider whether the district court legally
erred in its denial of the joint motion to exclude time pursuant
to the DPA. We conclude that it did.
     By rejecting the DPA based primarily on concerns about
the prosecution’s charging choices, the district court exceeded
its authority under the Speedy Trial Act. The Act excludes
any period of time “during which prosecution is deferred by
the attorney for the Government pursuant to written
agreement with the defendant, with the approval of the court,
for the purpose of allowing the defendant to demonstrate his
good conduct.” 18 U.S.C. § 3161(h)(2). While the exclusion
of time is subject to “the approval of the court,” there is no
ground for reading that provision to confer free-ranging
authority in district courts to scrutinize the prosecution’s
discretionary charging decisions. Rather, we read the statute
against the background of settled constitutional
understandings under which authority over criminal charging
decisions resides fundamentally with the Executive, without
                               10
the involvement of—and without oversight power in—the
Judiciary. So understood, the statute’s “approval of the court”
requirement did not empower the district court to disapprove
the DPA based on the court’s view that the prosecution had
been too lenient.
                               A.
     The Executive’s primacy in criminal charging decisions
is long settled. That authority stems from the Constitution’s
delegation of “take Care” duties, U.S. Const. art. II, § 3, and
the pardon power, id. § 2, to the Executive Branch. See
United States v. Armstrong, 517 U.S. 456, 464 (1996); In re
Aiken Cnty., 725 F.3d 255, 262-63 (D.C. Cir. 2013).
Decisions to initiate charges, or to dismiss charges once
brought, “lie[] at the core of the Executive’s duty to see to the
faithful execution of the laws.” Cmty. for Creative Non-
Violence v. Pierce, 786 F.2d 1199, 1201 (D.C. Cir. 1986).
The Supreme Court thus has repeatedly emphasized that
“[w]hether to prosecute and what charge to file or bring
before a grand jury are decisions that generally rest in the
prosecutor’s discretion.” United States v. Batchelder, 442
U.S. 114, 124 (1979); see Bordenkircher v. Hayes, 434 U.S.
357, 364 (1978).
     Correspondingly, “judicial authority is . . . at its most
limited” when reviewing the Executive’s exercise of
discretion over charging determinations. Pierce, 786 F.2d at
1201; see ICC v. Bhd. of Locomotive Eng’rs, 482 U.S. 270,
283 (1987). The decision whether to prosecute turns on
factors such as “the strength of the case, the prosecution’s
general deterrence value, the [g]overnment’s enforcement
priorities, and the case’s relationship to the [g]overnment’s
overall enforcement plan.” Wayte v. United States, 470 U.S.
598, 607 (1985). The Executive routinely undertakes those
assessments and is well equipped to do so. By contrast, the
                              11
Judiciary, as the Supreme Court has explained, generally is
not “competent to undertake” that sort of inquiry. Id. Indeed,
“[f]ew subjects are less adapted to judicial review than the
exercise by the Executive of his discretion in deciding when
and whether to institute criminal proceedings, or what precise
charge shall be made, or whether to dismiss a proceeding
once brought.” Newman v. United States, 382 F.2d 479, 480
(D.C. Cir. 1967). “Judicial supervision in this area” would
also “entail[] systemic costs.” Wayte, 470 U.S. at 608. It
could “chill law enforcement,” cause delay, and “impair the
performance of a core executive constitutional function.”
Armstrong, 517 U.S. at 465 (quotation omitted). As a result,
“the presumption of regularity” applies to “prosecutorial
decisions and, in the absence of clear evidence to the contrary,
courts presume that [prosecutors] have properly discharged
their official duties.” Id. at 464 (internal quotation marks,
quotation, and alterations omitted).
                              B.
     Those settled principles counsel against interpreting
statutes and rules in a manner that would impinge on the
Executive’s constitutionally rooted primacy over criminal
charging decisions. Of particular salience, Rule 48(a) of the
Federal Rules of Criminal Procedure requires a prosecutor to
obtain “leave of court” before dismissing charges against a
criminal defendant. Fed. R. Crim. P. 48(a). That language
could conceivably be read to allow for considerable judicial
involvement in the determination to dismiss criminal charges.
But decisions to dismiss pending criminal charges—no less
than decisions to initiate charges and to identify which
charges to bring—lie squarely within the ken of prosecutorial
discretion. See e.g., Newman, 382 F.2d at 480. To that end,
the Supreme Court has declined to construe Rule 48(a)’s
“leave of court” requirement to confer any substantial role for
courts in the determination whether to dismiss charges.
                              12
Rather, the “principal object of the ‘leave of court’
requirement” has been understood to be a narrow one—“to
protect a defendant against prosecutorial harassment . . . when
the [g]overnment moves to dismiss an indictment over the
defendant’s objection.” Rinaldi v. United States, 434 U.S. 22,
29 n.15 (1977). A court thus reviews the prosecution’s
motion under Rule 48(a) primarily to guard against the
prospect that dismissal is part of a scheme of “prosecutorial
harassment” of the defendant through repeated efforts to
bring—and then dismiss—charges. Id.
     So understood, the “leave of court” authority gives no
power to a district court to deny a prosecutor’s Rule 48(a)
motion to dismiss charges based on a disagreement with the
prosecution’s exercise of charging authority. For instance, a
court cannot deny leave of court because of a view that the
defendant should stand trial notwithstanding the prosecution’s
desire to dismiss the charges, or a view that any remaining
charges fail adequately to redress the gravity of the
defendant’s alleged conduct. See In re United States, 345
F.3d 450, 453 (7th Cir. 2003). The authority to make such
determinations remains with the Executive.
     The same considerations have informed our
understanding of the respective roles of the Executive and the
courts with regard to the acceptance of certain civil consent
decrees proposed by enforcement agencies. A provision of
the Antitrust Procedures and Penalties Act, known as the
Tunney Act, calls for a district court to enter a proposed
antitrust consent decree if “in the public interest.” 15 U.S.C.
§ 16(e). In United States v. Microsoft Corp., 56 F.3d 1448
(D.C. Cir. 1995), the Department of Justice filed a civil
antitrust complaint against Microsoft, together with a
proposed consent decree embodying the parties’ settlement of
the case. Id. at 1452. The district court denied approval of
the consent decree based on a belief that the complaint and
                               13
decree were inadequate in scope to address Microsoft’s
objectionable conduct. Id. at 1452-55. The court concluded
that the consent decree therefore failed to satisfy the statute’s
“public interest” standard.
     We reversed the district court and remanded for entry of
the proposed decree. The appellants argued that the district
judge had understood his authority under the statute’s “public
interest” provision unduly expansively, so as to enable him to
“bas[e] his rejection of the decree on considerations which
implicate the executive branch’s prosecutorial discretion.” Id.
at 1457. We agreed, explaining that the “public interest”
standard did not “empower[]” the district judge to reject “the
remedies sought” in the consent decree “merely because he
believed other remedies were preferable.” Id. at 1460.
Moreover, we indicated that the district “court was barred
from reaching beyond the complaint to examine practices the
government did not challenge.” Id. To be sure, a “district
judge is not obliged to accept” a proposed decree “that, on its
face and even after government explanation, appears to make
a mockery of judicial power.” Id. at 1462. But “[s]hort of
that eventuality,” we explained, “the Tunney Act cannot be
interpreted as an authorization for a district judge to assume
the role of Attorney General.” Id. Consequently, a district
court should not “reject a consent decree simply because it
believes the [g]overnment could have negotiated a more
exacting decree,” Massachusetts v. Microsoft Corp., 373 F.3d
1199 (D.C. Cir. 2004), or because it believes the government
“failed to bring the proper charges,” SEC v. Citigroup Global
Mkts., Inc., 752 F.3d 285, 297 (2d Cir. 2014).
     As we have since explained, we “construed the public
interest inquiry” under the Tunney Act “narrowly” in “part
because of the constitutional questions that would be raised if
courts were to subject the government’s exercise of its
prosecutorial discretion to non-deferential review.” Mass.
                              14
Sch. of Law at Andover, Inc. v. United States, 118 F.3d 776,
783 (D.C. Cir. 1997); see Swift v. United States, 318 F.3d 250,
253 (D.C. Cir. 2003). The upshot is that the “public interest”
language in the Tunney Act, like the “leave of court”
authority in Rule 48(a), confers no new power in the courts to
scrutinize and countermand the prosecution’s exercise of its
traditional authority over charging and enforcement decisions.
                              C.
    The same considerations govern our interpretation of the
Speedy Trial Act provision at issue here. That provision, as
noted, allows for excluding “[a]ny period of delay during
which prosecution is deferred by the attorney for the
Government pursuant to [a DPA], with the approval of the
court, for the purpose of allowing the defendant to
demonstrate his good conduct.” 18 U.S.C. § 3161(h)(2). As
with the “leave of court” language in Rule 48(a) and the
“public interest” authority in the Tunney Act, we construe the
“approval of the court” language in § 3161(h)(2) in a manner
that preserves the Executive’s long-settled primacy over
charging decisions and that denies courts substantial power to
impose their own charging preferences.
     As an initial matter, the context of a DPA, like that of
Rule 48(a), concerns the prosecution’s core prerogative to
dismiss criminal charges. While dismissal under a DPA
follows from the defendant’s adherence to agreed-upon
conditions over a specified period, the decision to seek
dismissal pursuant to a DPA—as under Rule 48(a)—
ultimately stems from a conclusion that additional prosecution
or punishment would not serve the public interest. Dismissal
in either situation thereby fulfills the Executive’s duty under
Article II to see that the laws are faithfully executed. See
Pierce, 786 F.2d at 1201.
                                15
     We see no reason to recognize a substantially broader
authority for courts to scrutinize prosecutorial charging
choices in the context of a DPA than in the context of Rule
48(a). Just as Rule 48(a)’s “leave of court” authority does not
allow a court to withhold approval of a motion to dismiss
charges based on a belief that more serious charges should be
brought against the defendant (or against a third party),
§ 3161(h)(2)’s “approval of the court” authority does not
permit a court to withhold approval of a motion to exclude
time under a DPA based on that same belief. In either
situation, the court’s withholding of approval would amount
to a substantial and unwarranted intrusion on the Executive
Branch’s fundamental prerogatives. And the Judiciary’s lack
of competence to review the prosecution’s initiation and
dismissal of charges, see Wayte, 470 U.S. at 607-08, equally
applies to review of the prosecution’s decision to pursue a
DPA and the choices reflected in the agreement’s terms. As
with conventional charging decisions, a DPA’s provisions
manifest the Executive’s consideration of factors such as the
strength of the government’s evidence, the deterrence value of
a prosecution, and the enforcement priorities of an agency,
subjects that are ill-suited to substantial judicial oversight. Id.
     To be sure, the criminal charges filed as part of a DPA
remain on the court’s docket throughout the time of the
agreement (i.e., pending assessment of whether the defendant
has satisfied the agreement’s conditions, upon which the
prosecution seeks dismissal of the charges). But the existence
of charges on the court’s docket suggests no greater power on
the part of the court to second-guess the underlying charging
decisions than under Rule 48(a): there, too, criminal charges
remain on the court’s docket until dismissed. The key point is
that, although charges remain pending on the court’s docket
under a DPA, the court plays no role in monitoring the
defendant’s compliance with the DPA’s conditions. For
                             16
instance, defendants who violate the conditions of their DPA
face no court-ordered repercussions.            Rather, the
prosecution—and the prosecution alone—monitors a
defendant’s compliance with the agreement’s conditions and
determines whether the defendant’s conduct warrants
dismissal of the pending charges. Just as is the case under
Rule 48(a), the prosecution, after taking stock of the
circumstances, concludes that continued pursuit of a criminal
conviction is unwarranted.
     A comparison to civil consent decrees is also instructive
in this regard. Civil consent decrees not only remain on a
court’s docket, but the court—unlike with a DPA—can
enforce the decree’s terms through exercise of the contempt
power. Even in the face of that enhanced judicial role, we
have narrowly construed a court’s “public interest” authority
to review a proposed antitrust consent decree under the
Tunney Act so as to avoid encroaching on the Executive’s
core discretion over enforcement decisions. See Microsoft, 56
F.3d at 1460-62; Massachusetts School of Law, 118 F.3d at
783. And as a general matter, Executive independence is
assumed to be even more pronounced in the context of
criminal charging decisions than in the context of civil
enforcement decisions. See In re Aiken County, 725 F.3d at
264-65 n.9. In that light, we perceive no basis for concluding
that courts have greater power to second-guess charging
decisions when reviewing the terms of a DPA than when
reviewing any other Executive exercise of criminal charging
authority, including dismissals of charges under Rule 48(a).
    The text of § 3161(h)(2) does not dictate any contrary
conclusion. The statutory language ties the “approval of the
court” requirement to the DPA’s “purpose of allowing the
defendant to demonstrate his good conduct.” 18 U.S.C.
§ 3161(h)(2).    We thus understand a court’s approval
authority for the exclusion of time under a DPA to have a
                              17
particular focus: i.e., to assure that the DPA in fact is geared
to enabling the defendant to demonstrate compliance with the
law, and is not instead a pretext intended merely to evade the
Speedy Trial Act’s time constraints. Whatever may be the
precise contours of that authority of a court to confirm that a
DPA’s conditions are aimed to assure the defendant’s good
conduct, it does not permit the court to impose its own views
about the adequacy of the underlying criminal charges.
Rather, as under Rule 48(a), those core charging decisions
remain the province of the Executive.
     The Senate Committee Report accompanying the Speedy
Trial Act reinforces that circumscribed understanding of a
district court’s “approval” authority under § 3161(h)(2). The
report describes the phrase, “with the approval of the court,”
as designed to “assure[] that the court will be involved in the
decision to divert and that the procedure will not be used by
prosecutors and defense counsel to avoid the speedy trial time
limits.” S. Rep. No. 93-1021, at 37 (1974). That statement
suggests that the judicial-approval requirement was not
intended to impinge on the Executive’s traditional
independence over charging decisions.             Rather, the
requirement enables courts to assure that a DPA does not exist
merely to allow evasion of speedy trial time limits, but instead
serves the bona fide purpose of confirming a defendant’s
good conduct and compliance with law.              The Senate
Committee Report further describes § 3161(h)(2) as generally
intended to “encourage the current trend among United States
attorneys” of holding criminal charges in abeyance while
defendants participate in rehabilitation programs. Id. at 36.
Interpreting § 3161(h)(2) to empower courts to scrutinize the
prosecution’s underlying charging decisions would tend to
discourage—not encourage—the prosecution’s use of DPAs,
contradicting the provision’s apparent overarching object.
                               18
                               D.
     In defending the notion that § 3161(h)(2)’s “approval of
the court” language gives district courts substantial authority
to second-guess the prosecution’s charging decisions, amicus
seeks to analogize a court’s review of a DPA under
§ 3161(h)(2) to a court’s review of a proposed plea agreement
under Rule 11 of the Federal Rules of Criminal Procedure.
That argument fails.
     To begin with, even in the context of reviewing a
proposed plea agreement under Rule 11, a district court lacks
authority to reject a proposed agreement based on mere
disagreement with a prosecutor’s underlying charging
decisions. Rule 11 states that a district court may “accept the
agreement, reject it, or defer a decision until the court has
reviewed the presentence report.”           Fed. R. Crim. P.
11(c)(3)(A).       Although “district courts must exercise
discretion in deciding whether to accept or reject a guilty plea,
that discretion is not unfettered.” United States v. Maddox, 48
F.3d 555, 556 (D.C. Cir. 1995). In particular, we have
explained, “trial judges are not free to withhold approval of
guilty pleas . . . merely because their conception of the public
interest differs from that of the prosecuting attorney.” United
States v. Ammidown, 497 F.2d 615, 622 (D.C. Cir. 1973).
     In addition, a district court’s authority to “accept” or
“reject” a proposed plea agreement under Rule 11 is rooted in
the Judiciary’s traditional power over criminal sentencing, as
the Rule itself indicates in permitting the court to “defer a
decision until the court has reviewed the presentence report.”
Fed. R. Crim. P. 11(c)(3)(A). Plea agreements can take
roughly two forms: (i) charge bargains, in which a defendant
agrees to plead guilty to certain charges in exchange for the
dismissal of other charges; and (ii) sentence bargains, in
which the defendant agrees to plead guilty to a particular
                              19
charge after the parties agree upon a sentence, which the
prosecution then recommends to the sentencing court. See
United States v. Robertson, 45 F.3d 1423, 1437 (10th Cir.
1995). In light of the Executive’s traditional power over
charging decisions and the Judiciary’s traditional authority
over sentencing decisions, see Ammidown, 497 F.2d at 619,
some of our sister circuits have concluded that district courts
have more limited authority to reject charge bargains than
sentence bargains. See Robertson, 45 F.3d at 1439; In re
Vasquez-Ramirez, 443 F.3d 692, 697-98 (9th Cir. 2006).
Regardless, even in the case of a charge bargain, the court
reviews the defendant’s admitted conduct and enters a
judgment of conviction, which in turn carries immediate
sentencing implications.
    The context of a DPA is markedly different. Unlike a
plea agreement—and more like a dismissal under Rule
48(a)—a DPA involves no formal judicial action imposing or
adopting its terms. Whereas a district court enters a judgment
of conviction and then imposes a sentence in the case of a
plea agreement, the court takes no such actions in the case of
a DPA. Rather, the entire object of a DPA is to enable the
defendant to avoid criminal conviction and sentence by
demonstrating good conduct and compliance with the law.
And a DPA’s provisions are agreed to by the parties, not the
court, with no occasion for the court to adopt the agreement’s
terms as its own. The court never exercises its coercive
power by entering a judgment of conviction or imposing a
sentence. It instead merely approves the prosecution’s
judgment that further pursuit of criminal charges is
unwarranted, as it does when it approves a prosecutor’s
motion to dismiss charges under Rule 48(a). And as is the
case when confronted with a motion to dismiss charges under
Rule 48(a), a district court lacks authority to disapprove a
                              20
DPA under § 3161(h)(2) on the ground that the prosecution
has been too lenient in its exercise of charging discretion.
                              E.
     Judged by those principles, the district court in this case
erred in denying the parties’ motion for exclusion of time
under § 3161(h)(2). There is no indication that the parties
entered into the DPA to evade speedy trial limits rather than
to enable Fokker to demonstrate its good conduct and
compliance with law. Rather, the district court denied the
exclusion of time based on its view that the prosecution
should have brought different charges or sought different
remedies. In doing so, the court exceeded its authority under
§ 3161(h)(2).
     From the first status conference concerning the DPA, the
district court repeatedly criticized the government for failing
to bring charges against individual company officers. See
Fokker Services, B.V., 79 F. Supp. 3d at 166; Tr. of Status
Conference (June 25, 2014), at 4; Tr. of Status Conference
(July 9, 2014), at 5. Noting its belief that illegal conduct had
been “orchestrated at the highest levels of the company,” 79
F. Supp. 3d at 166, and unpersuaded by the government’s
efforts to ground its charging decisions in traditional
prosecutorial considerations such as the strength of the
evidence and the value of pursuing of different charges, e.g.,
Gov’t Mem. in Support of DPA Reached with Fokker
Services, B.V., at 18-19, the district court questioned why no
individuals would be held separately accountable. 79 F.
Supp. 3d at 166. The court also faulted the government for
“not requiring Fokker Services to pay as its fine a penny more
than the $21 million in revenue it collected from its illegal
transactions.” Id. In addition, the court thought the
prosecution should have required an independent monitor as
                               21
part of the DPA’s terms. See id. The district court denied the
motion for the exclusion of time for those reasons.
     Even if the district court’s criticisms of the prosecution’s
exercise of charging authority were entirely meritorious—an
issue we have no occasion to address—the court should not
have “assume[d] the role of Attorney General,” Microsoft, 56
F.3d at 1462. Rather, the court should have confined its
inquiry to examining whether the DPA served the purpose of
allowing Fokker to demonstrate its good conduct, as
contemplated by § 3161(h)(2). There is no reason to question
the DPA’s bona fides in that regard, see Deferred Prosecution
Agreement, at 4-7; Gov’t Supp. Mem. in Support of DPA
Reached with Fokker Services, B.V., at 15-16, and the district
court made no suggestion otherwise. And insofar as a court
has authority to reject a DPA if it contains illegal or unethical
provisions, see United States v. Saena Tech Corp., 2015 WL
6406266, at *17-19 (D.D.C. Oct. 21, 2015); United States v.
HSBC Bank USA, N.A., 2013 WL 3306161, at *7 (E.D.N.Y.
July 1, 2013), the district court again made no such suggestion
here. The court instead denied the exclusion of time under
§ 3161(h)(2) based on a belief that the prosecution had been
unduly lenient in its charging decisions and in the conditions
agreed to in the DPA. The court significantly overstepped its
authority in doing so.
                              III.
     Having determined that the district court erred in denying
the motion to exclude time, we now decide whether to grant a
writ of mandamus to correct that error. See 28 U.S.C. § 1651;
Kellogg Brown & Root, 756 F.3d at 760. Mandamus is a
“drastic and extraordinary” remedy “reserved for really
extraordinary cases.” Cheney v. U.S. District Court for the
District of Columbia, 542 U.S. 367, 380 (2004). Before a
court may issue the writ, three conditions must be satisfied:
                                22
(i) the petitioner must have “no other adequate means to attain
the relief he desires”; (ii) the petitioner must show that his
right to the writ is “clear and indisputable”; and (iii) the court
“in the exercise of its discretion, must be satisfied that the writ
is appropriate under the circumstances.” Id. at 380-81. All of
those conditions are met here.
                                A.
    First, a mandamus petitioner must lack any “other
adequate means to attain the relief he desires.” Id. at 380.
That condition is satisfied in this case because interlocutory
appeal is unavailable, and appeal after final judgment would
be an inadequate form of relief.
     With respect to the possibility of interlocutory appeal, the
defendant in a criminal case generally has no ability to obtain
appellate review of an interlocutory order until she has been
convicted and sentenced.         See, e.g., United States v.
MacDonald, 435 U.S. 850, 857-61 (1978). The government
has a statutory right to contest certain kinds of interlocutory
orders in criminal cases. See 18 U.S.C. § 3731. But because
the denial of a speedy trial exclusion does not fall within that
statutory exception, we must determine whether the district
court’s order in this case is immediately appealable under the
collateral order doctrine. We find it is not.
      The collateral order doctrine serves as a limited exception
to the final judgment rule. See Cohen v. Beneficial Industrial
Loan Corp., 337 U.S. 541 (1949). In order to fall within the
exception, an interlocutory order must (i) “conclusively
determine the disputed question”; (ii) “resolve an important
issue completely separate from the merits of the action”; and
(iii) “be effectively unreviewable on appeal from a final
judgment” in the underlying action. Midland Asphalt Corp. v.
United States, 489 U.S. 794, 799 (1989) (quoting Coopers &
                               23
Lybrand v. Livesay, 437 U.S. 463, 468 (1978)). All three of
those conditions must be satisfied for the ruling in question to
qualify as an immediately-appealable collateral order. And
because delay can be “fatal to the vindication of the criminal
law,” Cobbledick v. United States, 309 U.S. 323, 324 (1940),
those conditions are applied “with the utmost strictness” in
criminal cases, Flanagan v. United States, 465 U.S. 259, 265
(1984).
     Here, we need not consider the first two prongs of the
collateral order test because the third one alone precludes
finding that the district court’s ruling is an immediately-
appealable collateral order. The Supreme Court has explained
that “only a narrow group of claims” satisfies the condition of
effective unreviewability. United States v. Hollywood Motor
Car Co., Inc., 458 U.S. 263, 270 (1982). Generally, the order
in question must implicate “an asserted right the legal and
practical value of which would be destroyed if it were not
vindicated before trial.” MacDonald, 435 U.S. at 860.
Fokker and the government have different interests at stake in
this appeal, and we consider them separately.
      Fokker asserts that it possesses a right not to be tried,
conferred by the DPA, which would be nullified if review
were postponed until after final judgment. It is well
established, however, that the “mere burden of submitting to
trial proceedings that will be wasted if the appellant’s position
is correct does not support collateral order appeal.” 15A
Wright, Miller & Cooper, Federal Practice and Procedure:
Jurisdiction § 3911.4 (2d ed. 1992). Because Fokker’s
purported right to avoid trial does not “rest[] upon an explicit
statutory or constitutional guarantee that trial will not occur,”
it does not satisfy the requirement of effective
unreviewability. Midland Asphalt Corp., 489 U.S. at 800-01;
see Digital Equipment Corp. v. Desktop Direct, Inc., 511 U.S.
863, 877-78 (1994).
                               24
      For its part, the government emphasizes that its right to
an immediate appeal is grounded in the separation of powers.
The government, however, fails to establish that the order
falls within the “narrow group of claims” that qualify as
effectively unreviewable. Hollywood Motor Car Co., 458
U.S. at 270. In most criminal cases, the government is barred
from appealing if it loses at trial and lacks standing to appeal
if it prevails. As a result, if an interlocutory order infringes
the separation of powers, the Executive ordinarily would be
unable to vindicate its prerogatives after final judgment.
     But here, the government could allow the speedy trial
clock to expire and then appeal the district court’s resulting
dismissal of charges on the ground that the court should have
granted the requested exclusion of time under § 3161(h)(2).
To the extent the government’s separation-of-powers claim is
meritorious—i.e., the district court acted impermissibly in
rejecting the DPA based on a disagreement with the
prosecutor’s charging decisions—the government could
vindicate its claim after final judgment. For that reason, the
district court’s order fails to qualify as “effectively
unreviewable on appeal from a final judgment,” Midland
Asphalt Corp., 489 U.S. at 799, for purposes of the collateral
order doctrine.
      While the availability of appeal after final judgment
precludes treating the district court’s order as an immediately-
appealable collateral order, the possibility of review after final
judgment is not an “adequate means to attain the relief [the
government] desires” so as to prevent the grant of mandamus.
Cheney, 542 U.S. at 380 (emphasis added). The district
court’s rejection of the DPA essentially left the government
with three options: renegotiate the agreement; proceed to a
trial or plea; or allow the speedy trial clock to run and then
appeal the resulting dismissal of charges. Neither of the first
two options would permit appellate review of the district
                              25
court’s rejection of the DPA. As for the third option,
allowing the speedy trial clock to run would enable review of
the order, but would come with its own attendant risks. If the
district court were to dismiss the case with prejudice, and the
government were unsuccessful on appeal, the government
might be unable to refile charges against an admittedly guilty
defendant. If the district court instead were to dismiss the
case without prejudice, the government still might be unable
to re-indict because of the statute of limitations. The
possibility that the government would be left with no remedy
against a culpable defendant inflicts an “irreparable injury”
that “will go unredressed” without mandamus relief. In re al-
Nashiri, 791 F.3d 71, 79 (D.C. Cir. 2015).

     Fokker (but not the government) separately argues that
the district court’s order amounts to an immediately-
appealable denial of an injunction for purposes of 28 U.S.C. §
1292(a)(1), which permits appeal of orders granting or
denying injunctions. By rejecting the DPA, Fokker claims,
the district court’s order had the practical effect of denying
injunctive relief. We disagree. To qualify as an injunction,
the requested relief must be “directed to a party, enforceable
by contempt, and designed to accord or protect some or all of
the substantive relief sought by a complaint.” I.A.M. Nat’l
Pension Fund Benefit Plan A v. Cooper Industries, Inc., 789
F.2d 21, 24 (D.C. Cir. 1986) (internal quotation marks
omitted); see 16 Wright, Miller & Cooper, Federal Practice
and Procedure: Jurisdiction § 3922 (3d ed. 2012). Here,
however, if the district court had granted the motion to
exclude time and accepted the DPA, its order would not have
been enforceable by contempt. Although Fokker could face
significant consequences—namely, criminal prosecution—if
it were to violate the terms of the DPA, it would not confront
judicial sanctions. As a result, the district court’s refusal to
exclude time did not amount to an immediately-appealable
                               26
denial of an injunction, leaving the government with no
“adequate means to attain the relief [it] desires” apart from
mandamus. Cheney, 542 U.S. at 380.
                               B.
     A mandamus petitioner must demonstrate that its right to
the writ is “clear and indisputable.” Cheney, 542 U.S. at 381.
For the reasons explained in Part II, we conclude that the
district court’s decision “constitutes a clear legal error.”
Kellogg Brown & Root, 756 F.3d at 762.
     It is true that, at the time of the district court’s action,
there was no appellate opinion specifically construing the
scope of a district court’s authority under 18 U.S.C.
§ 3161(h)(2). But we have never required the existence of a
prior opinion addressing the precise factual circumstances or
statutory provision at issue in order to find clear error
justifying mandamus relief. Indeed, the reason there is no
appellate opinion interpreting a district court’s authority under
§ 3161(h)(2) is that, before the decision under review, no
district court had denied a motion to exclude time based on a
mere disagreement with the prosecution’s charging decisions.
In fact, as far as we can tell, no district court had denied a
motion to exclude time under § 3161(h)(2) for any reason.
Conversely, numerous decisions of the Supreme Court and
this court made clear that courts generally lack authority to
second-guess the prosecution’s constitutionally rooted
exercise of charging discretion. See, e.g., Wayte, 470 U.S. at
607-08; Microsoft, 56 F.3d at 1460-63; Ammidown, 497 F.2d
at 621-22. Mandamus serves as a check on that kind of
“usurpation of judicial power.” Bankers Life & Cas. Co. v.
Holland, 346 U.S. 379, 383 (1953).
                              27
                              C.
    Finally, we “must be satisfied that the writ is appropriate
under the circumstances.” Cheney, 542 U.S. at 381. In this
case, the “totality of the circumstances” warrants granting
mandamus. Kellogg Brown & Root, 756 F.3d at 762.
     The order under review marks the first time a DPA
negotiated by the government has been subjected to judicial
scrutiny of the prosecution’s basic exercise of charging
discretion. DPAs have become an increasingly important tool
in the government’s efforts to hold defendants accountable.
They afford prosecutors an intermediate alternative between,
on one hand, allowing a defendant to evade responsibility
altogether, and, on the other hand, seeking a conviction that
the prosecution may believe would be difficult to obtain or
would have undesirable collateral consequences for the
defendant or innocent third parties. The agreements also give
prosecutors the flexibility to structure arrangements that, in
their view, best account for the defendant’s culpability and
yield the most desirable long-term outcomes.
    By rejecting a central component of the resolution
reached between a number of federal enforcement agencies
and the defendant company, the district court’s ruling “cannot
but have enormous practical consequences for the
government’s ability to negotiate future settlements,”
Microsoft, 56 F.3d at 1456, and could have “potentially far-
reaching consequences” for prosecutors’ ability to pursue—
and fashion the terms of—DPAs, Kellogg Brown & Root, 756
F.3d at 762. The order thus amounts to “an unwarranted
impairment of another branch in the performance of its
constitutional duties.” Cheney, 542 U.S. at 390. In short, the
“novelty of the District Court’s . . . ruling, combined with its
potentially broad and destabilizing effects in an important
area of law,” justify granting the government’s petition for a
                               28
writ of mandamus. Kellogg Brown & Root, 756 F.3d at 763
(quoting Cheney, 542 U.S. at 381).
                               IV.
     Fokker Services requests that we reassign this case to a
different district court judge. But the petition for a writ of
mandamus, which was brought by the government, does not
request reassignment. Although the party seeking the relief
we grant has not asked for reassignment, we briefly consider
the issue because appellate courts will on occasion reassign a
case sua sponte. See 28 U.S.C. §§ 455(a), 2106.
     Reassignment is warranted only in the “exceedingly rare
circumstance,” Kellogg Brown & Root, 756 F.3d at 763, in
which the district judge’s conduct is “so extreme as to display
clear inability to render fair judgment,” Liteky v. United
States, 510 U.S. 540, 551 (1994). This case does not
approach that high bar.           Although the district court
volunteered opinions about Fokker’s conduct on the basis of
facts presented during the proceedings, those sorts of “candid
reflections” concerning the judge’s assessment of a
defendant’s conduct “simply do not establish bias or
prejudice.” In re Barry, 946 F.2d 913, 914 (D.C. Cir. 1991);
see SEC v. First City Fin. Corp., 890 F.2d 1215, 1222 (D.C.
Cir. 1989). Nor do the district court’s observations suggest
“deep-seated . . . antagonism that would make fair judgment
impossible.” Liteky, 540 U.S. at 555. We see no reason to
doubt the district court’s ability to render fair judgment going
forward. We therefore decline to reassign the case.
                       *   *   *     *   *
    For the foregoing reasons, we vacate the district court’s
order and remand for further proceedings consistent with this
opinion.