Jeffrey J. BIGGS, Petitioner-Appellant, v. SECRETARY OF the CALIFORNIA DEPARTMENT OF CORRECTIONS AND REHABILITATION, Respondent-Appellee, 717 F.3d 678

OPINION BYBEE, Circuit Judge: After Jeffrey J. Biggs was convicted of murder and sentenced to twenty-five years to life in prison with the possibility of parole, California amended its constitution to give the Governor authority to review parole-board decisions for prisoners convicted of murder. Cal. Const, art. V, § 8(b). The parole board subsequently found Biggs suitable for parole, but then-Governor Arnold Schwarzenegger reversed the parole board’s decision. Biggs claims that retroactive application of the interim change to the California Constitution violates the Ex Post Facto Clause of the U.S. Constitution.

U.S. Const, art. I, § 10, cl. 1. We previously analyzed the constitutionality of the retroactive application of this same provision of the California Constitution in Johnson v. Gomez, where we upheld

article V, § 8(

b) against an Ex Post Facto Clause challenge. 92 F.3d 964, 965-68 (9th Cir.1996). After the Supreme Court’s most recent decision in the area, Garner v. Jones, 529 U.S. 244, 120 S.Ct. 1362, 146 L.Ed.2d 236 (2000), the California Supreme Court also upheld

article V, § 8(

b) against an Ex Post Facto Clause challenge. In re Rosenkrantz, 29 Cal.4th 616, 128 Cal.Rptr.2d 104, 59 P.3d 174, 191-201 (2002). Biggs claims that he is entitled to habeas relief because In re Rosenkrantz unreasonably applied Gamer in deciding that retroactive application of

article V, § 8(

b) did not violate the Ex Post Facto Clause of the U.S. Constitution, and the state court relied exclusively on Rosenkrantz in denying Biggs’ Ex Post Facto Clause claim. The district court denied Biggs’ habeas petition. We have jurisdiction pursuant to 28 U.S.C. §§ 1291 and 2253. We affirm. I In 1987, Biggs was convicted of murder and sentenced to twenty-five years to life in prison with the possibility of parole. In 1988, California passed Proposition 89, which added

section 8(

b) to

article V of the California Constitution, providing the Governor with authority to review parole-board decisions in any case in which the prisoner had been convicted of murder. See Cal. Const, art. V, § 8(b) (“No decision of the parole authority of this state with respect to the granting, denial, revocation, or suspension of parole of a person sentenced to an indeterminate term upon conviction of murder shall become effective for a period of 30 days, during which the Governor may review the decision subject to procedures provided by statute.

The Governor may only affirm, modify, or reverse the decision of the parole authority on the basis of the same factors which the parole authority is required to consider. The Governor shall report to the Legislature each parole decision affirmed, modified, or reversed, stating the pertinent facts and reasons for the action.”); see also Johnson, 92 F.3d at 965. In 2005, the parole board determined that Biggs was suitable for parole. Exercising the authority granted to him under

article V,

section 8(b), then-Governor Schwarzenegger reversed the parole board’s decision. Biggs unsuccessfully challenged the Governor’s decision through state habeas proceedings. The only state-court decision to address Biggs’ claim under the Ex Post Facto Clause was the Superior Court’s decision, which looked to the prior decision of the California Supreme Court in Rosenkrantz: The Rosenkrantz court considered at length the question of whether the Governor’s review and subsequent reversal of a grant of parole violated the ex post facto clauses of the state and federal Constitutions.

The court concluded that there was no ex post facto violation. Accordingly, the petition is denied as to the [ex post facto] claim. Order of Denial at 7, No. SC-14199A (Cal. App. Dep’t Super. Ct. Aug. 22, 2006) (internal citations omitted). Biggs then filed the instant federal habeas petition. While this federal habeas petition was pending, the parole board again found Biggs suitable for parole, and this time the Governor declined to review the parole board’s decision. Biggs was released on parole in August 2010, but the district court ruled that Biggs’ release did not render his habeas petition moot.

The district court denied habeas relief, and Biggs timely appealed. II We review a district court’s decision to grant or deny a state prisoner’s federal habeas petition de novo. Gill v. Ayers, 342 F.3d 911, 917 (9th Cir.2003). Because Biggs filed his federal habeas petition after April 24, 1996, the Antiterrorism and Effective Death Penalty Act of 1996 (“AED-PA”) applies. Lindh v. Murphy, 521 U.S. 320, 322, 327, 117 S.Ct. 2059, 138 L.Ed.2d 481 (1997).

Under AEDPA, we may only grant habeas relief if the adjudication of a claim on the merits in state court resulted in a state-court decision that (1) was “contrary to, or involved an unreasonable application of, clearly established Federal law, as determined by the Supreme Court of the United States,” or (2) was “based on an unreasonable determination of the facts in light of the evidence presented in the State court proceeding.” 28 Ü.S.C. § 2254(d). Biggs only challenges the state-court decision under the “unreasonable application” clause of 28 U.S.C. § 2254(d).

Under the “unreasonable application” clause, we may issue the writ “if the state .court correctly identifies the governing legal principle from [the Supreme Court’s] decisions but unreasonably applies it to the facts of the particular case.” Bell v. Cone, 535 U.S. 685, 694, 122 S.Ct. 1843, 152 L.Ed.2d 914 (2002). Even if we believe that the state court was incorrect in its application of Supreme Court law, we may not grant the writ unless the state court’s application of Supreme Court law was objectively unreasonable. Id.; see also Harrington v. Richter, — U.S. -, 131 S.Ct. 770, 785-86, 178 L.Ed.2d 624 (2011).

This is true even if the conclusion of the state court is inconsistent with, or even disagrees with, our own decisions. “Because AEDPA limits habeas relief to state decisions that offend clearly established federal law as set by the Supreme Court, a state court decision may not be overturned simply because of a conflict with circuit law.” Ortiz-Sandoval v. Clarke, 323 F.3d 1165, 1172 (9th Cir.2003) (internal quotation marks omitted); see also Marshall v.

Rodgers, — U.S. -, 133 S.Ct. 1446, 1450, 185 L.Ed.2d 540 (2013) (“[C]ircuit precedent may [not] be used to refine or sharpen a general principle of Supreme Court jurisprudence into a specific legal rule that th[e] Court has not announced.”). Ill The Ex Post Facto Clause of the U.S. Constitution prohibits the States from passing any “ex post facto Law.” U.S. Const, art. I, § 10, cl. 1. “In Collins v. Youngblood, ... [the Supreme Court] reaffirmed that the Ex Post Facto Clause incorporated ‘a term of art with an established meaning at the time of the framing of the Constitution.’” Cal. Dep’t of Corr. v.

Morales, 514 U.S. 499, 504, 115 S.Ct. 1597, 131 L.Ed.2d 588 (1995) (quoting Collins v. Youngblood, 497 U.S. 37, 41, 110 S.Ct. 2715, 111 L.Ed.2d 30 (1990)). In Collins, the Court explained the original meaning of the Ex Post Facto Clause: “Legislatures may not retroactively alter the definition of crimes or increase the punishment for criminal acts.” 497 U.S. at 43, 110 S.Ct. 2715. Since Collins was decided, the Court has addressed the bounds of the prohibition on laws that retroactively increase the punishment for criminal acts in two cases related to increased periods between parole hearings:

(1) California Department of Corrections v. Morales, decided in 1995, and

(2) Garner v. Jones, decided in 2000. Biggs points to these decisions in alleging that the state-court decision in his case, and the California Supreme Court’s decision in Rosenkrantz on which it relied, unreasonably applied clearly established federal law. A Morales involved a challenge to the retroactive application of a change in the California Penal Code that authorized the parole board to increase the gap between .parole suitability hearings for individuals convicted of “more than one offense which involves the taking of a life,” from one year, as previously required, to as long as three years.

Morales, 514 U.S. at 503-04, 115 S.Ct. 1597. To justify such an increase, the parole board had to (1) find that it was not reasonable to expect parole to be granted during the years with no hearings, and (2) state its bases for that finding. Id. at 503, 115 S.Ct. 1597. Morales urged the Court to hold “that the Ex Post Facto Clause forbids any legislative change that has any conceivable risk of affecting a prisoner’s punishment,” including the increase in the gap between parole hearings at issue. Id. at 508, 115 S.Ct. 1597.

The Court rejected this approach, noting that under such an approach “the judiciary would be charged under the Ex Post Facto Clause with the micromanagement of an endless array of legislative adjustments to parole and sentencing procedures.” Id. The Court asserted that it had “long held that the question of what legislative adjustments will be held to be of sufficient moment to transgress the [Ex Post Facto Clause] must be a matter of ‘degree.’ ” Id. at 509, 115 S.Ct. 1597 (internal quotation marks omitted).

In considering whether a change in law is constitutional under the Ex Post Facto Clause, courts “must determine whether [the change] produces a sufficient risk of increasing the measure of punishment attached to the covered crimes.” Id.

The Court declined to establish a bright-line rule, asserting that the change at issue in Morales did not require the Court to “articulate a single ‘formula’ ” for identifying changes in law that produce a sufficient risk of increased punishment, because the change at issue “create[d] only the most speculative and attenuated possibility of producing the prohibited effect of increasing the measure of punishment for covered crimes, and such conjectural effects are insufficient under any threshold [the Court] might establish.” Id.

The Court justified this determination by noting that the change “appliefd] only to a class of prisoners for whom the likelihood of release on parole is quite remote”—prisoners convicted of multiple crimes involving the taking of a life, id. at 510, 115 S.Ct. 1597—and that the gap between hearings would only be increased after the parole board held an initial hearing and made the requisite findings to delay the next hearing, id. at 511, 115 S.Ct. 1597.

The Court also noted that, in the case of a prisoner who experiences “an unanticipated change that is sufficiently monumental to alter [the prisoner’s] suitability for release on parole” after having the gap between hearings increased, the prisoner might be able to seek an expedited hearing based on that change. Id. at 512, 115 S.Ct. 1597. Finally, the Court noted that actual release dates often come several years after a finding of suitability, and thus that the practical effect of the increased gap between hearings was not significant. Id. at 513, 115 S.Ct. 1597.

The Court held that because the change in law created only “speculative and attenuated risk” of increased punishment, retroactive application of the law did not violate the Ex Post Facto Clause. Id. at 514, 115 S.Ct. 1597. The Court in Morales did not base its decision on how the change in law was actually implemented or applied, but rather considered the risk of increased punishment for prisoners based on a facial analysis of the change in law.

B Gamer also involved a challenge to the retroactive application of a change in law permitting an extension of the interval between parole hearings, this time in Georgia. 529 U.S. at 246, 120 S.Ct. 1362. Under Georgia law, the parole board was statutorily required to consider parole for inmates serving life sentences first after seven years of incarceration; by rule, the parole board had provided that it would reconsider parole every three years thereafter. Id. at 247, 120 S.Ct. 1362.

After the prisoner in Gamer committed his crime, the parole board amended its rules so that the gap between hearings after the initial hearing could be as long as eight years. Id. The prisoner was denied parole in 1995, and the parole board scheduled the next hearing for 2003, consistent with the change in its rules. Id. at 247-48, 120 S.Ct. 1362. The prisoner brought an action under the Ex Post Facto Clause, challenging the retroactive application of the change in law. Id. at 248, 120 S.Ct. 1362.

The Court acknowledged that the retroactive application of certain laws governing parole could unconstitutionally increase punishment under the Ex Post Facto Clause, id. at 249-50, 120 S.Ct. 1362, but noted that Morales “emphasiz[ed] that not every retroactive procedural change creating a risk of affecting an inmate’s terms or conditions of confinement is prohibited,” and that questions about such changes depend on the “degree” of the risk. Id.

The Court reiterated Morales’s message that “the Ex Post Facto Clause should not be employed for ‘the micromanagement of an endless array of legislative adjustments to parole and sentencing procedures.’” Id. at 252, 120 S.Ct. 1362 (quoting Morales, 514 U.S. at 508, 115 S.Ct. 1597). “The controlling inquiry ... [is] whether retroactive application of the change in ... law created ‘a sufficient risk of increasing the measure of punishment attached to the covered crimes.’ ” Id. at 250, 120 S.Ct. 1362 (quoting Morales, 514 U.S. at 509, 115 S.Ct. 1597).

In conducting this risk-based inquiry, derived from Morales, the Court stated that the change in law at issue had to be considered in “the whole context of Georgia’s parole system,” id. at 252, 120 S.Ct. 1362, which provides the parole board with broad discretion, id. at 252-53, 120 S.Ct. 1362.

The Court added that “[w]hether retroactive application of a particular change in parole law respects the prohibition on ex post facto legislation is often a question of particular difficulty when the discretion vested in a parole board is taken into account,” id. at 250, 120 S.Ct. 1362, even though this “discretion does not displace the protections of the Ex Post Facto Clause,” id. at 253, 120 S.Ct. 1362. “The essence of respondent’s case,” according to the Court, “[was] not that discretion has been changed in its exercise but that, in the period between parole reviews, it will not be exercised at all.” Id. at 254, 120 S.Ct. 1362.

The Court, however, rejected this claim because (1) the parole board maintained discretion as to how often to set hearings—subsequent hearings would take place “at least every eight years”; and (2) there were processes in place for expedited reviews in cases of changed circumstances. Id. (quoting Ga. Comp. R. & Regs. 473-3-05(2) (1985)). Given this continued discretion, the Court suggested that the risk of increased punishment, at least facially, was not sufficient to support a violation of the Ex Post Facto Clause. Id. at 254-55,120 S.Ct. 1362. But the Court’s analysis did not end there.

It went on: When the rule does not by its own terms show a significant risk, the respondent must demonstrate, by evidence drawn from the rule’s practical implementation by the agency charged with exercising discretion, that its retroactive application will result in a longer period of incarceration than under the earlier rule.... In the case before [the Court], respondent must show that as applied to his own sentence the law created a significant risk of increasing his punishment.

This remains the issue in the case, though the general operation of the Georgia parole system may produce relevant evidence and inform further analysis on this point.... The Court of Appeals’ analysis failed to reveal whether the amendment ..., in its operation, created a significant risk of increased punishment for respondent. Respondent claims he has not been permitted sufficient discovery to make this showing. Id. at 255-57,120 S.Ct. 1362. The Court stated that the “matter of adequate discovery” was one for the lower courts, and accordingly remanded. Id. at 257, 120 S.Ct. 1362.

The Court in Gamer thus applied the risk-based test from Morales, but did not stop with facial analysis as in Morales. Instead, the Court asserted that prisoners could demonstrate the significance of the risk of increased punishment through evidence of the implementation of the change in law, and that the prisoner was entitled to further discovery to support such a demonstration. It is this Supreme Court endorsement of an as-applied analysis of the significance of the risk of increased punishment that Biggs claims the state court unreasonably applied'—or, really, failed to apply—in his case.

IV Where, as here, the state’s highest court does not issue a reasoned decision, we must look to the last reasoned state-court decision and then determine if that decision runs afoul of 28 U.S.C. § 2254(d). Ylst v. Nunnemaker, 501 U.S. 797, 804-06, 111 S.Ct. 2590, 115 L.Ed.2d 706 (1991). As discussed above, the only reasoned state-court decision with regard to Biggs’ claim under the Ex Post Facto Clause was the Superior Court’s decision, which—understandably—relied exclusively on the California Supreme Court’s decision in Rosenkrantz.

Thus, we must examine Rosenkrantz to determine whether the Superior Court’s decision unreasonably applied clearly established federal law. In Rosenkrantz, the California Supreme Court was faced with precisely the same question we face: whether retroactive application of gubernatorial review under

article Y, § 8(

b) violates the Ex Post Facto Clause of the U.S. Constitution. 128 Cal. Rptr.2d 104, 59 P.3d at 183. The Rosenkrantz court, analyzing Morales and Garner, held that California’s scheme did not violate the Ex Post Facto Clause. Id. First, the court said that under the ordinary meaning of the phrase “increased the punishment,” it could not be reasonably said that the gubernatorial reversal violated the Ex Post Facto Clause. Id. 128 Cal.Rptr.2d 104, 59 P.3d at 193.

The court emphasized that the term of the petitioner’s sentence was the same before and after the implementation of review, and that the factors to be considered in determining whether to grant parole were left unchanged. Id. “The only change effected ... is the institution of an additional level of discretionary review of the Board’s decision granting or denying parole, resulting merely in a change in the identity of the entity or official within the executive branch that may make the ultimate decision on parole.” Id.

After pointing out that the Supreme Court had never held a change in law comparable to the institution of gubernatorial review of parole decisions to constitute “the type of measure to which the ex post facto clause applies,” id. 128 Cal.Rptr.2d 104, 59 P.3d at 193-94, the court explained that the petitioner’s claim had been “squarely and uniformly rejected by each of the prior California and federal decisions that have addressed the issue.” Id. 128 Cal.Rptr.2d 104, 59 P.3d at 194. In support, the court cited our decision in Johnson v.

Gomez, and the California Court of Appeal’s decision in In re Arafiles, 6 Cal.App.4th 1467, 8 Cal.Rptr.2d 492 (1992). Rosenkrantz, 128 Cal.Rptr.2d 104, 59 P.3d at 194-95. Rosenkrantz asserted that these cases were inconsistent with the Supreme Court’s ruling in Gamer, which postdated both decisions. Id. 128 Cal.Rptr.2d 104, 59 P.3d at 196. But the court rejected this argument, stating that “Gamer did not involve a legislative or constitutional provision even remotely similar to

article V,

section 8(b), and nothing in Gamer questions either the validity of the Arafiles and Johnson decisions themselves or the high court decisions upon which [those decisions] relied.” Id.

In distinguishing the Ex Post Facto Clause challenge in Gamer from Rosenkrantz’s challenge to the retroactive application of gubernatorial review, the court emphasized that in Gamer the Supreme Court had “explained] ... [that t]he essence of respondent’s case ... [was] not that discretion [relating to the grant of parole] has been changed in its exercise but that, in the period between parole reviews, [discretion] will not be exercised at all.” Id. 128 Cal.Rptr.2d 104, 59 P.3d at 198 (quoting Garner, 529 U.S. at 254] 120 S.Ct. 1362) (emphasis' and internal quotation marks omitted).

Even in this limited situation—where retroactive application of a change in law results in the parole board not exercising its discretion at all for a time—the Court in Gamer rejected the facial Ex Post Facto Clause challenge. See id. 128 Cal.Rptr.2d 104, 59 P.3d at 198-200 (citing Garner, 529 U.S. at 254-55, 120 S.Ct. 1362).

But, the Rosenkrantz court acknowledged, the Court in Gamer, through the language at the end of its opinion, “left open the possibility that the petitioner in that case could establish on remand that the new policy permitting the significant postponement of parole hearing dates should be treated as an increase in punishment for purposes of the ex post facto clause.” Id. 128 Cal.Rptr.2d 104, 59 P.3d at 198-99; see Garner, 529 U.S. at 255, 120 S.Ct. 1362. Based on the language at the end of the Gamer opinion, Rosenkrantz contended that “whether the application of

article V,

section 8(b), in [his] case ... violates the ex post facto clause turns upon whether ‘application [of the provision] will result in a longer period of incarceration than under the earlier rule.’ ” Rosenkrantz, 128 Cal.Rptr.2d 104, 59 P.3d at 199 (quoting Garner, 529 U.S. at 255, 120 S.Ct. 1362). The Rosenkrantz court said that such a reading of Gamer would mean that “article V,

section 8(b)[] would violate the ex post facto clause in every case in which the Governor reverses a Board decision granting parole.” Id.

Moreover, the court said, were Rosenkrantz’s reading of Gamer to be accepted, “virtually any procedural change in the parole process ... would fall within the prohibition of the ex post facto clause in any instance in which the procedural change resulted in the denial of parole,” because “in any such case it could be said that application of the new rule ‘will result in a longer period of incarceration than under the earlier rule.’ ” Id. (quoting Garner, 529 U.S. at 255, 120 S.Ct. 1362).

This result, the Rosenkrantz court stated, demonstrates that such a broad reading of Gamer cannot be correct; nor is it correct even to read Gamer to apply to all procedural changes in law that could affect the duration of prisoners’ imprisonment. Id. 128 Cal.Rptr.2d 104, 59 P.3d at 199-200.

Rather, Gamer’s concern with the impact of a procedural rule in a particular case was limited to changes of the sort at issue in Gamer, and there was good reason to distinguish that sort of change from the change at issue in Rosenkrantz: A revision [like those in Gamer and Morales ] that significantly delays the date when the relevant state authority considers the parole eligibility of a prisoner is analogous to a substantive provision increasing the minimum period of time a defendant must be imprisoned before parole even may be considered. (As we have seen, the court in Gamer adverted to this point, explaining that “[t]he essence of respondent’s case, as we see it, is not that discretion has been changed in its exercise but that, in the period between parole reviews, it will not be exercised at all.”).

Because a provision that reduces the frequency of parole hearings is at least potentially comparable to a provision that increases the minimum term of a sentence, a measure that extends the time between parole hearings is one that reasonably might be characterized as bringing about an increase in sentence to which the ex post facto clause might apply.... [T]he court in Gamer determined that when such a provision “does not by its own terms” create a significant risk that a prisoner’s sentence will be increased, the prisoner may establish that application of the provision will violate the ex post facto clause, by demonstrating through “evidence drawn from the rule’s practical implementation ... that [the rule’s] retroactive application will result in a longer period of incarceration than under the earlier rule.” There is nothing in Gamer, however, suggesting that this standard was intended to apply to a provision, unlike a measure reducing the frequency of parole hearings, that cannot reasonably be viewed as falling within the category of legislative measures that increase the punishment for a crime.

Id. (quoting Garner, 529 U.S. at 254-55, 120 S.Ct. 1362). The Rosenkrantz court went on to reiterate that “[t]he only change made by

article V,

section 8(b), is the institution of a new level of review of parole decisions.” Id. 128 Cal.Rptr.2d 104, 59 P.3d at 200. “[T]he opinions of the United States Supreme Court make it clear that this type of change in procedure is not the type of change addressed by the ex post facto clause,” and Gamer does not question these decisions. Id. One of the Supreme Court cases cited by the court in support of this position— one of the cases the court said Gamer did not question—is Mallett v. North Carolina, 181 U.S. 589, 21 S.Ct. 730, 45 L.Ed. 1015 (1901).

In Mallett, two defendants convicted in a state criminal trial had succeeded in having their convictions overturned on appeal in favor of a new trial. Id. at 590, 21 S.Ct. 730. At the time of the commission of the defendants’ offense and at the time of their trial, the State had no right to appeal such a decision under state law, but by the time the defendants won their appeal, the state law had changed, and the State appealed the reversal and order for a new trial. Id.

The state supreme court sustained the State’s appeal, remanding for execution of the original guilty verdict, id., and the defendants appealed to the Supreme Court, arguing that their rights had been violated under the Ex Post Facto Clause, id. at 592, 21 S.Ct. 730. Citing to. a line of cases that distinguished “procedural” changes in law—to which the Ex Post Facto. Clause did not apply—from the changes in law to which the Ex Post Facto Clause was applicable, the Supreme Court rejected the defendants’ claim. Id. at 593-97, 21 S.Ct. 730.

Though Collins subsequently dismissed the notion that laws could be ruled outside the purview of the Ex Post Facto Clause merely because they had been labeled as “procedural” and overruled certain prior Supreme Court cases that applied that type of analysis, 497 U.S. at 45-52, 110 S.Ct. 2715, Collins did not overrule Mallett even though it identified Mallett as a case that applied the type of analysis being dismisséd, id. at 45, 110 S.Ct. 2715.

The Rosenkrantz court described the Supreme Court’s Mallett decision as holding that “the type of procedural provision involved in that case—i.e., the addition of a new level of review of a decision favorable to a criminal defendant that could work to the defendant’s detriment ...—was not the type of procedural change that fell within the aegis of the ex post facto clause.” Rosenkrantz, 128 Cal.Rptr.2d 104, 59 P.3d at 194.

Since the Rosenkrantz court concluded that the change in question—like the change in Mallett—was not the. type of procedural change addressed by the ex post facto clause) it asserted that the nature of the implementation of gubernatorial review was irrelevant, and Rosenkrantz’s Ex Post Facto Clause claim had no merit. Id. 128 Cal.Rptr.2d 104, 59 P.3d at 200-01.

V Biggs essentially makes one argument: that it was an unreasonable application of clearly established federal law for the Rosenkrantz court to conclude that the introduction of gubernatorial review of parole decisions was not the type of change addressed by the Ex Post Facto Clause and thus not the type of change to which the as-applied “significant risk” test endorsed in Gamer must be applied, and that it was therefore an unreasonable application of clearly established federal law for the state court in Biggs’ case to rely exclusively on Rosenkrantz, We disagree. A We are not writing on a blank slate.

We previously held in Johnson v. Gomez that

article V, § 8(

b) does not violate the Ex Post Facto Clause. 92 F.3d at 967. In Johnson, we reviewed relevant Supreme Court decisions on the Ex Post Facto Clause, including Calder v. Bull, 3 U.S. (3 Dall.) 386, 1 L.Ed. 648 (1798); Mallett, 181 U.S. 589, 21 S.Ct. 730; Collins, 497 U.S. 37, 110 S.Ct. 2715; and the then-recent decision in Morales, 514 U.S. 499, 115 5.Ct. 1597. 92 F.3d at 966-67. Johnson argued that the gubernatorial review provision had “the purpose and effect of ... lengthening] prison terms by making it more difficult for convicted murderers with indeterminate sentences to be released on parole.” Id. at 967.

We rejected this argument. We observed that “the law itself is neutral inasmuch as it gives the governor power to either affirm or reverse a [parole board]’s granting or denial of parole.” Id. The law “simply removes the final parole decisionmaking authority from the [parole board] and places it in the hands of the governor.” Id. This change in California law, we concluded, could not be “materially distinguished]” from the change in law at issue in Mallett. Id.; see also Rosenkrantz, 128 Cal.Rptr.2d 104, 59 P.3d at 200 & n. 9 (reaching a similar conclusion).

We thus concluded that the gubernatorial review process did not violate the Ex Post Facto Clause. Johnson, 92 F.3d at 967. We decided Johnson in the year AED-PA became effective, and our decision was not constrained by AEDPA. Our decision in Johnson was thus a straight-up decision on the constitutionality of

article V, § 8(

b) under the Ex Post Facto Clause. That makes it a far more powerful judgment than if we had merely opined under AED-PA that the California courts had not violated clearly established Supreme Court precedent. But it is also true that we decided Johnson before the Supreme Court issued its decision in Gamer. Under our law-of-the-circuit rule, we are bound by Johnson unless it is “clearly irreconcilable” with intervening Supreme Court precedent. See Miller v. Gammie, 335 F.3d 889, 899-900 (9th Cir.2003) (en banc).

Accordingly, we must determine whether our decision in Johnson is clearly irreconcilable with Garner, and particularly with Gamer’s apparent endorsement of as-applied analysis. If so, we must “reject [Johnson ] as having been effectively overruled;” if not, we are bound by Johnson. Id. at 900. Nothing in Gamer signals that it is a change in the Court’s approach. It does not overrule or even question any prior Ex Post Facto Clause decisions, and it cites extensively to Morales. See Himes, 336 F.3d at 855 n. 4.

Quoting Morales, Garner emphasized that “the Ex Post Facto Clause should not be employed for ‘the micromanagement of an endless array of legislative adjustments to parole and sentencing procedures.’ ” 529 U.S. at 252, 120 S.Ct. 1362 (quoting Morales, 514 U.S. at 508, 115 S.Ct. 1597). The Court declared that this. remained an “important concert ].” Id. On its face, the language in Gamer regarding the as-applied inquiry tells us nothing about the scope of its applicability.

Gamer certainly does not say that this as-applied analysis must be conducted even with regard to changes in law that have historically been considered outside the reach of the Ex Post Facto Clause. See Rosenkrantz, 128 Cal.Rptr.2d 104, 59 P.3d at 198-201 (discussing Gamer). As Rosenkrantz cogently explained, the Supreme Court’s 1901 decision in Mallett is still good law—in fact, it was explicitly considered by the Court in Collins but left undisturbed while other cases were overruled— and Mallett strongly suggests that the type of change in law at issue here cannot yield cognizable Ex Post Facto Clause claims.

Id. 128 Cal.Rptr.2d 104, 59 P.3d at 200 & n. 9; see also id. 128 Cal.Rptr.2d 104, 59 P.3d at 194. To the extent the Gamer opinion includes language that can be interpreted as being relevant to the scope of applicability of the as-applied requirement, that language suggests a limited scope. Particularly, the Court asserted that “[t]he essence of respondent’s case ... [was] not that discretion has been changed in its exercise but that, in the period between parole reviews, it will not be exercised at all.” Garner, 529 U.S. at 254, 120 S.Ct. 1362.

The Rosenkrantz court’s reading of this statement as indicative of the Court drawing a line between changes in law that are “potentially comparable to a provision that increases the minimum term of a sentence” and changes in law that are not comparable to such a provision is quite reasonable. See Rosenkrantz, 128 Cal.Rptr.2d 104, 59 P.3d at 200.

Biggs can only point to the Court’s suggestion in Gamer that “[w]hen the rule does not by its own terms show a significant risk, the [petitioner] must demonstrate, by evidence drawn from the rule’s practical implementation by the agency charged with exercising discretion, that its retroactive application will result in a longer period of incarceration than under the earlier rule.” 529 U.S. at 255, 120 S.Ct. 1362. But, as discussed above, the scope of applicability of this suggestion is entirely unclear. Other than this suggestion, Gamer was just following Morales.

Johnson, which applied Morales, is thus not “clearly irreconcilable” with Gamer on account of our failure in Johnson to conduct as-applied analysis. See Miller, 335 F.3d at 900. Johnson remains good law. B Even though Johnson remains good law, we could potentially be bound to hold that Gamer clearly established a requirement of as-applied analysis in a case like Biggs’ case if a prior decision by a panel of our court had so held.

Though our panel opinions do not themselves constitute “clearly established Federal law” for AEDPA purposes, we would be bound to respect a panel holding that the requirement of as-applied analysis is clearly established Supreme Court law in a given context, just like we would be bound to respect any other panel holding.

See Rodgers, 133 S.Ct. at 1450 (asserting that, for purposes of determining the extent of “clearly established Federal law,” “an appellate panel may, in accordance with its usual law-of-the-circuit procedures, look to circuit precedent to ascertain whether it has already held that the particular point in issue is clearly established by Supreme Court precedent”); Chambers v. McDaniel, 549 F.3d 1191, 1199 (9th Cir.2008); see also Miller, 335 F.3d at 899-900.

We cannot elevate our own precedent “to refine or sharpen a general principle of Supreme Court jurisprudence into a specific legal rule that th[e] Court has not announced,” Rodgers, 133 S.Ct. at 1450, but we would owe deference to a clear on-point holding that as-applied analysis is required under clearly established federal law in this context.

We have discussed Gamer's, requirement of as-applied analysis in various contexts, including, most notably, in two AEDPA-governed habeas cases where we granted habeas relief based on the petitioners’ Ex Post Facto Clause challenges to the retroactive application of changes to Oregon’s parole system. In Himes v.

Thompson, we considered an AEDPA-governed habeas challenge to the retroactive application of two changes to Oregon’s parole regulations: (1) a change in the list of factors to be considered in deciding whether there was “aggravation” such that rerelease should be denied after a prior revocation of parole, and (2) a change in the possible reincarceration periods resulting from a denial of rerelease based on an “aggravation” finding. 336 F.3d at 855-64.

We held that retroactive application of these two changes, taken together, created a sufficient risk of increased punishment that the state court’s decision that there was no Ex Post Facto Clause violation was an unreasonable application of clearly established Supreme Court law, and accordingly we granted the habeas petition. Id. at 863-64. Our analysis, however, was explicitly facial in nature. Id. at 855 n. 5 (“Because we hold that the risk of increased punishment is facially apparent, we do not consider [an as-applied claim].”).

Moreover, we specifically said that, though our opinion was “consistent with Garner” we were not applying Gamer because “[p Ire-Garner principles ... compelled] the resolution of th[e] case.” Id. at 855 n. 4. In Brown v. Palmateer, we considered an AEDPA-governed habeas challenge to the retroactive application of a change to an Oregon statute that altered the method and standard for the parole board’s consideration of postponing a prisoner’s parole date based on psychological issues. 379 F.3d 1089, 1094-96 (9th Cir.2004).

In discussing the “significant risk” test established by Morales, we said that “[t]he Supreme Court teaches us to examine the retroactive statute both on its face and in real-world practice,” citing the language at the end of the Gamer opinion. Id. at 1095 (citing Garner, 529 U.S. at 255, 120 S.Ct. 1362).

We added that “[w]e ha[d] previously stated that in the absence of a disadvantage that affects prisoners in general, ‘an individual will satisfy the detriment requirement if he shows that it can “be said with assurance” that he would have received less severe punishment under the prior scheme,’” quoting our decision in Nulph v. Faatz, which predates both Morales and Gamer. Id. (quoting Nulph v. Faatz, 27 F.3d 451, 456 (9th Cir.1994)). We then held that the change in law facially violated the Ex Post Facto Clause.

Id. (“When compared in toto with the pre1993 version of § 144.125(3), the pos1>-1993 statute creates a significant risk that prisoners will face longer periods of incarceration. ... The new standard gives the Board the discretion to find that a prisoner is a danger to the community despite the contrary conclusions of a psychological/psychiatric evaluation. This creates a ‘sufficient risk’ of increased punishment to constitute violation of the Ex Post Facto Clause.”).

Applying Nulph, we added that “[h]ad the Board applied the pre-1993 version of the statute, ... [i]t can ‘be said with assurance’ that Brown would have had a shorter period of incarceration.” Id. (quoting Nulph, 27 F.3d at 456) (internal quotation marks omitted). Though we discussed Gamer in granting habeas petitions in AEDPA-governed cases in Himes and Brown, in neither case did we hold that the as-applied analysis endorsed in Gamer is required under clearly established law in a context like Biggs’ case.

First, we did not even hold that Gamer’s as-applied analysis was required under clearly established law for changes in law like the changes at issue in Himes and Brown. In Himes, our analysis was facial in nature and we specifically asserted that we were not applying Garner. 336 F.3d at 855 nn. 4-5. In Brown, though we said that the Supreme Court teaches us to consider the “significant risk” test both “on its face and in real-world practice,” we only considered that test facially. 379 F.3d at 1095.

We added that it could “be said with assurance” that Brown would have had a shorter period of incarceration under the prior version of the statute, but in making this statement about the as-applied impact of the retroactive application of the new version of the statute, we were applying Nulph, not the “significant risk” test. See id. Second, even if we had held that Gamer’s as-applied analysis was required under clearly established law for changes in law like the changes at issue in Himes and' Brown, that would not be binding here, given the nature of the change in law at issue.

The change in law at issue here, unlike the changes at issue in Himes and Brown, is a purely procedural' change to the manner in which parole decisions are made. Moreover, the change in law at issue here implicates Mallett; the changes at issue in Himes and Brown did not. A holding that as-applied analysis is required under clearly established federal law for a change in law like the change at issue in Himes or Brown, if such' a holding existed, could be—and should be—distinguished with regard to the change in law at issue here.

Other cases where we have discussed Gamer fall short of holding that as-applied analysis is required under clearly established federal law, in Biggs’ case or otherwise. In Scott v. Baldwin, a habeas challenge was brought concerning the retroactive application of a change in Oregon law that resulted in fewer parole-board reviews for inmates labeled as dangerous offenders. 225 F.3d 1020, 1021-23 (9th Cir.2000). The case was not governed by AEDPA. See id.

We noted the Supreme Court’s endorsement of as-applied analysis in Gamer, but stated that “Scott’s argument is limited to a facial challenge of the rule,” and thus that “any argument that the practical implementation of the rule disadvantages Scott is not before us.” Id. at 1022 at n. 5. Because Scott was not governed by AEDPA, our discussion of Gamer’s endorsement of as-applied analysis could not amount to a conclusion that such analysis is required under clearly established federal law under 28 U.S.C. § 2254(d)(1). Moreover, any such conclusion would be dicta given that no as-applied challenge was before us.

Further, even if there were a holding that as-applied analysis is required under clearly established federal law, we would not be bound by such a holding here given the different context, as discussed above in regard to Himes and Brown. In Moor v. Palmer, we rejected, under AEDPA, an Ex Post Facto Clause challenge to the retroactive application of a law broadening the requirement of psychological review as a precondition for parole. 603 F.3d 658, 662-66 (9th Cir.2010). After concluding that Moor had failed to demonstrate a facial Ex Post Facto Clause violation, we noted Gamer’s endorsement of as-applied analysis.

Id. at 666. We asserted, however, that “Moor expressly argue[d] that further factfinding [wa]s unnecessary, choosing to rest his ex post facto claim on a facial challenge to the statute.” Id. Even if we were to construe Moor as asserting that Gamer clearly established a requirement of as-applied analysis in some cases, we would not be bound here because, as in Scott: (1) that assertion would be dicta given that no as-applied challenge was before us, and (2) even if it were a holding, we would not be bound by that holding given the different context. Finally, in Gilman v.

Schwarzenegger, we said that the plaintiffs could succeed on their Ex Post Facto Clause claim through an evidentiary demonstration that retroactive application of the change in law in question would result in increased incarceration time, citing Garner. 638 F.3d 1101, 1106 (9th Cir.2011). But Gilman was a § 1983 case, id. at 1105, and thus contained no holding about clearly established federal law. Thus, because none of our cases discussing Gamer hold that as-applied analysis is required by clearly established federal law, we have no reason not to follow Johnson.

C Our Johnson decision undermines Biggs’ argument that Rosenkrantz, and thus the Superior Court’s opinion in his case, unreasonably applied clearly established federal law. As we have noted, under AED-PA, a bare conflict between a state-court decision and our precedent does not mean that the state-court decision is an unreasonable application of clearly established federal law; a state-court decision does not have to yield to our decision unless our decision itself rests on clearly established Supreme Court precedent. See Rodgers, 133 S.Ct. at 1450-51.

But when a state court decision is consistent with our precedent, our precedent must be taken as persuasive evidence that the state-court decision is correct; more importantly, when a state court’s application of Supreme Court law is the same as our application, our precedent must be accepted as conclusive proof that the state-court decision is not an “unreasonable application of[] clearly established Federal law, as determined by the Supreme Court,” 28 U.S.C. § 2254(d)(1), unless we are prepared to “reject [our precedent] as having been effectively overruled.” Miller, 335 F.3d at 900.

As discussed above, we see no reason to reject or not follow Johnson, so we cannot conclude that the state court here unreasonably applied Supreme Court law in coming to the same conclusion we reached in Johnson for materially congruent reasons. Moreover, even if we were not bound by Johnson, for largely the same reasons that Johnson is not clearly irreconcilable with Gamer, Rosenkrantz is not an unreasonable application of clearly established federal law under Gamer. The lack of clarity in Gamer itself pervades both inquiries.

Gamer tells us nothing about the scope of the applicability of any requirement of as-applied analysis, so, just as we could not say that Johnson is clearly irreconcilable with Gamer, we cannot say that the state court unreasonably applied Gamer in Biggs’ case. It is true that “the lack of a Supreme Court decision on nearly identical facts does not by itself mean that there is no clearly established federal law,” Rodgers, 133 S.Ct. at 1449 (citing Yarborough v. Alvarado, 541 U.S. 652, 664, 124 S.Ct. 2140, 158 L.Ed.2d 938 (2004)), but there is more than a mere factual distinction here.

Mallett is implicated here where it was not in Gamer; and, as the Rosenkrantz court reasonably asserted, the change in law at issue here is distinguishable from the changes in law at issue in Gamer and Morales and indistinguishable from the change at issue in Mallett. Whether or not the language in Gamer can be interpreted as broadly requiring as-applied analysis or not, there is no question that it can reasonably be interpreted as requiring no such analysis in the present case. Thus, even in the absence of Johnson, we would find no AEDPA violation here.

VI The Supreme Court did not clearly establish in Gamer that an as-applied analysis of the significance of the risk of increased punishment is required with regard to the retroactive application of a change in law like California’s gubernatorial review of parole board decisions. The California Supreme Court’s decision in Rosenkrantz was thus not an unreasonable application of clearly established federal law, and neither was the Superi- or Court’s decision in Biggs’ case that relied on it. AFFIRMED. .

A more detailed account of the facts of Biggs’ crime, his rehabilitation, and some earlier proceedings can be found in our prior published opinion regarding Biggs’ habeas proceedings. See Biggs v. Terhune, 334 F.3d 910, 912-13 (9th Cir.2003), overruled in part by Hayward v. Marshall, 603 F.3d 546, 555 (9th Cir.2010) (en banc). . Biggs seeks credit against his term of parole for time he served in prison due to the Governor’s reversal of the parole board's 2005 determination that he was suitable for parole. Before the district court, he asserted that the possibility of such relief militated against mootness.

Citing district-court opinions, the district court stated that whether the possibility of such relief prevents a parolee’s habeas petition from being moot depends on whether the term of parole is indeterminate. The district court explained that parolees who committed their crimes prior to January 1, 1983 face determinate parole terms. On the other hand, parolees who committed their crimes on or after January 1, 1983 are subject to a revised version of California Penal Code § 3000.1, under which parolees serve indeterminate parole terms unless they are released from parole.

Because Biggs committed his murder in September 1981 and his parole term is thus determinate, the district court held that the possibility of providing reliéf by crediting time against Biggs’ parole term prevents Biggs’ petition from being moot. The State does not raise mootness on appeal, so we do not consider it. . Biggs filed his federal habeas petition on September 20, 2006. .

The Court did say that “[tjhe California Supreme Court has noted that about 90% of all prisoners are found unsuitable for parole at the initial hearing, while 85% are found unsuitable at the second and subsequent hearings,” Morales, 514 U.S. at 510-11, 115 S.Ct. 1597, but these statistics were cited to support analysis of the legislative intent of the change in law, not to discuss how the change in law was implemented and the extent to which the implementation resulted in a risk of increased punishment. .

The Rosenkrantz court noted at the outset that Rosenkrantz was not in the best position to challenge gubernatorial review because the parole board had, left to its own devices, concluded that Rosenkrantz was not suitable for parole. 128 Cal.Rptr.2d 104, 59 P.3d at 191. It was only after a California court found that the parole board's decision was not supported by evidence and ordered the parole board to grant parole under the threat of contempt that the parole board did so. Id. The court said: "Accordingly, from a realistic perspective, petitioner cannot maintain persuasively that in this instance

article V,

section 8(b), has resulted in the denial of parole of an individual whom the Board, in the exercise of its independent judgment, has determined is suitable for parole.” Id. The court’s decision on Rosenkrantz's Ex Post Facto Clause challenge, however, did not turn on this unique factual circumstance. Immediately following the discussion of Rosenkrantz’s unique circumstance, the court said that "[t]he flaw in petitioner’s ex post facto claim ... is not confined to the particular circumstances of this case.” Id. .

Biggs also argues that it was unreasonable for the Superior Court to rely on the Rosenkrantz decision in his case because of factual differences between the two cases. Particularly, Biggs argues that it was unreasonable for the state court to rely on Rosenkrantz because, as discussed above, see supra note 5, the parole board only found Rosenkrantz suitable for parole after ordered to do so by the court, not based on its own independent review. This argument is a non-starter. First, it is hard to construe this as a cogent argument under 28 U.S.C. § 2254(d).

The Superi- or Court surely did not violate clearly established federal law by following a California Supreme Court decision that was materially on all fours factually. Second, the Rosenkrantz court specifically said its decision was not limited to cases where the parole board had not found the prisoner suitable for parole based on its own independent review, see supra note 5, so this factual distinction does not raise questions about reliance on Rosenkrantz. .

In Johnson, in addition to the arguments discussed above, we suggested that there was no Ex Post Facto Clause violation because (1) it could not be said "with certainty” that Johnson's sentence was increased by the gubernatorial reversal; and

(2) Johnson was unable to demonstrate that an increase in punishment "actually occurred.” 92 F.3d at 967. This language might be read to suggest that more certainty was required than the "sufficient” or "significant” risk contemplated in Gamer. Garner, 529 U.S. at 250, 255-57, 120 S.Ct. 1362. But we cannot say that Johnson is “clearly irreconcilable” with Garner on this basis given that Gamer did not alter the level of certainty required by Morales, see 514 U.S. at 509, 115 S.Ct. 1597; see also Himes v.

Thompson, 336 F.3d 848, 855 n. 4 (9th Cir.2003) (analyzing an AEDPA-governed Ex Post Facto Clause claim under Morales, but noting that "our analysis is also consistent with Gamer v. Jones "), and that the Johnson court was interpreting Morales. Unless we believe Johnson is otherwise clearly irreconcilable with Gamer, which, as we explain later in this

Section V.A, we do not, we cannot, as a three judge panel, decide that Johnson misinterpreted Morales.