OPINION

WERDEGAR, J. —

Plaintiff Timothy Sandquist and the various defendants here are parties to an arbitration agreement. A salient question is whether that agreement permits or prohibits arbitration on a classwide basis. Here we must answer a question one step removed — who decides whether the agreement permits or prohibits classwide arbitration, a court or the arbitrator? The question has divided the many state and federal courts to consider it.

We conclude no universal rule allocates this decision in all cases to either arbitrators or courts. Rather, who decides is in the first instance a matter of agreement, with the parties' agreement subject to interpretation under state contract law. Under state law, these parties' arbitration agreement allocates the decision to the arbitrator. Under federal arbitration law, no contrary presumption requires a different result, so the issue remains one for the arbitrator. Because the Court of Appeal arrived at a similar answer, we affirm.

FACTUAL AND PROCEDURAL BACKGROUND

The material facts are not in dispute. In 2000, plaintiff Timothy Sandquist was hired by defendants (collectively, Lebo or Lebo Automotive) to work as a salesperson at an automotive dealership.1 On Sandquist's first day, his manager gave him approximately 100 pages of preprinted forms with instructions to fill out and sign each document as quickly as possible so that Sandquist could begin work. The documents were not discussed with Sandquist, but he was required to sign them as a condition of employment. Included among the documents were three different form arbitration agreements. Under time pressure, Sandquist finished the paperwork as quickly as possible, without reviewing each document, and did not realize he was signing multiple arbitration agreements.

In 2012, Sandquist, who is African-American, sued Lebo Automotive. The operative complaint alleges Sandquist and other non-Caucasian employees were subjected to racial discrimination, harassment, and retaliation. The complaint seeks to bring claims on behalf of "a class of current and former employees of color." It includes an individual claim for constructive discharge and class claims for discrimination and creation of a hostile work environment under the California Fair Employment and Housing Act (Gov. Code, § 12940 et seq.) and unfair competition law (Bus. & Prof. Code, § 17200 et seq.). The complaint seeks injunctive and declaratory relief and damages.

Lebo Automotive moved to compel individual arbitration based on the arbitration agreements signed by Sandquist on his first day of work. (See Code Civ. Proc., § 1281.2.) Finding the agreements enforceable and not unconscionable and the instant dispute within their scope, the trial court granted the motion. The court also interpreted Stolt-Nielsen S. A. v. AnimalFeeds Int'l Corp. (2010) 559 U.S. 662 [176 L.Ed.2d 605, 130 S.Ct. 1758] and Kinecta Alternative Financial Solutions, Inc. v. Superior Court (2012) 205 Cal.App.4th 506 [140 Cal.Rptr.3d 347] as requiring the court to decide whether class arbitration was available. Because in the trial court's view the agreements did not permit class arbitration, it struck the class allegations under Code of Civil Procedure section 436. Although the trial court granted Sandquist leave to amend and time to find a substitute class representative, when he advised the court every employee at the dealership was subject to the same arbitration agreements, it dismissed the class claims with prejudice.

On appeal, the Court of Appeal reversed in part. It declined to address Sandquist's claim that the arbitration agreements were unconscionable because that ruling was not appealable (State Farm Fire & Casualty v. Hardin (1989) 211 Cal.App.3d 501, 505-506 [259 Cal.Rptr. 433]), but it considered his challenge to the dismissal of class allegations under the death knell doctrine (see In re Baycol Cases I & II (2011) 51 Cal.4th 751, 762 [122 Cal.Rptr.3d 153, 248 P.3d 681]). Disagreeing with the trial court's conclusion that existing precedent compelled the court to determine whether class arbitration was available, the Court of Appeal found the issue an open and unsettled one. It examined cases on each side of the existing divide and concluded the availability of class proceedings under an arbitration agreement is a question of contract interpretation for the arbitrator to decide in the first instance.

Lebo Automotive petitioned for review, contending the Court of Appeal's decision contributed to an existing state and federal split over who should decide whether an arbitration agreement permits class arbitration. We granted review.

DISCUSSION

I. State Law and the Parties' Arbitration Clauses

A. What the Arbitrator May Decide Is Initially a Matter of Agreement Under State Law

(1) The issue before us is not whether class arbitration is permissible here, but a matter antecedent to that issue: who should decide whether it is permissible, a court or an arbitrator. No universal one-size-fits-all rule allocates that question to one decision maker or the other in every case. Rather, "who decides" is a matter of party agreement. As the United States Supreme Court has explained in a closely related context, "[j]ust as the arbitrability of the merits of a dispute depends upon whether the parties agreed to arbitrate that dispute [citations], so the question `who has the primary power to decide arbitrability' turns upon what the parties agreed about that matter." (First Options of Chicago, Inc. v. Kaplan (1995) 514 U.S. 938, 943 [131 L.Ed.2d 985, 115 S.Ct. 1920].) And just as whether class arbitration is available depends on whether the parties agreed to allow or forbid it (Stolt-Nielsen S. A. v. AnimalFeeds Int'l Corp., supra, 559 U.S. at pp. 684-687), so the question who has the power to decide the availability of class arbitration turns upon what the parties agreed about the allocation of that power.

The United States Supreme Court's treatment of the issue confirms the parties' agreement as the mandatory starting point. In Green Tree Financial Corp. v. Bazzle (2003) 539 U.S. 444 [156 L.Ed.2d 414, 123 S.Ct. 2402] (Green Tree), the plurality and the principal dissent, although disagreeing about the ultimate "who decides" question, both agreed about where the analysis should begin. (See id. at p. 451 (plur. opn. of Breyer, J.) [concluding the question should be for the arbitrator because "[u]nder the terms of the parties' contracts, the question — whether the agreement forbids class arbitration — is for the arbitrator to decide"]; id. at p. 456 (dis. opn. of Rehnquist, C. J.) [agreeing that "the decision of what to submit to the arbitrator is a matter of contractual agreement by the parties"].) Similarly, in Stolt-Nielsen S. A. v. AnimalFeeds Int'l Corp., supra, 559 U.S. at page 680 and Oxford Health Plans LLC v. Sutter (2013) 569 U.S. ___, ___, fn. 2 [186 L.Ed.2d 113, 119, fn. 2, 133 S.Ct. 2064, 2068, fn. 2], the Supreme Court accepted — because the parties had so agreed — that an arbitrator should decide in the first instance whether class arbitration was available.

(2) Consequently, we must examine the parties' agreements to determine what they say concerning the "who decides" question. But under what body of law? Sandquist argues federal law governs exclusively, while Lebo Automotive urges state law does. We agree with Lebo: this examination must be conducted, at least initially, through the prism of state law. "When deciding whether the parties agreed to arbitrate a certain matter (including arbitrability), courts generally ... should apply ordinary state-law principles that govern the formation of contracts." (First Options of Chicago, Inc. v. Kaplan, supra, 514 U.S. at p. 944; see DIRECTV, Inc. v. Imburgia (2015) 577 U.S. ___, ___ [193 L.Ed.2d 365, 372, 136 S.Ct. 463, 468]; Volt Info. Sciences v. Leland Stanford Jr. U. (1989) 489 U.S. 468, 474 [103 L.Ed.2d 488, 109 S.Ct. 1248].) This default choice-of-law rule applies to the question whether the availability of class arbitration is for an arbitrator or a court; insofar as the question who decides "presents a disputed issue of contract interpretation," "state law, not federal law, normally governs such matters." (Green Tree, supra, 539 U.S. at p. 450 (plur. opn. of Breyer, J.); accord, id. at p. 454 (conc. opn. of Stevens, J.) [applying state law]; id. at pp. 457-458 (dis. opn. of Rehnquist, C. J.) [agreeing that "`[s]tates may regulate contracts, including arbitration clauses, under general contract law principles'" and "`[t]he interpretation of private contracts is ordinarily a question of state law'"].)

The parties do not disagree as to which state's law applies: California. The arbitration clauses were entered into in California, govern an employment relationship between a California resident and a company with its sole place of business in California, and invoke various provisions of California law throughout.

B. The Application of State Contract Law to These Agreements

Lebo Automotive had Sandquist sign three different form agreements. Each contains an arbitration clause. The language in each clause defining the scope of arbitrable matters varies slightly, but is materially the same.

The clause in the "Applicant's Statement & Agreement" provides in part: "I and the Company both agree that any claim, dispute, and/or controversy (including, but not limited to, any claims of discrimination and harassment, whether they be based on the California Fair Employment and Housing Act, as well as all other applicable state or federal laws or regulations) which would otherwise require or allow resort to any court or other governmental dispute resolution forum between myself and the Company (or its owners, directors, officers, managers, employees, agents, and parties affiliated with its employee benefit and health plans) arising from, related to, or having any relationship or connection whatsoever with my seeking employment with, employment by, or other association with the Company, whether based on tort, contract, statutory, or equitable law, or otherwise, (with the sole exception of claims arising under the National Labor Relations Act which are brought before the National Labor Relations Board, claims for medical and disability benefits under the California Workers' Compensation Act, and Employment Development Department claims) shall be submitted to and determined exclusively by binding arbitration."

The clause in the "Mandatory Arbitration Agreement" provides: "I agree that any claim, dispute, and/or controversy (including, but not limited to any claims of discrimination and harassment) which would otherwise require or allow resort to any court or other governmental dispute resolution forum, between me and the Company (or its owners, directors, officers, managers, employees[,] agents, and parties affiliated with its employee benefits and health plans) arising from, related to, or having any relationship or connection whatsoever with my seeking employment with, employment by, or other association with, the Company, whether based on tort, contract, statutory, or equitable law, or otherwise, shall be submitted to and determined exclusively by binding arbitration...."

Finally, the clause in the "Employee Acknowledgement and Agreement" provides: "I agree that any claim, or dispute, or controversy (including, but not limited to, any and all claims of discrimination and harassment) which would otherwise require or allow resort to any court or other governmental dispute resolution forum between myself and the Company (or its owners, directors, officers, managers, employees, agents, and parties affiliated with its employee benefit and health plans) arising from, related to, or having any relationship or connection whatsoever with my seeking employment with, employment by, or other association with the Company, whether based on tort, contract, statutory, or equitable law, or otherwise, (with the sole exception of claims arising under the National Labor Relations Act which are brought before the National Labor Relations Board, claims for medical and disability benefits under the California Workers Compensation Act, and Employment Development Department claims), shall be submitted to and determine[d] exclusively by binding arbitration...."

All three arbitration provisions share the same basic structure and much of the same language. All three contain two inclusive clauses that define the range of disputes that must be "submitted to and determined exclusively by binding arbitration." Two of the three add an exclusive clause that sets out a specific, limited set of disputes, otherwise covered by the clause's inclusive language, that are nevertheless withdrawn from the arbitrator's purview.

First, the provisions extend to "any claim, dispute, and/or controversy (including, but not limited to any [and all] claims of discrimination and harassment) which would otherwise require or allow resort to any court or other governmental dispute resolution forum, between [me/myself] and the Company." This language is comprehensive. If a dispute or controversy is between Sandquist and Lebo Automotive, as the one before us surely is, and if it might otherwise be permissibly submitted to a court, as the question whether class arbitration is available surely could be, this portion of the arbitration clause suggests a choice to have the decision made by an arbitrator.

Second, the provisions extend to all claims "arising from, related to, or having any relationship or connection whatsoever with my seeking employment with, employment by, or other association with the Company, whether based on tort, contract, statutory, or equitable law, or otherwise." (Italics added.) The underlying claims in the first amended complaint assert that Lebo Automotive harassed and discriminated against Sandquist on the basis of race in the course of his employment, created a hostile work environment, and ultimately constructively discharged him. They plainly arise from Sandquist's employment with Lebo Automotive. The procedural question those claims present — whether Sandquist may pursue his claims on a class basis — directly arises from his underlying claims. Given that the provisions are intended to sweep in disputes "having any relationship or connection whatsoever" with Sandquist's employment, that the issue before us arises from a lawsuit over Sandquist's employment would appear enough to satisfy this nexus requirement.

Finally, both the Applicant's Statement & Agreement and the Employee Acknowledgement and Agreement (although not the Mandatory Arbitration Agreement) contain an additional clause identifying specific disputes otherwise within the broad inclusive clauses of the arbitration provisions but intended not to be arbitrable. Every dispute within those inclusive clauses is for the arbitrator, "with the sole exception of claims arising under the National Labor Relations Act which are brought before the National Labor Relations Board, claims for medical and disability benefits under the California Workers' Compensation Act, and Employment Development Department claims." The drafter of these agreements might well have specified other matters not for the arbitrator, such as the availability of class arbitration at issue here, but did not.

(3) These features of the arbitration clauses suggest the "who decides" question is an arbitrable one, but they are by no means conclusive. In the presence of ambiguity, we turn to other principles applicable to the interpretation of arbitration clauses and contracts generally.

(4) When construing arbitration provisions, we must consider the parties' likely expectations about allocations of responsibility. (See Howsam v. Dean Witter Reynolds, Inc. (2002) 537 U.S. 79, 83, 86 [154 L.Ed.2d 491, 123 S.Ct. 588].) "Typically, those who enter into arbitration agreements expect that their dispute will be resolved without necessity for any contact with the courts." (Blanton v. Womancare, Inc. (1985) 38 Cal.3d 396, 402, fn. 5 [212 Cal.Rptr. 151, 696 P.2d 645], italics added.) In the many cases where the parties agree an underlying dispute is arbitrable and thus begin with a filing before an arbitrator, to resolve the "who decides" question in favor of a court would contravene that expectation and impose substantial additional cost and delay, requiring the parties to stay matters before the arbitrator, proceed to a courthouse for a construction of their arbitration agreement, perhaps continue through appellate review of that construction, and only then return back to arbitration for further dispute resolution. Here, the parties' arbitration provision declares a preference for arbitration because of its "reduced expense and increased efficiency." (See Hall Street Associates, L.L.C. v. Mattel, Inc. (2008) 552 U.S. 576, 588 [170 L.Ed.2d 254, 128 S.Ct. 1396] [touting "arbitration's essential virtue of resolving disputes straightaway"].) But "[r]eferring preliminary issues to the courts can cause `"serious delay and confusion, thus robbing the arbitration procedure of much of its value to the parties."'" (Ericksen, Arbuthnot, McCarthy, Kearney & Walsh, Inc. v. 100 Oak Street (1983) 35 Cal.3d 312, 323 [197 Cal.Rptr. 581, 673 P.2d 251].) We will not lightly assume Lebo, or Sandquist, would have expected or preferred a notably less efficient allocation of decisionmaking authority.

(5) Ultimately dispositive here are two other long-established interpretive principles. First, under state law as under federal law, when the allocation of a matter to arbitration or the courts is uncertain, we resolve all doubts in favor of arbitration. (Wagner Construction Co. v. Pacific Mechanical Corp. (2007) 41 Cal.4th 19, 26 [58 Cal.Rptr.3d 434, 157 P.3d 1029]; Engalla v. Permanente Medical Group, Inc. (1997) 15 Cal.4th 951, 971 [64 Cal.Rptr.2d 843, 938 P.2d 903]; see Mastrobuono v. Shearson Lehman Hutton, Inc. (1995) 514 U.S. 52, 62 & fn. 8 [131 L.Ed.2d 76, 115 S.Ct. 1212].) All else being equal, this presumption tips the scales in favor of allocating the class arbitration availability question to the arbitrator.

(6) Second, ambiguities in written agreements are to be construed against their drafters. (Civ. Code, § 1654; Rest.2d Contracts, § 206.) As the Restatement explains, "Where one party chooses the terms of a contract, he is likely to provide more carefully for the protection of his own interests than for those of the other party. He is also more likely than the other party to have reason to know of uncertainties of meaning. Indeed, he may leave meaning deliberately obscure, intending to decide at a later date what meaning to assert. In cases of doubt, therefore, so long as other factors are not decisive, there is substantial reason for preferring the meaning of the other party." (Rest.2d Contracts, § 206, com. a, p. 105; see Mastrobuono v. Shearson Lehman Hutton, Inc., supra, 514 U.S. at p. 63 ["The reason for this rule is to protect the party who did not choose the language from an unintended or unfair result."].)

(7) Thus, where, as here, the written agreement has been prepared entirely by the employer, it is a "well established rule of construction" that any ambiguities must be construed against the drafting employer and in favor of the nondrafting employee. (Pacific Lumber Co. v. Industrial Accident Com. (1943) 22 Cal.2d 410, 422 [139 P.2d 892]; accord, 24 Hour Fitness, Inc. v. Superior Court (1998) 66 Cal.App.4th 1199, 1215 [78 Cal.Rptr.2d 533].) Moreover, "[t]he rule requiring the resolution of ambiguities against the drafting party `applies with peculiar force in the case of a contract of adhesion. Here the party of superior bargaining power not only prescribes the words of the instrument but the party who subscribes to it lacks the economic strength to change such language.'" (Graham v. Scissor-Tail, Inc. (1981) 28 Cal.3d 807, 819, fn. 16 [171 Cal.Rptr. 604, 623 P.2d 165].) On the record before us, there is no dispute that the arbitration clauses were part of contracts of adhesion drafted by Lebo Automotive and imposed as conditions of employment.

This general principle of contract interpretation applies equally to the construction of arbitration provisions. (Mastrobuono v. Shearson Lehman Hutton, Inc., supra, 514 U.S. at pp. 62-63; Victoria v. Superior Court (1985) 40 Cal.3d 734, 739 [222 Cal.Rptr. 1, 710 P.2d 833]; 24 Hour Fitness, Inc. v. Superior Court, supra, 66 Cal.App.4th at p. 1215.) Where the drafter of a form contract has prepared an arbitration provision whose application to a particular dispute is uncertain, ordinary contract principles require that the provision be construed against the drafter's interpretation and in favor of the nondrafter's interpretation. (Victoria, at pp. 745-747.)

(8) Lebo Automotive could have written the description of matters within the arbitrator's purview less comprehensively. It could have prepared an arbitration provision that explicitly addressed any unstated desire to have the availability of class arbitration resolved by a court, notwithstanding the otherwise broad and all-encompassing language of the clause identifying matters for the arbitrator. It did not. To the extent that omission creates any ambiguity, it is Lebo that "drafted an ambiguous document, and ... cannot now claim the benefit of the doubt." (Mastrobuono v. Shearson Lehman Hutton, Inc., supra, 514 U.S. at p. 63.) We conclude, as a matter of state contract law, the parties' arbitration provisions allocate the decision on the availability of class arbitration to the arbitrator, rather than reserving it for a court.

C. There Is No Established Contrary State Law Presumption

Lebo Automotive argues that as a matter of state law these arbitration provisions should presumptively be read to allocate the class arbitration availability question to a court, absent any explicit commitment of the dispute to an arbitrator. Lebo rests this argument on our recent decision in City of Los Angeles v. Superior Court (2013) 56 Cal.4th 1086 [158 Cal.Rptr.3d 1, 302 P.3d 194]. We agree that the arbitration provisions, though most fairly read as allocating the class arbitration question to an arbitrator, do not explicitly do so. But we disagree that City of Los Angeles establishes any state law presumption that would in these circumstances require submission of the question to a court.

(9) City of Los Angeles involved the construction of an arbitration provision in a collectively bargained memorandum of understanding. The City of Los Angeles had responded to a fiscal crisis with unilateral furloughs. Various public employee unions filed grievances over the furloughs, followed by petitions to compel arbitration. No party disputed the proposition, well settled under both state and federal law, that absent the parties' commitment of the arbitrability decision to an arbitrator, disagreements over whether a particular dispute is within the scope of an arbitration provision are ordinarily the responsibility of a court. (City of Los Angeles v. Superior Court, supra, 56 Cal.4th at pp. 1093, 1096; see First Options of Chicago, Inc. v. Kaplan, supra, 514 U.S. at p. 944; Freeman v. State Farm Mut. Auto. Ins. Co. (1975) 14 Cal.3d 473, 479-480 [121 Cal.Rptr. 477, 535 P.2d 341].) City of Los Angeles broke no new ground on this point.

As Lebo Automotive asserts, City of Los Angeles confirms the strong presumption that courts should determine the jurisdiction of arbitrators. (See City of Los Angeles v. Superior Court, supra, 56 Cal.4th at p. 1096; Code Civ. Proc., § 1281.2 [on a petition to compel arbitration, "the court" should order arbitration "if it determines that an agreement to arbitrate the controversy exists"].) That is what we, a court, are tasked with doing here; consistent with the presumption, we are engaged in deciding whether a decision on the availability of class arbitration is for a court or an arbitrator.

(10) Lebo Automotive's argument rests on the assumption that as a matter of state law, the underlying question concerning the availability of class arbitration should itself be deemed a question of arbitrability for courts. In support, Lebo relies not only on City of Los Angeles but on the acknowledgement in Garcia v. DIRECTV, Inc. (2004) 115 Cal.App.4th 297, 298 [9 Cal.Rptr.3d 190] of pre-Green Tree state cases vesting the decision on the availability of class arbitration with trial courts. In the first of those earlier cases, this court held trial courts could, in their discretion, condition enforcement of an adhesive arbitration agreement on having the arbitration proceed on a class basis. (Keating v. Superior Court (1982) 31 Cal.3d 584, 608-614 [183 Cal.Rptr. 360, 645 P.2d 1192], revd. in part on other grounds sub nom. Southland Corp. v. Keating (1984) 465 U.S. 1 [79 L.Ed.2d 1, 104 S.Ct. 852].) But neither Keating nor cases following it (see Sanders v. Kinko's, Inc. (2002) 99 Cal.App.4th 1106, 1113-1114 [121 Cal.Rptr.2d 766]; Blue Cross of California v. Superior Court (1998) 67 Cal.App.4th 42, 60 [78 Cal.Rptr.2d 779]; Izzi v. Mesquite Country Club (1986) 186 Cal.App.3d 1309, 1319-1322 [231 Cal.Rptr. 315]; Lewis v. Prudential-Bache Securities, Inc. (1986) 179 Cal.App.3d 935, 945-946 [225 Cal.Rptr. 69]) acknowledged that the availability of class arbitration is in the first instance a matter of party consent,2 nor did they directly address the issue we face today and announce a state law pro-court, anti-arbitrator presumption. Neither City of Los Angeles nor any other state case establishes a presumption the availability of class arbitration is to be decided by courts.

(11) Nor need we decide, as a matter of first impression, whether state law embraces a particular pro-court or pro-arbitrator presumption. All three of the signed arbitration clauses here invoke the coverage of the Federal Arbitration Act. (9 U.S.C. § 1 et seq. (FAA).) While the parties have agreed to arbitrate "in conformity with the procedures of the California Arbitration Act," and while parties may elect to follow state procedures in lieu of the FAA's procedures (Volt Info. Sciences v. Leland Stanford Jr. U., supra, 489 U.S. at p. 477; DIRECTV, Inc. v. Imburgia, supra, 577 U.S. at p. ___ [193 L.Ed.2d at p. 372, 136 S.Ct. at p. 468]), the procedures the California Arbitration Act spells out do not specifically address the question of class arbitration availability (see Code Civ. Proc., § 1280 et seq.; cf. American Arbitration Assn., Supplementary Rules for Class Arbitration (2003) rule 3 [allocating the class availability question to the arbitrator]; JAMS, Class Action Procedures (2009) rule 2 [same]). Consequently, unlike in Volt Info. Sciences, this is not a case where the parties by agreement have expressly opted for the application of state law in lieu of any otherwise applicable federal rule. In the absence of such an opt out, the issue whether the availability of class arbitration is a question of arbitrability subject to a pro-court presumption arises in the shadow of federal law. Any state law presumption, were there one, would have to yield to whatever presumption the FAA establishes. We thus turn to federal law.

II. The FAA's Presumptions

(12) The crux of this case is whether the FAA imposes an interpretive presumption that, as a matter of federal law, preempts state law rules of contract interpretation and alters the conclusion state law would otherwise reach here. Congressional intent is the touchstone of any preemption analysis (e.g., Quesada v. Herb Thyme Farms, Inc. (2015) 62 Cal.4th 298, 318 [195 Cal.Rptr.3d 505, 361 P.3d 868]; Brown v. Mortenson (2011) 51 Cal.4th 1052, 1060 [126 Cal.Rptr.3d 428, 253 P.3d 522]), but neither the FAA's text nor its legislative history speaks to the question we face or offers insight as to that intent. Accordingly, we look to judicial understandings of the FAA.

The United States Supreme Court has directly addressed the "who decides" issue only once, in Green Tree, supra, 539 U.S. 444. The Green Tree parties' agreement submitted to the arbitrator "`[a]ll disputes, claims, or controversies arising from or relating to this contract or the relationships which result from this contract.'" (Id. at p. 451 (plur. opn. of Breyer, J.).) The plurality interpreted this language as allocating the class arbitration availability question to the arbitrator. (Id. at p. 452.) Furthermore, the same four justices explained, nothing in the FAA subjects the "who decides" question to any contrary pro-court presumption. (Green Tree, at pp. 452-453.) Accordingly, the plurality concluded "this matter of contract interpretation should be for the arbitrator, not the courts, to decide." (Id. at p. 453.)

(13) However, as the United States Supreme Court has twice reiterated in cases not directly addressing the "who decides" question, Green Tree contains no controlling view concerning what presumption, if any, the FAA requires when interpreting the parties' agreement as to who decides class arbitration availability. (See Stolt-Nielsen S. A. v. AnimalFeeds Int'l Corp., supra, 559 U.S. at pp. 678-679; Oxford Health Plans LLC v. Sutter, supra, 569 U.S. at p. ___, fn. 2 [186 L.Ed.2d at p. 119, fn. 2, 133 S.Ct. at p. 2068, fn. 2].) The issue thus remains unresolved. While accepting that extant Supreme Court cases do not decide the matter (dis. opn., post, p. 266), the dissent argues that from what they do decide, it is only a small leap to the conclusion that the availability of class arbitration is a question to presumptively withhold from the arbitrator (id. at p. 267). In the absence of controlling authority, however, we must resist the urge to take that leap without convincing evidence the FAA imposes an anti-arbitral presumption that overrides the state law reading of the parties' arbitration clause. In our view, the FAA does not.

The Supreme Court has interpreted the FAA as imposing two distinct presumptions, depending on the subject matter. "On the one hand, courts presume that the parties intend courts, not arbitrators, to decide ... disputes about `arbitrability,'" e.g., whether there is an enforceable arbitration agreement or whether it applies to the dispute at hand. (BG Group PLC v. Republic of Argentina (2014) 572 U.S. ___, ___ [188 L.Ed.2d 220, 228, 134 S.Ct. 1198, 1206]; see Howsam v. Dean Witter Reynolds, Inc., supra, 537 U.S. at pp. 83-84.) "On the other hand, courts presume that the parties intend arbitrators, not courts, to decide disputes about the meaning and application of particular procedural preconditions for the use of arbitration." (BG Group, at p. ___ [188 L.Ed.2d at p. 229, 134 S.Ct. at p. 1207]; see Howsam, at pp. 84-85.)

(14) These presumptions make sense; unless the parties specify otherwise, one would assume they should not be required to submit to an arbitrator the questions whether they have agreed to submit to an arbitration at all or arbitrate a given dispute. After all, because "[a]rbitration is strictly `a matter of consent'" (Granite Rock Co. v. Teamsters (2010) 561 U.S. 287, 299 [177 L.Ed.2d 567, 130 S.Ct. 2847], quoting Volt Info. Sciences v. Leland Stanford Jr. U., supra, 489 U.S. at p. 479), "`a party cannot be required to submit to arbitration any dispute which he has not agreed so to submit'" (Howsam v. Dean Witter Reynolds, Inc., supra, 537 U.S. at p. 83, quoting Steelworkers v. Warrior & Gulf Co. (1960) 363 U.S. 574, 582 [4 L.Ed.2d 1409, 80 S.Ct. 1347]). To presume arbitrability without first establishing, independently, consent to arbitration is to place the proverbial cart before the horse.

Once gateway questions of arbitrability have been settled, however, the FAA switches presumptions for issues affecting the manner in which an arbitration is to be conducted. "Thus `"procedural" questions which grow out of the dispute and bear on its final disposition' are presumptively not for the judge, but for an arbitrator, to decide." (Howsam v. Dean Witter Reynolds, Inc., supra, 537 U.S. at p. 84.) The pro-court presumption has only a "limited scope" (id. at p. 83), while the range of questions not qualifying as threshold gateway inquiries is correspondingly broad; for example, even a challenge to the enforceability of the entire contract in which an arbitration clause appears is presumptively for the arbitrator, not the court. (Buckeye Check Cashing, Inc. v. Cardegna (2006) 546 U.S. 440, 446 [163 L.Ed.2d 1038, 126 S.Ct. 1204].)

(15) The Green Tree plurality squarely considered in which category the question we face should fall, i.e., whether the availability of class arbitration should presumptively be for the court or the arbitrator. The plurality reasoned, persuasively in our view, that this question does not fall within the "narrow" class of questions subject to a pro-court presumption. (Green Tree, supra, 539 U.S. at p. 452.) Whether an agreement forbids class arbitration concerns "neither the validity of the arbitration clause nor its applicability to the underlying dispute between the parties." (Ibid.) It does not touch on any threshold matter necessary to establish as a condition precedent an agreement to arbitrate, but rather entails "what kind of arbitration proceeding the parties agreed to." (Ibid.) The question involves "contract interpretation and arbitration procedures. Arbitrators are well situated to answer that question." (Id. at p. 453.)

As the Green Tree plurality recognized, the availability or unavailability of class arbitration has nothing to do with whether the parties agreed to arbitrate, either in general or with respect to a specific dispute. Instead, the question is of the "what kind of proceeding" sort that arises subsequent to the gateway issue of whether to have an arbitral proceeding at all. (See AT&T Mobility LLC v. Concepcion (2011) 563 U.S. 333, 348-351 [179 L.Ed.2d 742, 131 S.Ct. 1740] [describing in detail how the shift to class arbitration involves a change in arbitral procedures].) Once the threshold matter of whether to arbitrate has been decided, assigning (presumptively, at least) any further questions turning on contract interpretation to the arbitrator is neither circular nor otherwise problematic. (See Employers Ins. of Wausau v. Century Indemnity (7th Cir. 2006) 443 F.3d 573, 578 [once it is established the parties have agreed to arbitrate a given dispute, questions about "the kind of arbitration proceeding" they have agreed to are "a matter of contract interpretation, which the arbitrator is well qualified to address"]; Shaw's Supermarkets v. United Food, Local 791 (1st Cir. 2003) 321 F.3d 251, 254 [once arbitrability is decided, propriety of consolidating multiple arbitrations "is a procedural matter for the arbitrator"].)

Notably, the principal Green Tree dissent did not disagree with the plurality on this point; that is, it did not contend that the availability of class arbitration is the sort of question that universally should be subject to a pro-court presumption.3 Rather, the principal dissent's rationale hinged on the language of a contract-specific provision relating to selection of the arbitrator that in their view should have required initial decisionmaking by a court. According to the dissent, the proper construction of a specific agreement concerning arbitrator selection is the sort of gateway matter that a court should resolve. (Green Tree, supra, 539 U.S. at p. 457 (dis. opn. of Rehnquist, C. J. [joined by O'Connor & Kennedy, JJ.]).) The arbitration agreement at issue in Green Tree contained a selection clause that, in the dissent's view, entitled the clause's drafter to choose a different arbitrator for every dispute, a right that allowing arbitration on a classwide basis before a single decision maker would have contravened. (Id. at p. 459.) Accordingly, under the specific circumstances of the Green Tree contract, the dissent contended class availability had to be resolved by a court.4

(16) One may understand the lines drawn by both the Green Tree plurality and the principal dissent as turning on which issues logically or necessarily must be resolved at the gateway, or threshold, before a case can feasibly go to an arbitrator. One such logical condition precedent is whether, in fact, the parties agreed to arbitrate at all; it makes no sense to compel parties to go before an arbitrator without first determining they agreed to do so. Another logically essential precedent matter is whether an instant dispute is within the scope of what the parties agreed to arbitrate; again, this question must be determined by someone other than the arbitrator — a court — before the dispute may be subject to arbitration. One might reasonably add to these, as the principal Green Tree dissent did, any disputes over how the arbitrator is to be chosen. (Green Tree, supra, 539 U.S. at p. 457 (dis. opn. of Rehnquist, C. J.).) What matters in each instance is not the stakes per se, but the logical relation of the issue as a condition precedent to an arbitration. No similar logical relation places the availability of classwide arbitration in the category of questions that must be resolved before any arbitrator will be able to entertain a dispute.

(17) Two other established FAA principles also weigh in favor of allocating the question to the arbitrator. First, a presumption that arbitrators decide the availability of class arbitration is more consistent with the desire for "expeditious results" that motivates many an arbitration agreement. (Mitsubishi Motors v. Soler Chrysler-Plymouth (1985) 473 U.S. 614, 633 [87 L.Ed.2d 444, 105 S.Ct. 3346]; see ante, pp. 246-247.) Congress intended through the FAA "to move the parties to an arbitrable dispute out of court and into arbitration as quickly and easily as possible." (Moses H. Cone Hospital v. Mercury Constr. Corp. (1983) 460 U.S. 1, 22 [74 L.Ed.2d 765, 103 S.Ct. 927]; accord, Preston v. Ferrer (2008) 552 U.S. 346, 357 [169 L.Ed.2d 917, 128 S.Ct. 978].) "Once it is determined ... that the parties are obligated to submit the subject matter of a dispute to arbitration, `procedural' questions which grow out of the dispute and bear on its final disposition should be left to the arbitrator." (John Wiley & Sons v. Livingston (1964) 376 U.S. 543, 557 [11 L.Ed.2d 898, 84 S.Ct. 909].) To reserve for a court more than the bare minimum inquiry necessary to confirm an enforceable agreement to arbitrate would entail a delay at odds with congressional pro-arbitration purposes. (Id. at p. 558; see Prima Paint v. Flood & Conklin (1967) 388 U.S. 395, 404 [18 L.Ed.2d 1270, 87 S.Ct. 1801] [the FAA embraces "the unmistakably clear congressional purpose that the arbitration procedure, when selected by the parties to a contract, be speedy and not subject to delay and obstruction in the courts"].)

(18) Second, among those principles most firmly established under the FAA is "that, as a matter of federal law, any doubts concerning the scope of arbitrable issues should be resolved in favor of arbitration." (Moses H. Cone Hospital v. Mercury Constr. Corp., supra, 460 U.S. at pp. 24-25.) Thus, if it is uncertain whether the issue is one for the courts or the arbitrator, we are well advised to allocate it to the arbitrator. This interpretive rule applies even here, where the uncertainty is over whether to place a question in the "presumptively for the court" or "presumptively for the arbitrator" category; if that question is close, as indeed it is, we resolve doubt in favor of placing the question in the "presumptively for the arbitrator" category. (See Green Tree, supra, 539 U.S. at p. 452 (plur. opn. of Breyer, J.); Shaw's Supermarkets v. United Food, Local 791, supra, 321 F.3d at pp. 254-255.)

In support of the opposite conclusion, Lebo Automotive relies on other recent appellate decisions that have concluded the availability of classwide arbitration is a "gateway question" for the court: Reed Elsevier, Inc. v. Crockett (6th Cir. 2013) 734 F.3d 594, 597 (Reed Elsevier), Opalinski v. Robert Half Internat. Inc. (3d Cir. 2014) 761 F.3d 326, 334 (Opalinski), and Garden Fresh Restaurant Corp. v. Superior Court (2014) 231 Cal.App.4th 678, 687 [180 Cal.Rptr.3d 89].

(19) Reed Elsevier reasons that the availability of class arbitration is an important question and is therefore a gateway question arbitrators should not be permitted to address unless the parties have expressly so provided. (Reed Elsevier, supra, 734 F.3d at p. 598 [identifying gateway questions for the court as those that are "fundamental" and subsidiary questions for an arbitrator as those that involve "details"].) But the distinction between gateway questions and nongateway questions, between questions for the court and for the arbitrator, is not one that focuses on a question's significance. Rather, the United States Supreme Court has drawn the line based not on importance, but on what the parties would have expected. (E.g., Howsam v. Dean Witter Reynolds, Inc., supra, 537 U.S. at pp. 83-84; First Options of Chicago, Inc. v. Kaplan, supra, 514 U.S. at pp. 944-945.) And as the First Circuit has said in applying this expectations test, "parties who have agreed to arbitrate a given subject most likely intend and expect that the arbitrator should resolve all issues that arise concerning that subject; if they do not, we think they would clearly express their contrary intent." (PaineWebber Inc. v. Elahi (1st Cir. 1996) 87 F.3d 589, 599.) Like the First Circuit, we decline to assume parties committing disputes to arbitration would have intended to confine arbitrators to the resolution of only subsidiary matters.

At oral argument, Lebo emphasized the higher stakes class arbitration may entail and the limited nature of appellate review (see AT&T Mobility LLC v. Concepcion, supra, 563 U.S. at pp. 348, 350; Reed Elsevier, supra, 734 F.3d at p. 598), and urged these features might reduce the likelihood that defendants, at least, would expect the availability of class arbitration to be resolved by an arbitrator. But every question resolved by an arbitrator is subject to limited review; forsaking broad appellate review is a routine part of the trade-off involved in opting for arbitration. (Mitsubishi Motors v. Soler Chrysler-Plymouth, supra, 473 U.S. at p. 628.) Nothing about the class arbitration availability question, among all the many questions that might be resolved by an arbitrator, supports presuming under federal law that parties otherwise eager to move dispute resolution out of the courtroom uniformly hold an unexpressed expectation, based on concerns about limited review, that the class availability question should be reserved for the courts. As for an increase in stakes, the procedural shift from multiple bilateral arbitrations to a single class arbitration does nothing to alter a defendant's potential aggregate liability. (Shady Grove Orthopedic Associates, P. A. v. Allstate Ins. Co. (2010) 559 U.S. 393, 408 [176 L.Ed.2d 311, 130 S.Ct. 1431] (plur. opn. of Scalia, J.).) To the extent the question whether such a shift is permitted by the parties' agreement may nevertheless be viewed as of exceeding importance, nothing in the FAA evinces a congressional purpose to keep higher stakes decisions from arbitrators, or to presume parties otherwise favorably disposed to arbitration would prefer that course.5

Garden Fresh Restaurant applies the same problematic logic as Reed Elsevier, reasoning from United States Supreme Court cases addressing the significance of the availability of classwide arbitration that the issue is one parties would prefer to keep from arbitrators absent express contrary agreement. (Garden Fresh Restaurant Corp. v. Superior Court, supra, 231 Cal.App.4th at pp. 686-687, discussing AT&T Mobility LLC v. Concepcion, supra, 563 U.S. at pp. 348-351 and Stolt-Nielsen S. A. v. AnimalFeeds Int'l Corp., supra, 559 U.S. at pp. 686-687; cf. Lee v. JPMorgan Chase & Co. (C.D.Cal. 2013) 982 F.Supp.2d 1109, 1113-1114 [explaining how the Supreme Court's class arbitration discussions in Concepcion and Stolt-Nielsen are relevant to the standard for determining the availability of class arbitration, rather than the prior question who decides whether class arbitration is available].) This assumption, that parties who enter an arbitration agreement diverting resolution of the merits of claims from courts to arbitrators would simultaneously be unwilling to have critical related procedural issues resolved by those same arbitrators, is one we do not embrace.6

The third case, Opalinski, offers slightly different reasoning. It observes that parties have the right to choose with whom they arbitrate, and courts typically resolve whether and how that right has been exercised. (Opalinski, supra, 761 F.3d at pp. 332-333, citing Stolt-Nielsen S. A. v. AnimalFeeds Int'l Corp., supra, 559 U.S. at pp. 683, 686.) The determination that an agreement allows for class arbitration has the potential to add many additional parties to the arbitration, and thus "affects whose claims may be arbitrated." (Opalinski, at p. 332.) Absent class members similarly have a right to determine with whom they arbitrate, and might not be bound absent an opt-in procedure evincing consent to have an arbitrator resolve their claims in a class proceeding. (Id. at p. 333, citing Oxford Health Plans LLC v. Sutter, supra, 569 U.S. at p. ___ [186 L.Ed.2d at p. 123, 133 S.Ct. at pp. 2071-2072] (conc. opn. of Alito, J.).) Consequently, Opalinski argues, the issue of class arbitration availability must be resolved first by a court.

(20) This argument fails to account for the fact the pro-court presumption for the two acknowledged gateway questions — the validity of the agreement and its application to the subject matter of a given dispute — already ensures court determination of the parties to any potential class arbitration. Even if an arbitrator were to conclude class arbitration is permitted, the arbitration would bring in only parties to the same form of arbitration agreement a court has already determined to be valid, and the same type of dispute a court has already determined to be within the scope of that agreement. (Cf. Stolt-Nielsen S. A. v. AnimalFeeds Int'l Corp., supra, 559 U.S. at p. 698 (dis. opn. of Ginsburg, J.) [where a proposed class consists of other parties to a standard arbitration agreement, observing that if arbitrators were to "certify the proposed class, they would adjudicate only the rights of persons `with whom' [petitioners] agreed to arbitrate, and only `issues' subject to arbitration"].) Here, for example, every employee at Lebo was apparently subject to the identical arbitration agreements upheld by the court. A determination by an arbitrator that those agreements permitted class proceedings would not subject Lebo to arbitration with anyone with whom it had not already agreed to arbitrate; it would simply consolidate actual or potential agreed-to arbitrations into a single proceeding. Protection of Lebo's right to arbitrate only with whom it chooses does not require that we read into the FAA a pro-court presumption on the question of class availability.

So, too, with the rights of would-be class members. Any potential binding arbitration award will arise only after the Lebo form arbitration agreement they too signed has been determined valid, and will govern only a dispute the trial court has found to be within the scope of that agreement. To the extent a would-be absent class member may believe he or she has unique defenses to enforcement of the arbitration agreement, the major arbitration entities provide clear notice and opt-out rights. (JAMS, Class Action Procedures, supra, rule 4; American Arbitration Assn., Supplementary Rules for Class Arbitration, supra, rule 5(b), (c).) Those same rules also require consideration of the full panoply of Federal Rules of Civil Procedure, rule 23 (28 U.S.C.) requirements before any class is certified. (JAMS, Class Action Procedures, supra, rule 3; American Arbitration Assn., Supplementary Rules for Class Arbitration, supra, rule 4.) There is no reason to assume a non-JAMS, non-American Arbitration Association arbitrator would refuse to afford similar protections — and if the arbitrator did refuse, the resulting award would doubtless not be binding. (See AT&T Mobility LLC v. Concepcion, supra, 563 U.S. at p. 349 ["For a class-action money judgment to bind absentees in litigation, class representatives must at all times adequately represent absent class members, and absent members must be afforded notice, an opportunity to be heard, and a right to opt out of the class. [Citation.] At least this amount of process would presumably be required for absent parties to be bound by the results of arbitration."].) Given prehearing notice, an absent class member can decide for him- or herself whether to abide by the prior court ruling that the arbitration agreement is enforceable and applicable, or opt out and take his or her chances in a separate proceeding. In neither instance will the absent class member be subjected to arbitration without consent or the opportunity to have a court assess the validity of his or her agreement to arbitrate.7

As a further basis for its conclusion, the Third Circuit interpreted the class/no-class question as affecting the type of dispute and thus analogous to the gateway judicial inquiry into whether a specific dispute was within the parties' agreement to arbitrate. (Opalinski, supra, 761 F.3d at pp. 333-334.) But the reason the court determines whether a given dispute is within an agreement to arbitrate is that, if it is not, then the parties to the controversy have not agreed to submit themselves to an arbitrator at all. Logically, parties cannot be compelled to go before an arbitrator without an independent decision maker — the court — first deciding they have consented to go before that decision maker. (Steelworkers v. Warrior & Gulf Co., supra, 363 U.S. at p. 582 ["For arbitration is a matter of contract and a party cannot be required to submit to arbitration any dispute which he has not agreed so to submit."].) In contrast, neither answer to the class/no class question implicates the parties' consent to submit the basic underlying controversy to arbitration. That arbitration is a matter of consent necessitates reserving issues of an agreement's validity and the scope of its subject matter for a court, but does not similarly dictate reservation of the class arbitration issue.

(21) Case law aside, Lebo Automotive contends as a policy matter that arbitrators have background incentives in the form of potential higher fees that would cause them to favor contract interpretations allowing for class arbitration, and therefore cannot be entrusted with the decision. Lebo's concerns are no different than those the dissent raised in Prima Paint v. Flood & Conklin, supra, 388 U.S. 395, when arguing that, notwithstanding the broad sweep of the parties' ambiguous arbitration clause, questions of contractual fraud in the inducement should be withdrawn from the arbitrator because of a possible implicit incentive to find no fraud and thereby increase potential compensation. (Id. at p. 416 (dis. opn. of Black, J.).) They did not carry the day with the Prima Paint majority, nor have they with this court; we too have presumptively allocated fraud in the inducement questions to arbitrators despite the theoretical potential for a conflict. (See Ericksen, Arbuthnot, McCarthy, Kearney & Walsh, Inc. v. 100 Oak Street, supra, 35 Cal.3d at p. 323.) Also generally uncognizable is the belief that arbitrators might over time be biased toward the repeat players that bring them business, and that the arbitral forum thus inherently favors such repeat players. We may not presume categorically that arbitrators are ill-equipped to disregard such institutional incentives and rule fairly and equitably; the FAA requires that we treat arbitration as a coequal forum for dispute resolution. (9 U.S.C. § 2; see, e.g., Gilmer v. Interstate/Johnson Lane Corp. (1991) 500 U.S. 20, 30-32 [114 L.Ed.2d 26, 111 S.Ct. 1647].)8

As justification for reaching the merits of the availability of class arbitration, the trial court relied on Stolt-Nielsen S. A. v. AnimalFeeds Int'l Corp., supra, 559 U.S. 662 and Kinecta Alternative Financial Solutions, Inc. v. Superior Court, supra, 205 Cal.App.4th 506. But neither case has any bearing on the threshold issue, whether a court or arbitrator should address that question. Stolt-Nielsen expressly does not address it. (Stolt-Nielsen, at p. 680.) Nor does Kinecta address the issue; instead, it simply assumes a court should decide the question, without referring to the parties' agreement to determine what allocation that agreement makes. (Kinecta, at pp. 517-519.)

(22) In sum, neither appellate decisions subsequent to Green Tree, nor the policy concerns Lebo Automotive raises, persuade us the Green Tree plurality was wrong. To the contrary, we agree with the plurality that the determination whether a particular agreement allows for class arbitration is precisely the kind of contract interpretation matter arbitrators regularly handle. Along with the Green Tree plurality, we find nothing in the FAA or its underlying policies to support the contrary presumption, that this question should be submitted to a court rather than an arbitrator unless the parties have unmistakably provided otherwise.9

III. Harmless Error

In the alternative, Lebo Automotive argues that any error by the trial court in taking for itself the interpretation of the parties' arbitration agreement on matters of classwide arbitration was harmless because the trial court's interpretation of the contract was substantively correct. Lebo argues that the Court of Appeal erred by ordering reversal without first deciding for itself what the parties' contract calls for, and that we likewise should examine the parties' agreement and decide whether it provides for classwide arbitration.

(23) We decline to do so. It is no response to a court's misapprehension concerning the proper allocation of an issue of contract interpretation to an arbitrator to say that the error is not reversible unless yet a second court, also not entrusted with the issue of interpretation, disagrees on the merits with the first improper decision maker. The remedy when an issue has erroneously been addressed by a court rather than an arbitrator is to remand with instructions that the correct decision maker consider the issue anew. (See Green Tree, supra, 539 U.S. at pp. 453-454 (plur. opn. of Breyer, J.); Garcia v. DIRECTV, Inc., supra, 115 Cal.App.4th at p. 304.) As in Green Tree, "the parties have not yet obtained the arbitration decision that their contracts foresee" (Green Tree, at p. 453); remanding "enforc[es] the parties' arbitration agreements according to their terms" (id. at p. 454).

Lebo Automotive argues that state law requires harmless error review in all cases before reversal will follow. (Cal. Const., art. VI, § 13; Code Civ. Proc., § 475.) Not so; some errors are reversible per se. The error here falls within that class requiring automatic reversal because its effects are "`unmeasurable'" and "`def[y] analysis by "harmless-error" standards.'" (People v. Blackburn (2015) 61 Cal.4th 1113, 1135 [191 Cal.Rptr.3d 458, 354 P.3d 268]; see People v. Collins (2001) 26 Cal.4th 297, 311 [109 Cal.Rptr.2d 836, 27 P.3d 726]; Martin v. County of Los Angeles (1996) 51 Cal.App.4th 688, 698 [59 Cal.Rptr.2d 303].) We cannot say whether an arbitrator would have decided the issue the same or differently. Indeed, to deny remand by insisting an arbitrator would surely have agreed with the trial court's view or our view of the merits of class availability is to recommit the very error complained of — deprivation of a decision by a contractually agreed-upon decision maker. The denial of the parties' right to their agreed-upon decision maker is thus the sort of miscarriage of justice that requires reversal without further harmless error analysis.

DISPOSITION

For the foregoing reasons, we affirm the Court of Appeal.

Cantil-Sakauye, C. J., Liu, J., and Cuéllar, J., concurred.

KRUGER, J., Dissenting. —

"Class arbitration is a matter of consent: An arbitrator may employ class procedures only if the parties have authorized them." (Oxford Health Plans LLC v. Sutter (2013) 569 U.S. ___, ___ [186 L.Ed.2d 113, 133 S.Ct. 2064, 2066] (Oxford Health).) In this case, the trial court held that the parties' agreement did not authorize class arbitration. Plaintiff Timothy Sandquist argues that the arbitrator, rather than the court, should have made that determination. The focus of my disagreement with the majority involves the question of federal law at the center of this case: Whether, under the Federal Arbitration Act (9 U.S.C. § 1 et seq.; FAA), the availability of class arbitration under the parties' agreement is a "gateway question of arbitrability" that is presumptively for a court to decide, or whether it is instead a matter presumptively reserved for the arbitrator.

In Green Tree Financial Corp. v. Bazzle (2003) 539 U.S. 444, 451-452 [156 L.Ed.2d 414, 123 S.Ct. 2402] (Green Tree), a plurality of the United States Supreme Court took the view that the classwide arbitrability question is a procedural matter, rather than a gateway question of arbitrability. The majority of this court today draws on the reasoning of the plurality opinion to reach the same conclusion. But in more recent years, the United States Supreme Court has not only disavowed any notion that Green Tree decided the issue, it has also, as another court put it, "given every indication, short of an outright holding, that classwide arbitrability is a gateway question rather than a subsidiary one." (Reed Elsevier, Inc. v. Crockett (6th Cir. 2013) 734 F.3d 594, 598 (Reed Elsevier), cert. denied (2014) 572 U.S. ___ [189 L.Ed.2d 173, 134 S.Ct. 2991].) In light of these post-Green Tree developments, every federal court of appeals to consider the issue on the merits has concluded — in contrast to the majority's holding today — that whether an arbitration agreement permits class arbitration is presumptively a question for the court, rather than the arbitrator. (See pp. 266-267, post.) Because I, too, read the high court's cases as indicating that classwide arbitrability is a gateway question for purposes of the FAA, I would affirm the trial court's decision in this case, and respectfully part company with my colleagues in the majority.

I.

"While the interpretation of an arbitration agreement is generally a matter of state law, [citations], the FAA imposes certain rules of fundamental importance, including the basic precept that arbitration `is a matter of consent, not coercion,' [citation]." (Stolt-Nielsen S. A. v. AnimalFeeds Int'l Corp. (2010) 559 U.S. 662, 681 [176 L.Ed.2d 605, 130 S.Ct. 1758] (Stolt-Nielsen).) Once a court has determined that the parties have agreed to submit a dispute to arbitration, the law presumes that the parties have also agreed to allow the arbitrator to decide subsidiary questions concerning arbitration procedure, resolving any doubts in favor of the "`liberal federal policy favoring arbitration agreements.'" (Howsam v. Dean Witter Reynolds, Inc. (2002) 537 U.S. 79, 83 [154 L.Ed.2d 491, 123 S.Ct. 588] (Howsam).) But the law makes an exception to this rule for what the United States Supreme Court refers to as "`question[s] of arbitrability'" (ibid.), which "include certain gateway matters, such as whether the parties have a valid arbitration agreement at all or whether a concededly binding arbitration clause applies to a certain type of controversy" (Green Tree, supra, 539 U.S. at p. 452). Courts will not assume that the parties agreed to arbitrate such gateway matters "unless there is `clea[r] and unmistakabl[e]' evidence" to that effect. (First Options of Chicago, Inc. v. Kaplan (1995) 514 U.S. 938, 944 [131 L.Ed.2d 985, 115 S.Ct. 1920] (First Options).) This, the high court has explained, is because to interpret "silence or ambiguity" in the parties' agreement as consent to submit the matter to an arbitrator "might too often force unwilling parties to arbitrate a matter they reasonably would have thought a judge, not an arbitrator, would decide." (First Options, supra, 514 U.S. at p. 945.) The rule governing "`question[s] of arbitrability'" thus applies in "circumstance[s] where contracting parties would likely have expected a court to have decided the gateway matter, where they are not likely to have thought that they had agreed that an arbitrator would do so, and, consequently, where reference of the gateway dispute to the court avoids the risk of forcing parties to arbitrate a matter that they may well not have agreed to arbitrate." (Howsam, supra, 537 U.S. at pp. 83-84.)

In accordance with these principles, the resolution of a "`question of arbitrability'" is reviewable by courts de novo. (Oxford Health, supra, 569 U.S. at p. ___, fn. 2 [186 L.Ed.2d at p. 119, fn. 2, 133 S.Ct. at p. 2068, fn. 2].) By contrast, under the FAA, decisions assigned to an arbitrator are reviewable for "misconduct rather than mistake." (AT&T Mobility LLC v. Concepcion (2011) 563 U.S. 333, 350-351 [179 L.Ed.2d 742, 131 S.Ct. 1740] (Concepcion).) When an arbitrator has been given the authority to decide whether a party's agreement provides for class arbitration, his or her decision generally will stand "regardless of a court's view of its (de)merits." (Oxford Health, supra, 569 U.S. at p. ___ [186 L.Ed.2d 113, 133 S.Ct. at p. 2068]; see 9 U.S.C. § 10.)

In Green Tree, a plurality of the United States Supreme Court reasoned that whether an arbitration agreement authorizes class arbitration does not fall into this category of gateway questions of arbitrability because "the question is not whether the parties wanted a judge or an arbitrator to decide whether they agreed to arbitrate a matter," but "what kind of arbitration proceeding the parties agreed to." (Green Tree, supra, 539 U.S. at p. 452.) This view did not garner a majority, as the high court has since emphasized. (See Stolt-Nielsen, supra, 559 U.S. at pp. 678-679, 680; accord, Oxford Health, supra, 569 U.S. at p. ___, fn. 2 [186 L.Ed.2d at p. 119, fn. 2, 133 S.Ct. at p. 2068, fn. 2].) And although the United States Supreme Court has not had occasion to revisit the question of who decides classwide arbitrability, subsequent decisions of the court have undermined Green Tree's premise that questions of classwide arbitrability merely concern the "`procedural mode'" available for the presentation of the plaintiff's claims. (Stolt-Nielsen, supra, 559 U.S. at p. 687.)

In Stolt-Nielsen, the court considered whether an arbitration panel erred in ordering class arbitration when the parties' agreement was "`silent'" on the question and the parties had stipulated that they had never reached an agreement on the issue. (Stolt-Nielsen, supra, 559 U.S. at pp. 666, 668-669.) Answering that question in the affirmative, the court rejected the arbitration panel's class arbitration decision as "fundamentally at war with the foundational FAA principle that arbitration is a matter of consent." (Id. at p. 684.) The court acknowledged that, "[i]n certain contexts, it is appropriate to presume that parties that enter into an arbitration agreement implicitly authorize the arbitrator to adopt such procedures as are necessary to give effect to the parties' agreement. [Such] ... `"`procedural' questions ... grow out of the dispute and bear on its final disposition" [and] are presumptively not for the judge, but for an arbitrator, to decide.'" (Id. at pp. 684-685, quoting Howsam, supra, 537 U.S. at p. 84.) But classwide arbitrability, the court concluded, does not fall within that category "because class-action arbitration changes the nature of arbitration to such a degree that it cannot be presumed the parties consented to it by simply agreeing to submit their disputes to an arbitrator." (Stolt-Nielsen, at p. 685.)

The court described the "fundamental" differences between bilateral and class arbitration — and has later expounded on them — in two ways that are particularly relevant here. (Stolt-Nielsen, supra, 559 U.S. at p. 686.) First, as the court described it, the switch from bilateral to class arbitration is one that strikes at the heart of the bargain the parties make when they agree to "forgo the procedural rigor and appellate review of the courts" in favor of the "lower costs" and "greater efficiency" of private dispute resolution. (Id. at p. 685.) For one thing, the court pointed out, class arbitration dramatically increases the risks to defendants: "[T]he commercial stakes of class-action arbitration are comparable to those of class-action litigation, [citation], even though the scope of judicial review is much more limited, [citation]." (Id. at pp. 686-687.) And for another, as the court later elaborated in Concepcion, "the switch from bilateral to class arbitration sacrifices the principal advantage of arbitration — its informality — and makes the process slower, more costly, and more likely to generate procedural morass than final judgment." (Concepcion, supra, 563 U.S. at p. 348; see 14 Penn Plaza LLC v. Pyett (2009) 556 U.S. 247, 269 [173 L.Ed.2d 398, 129 S.Ct. 1456] ["[T]he relative informality of arbitration is one of the chief reasons that parties select arbitration."].) Thus, "the relative benefits of class-action arbitration are much less assured, giving reason to doubt the parties' mutual consent to resolve disputes through class[ ] arbitration." (Stolt-Nielsen, at pp. 685-686.) In support of the point, the court made express reference to its "questions of arbitrability" jurisprudence, citing First Options for the proposition that "`one can understand why courts might hesitate to interpret silence or ambiguity on the "who should decide arbitrability" point as giving the arbitrators that power, for doing so might too often force unwilling parties to arbitrate' contrary to their expectations." (Id. at p. 686, quoting First Options, supra, 514 U.S. at p. 945.)

Second, the high court noted, the switch from bilateral to class adjudication based on the parties' agreement necessarily calls into question which parties, precisely, are involved, since "[t]he arbitrator's award no longer purports to bind just the parties to a single arbitration agreement, but adjudicates the rights of absent parties as well." (Stolt-Nielsen, supra, 559 U.S. at p. 686.) Stated differently, the classwide arbitrability decision does not merely concern the procedural mode for presentation of plaintiff's claims, but whether the claims of other, absent plaintiffs, may be arbitrated as well. In that fashion, the question touches on the basic principle that parties must be permitted to specify not only what issues they will arbitrate, but "with whom" they will do so. (Id. at p. 683.)

For these reasons, among others, the court in Stolt-Nielsen rejected the idea that the availability of class arbitration is purely a matter of procedure. If that were so, the court explained, "there would be no need to consider the parties' intent with respect to class arbitration" because procedural questions of that sort are "committ[ed] ... presumptively to the arbitrator's discretion." (Stolt-Nielsen, supra, 559 U.S. at p. 687, citing Howsam, supra, 537 U.S. at p. 84.) But, in the court's estimation, "the differences between bilateral and class-action arbitration are too great for arbitrators to presume, consistent with their limited powers under the FAA, that the parties' mere silence on the issue of class-action arbitration constitutes consent to resolve their disputes in class proceedings." (Stolt-Nielsen, at p. 687, fn. omitted.)

Later, in Concepcion, the court cited much the same set of considerations in holding that "the FAA prohibits States from conditioning the enforceability of certain arbitration agreements on the availability of class[ ] arbitration procedures." (Concepcion, supra, 563 U.S. at p. 336.) Class arbitration, the court reasoned, necessarily results in slower dispositions and higher costs. (Id. at p. 348.) Moreover, because class arbitration implicates the rights of absent parties, due process "requires procedural formality." (Id. at p. 349 ["For a class-action money judgment to bind absentees in litigation, class representatives must at all times adequately represent absent class members, and absent members must be afforded notice, an opportunity to be heard, and a right to opt out of the class. [Citation.] At least this amount of process would presumably be required for absent parties to be bound by the results of arbitration."].) The court thought it "unlikely" that in passing the FAA in 1925 Congress meant to entrust arbitrators "with ensuring that third parties' due process rights are satisfied," particularly since "class arbitration is a `relatively recent development.'" (Id. at pp. 349-350.) And finally, the court reiterated that "class arbitration greatly increases risks to defendants.... The absence of multilayered review makes it more likely that errors will go uncorrected. Defendants are willing to accept the costs of these errors in [classical] arbitration, since their impact is limited to the size of individual disputes, and presumably outweighed by savings from avoiding the courts. But when damages allegedly owed to tens of thousands of potential claimants are aggregated and decided at once, the risk of an error will often become unacceptable." (Id. at p. 350.) The court concluded: "We find it hard to believe that defendants would bet the company with no effective means of review, and even harder to believe that Congress would have intended to allow state courts to force such a decision." (Id. at p. 351, fn. omitted.)

None of this, to be sure, amounts to a holding that the availability of class arbitration is presumptively a decision for a court, rather than an arbitrator. As the majority correctly notes (maj. opn., ante, at p. 260), that particular question was not before the court in Stolt-Nielsen, since the parties in that case had expressly assigned the determination to the arbitration panel (see Stolt-Nielsen, supra, 559 U.S. at p. 680). And certainly no such question was before the court in Concepcion, in which the parties' agreement expressly disallowed class arbitration. (Concepcion, supra, 563 U.S. at p. 336.) But the import of the decisions seems fairly inescapable. If, because of the fundamental differences between bilateral and class arbitration, the court is unwilling to treat classwide arbitrability as a mere procedural matter — and thus unwilling to presume that parties' "silence on the issue of class-action arbitration constitutes consent to resolve their disputes in class proceedings" (Stolt-Nielsen, 559 U.S. at p. 687, fn. omitted) — it seems rather unlikely that the court would be willing to presume that the parties have consented to allow an arbitrator to make an essentially unreviewable determination to the same effect. That is particularly true because, as the court has emphasized, the decision is one that does not affect the named plaintiff and defendant alone, but implicates whether other, absent plaintiffs will also be required to submit their claims to arbitration.

Given these recent developments in the high court's jurisprudence, it is unsurprising that, at least to date, every federal court of appeals to consider the issue on the merits has held that the availability of class arbitration is a question of arbitrability for a court, rather than an arbitrator, "`unless the parties clearly and unmistakably provide otherwise.'" (Dell Webb Communities, Inc. v. Carlson (4th Cir. 2016) 817 F.3d 867, 873-877; Opalinski v. Robert Half Internat. (3d Cir. 2014) 761 F.3d 326, 330, cert. denied (2015) 575 U.S. ___ [191 L.Ed.2d 558, 135 S.Ct. 1530]; Reed Elsevier, supra, 734 F.3d at pp. 597-599; but cf. Robinson v. J & K Administrative Management Services, Inc. (5th Cir. 2016) 817 F.3d 193, 197 [acknowledging Stolt-Nielsen, but concluding the panel was bound by the "rule of orderliness" to apply pre-Stolt-Nielsen circuit precedent following the plurality opinion in Green Tree].) As these courts have explained, the high court's characterization of class arbitration in Stolt-Nielsen and Concepcion is difficult to reconcile with a view that would instead treat classwide arbitrability as simply a question of "what kind of ... proceeding" the parties have agreed to. (Green Tree, supra, 539 U.S. at p. 452, italics omitted; see maj. opn., ante, at pp. 252-253.)

The majority criticizes these decisions for confusing two separate issues: the importance of the classwide arbitrability question, on the one hand, with the parties' expectations about who will resolve it, on the other. (Maj. opn., ante, at pp. 255-256.) The majority is correct, of course, that not every question that is "important" to the parties is necessarily a gateway question that is presumptively reserved for judicial determination. So, for example, in the First Circuit case on which the majority relies, the parties may have cared very much indeed about whether the claim at issue was time-barred under applicable arbitration rules, but the caring alone does not render the interpretation of the time limit rule a question of arbitrability for the court to decide. (PaineWebber Inc. v. Elahi (1st Cir. 1996) 87 F.3d 589, 599; accord, Howsam, supra, 537 U.S. at p. 85.) But Stolt-Nielsen rests on the premise that classwide arbitrability is a qualitatively different kind of question from a question about the proper interpretation of a timing rule because it so radically changes the nature of the bargain the parties strike when they agree to submit their disputes to arbitration. If, as the high court says, the nature of these changes "giv[es] reason to doubt the parties' mutual consent to resolve disputes through class[ ] arbitration" from their general agreement to arbitrate (Stolt-Nielsen, supra, 559 U.S. at pp. 685-686), it is not much of a leap to conclude that these changes also provide reason to doubt the parties' consent to permit an arbitrator to decide, in his or her essentially unreviewable judgment, whether the parties' silence or ambiguous contract language constitutes consent to class arbitration.

The majority also dismisses the cases' concerns about permitting an arbitrator to bring absent plaintiffs into the dispute, noting that "[e]ven if an arbitrator were to conclude class arbitration is permitted, the arbitration would bring in only parties to the same form of arbitration agreement a court has already determined to be valid, and the same type of dispute a court has already determined to be within the scope of that agreement." (Maj. opn., ante, at p. 257.) But as the majority elsewhere acknowledges, there is no requirement or guarantee that a court will ever make such a determination. (Id. at p. 247.) If the named plaintiff chooses not to contest the validity of the agreement or its application to a given dispute, then no court would have occasion to consider whether absent class members had also given their consent to arbitrate. It may be, as the majority suggests, that the answer is simply that arbitrators must provide absent plaintiffs with an adequate opportunity to opt out, or the arbitral award will not be binding on them. (Id. at p. 258; but see Oxford Health, supra, 569 U.S. at p. ___ [186 L.Ed.2d 113, 133 S.Ct. at p. 2071] (conc. opn. of Alito, J.) [arguing that an arbitral award cannot be binding unless absent class members have been required to opt in, since "`[a]rbitration is simply a matter of contract between the parties,' [citation], and an offeree's silence does not normally modify the terms of a contract [citation].") But again, if the court thinks it "unlikely" that the Congress that enacted the FAA would have entrusted arbitrators with the protection of third parties' due process rights (Concepcion, supra, 563 U.S. at p. 349), it is unclear that the court would consider it any more likely that the parties would have intended to assign to an arbitrator the determination whether class arbitration is available. And in any event, the remedy for arbitral error — to permit collateral attacks on the award — is one that, as Justice Alito has noted, may reinforce doubts about the parties' likely intent at the time of contract formation. (See Oxford Health, supra, 569 U.S. at p. ___ [186 L.Ed.2d 113, 133 S.Ct. at p. 2072] (conc. opn. of Alito, J.) ["Class arbitrations that are vulnerable to collateral attack allow absent class members to unfairly claim the `benefit from a favorable judgment without subjecting themselves to the binding effect of an unfavorable one,' [citation].... [T]his possibility should give courts pause before concluding that the availability of class arbitration is a question the arbitrator should decide."].)

II.

It may well be that further developments in the United States Supreme Court will shed new light on the issue before us. But unless and until the court revisits the issue, I would follow where the court has led. Because the majority today charts a different path, I must respectfully dissent.

Chin, J., and Corrigan, J., concurred.