Appeal from the United States District Court for the Northern District of California William Alsup, District Judge, Presiding

Argued and Submitted March 18, 2025 San Francisco, California

On Remand from the United State Supreme Court

Filed October 2, 2025

Before: Jay S. Bybee and Ryan D. Nelson, Circuit Judges, and Susan R. Bolton, * District Judge.

Page 2Opinion by Judge Bybee; Dissent by Judge R. Nelson

SUMMARY **

On remand from the United States Supreme Court, the panel (1) affirmed the district court's holding that the National Bank Act (NBA) did not preempt a class of borrowers' claim that Flagstar Bank, FSB, failed to pay interest on their escrow accounts as required by California Civil Code § 2954.8(a); and (2) vacated and remanded the district court’s judgment and class certification order for the district court to modify the class definition date and the judgment amount.

In Lusnak v. Bank of America, N.A. , 883 F.3d 1185, 1188 (9th Cir. 2018), this court held that California’s interest-onescrow rule was not preempted by the NBA. The panel held here that, under the standards described in Miller v. Gammie , 335 F.3d 889 (9th Cir. 2003) (en banc), it does not have the authority to overrule Lusnak in light of the Supreme Court's intervening decision in Cantero v. Bank of America, N.A. , 602 U.S. 205 (2024), because Cantero is not clearly irreconcilable either with the reasoning or the result in Lusnak .

Page 3Judge R. Nelson dissented because he viewed Cantero as clearly irreconcilable with Lusnak , since Lusnak did not apply the comparative analysis required by Cantero . As an intermediate court, this court must follow Supreme Court precedent, and Lusnak has been effectively overruled by Cantero . While Miller v. Gammie constrains a three-judge panel's authority to overrule circuit precedent, it does not allow the panel to apply precedent inconsistent in theory or reasoning with intervening Supreme Court precedent. Applying Cantero , the NBA preempts California’s interest-on-escrow law.

Peter B. Fredman (argued), Law Offices of Peter B. Fredman PC, Berkeley, California; Steve Berman, Thomas E. Loeser, and Craig R. Spiegel, Hagens Berman Sobol Shapiro LLP, Seattle, Washington; for Plaintiffs-Appellees.

Jonathan Y. Ellis (argued), McGuireWoods LLP, Raleigh, North Carolina; Kathryn M. Barber and Brian D. Schmalzbach, McGuireWoods LLP, Richmond, Virginia; Carolee A. Hoover and David C. Powell, McGuireWoods LLP, San Francisco, California; for Defendant-Appellant.

James R. McGuire, Buckeley LLP, San Francisco, California; Jeffrey P. Naimon, John R. Coleman, and Caroline M. Stapleton, Buckeley LLP, Washington, D.C.; Matthew A. Schwartz, H. Rodgin Cohen, and Shane M. Palmer, Sullivan & Cromwell LLP, New York, New York; Gregg L. Rozansky and Tabitha Edgens, The Bank Policy Institute, Washington, D.C.; David Pommerehn, Consumer Bankers Association, Washington, D.C.; Thomas Pinder andPage 4 Andrew Doersam, The American Bankers Association, Washington, D.C.; Jonathan D. Urick, Tyler S. Badgley, and Janet Galeria, U.S. Chamber Litigation Center, Washington, D.C.; Justin Wiseman, Michael W. Briggs, and Alisha Sears, Mortgage Bankers Association, Washington, D.C.; for Amici Curiae The Bank Policy Institute, American Bankers Association, The Chamber of Commerce of the United States of America, Consumer Bankers Association, and Mortgage Bankers.

Stefan Jouret, Jouret LLC, Boston, Massachusetts; Matthew Lambert Conference of State Bank Supervisors, Washington, D.C.; Arthur E. Wilmarth Jr., George Washington University Law School, Washington, D.C.; for Amici Curiae Conference of State Bank Supervisors and American Association of Residential Mortgage Regulators.

Page 5BYBEE, Circuit Judge:

In Lusnak v. Bank of America, N.A. , 883 F.3d 1185, 1188 (9th Cir. 2018), we held that California’s interest-on-escrow rule was not preempted by the National Bank Act. The issue in this case is whether, under the standards described in Miller v. Gammie , 335 F.3d 889 (9th Cir. 2003) (en banc), this panel has the authority to overrule Lusnak in light of the Supreme Court’s intervening decision in Cantero v. Bank of America, N.A. , 602 U.S. 205 (2024). We hold that we do not.

Adopted in 1864, the National Bank Act (NBA) “establish[ed] the system of national banking still in place today. . . . The Act vested in nationally chartered banks enumerated powers and ‘all such incidental powers as shall be necessary to carry on the business of banking.’” Watters v. Wachovia Bank, N.A. , 550 U.S. 1, 10–11 (2007) (quoting 12 U.S.C. § 24 Seventh). Although '[f]ederally chartered banks are subject to state laws of general application in their daily business to the extent such laws do not conflict with the letter or the general purposes of the NBA,” the Court has “repeatedly made clear that federal control shields national banking from unduly burdensome and duplicative state regulation.” Id . at 11 (citations omitted). In Barnett Bank of Marion County, N.A. v. Nelson , the Court stated that state laws are preempted where they 'forbid, or . . . impair significantly, the exercise of a power that Congress explicitly granted.” 517 U.S. 25, 33 (1996). At the same time, the Court adverted that “this is not to deprive States of the power to regulate national banks, where . . . doing so does notPage 6 prevent or significantly interfere with the national bank's exercise of its powers.' Id.

In the Dodd-Frank Wall Street Reform and Consumer Protection Act, Pub. L. No. 111-203, 124 Stat. 1376 (2010), Congress clarified the state law preemption standards for national banks and expressly incorporated Barnett Bank in the statute. As relevant here, the statute provides:

State consumer financial laws are preempted, only if—

. . . (B) In accordance with the legal standard for preemption in the decision of the Supreme Court of the United States in Barnett Bank of Marion County, N.A. v. Nelson, . . . the State consumer financial law prevents or significantly interferes with the exercise by the national bank of its powers . . . .

12 U.S.C. § 25b(b)(1).

Since 1976, “[e]very financial institution” in California that makes certain home mortgage loans and sets up an escrow account “shall pay interest on the amount so held to the borrower . . . of at least 2 percent simple interest per annum.” Cal. Civ. Code § 2954.8(a). In 2018, we held that the NBA does not preempt § 2954.8(a). Lusnak , 883 F.3d at 1188. We acknowledged that 'Dodd-Frank significantly altered the regulatory framework governing financial institutions,' but we found that 'with respect to NBA preemption, it merely codified the existing standard established in Barnett Bank [.]' Id. Applying Barnett Bank , we held that § 2954.8(a) “does not prevent or significantly interfere with Bank of America's exercise of its powers.' Id.

Page 7at 1194. We held open the possibility, however, that if a state set “punitively high rates,” a bank could bring a successful claim that the rate “prevent[ed] or significantly interfere[d] with a bank’s ability to engage in the business of banking.” Id. at 1195 n.7.

In this case, William Kivett and Bernard and Lisa Bravo represent a class of borrowers for whom Flagstar maintained escrow accounts for payment of property taxes and insurance premiums. Flagstar acknowledged that it did not pay interest on any escrow accounts until 2017—when it began paying interest on subserviced escrow accounts only—as required by California Civil Code § 2954.8(a). These accounts did not include the class. Flagstar took the position that the NBA preempted § 2954.8(a) and, because the California law was invalid, Flagstar was not required to pay interest on funds held in escrow accounts. Relying on our decision in Lusnak , the district court granted summary judgment in favor of the plaintiffs and ordered Flagstar to pay the class $8 million in restitution, plus prejudgment interest. Kivett v. Flagstar Bank, FSB , 506 F. Supp. 3d 749, 754, 762, 767–68 (N.D. Cal. 2020).

Flagstar appealed, and we affirmed, concluding that Lusnak foreclosed Flagstar's preemption argument. Kivett v. Flagstar Bank, FSB , No. 21-15667, 2022 WL 1553266, at *1 (9th Cir. May 17, 2022). We rejected Flagstar’s invitation to overrule Lusnak as wrongly decided. Id. But because the district court incorrectly tolled the statute of limitations and misstated the award, we remanded to the district court to modify its order. Id. at *2.

Meanwhile, a similar challenge to New York’s intereston-escrow law was making its way to the Second Circuit. Similar to California, New York requires banks to pay Page 8interest on escrow of “not less than two per centum per year . . . or a rate prescribed by the superintendent of financial services[.]' N.Y . Gen. Oblig. L. § 5-601. In Cantero v. Bank of America, N.A. , the Second Circuit reasoned that it was “the nature of an invasion into a national bank's operations— not the magnitude of its effects—that determines whether a state law purports to exercise control over a federally granted banking power and is thus preempted.” 49 F.4th 121, 131 (2d Cir. 2022), vacated and remanded by 602 U.S. 205, 220– 21 (2024). Accordingly, “[t]he issue is not whether [New York’s] rate of 2% is so high that it undermines the use of such accounts, or . . . it substantially impacts national banks’ competitiveness. The power to set minimum rates is the ‘power to control,’ and [that] is the ‘power to destroy.’” Id. at 134-35 (quoting McCulloch v. Maryland , 17 U.S. (4 Wheat.) 316, 431 (1819)). The Second Circuit held that the New York law was preempted. Id. at 125.

Both the Cantero plaintiffs and Flagstar petitioned the Supreme Court for a writ of certiorari. The Court granted certiorari in Cantero and rejected the Second Circuit's preemption approach, holding that the Second Circuit “did not analyze preemption in a manner consistent with [the] Dodd-Frank [Act] and Barnett Bank [.]' Cantero , 602 U.S. at 221. The Court held that the Second Circuit had employed “a categorical test that would preempt virtually all state laws that regulate national banks[.]” Id. at 220-21. The Court commented that the parties might wish for a 'clearer preemption line one way or the other. But . . . Barnett Bank did not draw a bright line.' Id . at 221. Rather than deciding the preemption question, the Court remanded to the Second Circuit to conduct a “nuanced comparative analysis” of the Court's prior preemption cases and 'make a practical Page 9assessment of the nature and degree of the interference caused by [the] state law.' Id. at 219-20.

The Supreme Court then granted Flagstar's petition, vacated our judgment, and remanded “for further consideration in light of Cantero [.]” Flagstar Bank, N.A. v. Kivett , 144 S. Ct. 2628 (2024). In a revised memorandum disposition, we again stated that in Lusnak '[w]e properly applied the test for preemption from Barnett Bank ' and we observed that 'the Supreme Court's decision in Cantero suggests that Lusknak was correctly decided[.]' Kivett v. Flagstar Bank, FSB , No. 21-15667, 2024 WL 3901188, at *2 (9th Cir. Aug. 22, 2024). We concluded, in any event, that we could not overrule Lusnak unless it was 'clearly irreconcilable with the reasoning or theory of intervening higher authority.' Id. (quoting Miller , 335 F.3d at 893). We thus reaffirmed the district court's decision. Id. (citation omitted).

Flagstar filed a petition for panel rehearing, which we granted so that we could consider “whether California Civil Code § 2954.8(a) is preempted by the National Bank Act under the standard and methodology described in Cantero [.]' Kivett v. Flagstar Bank, FSB , No. 21-15667, 2024 WL 5206133, at *1 (9th Cir. Dec. 24, 2024). We asked the parties for further briefing and heard oral argument.

'We review de novo the district court's decision on cross motions for summary judgment.” Csutoras v. Paradise High Sch. , 12 F.4th 960, 965 (9th Cir. 2021). '[V]iewing the evidence in the light most favorable to the nonmoving party,' we consider 'whether there are genuine issues of material fact and whether the district court correctly applied the relevant substantive law.' Id. (citation omitted). WePage 10 review de novo questions of statutory interpretation and preemption. McShannock v. JP Morgan Chase Bank NA , 976 F.3d 881, 887 (9th Cir. 2020) (citation omitted).

This case raises two separate questions: First, whether our decision in Lusnak is so “clearly irreconcilable” with the Supreme Court’s “theory or reasoning” in Cantero that it has been “effectively overruled” and is no longer binding on this panel. Miller , 335 F.3d at 899–90. Second, if so, whether § 2954.8(b) is preempted under the standards described in 12 U.S.C. § 25b(b)(1)(B). We conclude that Cantero is not clearly irreconcilable either with the reasoning or the result in Lusnak , so we decline to reach the second question. We do not hold that Lusnak was correctly decided, only that we have no authority to overrule it. Correction in this court, if any is warranted, is only appropriate through our en banc procedures.

Miller v. Gammie is our en banc decision providing the standards under which a three-judge panel may overrule a prior circuit decision. In that decision, we emphasized that as an intermediate appellate court “[a] goal of our circuit’s decisions, including panel and en banc decisions, must be to preserve the consistency of circuit law.” Miller , 335 F.3d at 900. Our starting point is the general principle that a threejudge panel may not overrule a prior court decision. Nevertheless, we have acknowledged the reality that consistency in our court decisions “must not be pursued at the expense of creating an inconsistency between our circuit decisions and the reasoning of state or federal authority embodied in a decision of a court of last resort.” Id. Just as we are bound by our own decisions, we are also bound by Page 11the decisions of the Supreme Court (or a state supreme court on an issue of state law) and our need for consistency in our own decisions must sometimes yield to the reality that a decision has become untenable because of the “holding[] . . . [or] 'mode of analysis'' of the higher court. Id. (quoting Antonin Scalia, The Rule of Law as a Law of Rules , 56 U. Chi. L. Rev. 1175, 1177 (1989)). In Miller , we decided that “issues decided by the higher court need not be identical in order to be controlling. Rather, the relevant court of last resort must have undercut the theory or reasoning underlying the prior circuit precedent in such a way that the cases are clearly irreconcilable.' Id.

Since Miller , we have added to our understanding of what constitutes “clear irreconcilability.” The requirement is a “high standard.” Rodriguez v. AT & T Mobility Servs. LLC , 728 F.3d 975, 979 (9th Cir. 2013). The presumption is against overruling our prior decision: “if we can apply our precedent consistently with that of the higher authority, we must do so.” FTC v. Consumer Def., LLC , 926 F.3d 1208, 1213 (9th Cir. 2019). “It is not enough for there to be ‘some tension’ between the intervening higher authority and prior circuit precedent or for the intervening higher authority to ‘cast doubt’ on the prior circuit precedent.” Lair v. Bullock , 697 F.3d 1200, 1207 (9th Cir. 2012) (first quoting United States v. Orm Hieng , 679 F.3d 1131, 1140-41 (9th Cir. 2012); then quoting United States v. Delgado-Ramos , 635 F.3d 1237, 1239 (9th Cir. 2011)). Even if a Supreme Court decision 'contain[s] language that might persuade us to decide [the prior case] differently if presented to us today[,] . . . the fact that we might decide a case differently than a prior panel is not sufficient grounds for deeming the case overruled. Nothing short of ‘clear irreconcilability’ will Page 12do.” Close v. Sotheby’s, Inc. , 894 F.3d 1061, 1073–74 (9th Cir. 2018) (citations omitted).

The Supreme Court presumably took Cantero to resolve a circuit split between the Second Circuit's decision in Cantero and our decision in Lusnak . It did not resolve that split. Instead, the Supreme Court disapproved of the Second Circuit's 'categorical test that would preempt virtually all state laws that regulate national banks[.]” Cantero , 602 U.S. at 220-21. That is the Court's holding-that there is no categorical test for determining when a state banking regulation is preempted. The Court held that the Barnett Bank standard-which was explicitly incorporated into the Dodd-Frank revisions to 12 U.S.C. § 25b-meant that 'some (but not all) non-discriminatory state laws that regulate national banks are preempted.' Id. at 221. ' Barnett Bank did not draw a bright line,' but 'navigate[d] th[e] Court's prior bank preemption cases.' Id. What Barnett Bank requires is 'a practical assessment of the nature and decree of the interference caused by a state law,” id . at 219– 20, a “nuanced comparative analysis” of the Court’s prior cases, id . at 220, “based on the text and structure of the laws, comparison to other precedents, and common sense,” id. at 220 n.3.

What the Court did, without reaching its own conclusion whether interest-on-escrow laws are preempted, was to review how Barnett Bank addressed six of the Court’s bankpreemption cases issued between 1870 and 1982. Three of those casesFidelity Federal Savings & Loan Association v. de la Cuesta , 458 U.S. 141 (1982); Franklin National Bank of Franklin Square v. New York , 347 U.S. 373 (1954); and First National Bank of San Jose v. California , 262 U.S.

Page 13366 (1923)—held that the state regulation was preempted. The other three cases— Anderson National Bank v. Luckett , 321 U.S. 233 (1944); McClellan v. Chipman , 164 U.S. 85 (1896); and National Bank v. Commonwealth , 76 U.S. (9 Wall.) 353 (1870)-held that the state regulation was not preempted. The Court then advised:

If the state law's interference with national bank powers is more akin to the interference in cases like Franklin , Fidelity , First National Bank of San Jose , and Barnett Bank itself, then the state law is preempted. If the state law's interference with national bank powers is more akin to the interference in cases like Anderson , National Bank v. Commonwealth , and McClellan , then the state law is not preempted.

Cantero , 602 U.S. at 220.

Because Cantero did not decide whether the NBA preempts state interest-on-escrow laws, the result is not inconsistent with Lusnak 's judgment. Nor is Cantero ’s holding-that the Second Circuit erred in applying a categorial test for preemption-inconsistent with Lusnak . Had the Lusnak panel adopted a categorical test that 'would preempt virtually no non-discriminatory state laws that apply to both state and national banks,” id. at 221, we would conclude that Lusnak is clearly irreconcilable with Cantero . But we did not apply anything close to a categorical test, and we left open the possibility for a future as-applied challenge to California's interest-on-escrow rule. Lusnak , 883 F.3d at 1195 n.7.

Page 14Nevertheless, in Miller , we emphasized that it was not only results, but 'the theory or reasoning'-the Court's 'explications of the governing rules of law'-that count. Miller , 335 F.3d at 900 (quoting County of Allegheny v. ACLU Greater Pittsburgh Chapter , 492 U.S. 573, 668 (1989) (Kennedy, J., concurring in part and dissenting in part)). That will require a closer comparison of Barnett Bank and the preemption cases cited by the Court and our methodology in Lusnak .

We will start with some observations about how we proceeded in Lusnak . First, we identified the 'central question' in the case as 'whether the NBA preempts California Civil Code § 2954.8(a).' Lusnak , 883 F.3d at 1190. Second, we reviewed briefly the background of the NBA and the changes brought about by Dodd-Frank in 2010. We quoted the NBA's preemption provision, 12 U.S.C. § 25b(b)(1)(B), and noted that although “there is no presumption against preemption,” the burden of proof rested with the bank challenging the statute. Id. at 1191. Third, we discussed Barnett Bank in some detail, both for its influence on Dodd-Frank, id. at 1191-94 and for its substantive rule of preemption, id. at 1194-95.

Fourth, and perhaps most important for the issue now before us, although we addressed Barnett Bank , we did not cite or discuss the six preemption cases discussed in Barnett Bank and recounted in Cantero . We did, however, consider whether the interest-on-escrow rule “prevent[ed] or significantly interfere[d] with Bank of America’s exercise of its powers.' Id. at 1194. One reason we concluded that § 2954.8(a) did not do so was because in Dodd-Frank, Congress amended the Truth in Lending Act (TILA), to Page 15require interest-on-escrow in certain accounts “[i]f prescribed by applicable State or Federal law.” 15 U.S.C. § 1639d(g)(3). We accepted this provision as “express[ing] Congress's view that [interest-on-escrow] laws would not necessarily prevent or significantly interfere with a national bank's operations.' Lusnak , 883 F.3d at 1194-95. We found that our reading of § 1639d was reinforced by the DoddFrank legislative history which noted that the new provision would require servicers to “‘mak[e] interest payments on the escrow account if required under [state or federal] laws.’” Id. at 1196 (quoting H.R. Rep. No. 111-94, at 91). 'This passage shows Congress's view that creditors, including large corporate banks like Bank of America, can comply with state escrow interest laws without any significant interference with their banking powers.” Id.

Is Cantero ’s methodology clearly irreconcilable with our methodology in Lusnak ? We think not. Two points are critical here. First, Cantero admonishes courts to consider Barnett Bank and the six cases that Barnett Bank cited. We will not review all of those cases here, because the Court reviewed them carefully in Cantero . 602 U.S. at 214-19. We will review the two cases that we think most applicable to this case. The first is Franklin National , which presents the best case for preemption. In that case, which Cantero described as the 'paradigmatic example of significant interference,” id. at 216, the state law at issue prohibited commercial banks “from using the word ‘saving’ or ‘savings’ in their advertising or business,” Franklin National , 347 U.S. at 374. The state law did not prevent national banks from taking savings deposits or advertising that they did so—national banks just could not use the word “savings” in their advertising. Id. at 378. Federal law, in Page 16contrast, provided that national banks could receive savings deposits “without qualification or limitation” and that national banks possessed “all such incidental powers as shall be necessary to carry on the business of banking[.]” Id. at 375–76. The Court held that New York’s restriction on the use of the word “savings”—a word that “aptly describe[d] . . . the type of business carried on by these national banks'-created a 'clear conflict' between state and federal law. Id. at 378. Accordingly, federal law preempted the state law. Id. at 378–79.

The second case is Anderson National Bank , which Cantero called “the primary example of a case where state law was not preempted.' Cantero , 602 U.S. at 217. Anderson involved a Kentucky law that permitted Kentucky to confiscate abandoned bank deposits. Anderson , 321 U.S. at 236. The Court had previously held in First National Bank of San Jose , that a California law that authorized California to claim bank deposits that customers left unclaimed for more than 20 years interfered with the national bank's powers and success, so federal law preempted the state law. 262 U.S. at 370. The Court had reasoned in First National Bank of San Jose that “[t]he success of almost all commercial banks depends upon their ability to obtain loans from depositors, and [banks] might well hesitate to subject their funds to possible confiscation.' Id. But unlike the California law, the Kentucky law in Anderson required the state to produce proof that seized accounts had been abandoned. Anderson , 321 U.S. at 250. The Court reasoned that “the escheat or appropriation by the state of property in fact abandoned or without an owner is . . . as old as the common law itself” and would not “deter [bank customers] from placing their funds in national banks.” Id. at 251–52. The law did “not discriminate against national banks,” id . atPage 17 247, and “the mere fact that the depositor’s account is in a national bank” and was subject to Kentucky law did “[not] impose an undue burden on the performance of the banks’ functions,” id. at 248; see id. at 252 ('[W]e can perceive . . . no unlawful encroachment on the rights and privileges of national banks.”). The Court concluded that Kentucky’s law was not preempted.

We are not sure what additional information we should glean from these cases. Neither case seems particularly applicable to the interest-on-escrow law. As in Franklin National , we recognize that national banks possess “all such incidental powers as shall be necessary to carry on the business of banking,” and that should include the power to create escrow accounts. 347 U.S. at 376. The interest-onescrow accounts raises the cost to national banks to use escrow accounts and may discourage them from issuing and servicing loans. That certainly “interferes” with the banks’ unfettered exercise of their statutory powers, and a court might reasonably determine that it “significantly interferes” and, for that reason, is preempted under Barnett Bank . On the other hand, the interest-on-escrow rule is a rule of general application in California, applying equally to state and federal banks, and “state laws c[an] apply to national banks as long as the state laws d[o] not ‘in any way impai[r] the efficiency of national banks or frustrate[e] the purpose for which they were created.'' Cantero , 602 U.S. at 219 (quoting McClellan , 164 U.S. at 358). A court might reasonably conclude that such a general rule is not preempted because Barnett Bank ruled that 'some (but not all) non-discriminatory state laws that regulate national banks are preempted.” Id. at 221.

This leaves us in equipoise. Neither line of cases seems to compel the result here. Which leads to our second point:

Page 18Lusnak relied on other tools of statutory interpretation and nothing in Cantero suggests that the 'nuanced comparative analysis,” id. at 220, of the cases cited in Barnett Bank and reviewed in Cantero is the sole method for determining preemption. Indeed, Cantero noted that there were other issues that the lower courts “may address as appropriate on remand.” Id. at 221 n.4. More importantly, Barnett Bank applied “ordinary legal principles of [preemption],” which included considering 'the Federal statute's background or history.” Barnett Bank , 517 U.S. at 37. Nothing in Cantero casts doubt on our power to use the full array of interpretive tools in preemption analysis.

We do not think that Cantero has prescribed a mode of analysis that is clearly irreconcilable with what we did in Lusnak . That means that we have no warrant as a threejudge panel to declare Lusnak overruled and decide the question of California's interest-on-escrow rule anew. In reaching this conclusion, we hold only that Lusnak remains good law. Whether we would have reached the same conclusion is irrelevant. Until an en banc court or the Supreme Court tells us otherwise, we are bound by our prior decision.

Because Lusnak is not clearly irreconcilable with Cantero , we cannot overrule it. Lusnak controls this case, and under Lusnak , 'the NBA does not preempt California Civil Code § 2954.8(a).' 883 F.3d at 1197. The district court did not err in granting summary judgment for the plaintiffs, except that, as we previously concluded, it incorrectly tolled the statute of limitations and therefore misstated the award. Accordingly, we AFFIRM the district court’s preemption holding. We VACATE and REMANDPage 19 the judgment and class certification order for the district court to modify the class definition date from April 18, 2018, to August 22, 2018, and the judgment amount from $9,262,769.24 to $9,180,580.15.

R. NELSON, Circuit Judge, dissenting:

This was at one point a straightforward case. In Lusnak v. Bank of America, N.A. , we held that federal law does not preempt California Civil Code § 2954.8(a), which requires financial institutions to pay at least 2% interest annually on certain mortgage escrow accounts. 883 F.3d 1185, 1194 (9th Cir. 2018). So when Flagstar Bank argued that California’s law is preempted as to federally chartered banks, we rejected its argument as foreclosed by our precedent. Kivett v. Flagstar Bank, FSB ( Kivett I ), No. 21-15667, 2022 WL 1553266, at *1 (9th Cir. May 17, 2022).

Since then, the landscape has changed. Last year, in Cantero v. Bank of America, N.A. , the Supreme Court clarified the standard for federal preemption of state banking laws. 602 U.S. 205 (2024). Under Cantero , a state law significantly interferes with national banking powers, and is thus preempted, if a “nuanced comparative analysis” of the Supreme Court's precedents shows that the law interferes like other laws that the Court has considered preempted. Id. at 220. By contrast, if the state law’s interference is more like the interference from laws that the Court has upheld, the state law is not preempted. Id.

After rehearing, I view Cantero as “clearly irreconcilable” with Lusnak , since Lusnak did not apply the comparative analysis required by Cantero . Miller v.

Page 20Gammie , 335 F.3d 889, 900 (9th Cir. 2003) (en banc). Instead, Lusnak derived from an inapplicable statute a categorical anti-preemption rule that Cantero declined to endorse and which is inconsistent with a fortified application of Cantero ’s reasoning. As an intermediate court, we must follow Supreme Court precedent. Lusnak has therefore been 'effectively overruled.' Id. While Miller v. Gammie constrains a three-judge panel’s authority to overrule circuit precedent, it does not allow us to apply precedent inconsistent in theory or reasoning with intervening Supreme Court precedent.

Setting Miller v. Gammie aside, Lusnak was wrongly decided. Under Cantero ’s comparative framework, California’s law is preempted: its interference with national banking powers is “more akin” to the interference stemming from state laws that the Court has already deemed preempted. Cantero , 602 U.S. at 220. And its interference is less analogous to the interference in cases where a state law was not preempted. Id. The majority does not conclude otherwise. I respectfully dissent.

Two centuries ago, the Supreme Court famously held in McCulloch v. Maryland that federal law supersedes state law in matters involving the national banking system. 17 U.S. (4 Wheat.) 316, 430–31 (1819); see Watters v. Wachovia Bank, N.A. , 550 U.S. 1, 10 (2007). When Maryland levied a tax on “all banks or branches thereof” not “chartered by the [state] legislature,' James McCulloch, a cashier at the federally chartered Second Bank of the United States, did not pay. 17 U.S. at 318–19. Chief Justice Marshall wrote that Maryland could not tax the federal bank, noting the “plain repugnance”Page 21 in giving state governments a 'power to control the constitutional measures” of the federal government. Id. at 431. “[T]he power to tax,” after all, “involves the power to destroy.” Id.

Though the federal bank in McCulloch no longer exists, the United States still “maintains a dual system of banking, made up of parallel federal and state banking systems” that “co-exist and compete.” Cantero v. Bank of Am., N.A. , 602 U.S. 205, 209-10 (2024). In 1863, Congress enacted the National Bank Act (NBA), which created today’s uniform national banking system. Under that system, privately owned banks may choose a federal or state charter. Banks with federal charters—so-called “national” banks—are governed mainly by federal law, while state-chartered banks are subject to additional state regulation. For national banks, the NBA confers certain enumerated powers and “all such incidental powers as shall be necessary to carry on the business of banking.” 12 U.S.C. § 24 Seventh.

In Barnett Bank of Marion County, N.A. v. Nelson , the Supreme Court explained that the NBA’s grants of authority, both enumerated and incidental, are “not normally limited by, but rather ordinarily pre-empt[], contrary state law.” 517 U.S. 25, 32 (1996). As the Court explained, 'Congress would not want States to forbid, or to impair significantly, the exercise of a power that Congress explicitly granted.' Id. at 33. So while a “presumption against federal preemption of state law” sometimes applies, that principle “is inapplicable to federal banking regulation.” Rose v. Chase Bank USA, N.A. , 513 F.3d 1032, 1037 (9th Cir. 2008) (quotation omitted); see Cuomo v. Clearing House Ass'n, L.L.C. , 557 U.S. 519, 554–55 (2009) (Thomas, J., concurring in part and dissenting in part).

Page 22Still, there is some room for state regulation of national banks. In Barnett Bank , the Court held that states may regulate national banks where “doing so does not prevent or significantly interfere with the national bank’s exercise of its powers.” 517 U.S. at 33. For instance, national banks are not exempt from “state laws of general application . . . to the extent such laws do not conflict with the letter or the general purposes of the NBA.” Watters , 550 U.S. at 11. But that is the exception, not the rule. When a state law “significantly impair[s] the exercise of authority, enumerated or incidental under the NBA,' it 'must give way.' Id. at 12 (citing Barnett Bank , 517 U.S. at 32-34).

Barnett Bank applied its significant-interference test to a Florida law that prohibited national banks from selling insurance in small towns. To determine whether Florida’s insurance law was preempted, the Court looked at prior cases where a state law was preempted, as well as some where it was not. 517 U.S. at 32–37. For example, the Court pointed to Franklin National Bank of Franklin Square v. New York , 347 U.S. 373, 377–78 (1954), which held that a New York law forbidding national banks from using the word “savings” in advertising significantly interfered with the national bank's ability to advertise efficiently . Id. at 33. In part because New York's 'quite similar' law interfered with national banking powers in a manner akin to Florida's insurance law, the Florida law was preempted. Id.

Besides analyzing other pro-preemption cases, the Barnett Bank Court addressed cases where the state-law interference was not significant. Id. at 33–34. It cited, for example, Anderson National Bank v. Luckett , 321 U.S. 233, 252 (1944), where a Kentucky statute governing abandoned deposit accounts did not 'unlawful[ly] encroac[h] on the rights and privileges of national banks.” Id. at 33 (quoting Page 23Anderson , 321 U.S. at 247–52). In the end, though, Florida’s insurance law more significantly interfered with national banking powers than the laws in those other decisions.

The takeaway from Barnett Bank is simple: there is no one-size-fits-all approach for deciding when a state law “significantly interfere[s] with the national bank’s exercise of its powers.” Id. Courts must instead look to the laws in prior bank preemption precedents, comparing the nature and degree of interference caused by those laws with the state regulation under review.

In 2010, Congress codified Barnett Bank 's comparative analysis in the Dodd-Frank Wall Street Reform and Consumer Protection Act, Pub. L. No. 111-203, 124 Stat. 1376. Today, a state law is preempted if—and “only if”—it “prevents or significantly interferes with the exercise by the national bank of its powers” “in accordance with the legal standard for preemption in the decision of the Supreme Court of the United States in Barnett Bank of Marion County, N.A. v. Nelson, Florida Insurance Commissioner, et al. , 517 U.S. 25 (1996).” 12 U.S.C. § 25b(b)(1)(B).¹

Among the NBA’s enumerated powers is the authority to “make, arrange, purchase or sell loans or extensions of credit secured by liens on interests in real estate”—put simply, to administer home mortgage loans. 12 U.S.C. § 371(a); see also 12 C.F.R. § 34.3(a). Incidental to that authority, national banks can provide and service escrow accounts to Page 24aid in home mortgage lending. Office of the Comptroller of the Currency (OCC) Inter. Ltr. 1041, 2005 WL 3629258, at *2 (Sept. 28, 2005).

Mortgage escrow accounts are an important tool for lenders and borrowers alike. See Cantero , 602 U.S. at 210– 11. Borrowers make installment payments into escrow, which lenders then use to pay property taxes, insurance premiums, and other charges on the borrower’s behalf. This simplifies the budgeting process for borrowers and mitigates risk for lenders, who can avoid tax liens and lapses in insurance coverage on the property. For these reasons, mortgage escrow accounts have taken hold in the American residential mortgage market, with most loan originations including an escrow account. And several federal agencies, like the Department of Agriculture’s Rural Housing Service, require escrow accounts for government-sponsored housing programs. See 7 C.F.R. § 3555.252(b)(1).

Lenders will sometimes pay interest to borrowers on the balances of their mortgage escrow accounts. That remains true even though the federal statute that regulates mortgage escrow accounts-the Real Estate Settlement Procedures Act (RESPA)-does not require national banks to pay interest on funds held in escrow. Congress has considered such a requirement three times-and has rejected it each time. See H.R. 27, 103d Cong. (1993); H.R. 3542, 102d Cong. (1991); Gov't Accountability Off., Study of the Feasibility of Escrow Accounts on Residential Mortgages Becoming Interest Bearing (1973), https://perma.cc/W4WQ-JU7J.

There is one limited exception, however. Under § 1639d of the Truth in Lending Act (TILA), national banks must establish escrow accounts for certain high-priced mortgages.

Page 25See 15 U.S.C. § 1639d(a)–(b). And contrary to RESPA, a lender who maintains a TILA account must pay interest on the funds held in escrow “[i]f prescribed by applicable State or Federal law,' 'in the manner as prescribed by that applicable State or Federal law.” Id. § 1639d(g)(3). This ensures that borrowers with weak credit receive the benefits that escrow accounts provide.

Apart from federal law, the states administer their own web of interest-on-escrow (IOE) laws. These laws require lenders to pay interest on funds that borrowers must deposit in mandatory mortgage escrow accounts. At least 12 states have such laws. See, e.g. , Or. Rev. Stat. §§ 86.205, 86.245; Minn. Stat. § 47.20, subdiv. 9(a). And each differs from the others-whether by imposing unique interest rates or restricting their application to certain properties.

The question here is whether national banks must comply with state IOE laws. Seven years ago, in Lusnak v. Bank of America, N.A. , we answered yes. 883 F.3d 1185, 1194 (9th Cir. 2018). According to Lusnak , the NBA does not preempt California's IOE law-the same law Flagstar challenges. Id. Without more, this appeal is open and shut, as we first found. Kivett v. Flagstar Bank, FSB ( Kivett I ), No. 21-15667, 2022 WL 1553266, at *1 (9th Cir. May 17, 2022).

But doubt remained about whether Lusnak was correctly decided. According to the Nation’s top banking regulator, Lusnak “comprehensively misinterpreted” Barnett Bank by “invert[ing]” its expectation that the NBA’s enumerated and incidental powers will ordinarily preempt contrary state law. Br. of Amicus Curiae Off. of the Comptroller of the Currency in Support of Appellee’s Pet. for Reh’g En Banc at 1, 10–11, Page 26Lusnak , 883 F.3d 1185 (9th Cir. 2018) (No. 14-56755). And three years ago, the Second Circuit expressly rejected Lusnak in concluding that the NBA preempts New York's IOE law. Cantero v. Bank of Am., N.A. , 49 F.4th 121, 137– 38 (2d Cir. 2022), vacated and remanded by 602 U.S. 205 (2024).

This resulted in a split between the Second Circuit's decision in Cantero and our decision in Lusnak . Though the Supreme Court granted certiorari in the Second Circuit case, it did not resolve the split. Instead, the Court returned the case to the Second Circuit because it 'did not analyze preemption in a manner consistent with Dodd-Frank and Barnett Bank .' Cantero , 602 U.S. at 221. In doing so, the Court reminded us of what Barnett Bank requires: a 'practical assessment of the nature and degree of the interference caused by a state law,' based on a 'nuanced comparative analysis” of the Court’s prior preemption cases. Id. at 219-20.

PostCantero , we too were told to reconsider our initial decision under Lusnak . Flagstar Bank, N.A. v. Kivett , 144 S. Ct. 2628 (2024). We first followed our prior decision applying Lusnak . See Kivett v. Flagstar Bank, FSB ( Kivett II ), No. 21-15667, 2024 WL 3901188, at *1-2 (9th Cir. Aug. 22, 2024). We granted rehearing to consider again whether Cantero so undermines Lusnak that it no longer binds this panel. Kivett v. Flagstar Bank, FSB , No. 21-15667, 2024 WL 5206133, at *1 (9th Cir. Dec. 24, 2024).

Today, the majority solidifies Lusnak in our circuit law. I would not. Under our precedent, we must treat as “effectively overruled” any circuit decision that is “clearly irreconcilable' with the 'theory or reasoning' of new Page 27Supreme Court authority. Miller v. Gammie , 335 F.3d 889, 900 (9th Cir. 2003) (en banc). That standard, though rigorous, is satisfied here. Lusnak embraced a categorical preemption test that is nothing like Cantero 's comparative analysis. And Cantero , standing alone, compels us to hold that the NBA preempts California’s IOE law.

Start with the first question: Is Lusnak good law after Cantero ? If so, Lusnak still dictates this case. See Kivett I , 2022 WL 1553266, at *1.

1

In Lusnak , the plaintiff argued on behalf of a putative class that a national bank’s noncompliance with California’s IOE law violated the State's Unfair Competition Law (UCL). 883 F.3d at 1190. Citing Barnett Bank and DoddFrank for the significant-interference test, we held that California’s IOE law was not preempted because it did not prevent or significantly interfere with the exercise of national banking powers. Id. at 1194 (citing § 25b(b)(1)(B)).

But rather than analogize to prior preemption cases, we based our conclusion on a statute that was never mentioned in Barnett Bank or any other preemption precedent: TILA's § 1639d(g)(3). See Lusnak , 883 F.3d at 1194-95. Recall that § 1639d(g)(3) requires national banks to pay interest on escrowed funds “[i]f prescribed by applicable State . . . law,” but only for certain mortgages. 15 U.S.C. § 1639d(g)(3). The statute did not apply to the mortgage in Lusnak . 883 F.3d at 1197. Nor have Plaintiffs ever maintained that their mortgages fall under § 1639d(g)(3).

Still, Lusnak reasons that § 1639d(g)(3)’s state-law payment requirement “expresses Congress’s view that [state Page 28IOE] laws would not necessarily prevent or significantly interfere with a national bank's operations.' Id. at 1194-95. As the logic goes, by requiring national banks to pay interest on escrowed funds for a different subset of mortgages, Congress saw no conflict between state IOE laws and the powers of national banks. Thus, Congress did not intend for the NBA to preempt such laws.

This reasoning was apparently “confirm[ed]” by a House Report that “explains Congress’s purpose” behind § 1639d(g)(3). Id. at 1195–96 (citing H.R. Rep. No. 111-94 (2009)). The Report states that mortgage servicers must administer escrow accounts in accordance with “applicable” state laws, 'including making interest payments on the escrow account if required under such laws.' H.R. Rep. No. 111-94, at 91. From this unenacted legislative history we derived 'Congress's view' that 'creditors, including large corporate banks[,]' can comply with state IOE laws 'without any significant interference with their banking powers.' Lusnak , 883 F.3d at 1196.

In a footnote, we also mentioned that the NBA may preempt a state law “setting punitively high [interest] rates.” Id. at 1195 n.7. But we did not explain what such laws might be. Instead, we repeated-rather categorically-that 'no legal authority establishes that state escrow interest laws prevent or significantly interfere with the exercise of national bank powers.' Id. at 1197. “Congress itself . . . has indicated that they do not.' Id. So California's IOE law was not preempted.

Now consider Cantero . The Second Circuit’s decision, like our decision in Lusnak , adopted a categorical preemption rule-just to the opposite extreme. Whereas Page 29Lusnak allows for virtually no preemption, on the Second Circuit's read, cases stretching back to McCulloch show that the NBA preempts any state law that “exert[s] control over a banking power.” Cantero , 49 F.4th at 132.

Rejecting the Second Circuit’s interpretation, the Court emphasized that clear-cut standards have no place under Barnett Bank . The Second Circuit 'distill[ed] a categorical test that would preempt virtually all state laws that regulate national banks.” Cantero , 602 U.S. at 220–21. Meanwhile, the Cantero plaintiffs “would yank the preemption standard” in the other direction and 'preempt virtually no nondiscriminatory state laws that apply to both state and national banks.” Id. at 221. The Court acknowledged the desire for a 'clearer preemption line.' Id. But its hands were tied. 'Congress expressly incorporated Barnett Bank into the U.S. Code,' and ' Barnett Bank did not draw a bright line.' Id.

Nor did Cantero draw any categorical inferences from § 1639d(g)(3). The Cantero plaintiffs invoked Lusnak 's theory, arguing that § 1639d(g)(3) provides “strong evidence' that Congress 'sees no irreconcilable conflict' between state IOE laws and national banking powers. Pet. for Writ of Cert. at 23, Cantero , 602 U.S. 205 (2024) (No. 22-529). The Court implicitly rejected that argument, simply noting, as was true in Lusnak , 'that § 1639d does not apply to the mortgages in this case.” 602 U.S. at 211 n.1. Rather than rely on an inapplicable law as we did in Lusnak , the Court focused its preemption analysis solely on Barnett Bank . And Barnett Bank said nothing about § 1639d(g)(3).

Instead, Barnett Bank “sought to carefully account for and navigate [the Supreme] Court’s prior bank preemption cases.” Id. at 221. And as Dodd-Frank makes clear, courts may find a state law preempted 'only if' it 'prevents or Page 30significantly interferes' with national banking powers 'in accordance with the legal standard” from Barnett Bank . Id. (quoting § 25b(b)(1)(B)). So we too “must do as Barnett Bank did and likewise take account of” the Court’s preemption precedents. Id. at 215-16.

To aid in this analysis, Cantero identified six relevant precedents, each cited in Barnett Bank . Id. at 219-20. Three show the kinds of preempted state laws that significantly interfere with national banking powers. Fidelity Fed. Sav. & Loan Ass’n v. de la Cuesta , 458 U.S. 141 (1982); Franklin , 347 U.S. at 373; First Nat’l Bank of San Jose v. California , 262 U.S. 366 (1923). Another three reflect the degree of state-law interference that is not significant enough to warrant preemption. Anderson , 321 U.S. at 233; McClellan v. Chipman , 164 U.S. 347 (1896); Nat’l Bank v. Commonwealth , 76 U.S. (9 Wall.) 353 (1870). In all six, the Court assessed the 'nature and degree' of the state laws' interference 'based on the text and structure of the laws, comparison to other precedents, and common sense.' Cantero , 602 U.S. at 220 n.3.

The Court then articulated the preemption test. In assessing significant interference, “courts may consider the interference caused by the state laws in Barnett Bank , Franklin , Anderson , and the other precedents on which Barnett Bank relied.' Id. at 220. If a state law's interference with national banking powers is 'more akin' to the interference in the first set of cases, the law is preempted. Id. If the state law’s interference is 'more akin' to the interference in the second set of cases, it is not preempted. Id. Because the Second Circuit “did not conduct that kind of nuanced comparative analysis,' remand was required. Id.

Page 31Lusnak did not conduct a comparative analysis, either. Though Lusnak pointed to Barnett Bank for the significantinterference test, we did not cite, much less analyze, any of the bank preemption precedents identified in Cantero . Instead, we extracted Lusnak 's holding-state IOE laws are generally not preempted—from a statute that Barnett Bank never addressed and on which Cantero did not rely. Lusnak 's reasoning and holding is thus incompatible with the “theory or reasoning” of Cantero . Miller , 335 F.3d at 900.

Start with Lusnak 's reliance on § 1639d(g)(3). As the majority notes, Cantero holds at least that 'there is no categorical test for determining when a state banking regulation is preempted.” Maj. Op. at 12. Lusnak violates that principle in spades. We held that § 1639d(g)(3) expressed Congress’s understanding that state IOE laws did not significantly interfere with the exercise of national banking powers. 883 F.3d at 1194-95, 1197. That is a categorical test. By using § 1639d(g)(3) to resolve the preemption question in one fell swoop, Lusnak did precisely what Cantero forbids.

Section 1639d(g)(3) is not a broad congressional pronouncement on preemption of state IOE laws, but a limited exception to the default rule that national banks need not pay interest on escrowed funds. Again, the statute applies only to a subset of high-priced mortgages, none of which appear in this case, Cantero , or Lusnak . See 15 U.S.C. § 1639d(b), (f) (preserving lenders’ authority to set the terms of non-covered escrow accounts). So § 1639d(g)(3) shows at most that Congress wanted national banks to comply with state IOE laws when administering the kinds of mortgage escrow accounts addressed in the statute. See Cantero , 49Page 32 F.4th at 140 (Pérez, J., concurring) (“Congress did intend to subject national banks” to state IOE laws, but only “when financing certain [covered] mortgage loans.”). Otherwise, there is no requirement that national banks pay interest on escrowed funds.

Lusnak took TILA’s carveout to the extreme. It seized on Congress's exception for high-priced mortgages and extrapolated a general legislative intent that all state IOE laws (even those untouched by TILA) pass muster under Barnett Bank 's significant-interference test. That is not how we read statutes. We do not look at statutory exceptions, refashion them as a broader, universal “intent” of Congress, and then deduce “further exceptions from there.” Id. at 138 (maj. op.) (noting that the Supreme Court's oft-maligned decision in Holy Trinity Church v. United States , 143 U.S. 457 (1892), applied similar reasoning). When Congress takes care to enumerate certain exceptions, additional exceptions are rarely implied. See TRW Inc. v. Andrews , 534 U.S. 19, 28 (2001). It makes more sense to read § 1639d(g)(3) 'as a decision by Congress to carve out an exception from its general [preemptive] rule, rather than expressly imposing a burden on some mortgage loans in order to impliedly impose a burden on all of them.” Cantero , 49 F.4th at 138.

In that vein, the Supreme Court in Cantero declined to infer from § 1639d(g)(3) a general intent of Congress with respect to preemption of state IOE laws, noting that the statute did not apply to the mortgages at issue. 602 U.S. at 211 n.1. Nor did the statute apply to the mortgage in Lusnak . 883 F.3d at 1194–95, 1197. In Cantero , the Court could have followed our lead and held that § 1639d(g)(3) expresses Congress’s understanding that state IOE laws do not significantly interfere with national banking powers. That Page 33theory was presented to the Court. See supra , at 31. Yet Cantero says nothing about § 1639d(g)(3)’s relevance to the preemption inquiry.

It makes sense why: courts can apply Barnett Bank 's preemption standard while still honoring § 1639d(g)(3). In instructing national banks to comply with state IOE laws when financing some covered mortgage loans, Congress did not abandon the statutory mandate that preemption of state IOE laws with respect to all other mortgages must be assessed “in accordance with” Barnett Bank 's comparative analysis. See 12 U.S.C. § 25b(b)(1)(B); contra Lusnak , 883 F.3d at 1197. And as Cantero makes clear, Barnett Bank 's comparative analysis—codified in Dodd-Frank—is diametrically opposed to Lusnak 's categorical reliance on § 1639d(g)(3).

The majority asserts that Lusnak 'did not apply anything close to a categorical test.' Maj. Op. at 13. Not so. Lusnak is unequivocal: '[N]o legal authority establishes that state escrow interest laws prevent or significantly interfere with the exercise of national bank powers, and Congress itself, in enacting Dodd-Frank, has indicated that they do not.” 883 F.3d at 1197. Also consider Lusnak 's reading of § 1639d(g)(3)'s legislative history. The House Report 'shows Congress's view that creditors, including large corporate banks . . . can comply with state escrow interest laws without any significant interference with their banking powers.” 883 F.3d at 1196 (emphasis added). PreCantero , we saw Lusnak 's language for what it is: 'unqualified.' Kivett I , 2022 WL 1553266, at *1. It does not get more categorical than that.

And the United States reads Lusnak the same way. While both this case and Cantero were pending at the certiorari Page 34stage, the United States agreed with Flagstar that Lusnak was wrongly decided. Br. for U.S. as Amicus Curiae at 19–20, Flagstar Bank, N.A. v. Kivett , 144 S. Ct. 2628 (2024) (No. 22-349). Lusnak 'erred in treating Section 1639d as determinative of the preemption question.' Id. at 20. In other words, Lusnak 'elided' the 'practical, degree-ofinterference assessment” that Dodd-Frank “requires.” Id. at 19.

As the majority notes, Lusnak mentioned in a footnote that a state law setting “punitively high” interest rates could theoretically prevent or significantly interfere with a national bank’s powers. Maj. Op. at 7 (citing Lusnak , 883 F.3d at 1195 n.7). Lusnak never explained what constitutes a punitively high interest rate.² And Lusnak never squared that caveat with its otherwise categorical reasoning. But even taking Lusnak at its word that a state IOE law may be preempted in the right circumstances, our decision still does not align with Cantero . In the same footnote, Lusnak reiterated that § 1639d(g)(3) 'reflects a determination that state [IOE] laws do not necessarily prevent or significantly interfere with a national bank’s business.” 883 F.3d at 1195 n.7. Yet again, that is not the comparative analysis that Cantero prescribes.

In fairness, the majority acknowledges that Lusnak “did not cite or discuss' the six preemption cases addressed in Barnett Bank , even though Cantero 'admonishes' courts to Page 35do just that. Maj. Op. at 14–15. The majority then conducts a preliminary comparative analysis based on the two precedents— Franklin and Anderson -that it thinks are 'most applicable' to this case. Id. at 15-18. But the majority finds itself in 'equipoise,' concluding that comparison to past preemption cases (i.e., what Cantero demands) does not 'seem[] to compel the result here.' Id. at 17. From that the majority suggests that Lusnak -which the majority concedes did not conduct a comparative analysis-is not clearly irreconcilable with Cantero .

But we consider whether the new authority “undercut[s] the theory or reasoning underlying the prior circuit precedent in such a way that the cases are clearly irreconcilable.” Miller , 335 F.3d at 900 (emphasis added); see Maj. Op. at 14. We are bound by the Supreme Court’s 'mode of analysis,' not just its holdings. Miller , 335 F.3d at 900 (quoting Antonin Scalia, The Rule of Law as a Law of Rules , 56 U. Chi. L. Rev. 1175, 1177 (1989)); see id. (lower courts must adhere to the Supreme Court’s “explications of the governing rules of law” (quoting Cnty. of Allegheny v. ACLU Greater Pittsburgh Chapter , 492 U.S. 573, 668 (1989) (Kennedy, J., concurring in part and dissenting in part))). A three-judge panel can hold that the Court's articulation of a legal standard is clearly irreconcilable with circuit precedent, while reaching the same result as it would have otherwise. See id.

The majority ultimately holds that Lusnak remains binding precedent because 'nothing in Cantero suggests” that its nuanced comparative analysis is “the sole method for determining preemption.” Maj. Op. at 18. More to the point, the majority concludes that '[n]othing in Cantero casts doubt on our power to use the full array of interpretive tools in preemption analysis.” Id.

Page 36Cantero says otherwise. “Under Dodd-Frank . . . courts may find a state law preempted ‘ only if ,’ ‘in accordance with the legal standard’ from Barnett Bank , the law 'prevents or significantly interferes with the exercise by the national bank of its powers.’” 602 U.S. at 221 (emphasis added) (quoting § 25b(b)(1)(B)). What is the legal standard from Barnett Bank ? A 'nuanced comparative analysis' based on the Court's prior bank preemption cases. Id. at 220.

Cantero does not allow for anything else, as the majority claims. Quite the opposite. The Court repeated that a “court applying [the] Barnett Bank standard must make a practical assessment of the nature and degree of the interference caused by a state law.” Id. at 219–20. And “[g]iven DoddFrank's direction to identify significant interference 'in accordance with’ Barnett Bank , courts addressing preemption questions in this context must do as Barnett Bank did and likewise take account of those prior decisions of [the Supreme] Court and similar precedents.' Id. at 215-16 (emphasis added) (quoting § 25b(b)(1)(B)). The Court never suggested that we can bypass this analysis-which Congress enshrined in the U.S. Code 3 —for other “interpretive tools.” Maj. Op. at 18. By allowing preemption “only if” it is “in accordance with” the standard Page 37from Barnett Bank , Congress did not permit us to use whatever interpretive tools we find most relevant. See 12 U.S.C. § 25b(b)(1)(B). And if the Court thought that DoddFrank allows for the 'full array' of preemption analyses, Maj. Op. at 18, then why explain what courts 'must do' when 'addressing preemption questions in this context'? Cantero , 602 U.S. at 215–16.⁴

The majority notes that Cantero identified other matters that the Second Circuit 'may address as appropriate on remand.” Maj. Op. at 18 (quoting 602 U.S. at 221 n.4). The implication is that Cantero did not adopt the comparative analysis as the preemptive be-all-and-end-all. The majority, however, leaves out what those matters are. The Second Circuit was free to address the OCC’s preemption rules and a Dodd-Frank provision that allows preemption of state law by federal provisions outside of 'title 62 of the Revised Statutes.' See 12 U.S.C. § 25b(b)(1)(B)–(C). Those matters prove my point. Both derive from Dodd-Frank’s codified preemption standard, not some extra-textual interpretive device that Congress has not approved for use in banking preemption cases. The OCC can make 'preemption determination[s]' on a 'case-by-case basis,' but only 'in accordance with the legal standard” from Barnett Bank . Id. § 25b(b)(1)(B). And § 25b(b)(1)(C) authorizes an additional Page 38statutory pathway (besides under Barnett Bank ) for federal preemption of state banking laws. Id. § 25b(b)(1)(C) (“State consumer financial laws are preempted, only if . . . the State consumer financial law is preempted by a provision of Federal law other than title 62 of the Revised Statutes.'); see also supra , at 23 n.1. Thus, Cantero 's reference to other issues on remand only supports its holding that preemption determinations must conform to what Congress authorized in Dodd-Frank- a nuanced comparative analysis under Barnett Bank .

We should have treated Lusnak as effectively overruled. Granted, this is a “high standard” to meet. Lair v. Bullock , 697 F.3d 1200, 1207 (9th Cir. 2012) (quotation omitted). But this is not a situation where there is merely “some tension” between the cases or where a new Supreme Court decision simply 'cast[s] doubt' on our precedent. Id. (quotations omitted). Nor is this a matter of 'decid[ing] a case differently than a prior panel,” as the majority alleges. See Maj. Op. at 11–12 (quoting Close v. Sotheby' s, Inc. , 894 F.3d 1061, 1073–74 (9th Cir. 2018)). The question is whether we can apply Lusnak 'without running afoul' of Cantero . Lair , 697 F.3d at 1207 (cleaned up). Because Lusnak did not “do as Barnett Bank did and likewise take account of' the Court's prior bank preemption cases, the answer is no. 5 Cantero , 602 U.S. at 215–16.

Page 39B

As discussed above, Lusnak is clearly irreconcilable with Cantero . Regardless of whether Cantero satisfies Miller v. Gammie , however, Lusnak was wrongly decided. And Cantero explains why. The NBA preempts California’s IOE law under Cantero ’s comparative framework. That analysis is straightforward: California’s IOE law significantly interferes with national banking powers like the preempted state laws in Barnett Bank . Thus, California's IOE law is preempted in line with Cantero ’s comparative methodology.

Begin with Franklin , which Cantero called the 'paradigmatic example of significant interference.' 602 U.S. at 216. New York prohibited most banks “from using the word 'saving' or ‘savings’ in their advertising or business.' Id. (quoting Franklin , 347 U.S. at 374). As Cantero explained, New York's advertising restriction 'significantly interfered' with national banks' statutory powers because it prevented the use of a ''commonly understood description which Congress has specifically selected’ to describe [the banks’] activities: receiving savings deposits.” Id. (quoting Franklin , 347 U.S. at 378). Because New York 'could not interfere with the nationalPage 40 bank’s ability to [advertise] efficiently,” its law was preempted. Id.

California’s IOE law is far more disruptive to national banking powers than the law in Franklin . The interference stemming from an advertising restriction pales in comparison to a state law that dictates a national bank's pricing of its mortgage products. Cf. Monroe Retail, Inc. v. RBS Citizens, N.A. , 589 F.3d 274, 283 (6th Cir. 2009) (“[T]he level of ‘interference’ that gives rise to preemption under the NBA is not very high.” (quotation omitted)). A national bank operating in California can only offer mortgage escrow accounts if it pays at least 2% interest on escrowed funds, a rate far higher than what the market may otherwise demand. See Cal. Civ. Code § 2954.8(a); see also supra , at 34 n.2. A state law requiring national banks to pay extra to conduct mortgage lending in the state no doubt interferes with both the enumerated power to administer home mortgage loans, 12 U.S.C. § 371(a), and the incidental power to provide and service mortgage escrow accounts, OCC Inter. Ltr. 1041, 2005 WL 3629258, at *2. See Watters , 550 U.S. at 13 (states “may not curtail or hinder a national bank’s efficient exercise of any . . . power, incidental or enumerated under the NBA” (emphasis added)).

It is hard to see how preventing national banks from setting their own prices is not a significant interference with their enumerated or incidental powers. If state IOE laws make it more expensive for national banks to administer escrow accounts, then banks must offset the costs to ensure sufficient returns, either by charging higher interest rates on mortgage loans or requiring larger down payments. And by increasing underwriting costs, California's IOE law may 'reduce lending' by national banks, 'particularly to highrisk borrowers.' McShannock v. JP Morgan Chase Bank, Page 41N.A. , 976 F.3d 881, 893–94 (9th Cir. 2020) (California’s IOE law is preempted under the “less onerous” standard for the federal Home Owners’ Loan Act). A state law that alters a national bank’s pricing almost by definition interferes more with the bank’s powers than a simple advertising restriction.

Indeed, our court and others regularly find federal preemption in cases involving national banks’ pricing schemes. In Martinez v. Wells Fargo Home Mortgage, Inc. , we found preempted a plaintiff’s claim that a national bank overcharged underwriting and tax service fees in violation of California's UCL. 598 F.3d 549, 555-56 (9th Cir. 2010). And in Bank of America v. City & County of San Francisco , we held that the NBA preempted municipal ordinances prohibiting banks from charging ATM fees to nondepositors. 309 F.3d 551, 561–64 (9th Cir. 2002); see also Baptista v. JPMorgan Chase Bank, N.A. , 640 F.3d 1194, 1198 (11th Cir. 2011) (preemption where the 'state's prohibition on charging fees to non-accountholders . . . substantial[ly] conflict[ed] with federal authorization to charge such fees”). These cases all involve efforts to dictate how much a national bank can charge for its banking products. So too here.

There is also Fidelity , another case that Cantero identified as a pro-preemption example of significant interference with national banking powers. Federal law authorized federal savings and loans to include due-on-sale clauses in their contracts. Fidelity , 458 U.S. at 154. Yet California law “limited” that right to times when “enforcing the due-on-sale clause was reasonably necessary.” Cantero , 602 U.S. at 216–17 (quoting Fidelity , 458 U.S. at 154–55). The Court considered the California law preempted because it barred a federal savings and loan from exercising a dueon-sale clause 'solely at its option,' even though it could Page 42comply with both the state and federal laws. Id. at 217 (quoting Fidelity , 458 U.S. at 155). The state law 'thus interfered with ‘the flexibility given’ to the savings and loan by federal law.' Id. (quoting Fidelity , 458 U.S. at 155).

California’s IOE law similarly interferes with 'the flexibility given” to national banks in the administration of mortgage escrow accounts. See id. (quoting Fidelity , 458 U.S. at 155). RESPA does not require lenders to pay interest on escrowed funds. And even TILA recognizes that lenders can decide the terms of escrow accounts for mortgage loans that do not require escrow-interest payments under § 1639d(g)(3). See 15 U.S.C. § 1639d(f)(1)-(2) (for nonTILA mortgages, nothing in the statute precludes establishing escrow accounts “on terms mutually agreeable to the parties to the loan” or “at the discretion of the lender or servicer”). By mandating interest payments on escrowed funds, California’s law undercuts the flexibility that federal law affords to national lenders who offer mortgage escrow accounts. Like Franklin , Fidelity favors preemption here.

First National Bank of San Jose —the final propreemption case identified in Cantero -cuts both ways. California law allowed the State to seize unclaimed deposits after 20 years without proof of abandonment. 262 U.S. at 366. The law therefore “attempt[ed] to qualify in an unusual way agreements between national banks and their customers.” Cantero , 602 U.S. at 218 (quoting First Nat’l , 262 U.S. at 370). As Cantero explained, this qualification “could cause customers to ‘hesitate’ before depositing funds at the [national] bank—and thus interfere with the ‘efficiency’ of the national bank in receiving deposits.” Id. (quoting First Nat’l , 262 U.S. at 369–70).

Page 43California's IOE law undermines the 'efficiency' of Flagstar’s exercise of its national banking powers by requiring the bank to comply with IOE requirements for some, but not all its mortgages (i.e., only for properties located in California). And California’s law, like the law in First National , qualifies the terms of mortgage lending agreements between national banks and their customers. But First National focused on the California law's 'deterrent effect” on potential national bank customers. Id. An IOE law that benefits borrowers by requiring lenders to pay interest on escrowed funds lacks a similar deterrent effect. So First National is a draw under Cantero ’s comparative analysis.

Even so, the significant interference caused by California’s IOE law is still more analogous to the state laws in Barnett Bank 's pro-preemption cases. On the other hand are cases like Anderson , the 'primary example of a case where state law was not preempted.' Id. at 217. The Kentucky law in Anderson required banks 'to turn over abandoned deposits to the State.” Id. (citing Anderson , 321 U.S. at 236). The Court explained that 'an inseparable incident” of a national bank’s power to accept deposits is the “obligation to pay” those deposits “to the persons entitled to demand payment according to the law of the state where [the bank] does business.” Id. (quoting Anderson , 321 U.S. at 248–49). “And [the] Kentucky law simply allowed the State to ‘demand payment of the accounts in the same way and to the same extent that the depositors could’ after the depositors abandoned the account.” Id. at 217–18 (quoting Anderson , 321 U.S. at 249). Put simply, Kentucky's law could not 'impose an undue burden' on the operations of national banks because it reflected a rule “as old as the common law itself.” Anderson , 321 U.S. at 248, 251. The law requiredPage 44 'nothing more than performance of a duty by the bank [already] imposed by the federal banking laws.” Id. at 252.

California’s IOE law does not require national banks to perform a duty already required by federal law. Federal law does the opposite. Unless TILA’s exception applies, federal law does not mandate interest payments on escrowed funds. Nor is an IOE requirement “as old as the common law itself.” Id. at 251. So Anderson is not analogous.

Neither is Commonwealth , another case that Barnett Bank cited as an example of a state law that could apply to national banks. The generally applicable law there “taxed the shareholders of all banks (including national banks) on their shares of bank stock.” Cantero , 602 U.S. at 218-19 (citing Commonwealth , 76 U.S. at 360). The Court 'explained that national banks are 'exempted from State legislation, so far as that legislation may interfere with, or impair their efficiency in performing the functions' that federal law authorizes them to perform.' Id. at 219 (quoting Commonwealth , 76 U.S. at 362). But, as Cantero emphasized, national banks 'remain subject to state law governing ‘their daily course of business’ such as generally applicable state contract, property, and debt-collection laws.” 6 Id. (quoting Commonwealth , 76 U.S. at 361-62). Because the Commonwealth law ''in no manner hinder[ed]' the national bank's banking operations, and produced 'no greater interference with the functions of the bank than any Page 45other’ law governing businesses, the law was not preempted.” Id. (quoting Commonwealth , 76 U.S. at 362– 63).

The same cannot be said of California’s IOE law. Unlike the law in Commonwealth , and contrary to the majority's suggestion, California’s IOE law is not a generally applicable business statute. Maj. Op. at 17. Rather, California’s law is a banking-specific provision that hinders a national bank’s exercise of its banking powers. See Cal. Civ. Code § 2954.8(a) (limiting the statute to 'financial institution[s]”). So Commonwealth is inapt.

McClellan is much the same. That case involved another generally applicable law that voided any transfer of property by any person or entity in cases of insolvency. 164 U.S. at 348-49. The McClellan Court recognized that generally applicable state contract laws 'could be said to act as 'a restraint upon the power of a national bank within the State to make such contracts.’” Cantero , 602 U.S. at 219 (quoting McClellan , 164 U.S. at 358). Still, 'such state laws could apply to national banks as long as the state laws did not ‘in any way impai[r] the efficiency of national banks or frustrat[e] the purpose for which they were created.'' Id. (quoting McClellan , 164 U.S. at 358).

Again, California’s IOE law is not a “generally applicable contract law' like the law in McClellan . Id. And California’s IOE law—by making it more costly for national banks to offer and service mortgage escrow accounts for properties located in California-'frustrat[es] the purpose for which' the national banking system was created. Id. (quotation omitted). If the NBA did not preempt IOE laws like California’s, national banks would be subject to '[d]iverse and duplicative' state regulation of mortgage Page 46escrow accounts, which is 'precisely what the NBA was designed to prevent.” Watters , 550 U.S. at 13–14; see also Easton v. Iowa , 188 U.S. 220, 229 (1903) (the national banking system was meant to be “independent” of legislation which “might impose limitations and restrictions as various and as numerous as the states”). As the Supreme Court has routinely explained, Congress did not intend “to leave the field open for the States to attempt to promote the welfare and stability of national banks by direct legislation,” given the “[c]onfusion [that] would necessarily result from control possessed and exercised by two independent authorities.' Watters , 550 U.S. at 14 (quoting Easton , 188 U.S. at 231– 32). On these facts, McClellan points to preemption.

This is the analysis that we should have done. Using the language of Cantero , the significant interference stemming from California’s IOE law is “more akin” to the interference in Franklin and Fidelity . 602 U.S. at 220. And it is less analogous to the interference in Anderson , Commonwealth , and McClellan . That means California's IOE law is preempted.

Cantero controls this case. Its comparative methodology bears no resemblance to Lusnak 's categorical test, so much so that Lusnak has been 'effectively overruled.' Miller , 335 F.3d at 900. Regardless, following Cantero ’s comparative analysis, we should have held that the NBA preempts California’s IOE law. I respectfully dissent.