CALABRESI, Circuit Judge:

Plaintiffs below appeal from the decision of the United States District Court for the Southern District of New York (Stein J.) granting Defendants' motion for summary judgment and denying Plaintiffs' motion for summary judgment. The District Court's decision is VACATED and REMANDED for further proceedings consistent with this opinion.

BACKGROUND

Plaintiffs-Appellants Elaine Clark, Raymond Giangrasso, Tony Gonzales, Johnny L. Heatherman, and Monell White brought suit in the United States District Court for the Southern District of New York (Stein, J.) against Defendants-Appellees the Social Security Administration and its Commissioner, Michael J. Astrue, alleging that the Administration has suspended Old-Age, Survivor, and Disability Insurance and Supplemental Security Income benefits in a manner inconsistent with the authorizing statute. The District Court granted Defendants' motion for summary judgment and denied Plaintiffs' motion for summary judgment.

I. A.

Title XVI and Title II of the Social Security Act provide benefits for qualifying individuals. 42 U.S.C. §§ 402, 1382. Title XVI's Social Security Income (SSI) program provides benefits to "each aged, blind, or disabled individual who does not have an eligible spouse" and whose income and resources fall below a certain level. 42 U.S.C. § 1382(a). Title IPs Old-Age, Survivor, and Disability Insurance (OASDI) program provides benefits to insured individuals, independent of need, who are, among other things, at least 62 years of age. 42 U.S.C. § 402.

Congress has passed two provisions authorizing the Social Security Administration to suspend the benefits of an individual who is violating the terms of her probation or parole (hereinafter "suspension provisions"). As part of the Personal Responsibility and Work Opportunity Reconciliation Act of 1996, Pub.L. No. 104-193, § 202, 110 Stat. 2105, 2185-86, Congress passed the SSI program's suspension provision, which states:

No person shall be considered an eligible individual or eligible spouse for purposes of this subchapter with respect to any month if during such month the person is —

(i) fleeing to avoid prosecution, or custody or confinement after conviction, under the laws of the place from which the person flees, for a crime, or an attempt to commit a crime, which is a felony under the laws of the place from which the person flees, or, in jurisdictions that do not define crimes as felonies, is punishable by death or imprisonment for a term exceeding 1 year regardless of the actual sentence imposed; or

(ii) violating a condition of probation or parole imposed under Federal or State law.

42 U.S.C. § 1382(e)(4)(A) (emphasis added). Congress passed a similar provision for the OASDI program in the Social Security Protection Act of 2004, Pub.L. No. 108-203, § 203, 118 Stat. 493, 509:

(1)(A) Notwithstanding any other provision of this subchapter, no monthly benefits shall be paid under this section or under section 423 of this title to any individual for any month ending with or during or beginning with or during a period of more than 30 days throughout all of which such individual —

(i) is confined in a jail, prison, or other penal institution or correctional facility pursuant to his conviction of a criminal offense,

(ii) is confined by court order in an institution at public expense in connection with —

(I) a verdict or finding that the individual is guilty but insane, with respect to a criminal offense,

(II) a verdict or finding that the individual is not guilty of such an offense by reason of insanity,

(III) a finding that such individual is incompetent to stand trial under an allegation of such an offense, or

(IV) a similar verdict or finding with respect to such an offense based on similar factors (such as a mental disease, a mental defect, or mental incompetence),

(iii) immediately upon completion of confinement as described in clause (i) pursuant to conviction of a criminal offense an element of which is sexual activity, is confined by court order in an institution at public expense pursuant to a finding that the individual is a sexually dangerous person or a sexual predator or a similar finding,

(iv) is fleeing to avoid prosecution, or custody or confinement after conviction, under the laws of the place from which the person flees, for a crime, or an attempt to commit a crime, which is a felony under the laws of the place from which the person flees, or, in jurisdictions that do not define crimes as felonies, is punishable by death or imprisonment for a term exceeding 1 year regardless of the actual sentence imposed, or

(v) is violating a condition of probation or parole imposed under Federal or State law.

42 U.S.C. § 402(x)(1)(A) (emphasis added).

In 2004, Congress also provided for circumstances in which suspended benefits should be reinstated (hereinafter "good-cause provisions"). See Social Security Protection Act of 2004, Pub.L. 108-203, § 203, 118 Stat, at 509. The good-cause provision as it applies to the OASDI's suspension provision reads as follows:

(iii) Notwithstanding subparagraph (A), the Commissioner shall, for good cause shown, pay the individual benefits that have been withheld or would otherwise be withheld pursuant to clause (iv) or (v) of subparagraph (A) if the Commissioner determines that —

(I) a court of competent jurisdiction has found the individual not guilty of the criminal offense, dismissed the charges relating to the criminal offense, vacated the warrant for arrest of the individual for the criminal offense, or issued any similar exonerating order (or taken similar exonerating action), or

(II) the individual was erroneously implicated in connection with the criminal offense by reason of identity fraud.

42 U.S.C. § 402(x)(1)(B)(iii) (emphasis added). The SSI program's good-cause provision is to the same effect:

(B) Notwithstanding subparagraph (A), the Commissioner shall, for good cause shown, treat the person referred to in subparagraph (A) as an eligible individual or eligible spouse if the Commissioner determines that —

(i) a court of competent jurisdiction has found the person not guilty of the criminal offense, dismissed the charges relating to the criminal offense, vacated the warrant for arrest of the person for the criminal offense, or issued any similar exonerating order (or taken similar exonerating action), or

(ii) the person was erroneously implicated in connection with the criminal offense by reason of identity fraud. 42 U.S.C. § 1382(e)(4)(B) (emphasis added).

B.

On June 30, 2000, the Social Security Administration enacted a regulation implementing the 1996 suspension provision. 65 Fed.Reg. 40, 493-94. That regulation states:

(1) Suspension of benefit payments because an individual is a fugitive as described in paragraph (a)(1) or (a)(2) of this section or a probation or parole violator as described in paragraph (a)(3) of this section is effective with the first day of whichever of the following months is earlier —

(i) The month in which a warrant or order for the individual's arrest or apprehension, an order requiring the individual's appearance before a court or other appropriate tribunal (e.g., a parole board), or similar order is issued by a court or other duly authorized tribunal on the basis of an appropriate finding that the individual —

(A) Is fleeing, or has fled, to avoid prosecution as described in paragraph (a)(1) of this section;

(B) Is fleeing, or has fled, to avoid custody or confinement after conviction as described in paragraph (a)(2) of this section;

(C) Is violating, or has violated, a condition of his or her probation or parole as described in paragraph (a)(3) of this section[.]

20 C.F.R. § 416.1339(b). The Administration has not adopted an analogous implementing regulation for the OASDI suspension provision.

C.

The Social Security Administration further construed Congress's suspension provisions in its sub-regulatory Program Operations Manual System (POMS). The POMS states that SSI benefits should be suspended under the following circumstances:

The month in which a warrant, a court order or decision, or an order or decision by an appropriate agency in the United States (e.g. parole board) is issued which finds that the individual is wanted in connection with a crime that is a felony or for attempting to commit a crime that is a felony or is violating a condition of his or her probation or parole[.]

POMS § SI 00530.010(A). Suspended SSI benefits may be reinstated "effective with the first month throughout which the individual does not have an active warrant pending for a . . . violation of a condition of his or her parole or probation." Id. at § 00530.020(A).

With respect to the OASDI suspension provision, the POMS states:

Unless good cause is found . . . payment is prohibited for any month in which a beneficiary (including juveniles) has an unsatisfied warrant for more than 30 continuous days for: . . . [v]iolation of a condition of Federal or State probation/parole.

POMS § GN 02613.010(A)(1). And as is the case for SSI benefits, OASDI benefits may be reinstated "effective with the first month throughout which the individual does not have: . . . a warrant pending for violation of a condition of parole or probation." POMS § GN 02613.500(A)(1).

D.

Defendant Michael J. Astrue explains in his brief to this Court the process by which the Administration suspends SSI and OASDI benefits based on a determination that an individual is violating the terms of her probation or parole. The Administration first collects warrant data from federal, state, and local law enforcement agencies. Using the Federal Bureau of Investigation's National Crime Information Center uniform offense codes, the Administration identifies those warrants issued on the basis of an alleged parole or probation violation. Next, the Administration verifies the identities of the individuals against whom the warrants were issued and, once verified, matches those names with individuals receiving SSI and OASDI benefits. The names are forwarded to the Office of the Inspector General (OIG). The OIG then sends to the law enforcement agencies a form that requests verification (1) of the code offense, (2) that "reasonable measures" were used in the warrant's issuance to avoid identity fraud, and (3) that the warrant remains open.

After the Administration verifies a warrant, it sends a notice to the benefits recipient. For OASDI recipients, the notice states that the recipients have 30 days to advise the Administration that they wish to protest the suspension by showing good cause why benefits should not be suspended. For SSI beneficiaries, the notice states that recipients have 10 days to advise the Administration that they wish to show good cause why their benefits should not be suspended. As to both OASDI and SSI benefits, if the recipients do not respond, the Administration suspends their benefits. If the recipients do respond, they have 90 days to show "good cause." POMS §§ GN 002613.025(B)(3)(c), GN 02613.030, SI 00530.015(B)(3)(c). Benefits will continue for 90 days or until the Administration determines that the recipient has failed to show good cause to keep benefits from being suspended. Although recipients may appeal the initial good-cause determination through multiple levels of administrative review, and ultimately to a federal court, see 42 U.S.C. § 405(g), 1383(c)(3); 20 C.F.R. § 404.981, 416.1481, benefits remain suspended throughout the appeal process.

Individuals with an arrest warrant for an alleged parole or probation violation can also have their benefits reinstated by showing good cause.

Two types of good cause exist: mandatory and discretionary. A recipient can establish mandatory good cause by showing, among other things, a document from the issuing court or tribunal indicating that a warrant was issued in error or that the warrant was vacated. POMS §§ GN 02613.025(B), SI 00530.015(B). Discretionary good cause, in contrast, does not require showing that the warrant has been vacated or quashed. To be eligible, recipients must first demonstrate (1) that the underlying offense and the subsequent parole or probation violation were nonviolent and not drug related; and (2) that they have not committed any felony crimes since the warrant was issued. In addition, they have to satisfy one of two other sets of requirements. Recipients can either provide evidence that the law enforcement agency that issued the warrant is unwilling to act on the warrant, or recipients can show (a) that the warrant is the only existing warrant and is at least 10 years old; and (b) that the beneficiary lacks the mental capacity to resolve a warrant, is incapable of managing payments, is legally incompetent, has appointed a representative payee to handle his payments, or is residing in a long-term care facility. Id.

II.

Plaintiffs are all SSI or OASDI benefits recipients whose benefits were suspended on the basis of an arrest warrant asserting that they were violating a term of their probation or parole. Three of the five plaintiffs were ultimately found not to have been violating a condition of their probation or parole. One plaintiffs case remains unresolved. The fifth plaintiff died during the course of these proceedings, also with her case unresolved.

III.

Plaintiffs filed this suit in the United States District Court for the Southern District of New York (Stein, J.) on December 28, 2006. Shortly after the filing of the complaint, two plaintiffs moved for a temporary restraining order and all the plaintiffs moved for a preliminary injunction. The District Court denied the motions and concluded with respect to the latter motion that Plaintiffs were not likely to succeed on the merits.

Following discovery, the parties cross-moved for summary judgment, which the District Court granted to Defendants. See Clark v. Astrue, No. 06 Civ. 15521, 2008 WL 4387709 (S.D.N.Y. Sept.22, 2008). The Court concluded that "[t]he suspension provisions themselves are silent as to how the [Administration] is to determine whether a Social Security recipient `is violating' probation or parole." Id. at *7. But it found support for Defendants' position in the 2004 good-cause provisions. The Court reasoned that because Congress enacted the good-cause provisions, which authorize the Administration to reinstate suspended benefits when the warrant for a criminal offense is vacated, "it follows that Congress also contemplated that a warrant constitutes a sufficient basis on which to suspend benefits in the first instance." Id. The District Court rejected Plaintiffs' argument that because the good-cause provisions apply only to warrants for a "criminal offense," they do not apply to suspensions on the basis of a parole or probation violation, as these do not constitute an independent criminal offense. The District Court stated that "there is . . . no indication in the statute that Congress intended to draw a distinction between warrants for criminal offenses and warrants for probation and parole violations." It also stated that under Plaintiffs' reading, the good-cause provisions would be almost entirely inapplicable to benefits suspended on the basis of parole or probation violations. Id.

The District Court further found that this Court's decision in Fowlkes v. Adamec, 432 F.3d 90 (2d Cir. 2005), somewhat paradoxically, supported Defendants' interpretation of the suspension provisions. Fowlkes addressed a different ground on which the Administration can suspend SSI and OASDI benefits: when a recipient is fleeing prosecution, custody, or confinement for a felony criminal offense. Prior to Fowlkes, the Administration had a policy of suspending benefits solely on the basis of the issuance of an arrest warrant for a felony offense. In Fowlkes, we held that the Administration's practice, as embodied in its POMS, is contrary to the plain meaning of the Act because the Act's use of the word "fleeing" "is understood to mean the conscious evasion of arrest or prosecution." Id. at 96. Because an arrest warrant can issue without a finding that one has such intent, the fact of a warrant is an insufficient basis on which to suspend benefits. See id. at 97. In contrast with Plaintiffs' reliance on Fowlkes, the District Court distinguished the case on the ground that the warrant here is not over-inclusive, and that in Fowlkes, "the Second Circuit gave no indication that it would take issue with the [Administration's] suspension of benefits on the basis of a warrant that did indicate that the recipient was fleeing to avoid prosecution." Clark, 2008 WL 4387709, at *10.

The District Court additionally held that even if the Administration's interpretation is not compelled by the plain language of the Act, its interpretation of the Act, whether embodied in a regulation or in its POMS, is entitled to "substantial deference." Id. at *8. Because the Court found that the Administration's practice is consistent with its POMS and implementing regulation, both of which reasonably interpret the Act, it concluded that Plaintiffs' claims must fail. Id. at *9-11.

DISCUSSION I.

The Administration's current practice takes a warrant alleging a probation or parole violation as sufficient and irrebuttable evidence that one is violating the conditions of probation or parole. We review an agency's interpretation of its own authorizing statute according to the two-step framework from Chevron U.S.A., Inc. v. Natural Res. Def. Council, Inc., 467 U.S. 837, 104 S.Ct. 2778, 81 L.Ed.2d 694 (1984). We recently explained this analysis as follows:

At step one, we consider whether Congress has directly spoken to the precise question at issue. If the intent of Congress is clear, that is the end of the matter; for the court, as well as the agency, must give effect to the unambiguously expressed intent of Congress. To ascertain Congress's intent, we begin with the statutory text because if its language is unambiguous, no further inquiry is necessary. Only if we determine that Congress has not directly addressed the precise question at issue will we turn to canons of construction and, if that is unsuccessful, to legislative history to see if those interpretative clues permit us to identify Congress's clear intent.

If, despite these efforts, we still cannot conclude that Congress has directly addressed the precise question at issue, we will proceed to Chevron step two, which instructs us to defer to an agency's interpretation of the statute it administers, so long as it is reasonable.

Snell Island SNF LLC v. NLRB, 568 F.3d 410, 415 (2d Cir. 2009) (quoting N.Y. State Office of Children Family Servs. v. U.S. Dep't of Health Human Servs. Admin. for Children Families, 556 F.3d 90, 97 (2d Cir. 2009)).

In the suspension provisions, Congress authorizes the Social Security Administration to suspend SSI and OASDI benefits upon certain factual triggers. But what are those factual triggers and upon what showing can one determine that the facts underlying those triggers exist? Two triggers are relevant to this case. The trigger we addressed in Fowlkes requires a determination that one is, among other things, "fleeing to avoid prosecution . . . for a crime . . . which is a felony" (hereinafter "fugitive felon provision"). 42 U.S.C. §§ 1382(e)(4)(A)(i), 402(x)(1)(A)(iv). The trigger at issue here requires a determination that one "is violating a condition of probation or parole" (hereinafter "probation or parole violation provision"). 42 U.S.C. §§ 1382(e)(4)(A)(ii), 402(x)(1)(A)(v). The issue before us is whether the fact of a warrant, issued on the basis of "probable cause" or "reasonable suspicion" to believe that one is violating a condition of probation or parole, is equivalent to a determination that one is in fact violating a condition of probation or parole. We find that it is not and therefore that the Administration's practice is contrary to the plain meaning of the Act.

The law often speaks of facts, but it usually operates on probabilities. In the civil context, a finding that X is more likely than not true is the equivalent to a finding that X is true. See, e.g., Pulsifer v. Walker, 85 N.H. 434, 159 A. 426, 427 (1932) ("The court . . . found that `the probabilities are that the option of renewal clause was in the original draft as appears in the document as signed.' In legal effect this is equivalent to a definite finding that the proviso was in the original draft."). By contrast, "probable cause is a lower standard than preponderance of the evidence." United States v. Juwa, 508 F.3d 694, 701 (2d Cir. 2007); cf. Illinois v. Gates, 462 U.S. 213, 243 n. 13, 103 S.Ct. 2317, 76 L.Ed.2d 527 (1983) ("[P]robable cause requires only a probability or substantial chance of criminal activity, not an actual showing of such activity. By hypothesis, therefore, innocent behavior frequently will provide the basis for a showing of probable cause[.]").

The Administration argues that, in the context before us, it may treat something as true based on a factual probability below a preponderance of the evidence. Specifically, it asserts that pursuant to these statutes, Congress's requirement that X be true can be reduced to a requirement that one have probable cause to believe, or a reasonable suspicion that, X is true. While Congress may in some instances set a lower probabilistic threshold, we disagree that it has done so here. And we hold, more generally, that unless it specifies clearly to the contrary, when Congress provides that a fact triggers civil legal consequences, it is requiring a finding that the fact is more likely than not true.

When Congress has intended a lower probabilistic showing to trigger legal consequences, it has done so explicitly. For example, in a revocation-of-parole provision in the Parole Commission Phaseout Act of 1996, Pub.L. No. 104-232, 110 Stat. 3055, Congress expressly conditioned legal consequences on a determination of "probable cause to believe that [one] has violated a condition of [one's] parole." 18 U.S.C. § 4214(a)(1)(A) (expires in 2011); see also id. (stating that the Commission shall prepare a digest "upon a finding of probable cause"). Courts have tended to do the same. See, e.g., Morrissey v. Brewer, 408 U.S. 471, 483-88, 92 S.Ct. 2593, 33 L.Ed.2d 484 (1972) (distinguishing "an appropriate determination that the individual has in fact breached the conditions of parole" from the "`preliminary hearing' to determine whether there is probable cause or reasonable ground to believe that the arrested parolee has committed acts that would constitute a violation of parole conditions"); Calhoun v. N.Y. State Div. of Parole Officers, 999 F.2d 647, 652-54 (2d Cir. 1993) (finding it a violation of due process to detain a defendant based on a "finding of probable cause for a parole violation" but before a final revocation hearing).

The Administration does not argue that, in practice, warrants, like those before us, are generally based on evidence sufficient to satisfy a more-likely-than-not probabilistic threshold. As a result, we do not address what relevance, if any, such a showing would have. Plaintiffs, moreover, provide evidence to the contrary. They cite to the statements of several attorneys from various jurisdictions to support their claim that warrants are issued on the basis of "nothing more than an allegation." Appellants' Brief at 17-18 (discussing procedures in New York, Ohio, Florida, and California). And, Amici Curiae Empire Justice, et al., proffer numerous explanations for why a warrant, although perhaps adequately supported by probable cause, might not satisfy the higher more-likely-than-not evidentiary burden: A recipient might receive permission from her probation or parole officer to leave the state, but that permission might never have been properly recorded; probation-or parole-related documents might have been sent to the wrong address; fees duly paid by a recipient might not have been properly recorded; or the recipient might have qualified for a fee waiver. See Brief for Empire Justice, et al., as Amici Curiae Supporting Plaintiffs-Appellants, at 16-17, 20. Under any of these circumstances, the warrant might well be valid but the Administration would not be justified in suspending benefits.

It is nonetheless possible that a warrant, coupled with adequate procedural safeguards, could satisfy a more-likely-than-not probabilistic showing, and therefore constitute a "finding" within the meaning of the Act. A warrant might, in some circumstances, be sufficient to create a presumption of an actual violation, in which case the burden of showing no violation might shift to the recipient. The recipient, having been informed of the basis for the warrant and given an opportunity to present evidence, might then be required to demonstrate that, despite the warrant, she is not violating the conditions of her probation or parole. We do not address the sufficiency or necessity of these procedures here. We mention them only in response to the Administration's argument that, pursuant to Plaintiffs' interpretation of the Act, benefits may only be suspended after a final revocation hearing. Today's holding does not go so far.

II.

Defendants make several arguments why the Social Security Act should not be read to require a finding that it is more likely than not that one is violating a condition of probation or parole before benefits can be suspended: (1) such a requirement is more difficult to administer; (2) the suspension provisions' language suggests that Congress desired the Administration to act promptly; (3) Congress approved of the Administration's current practice in the 2004 good-cause provisions; and (4) we approved of the Administration's current practice in Fowlkes. We address these arguments in turn.

A.

Defendants argue that "the most objective and clearly ascertainable indication that a beneficiary `is violating' applicable conditions is the issuance of a violation warrant" and that the Social Security Administration is "not equipped to independently investigate and adjudicate whether each beneficiary . . . `is violating' any applicable conditions." Appellees' Br. at 29-30. We do not doubt that a warrant may be the most easily ascertainable indication that one is violating the conditions of her probation or parole. But if, as discussed above, that indication is insufficient, then the fact that it is easily ascertainable does not cure its insufficiency. And while it may be the case that the Administration is not equipped to determine on a case-by-case basis whether an individual is violating a condition of probation or parole, that is precisely what, we conclude, Congress has required it to do before it may suspend benefits. That Congress requires the Administration to do something it cannot easily do does not change the fact that Congress requires it.

B.

Defendants further argue that Congress's use of the present tense "is" contemplates prompter suspension than our interpretation of the suspension provisions permits. This argument is circular. True enough, as soon as one knows that a recipient is in violation, benefits should be suspended. But the question before us is at what point does one know that a violation has occurred? If under the relevant statute, read in an ordinary way, one cannot know that a violation has occurred until there is a determination that it is more likely than not that a violation has occurred, then to act sooner is to act prematurely. Once one determines that a recipient is more likely than not violating a condition of probation or parole, benefits can be suspended retroactively as of the date that the recipient began violating probation or parole. But prompter action is not appropriates.

C.

The District Court relied heavily on Congress's 2004 good-cause provisions, 42 U.S.C. §§ 402(x)(1)(B)(iii), 1382(e)(4)(B). These provisions require the Administration to reinstate benefits suspended on the basis of the fugitive felon provision or the probation or parole violation provision when certain conditions are met. Significantly, the good-cause provisions do not specify which "good causes" apply to benefits suspended on the basis of the fugitive felon provision and which apply to benefits suspended on the basis of the probation or parole violation provision. The good causes requiring reinstatement of benefits are:

(I) a court of competent jurisdiction has found the individual not guilty of the criminal offense, dismissed the charges relating to the criminal offense, vacated the warrant for arrest of the individual for the criminal offense, or issued any similar exonerating order (or taken similar exonerating action), or

(II) the individual was erroneously implicated in connection with the criminal offense by reason of identity fraud.

Id. Defendants argue that by including among the good causes situations in which a court has "vacated the warrant for arrest of the individual for the criminal offense," it necessarily approved of the Administration's practice of suspending benefits on the basis of a warrant that alleged a probation or parole violation. We disagree.

First, as discussed above, a warrant is certainly evidence that one is violating a condition of probation or parole. And in some circumstances a warrant might even be sufficient (although not irrebuttable) evidence. Because a warrant is relevant evidence of a violation, benefits may, in appropriate circumstances, be suspended based, either entirely or in part, on the existence of a warrant. The subsequent vacatur of such a warrant would of course undermine the ground on which the benefits were suspended. And this both explains and justifies Congress's decision to allow reinstatement of benefits upon the vacatur of the warrant. It does not follow from this, however, that Congress meant that all warrants are sufficient bases on which to suspend benefits.

Second, it is far from clear that the language on which Defendants and the District Court seize applies to probation or parole violations. Violating a condition of probation or parole is not generally an independent criminal offense. As the Supreme Court has stated:

[R]evocation of parole is not part of a criminal prosecution and thus the full panoply of rights due a defendant in such a proceeding does not apply to parole revocations. Parole arises after the end of the criminal prosecution, including imposition of sentence. Supervision is not directly by the court but by an administrative agency, which is sometimes an arm of the court and sometimes of the executive. Revocation deprives an individual, not of the absolute liberty to which every citizen is entitled, but only of the conditional liberty properly dependent on observance of special parole restrictions.

Morrissey, 408 U.S. at 480, 92 S.Ct. 2593 (internal citation omitted); see also United States v. Aspinall, 389 F.3d 332, 342 (2d Cir. 2004) (finding that "it has long been established that probation revocation, like parole revocation, is not a stage of a criminal prosecution" (internal quotation marks omitted)), abrogated on other grounds by United States v. Booker, 543 U.S. 220, 125 S.Ct. 738, 160 L.Ed.2d 621 (2005).; Knight v. Estelle, 501 F.2d 963, 964 (5th Cir. 1974) ("A parole revocation hearing is not a criminal proceeding. Its purpose is not to assess guilt or to assign blameworthy acts to the various discrete pigeonholes of the criminal laws. Rather it is held to determine whether the attempt by parole to restore the parolee to the ranks of the carriers and remove him from those of the carried has failed." (footnote omitted)).

For this reason, the warrants referred to in the good-cause provisions are probably best read to refer to warrants for felony criminal offenses, thereby providing the basis for reinstating benefits suspended under the fugitive felon provision. The District Court disagreed, noting, "Under plaintiffs' reading, the good-cause provisions would either be entirely inapplicable to persons who violated their probation or parole because the provisions refer exclusively to `criminal offenses' throughout, or else would apply only in the rare instances in which such persons were able to have their underlying convictions vacated." 2008 WL 4387709, at *8. We agree that the nearly exclusive concern of the good-cause provisions with fugitive felons is odd. But it is not so odd that it justifies reading out of the statute the word "criminal," which Congress used repeatedly in the relevant parts of the statutes.

D.

Finally, our decision in Fowlkes does not support Defendants' interpretation of the suspension provisions. Defendants argue, and the District Court reasoned, that because the defect in the warrant that this Court pointed out in Fowlkes does not apply to probation or parole violations, Fowlkes can be read as implicitly approving of Defendants' use of warrants as sufficient and irrebuttable evidence in such cases. This logic is problematic for two reasons. First, just because Fowlkes is distinguishable from this case does not mean that it supports Defendants' position. A court is not obligated to identify every problem in a problematic agency practice, nor every defect in a problematic warrant.

Second, the factual trigger in Fowlkes is significantly different from the factual trigger in this case. The trigger for the probation or parole violation provision is the actual violation of a condition of one's probation or parole. The trigger for the fugitive felon provision is, among other things, the act of fleeing a felony prosecution; it is not the act of committing the felony. Because of this difference, a warrant means something quite different in the two contexts. In the probation or parole violation context, a warrant is evidence — amounting perhaps to probable cause — that one is in fact violating a condition of one's probation or parole. In the fugitive felon context, an outstanding arrest warrant for a felony is effectively an element of the offense; it is arguably direct evidence that the recipient is in fact "wanted in connection with a crime that is a felony." POMS § SI 00530.010. We of course need not address whether, but for the defect in the Administration's interpretation of the provisions noted in Fowlkes, such an outstanding arrest warrant would always constitute more-likely-than-not evidence that one is in fact fleeing a felony prosecution. We simply note that the Administration, and this Court, might reasonably treat warrants differently in the two contexts.

III.

Because we find that the Social Security Administration's practice is contrary to the plain meaning of the Social Security Act, we do not address whether the Administration's practice is consistent with its own implementing regulation. We similarly do not need to address what degree of deference would be accorded to an agency's interpretation of an act when that interpretation is, as is arguably true in this case, expressed only in its operations manual and not in regulations implementing the act.

CONCLUSION

For the foregoing reasons, we hold that the Social Security Administration's practice of treating a warrant alleging that a recipient is violating a condition of probation or parole as sufficient and irrebuttable evidence that the recipient is in fact violating a condition of probation or parole is inconsistent with the plain meaning of the Social Security Act. We therefore vacate the District Court's grant of summary judgment for Defendant and its denial of summary judgment for Plaintiffs and remand for further proceedings consistent with this opinion.