The issue before the Supreme Court of Canada in Uber Technologies v Heller (“Heller“) was deceptively simple: who decides the validity of an arbitration clause?  Mr. Heller’s agreement with Uber directed all disputes to a Dutch arbitrator under the International Chamber of Commerce’s arbitration rules (the “ICC Rules”) sitting in Amsterdam and applying Dutch law. Rather than sending the dispute, including the validity issue, to arbitration, the Supreme Court held the arbitration agreement was invalid on the basis it was unconscionable.  The decision paved the way for a significant Canadian class action against Uber to proceed in the courts of Ontario.  The decision will likely have important consequences for drafting, obtaining agreements to, and enforcing arbitration agreements in domestic and international arbitration.

The implications of Heller for the law of unconscionability in Canada are the subject of our colleagues’ Bruce and Nicholas Reynolds’ recent article.


The Facts

Mr. Heller entered into an online standard-form agreement with Uber in order to work as an Uber food delivery driver.  The agreement between Mr. Heller and an Uber subsidiary incorporated in the Netherlands contained a clause requiring all unresolved disputes under the agreement to proceed to mediation and, if that failed, arbitration pursuant to the ICC rules in Amsterdam. The agreement was to be “governed by and construed in accordance with the laws of the Netherlands”, excluding its conflicts of laws rules.  Under the ICC Rules, commencing arbitration required up-front administrative and filing fees of USD $14,500.  That cost excluded any legal fees and other costs of participation. Mr. Heller earned between CAD $400 to $600 per week.

Rather than pursuing mediation and subsequent arbitration under the agreement, Mr. Heller initiated a class action in Ontario claiming CAD $400 million on behalf of Uber drivers on the basis that an Uber driver is entitled to the benefits of being an “employee” within the meaning of Ontario’s Employment Standards Act, 2000, S.O. 2000, c. 41 (“ESA”).  Uber sought a stay of the class action in favour of the Dutch arbitration.  Mr. Heller argued the arbitration agreement was invalid because it was unconscionable and because was contrary to the mandatory provisions of the ESA.


The Decision of the Courts Below

The Ontario Superior Court of Justice determined that the “competence-competence” principle, under which arbitrators are competent to determine their own jurisdiction, applied to the validity of the arbitration agreement. The Court therefore issued a stay of Mr.  Heller’s action.

The Court of Appeal unanimously reversed and lifted the stay.  It held it could determine the arbitration agreement to be invalid, and the Court did so on the basis the arbitration agreement was unconscionable: that is, it was obtained through significantly unequal bargaining power and it was improvident because of the cost of arbitration to Mr. Heller.  The Court of Appeal also found that the agreement purported to contract out of the mandatory provisions of the ESA.


The Supreme Court’s decision

The Supreme Court agreed with the Ontario Court of Appeal.  A seven-judge majority of the Supreme Court, whose reasons for judgment were authored by Abella and Rowe, agreed it could determine the validity of the arbitration agreement and held it was invalid because it made it “impossible for one party to arbitrate”.

The first question was what validity test to apply.  The answer turned on the nature of the arbitration: was it international or domestic?  While in both cases a stay was mandatory, there are differently expressed bases for refusing a stay.  The majority held the correct test was ‘invalidity’ under the domestic Ontario Arbitration Act, because Ontario’s International Commercial Arbitration Act (“ICAA“), which incorporates the UNCITRAL Model Law on International Commercial Arbitration into law in Ontario, did not apply.

Under the ICAA, if a court finds the arbitration agreement is ‘null and void, inoperative or incapable of being performed’, it need not stay the court proceeding. There are different exceptions under Ontario’s Arbitration Act. The majority held, in accordance with prior Ontario case law and secondary authorities, that employment law issues did not fall within the ICAA, so Mr. Heller’s action would be stayed under the Arbitration Act unless the arbitration agreement was ‘invalid’ under the applicable exception.

This was the most favorable outcome for Mr. Heller.  Traditionally, ‘invalidity’ has been interpreted narrowly in this context as issues going to enforceability of the arbitration agreement: for example, where there was no consensus ad idem or non est factum applied, or where there was a lack of authority to enter into the contract, or illegality. Unconscionability may be such an issue. Under the ICAA, ‘null and void’ has typically been interpreted as focusing on lack of consent: fraud, undue influence, duress, misrepresentation or incapacity which would mean there was no agreement right from the beginning.  An unconscionable agreement may not be ‘null and void’.

To determine if an arbitration agreement is invalid, the majority and the parties agreed, normally such matters would be referred to the arbitrator. However, notwithstanding an arbitrator’s fundamental jurisdiction to determine his or her own jurisdiction, the Supreme Court noted that it had previously held courts could engage in and determine invalidity if it involved questions of law or questions of mixed law and fact that required only superficial examination of the factual record.  In this regard, the majority referred to the Supreme Court’s prior decisions in Dell Computer, Dell’s companion case Rogers Wireless (both 2007) and Seidel v TELUS (2011).The majority concluded that, here, a superficial review of the record led ineluctably to the conclusion that a stay would result in, or at least there was a ‘real prospect’ of, the claim never proceeding to arbitration.

However, the majority apparently went further than this.  They suggested it would be appropriate to depart from the general rule of systematic referral on the basis that the facts raised an access to justice issue that was not present in Dell and was not contemplated by the Ontario Arbitration Act.  The majority characterized the case as ‘abnormal’.

Having said that, the majority concluded the arbitration agreement was invalid because of the access to justice issue raised by the factual record: the cost to Mr. Heller of the arbitration was prohibitive, Mr. Heller could not reach the physical location of the arbitration, and the choice of Dutch law to determine the dispute would oust Ontario’s public policy as embodied in the ESA.  That added up, in the minds of the majority, to denying the possibility of relief for the claim by insulating the arbitration agreement from meaningful challenge.

The majority’s reasoning is opaque.  The majority suggested the ‘normal’ framework outlined by Dell did not apply.  However, the majority’s analysis proceeded apparently within the Dell framework to determine a mixed question of law and fact: whether the arbitration agreement was invalid.  The majority’s reasons are probably best understood as an expansion of the concept of invalidity under the Ontario Arbitration Act.  Either way, whether or not the ‘normal’ framework applies, the Supreme Court appears to have taken a significant step away from the systematic referral rule.  Even if it has left the Dell framework intact, determination of the issue of invalidity under the Ontario Arbitration Act may now entail analysis, at least at the superficial level permitted by this case and Dell, to determine if access to justice or other issues raise extrinsic factors which may similarly prevent access to justice being achieved through enforcement of an arbitration agreement.

The majority was alive to the possibility that permitting such issues to be raised on a stay application may increase spurious delay tactics.  While the majority said they did not wish to encourage ‘mini-trials’ of this issue, they concluded that courts are equipped to distinguish ‘genuine challenges’ from delay or lack of bona fides and that a court’s usual powers to award costs would be sufficient to control litigants.

Importantly, the majority noted two questions of fact, either of which would have prevented the court from determining the validity of the arbitration agreement.  First, Uber chose not to prove the applicable Dutch law concerning the validity of the arbitration agreement.  Second, there were no disputed facts in this case.  Under Dell’s framework either would have required a stay in favor of arbitration.

With respect to the Dutch law issue, the majority recognized this left a significant ‘loophole’ for contract drafters.  Choosing a foreign law increases the likelihood of sending questions of validity to the arbitrator because proof of foreign law in Canada requires hearing evidence in order to make findings as to the content of foreign law, something that a court would not ordinarily contemplate in a superficial review of the record unless there is no dispute as to that law.

Two justices wrote separate reasons for judgment.

Brown, in his concurring reasons for judgment would have refused to grant a stay on the basis that the arbitration clause was effectively an agreement not to arbitrate and as such was invalid on public policy grounds.  He relied on cases limiting the operation of forum selection clauses and exclusion clauses, which raise concerns relating to the administration of justice, and limiting the operation of restrictive covenants (Douez, Tercon, Elsley, and Shafron), all of which were determined on a public policy basis.  In his view, the key public policy consideration was the cost of access to arbitration and the parties’ relative bargaining positions at the time the arbitration agreement was entered into.  He recognized that in some cases, significant up-front costs far greater than those required to commence a court action may be warranted in light of the parties’ relationship and the timely resolution that arbitration can provide.  But in rare cases, like this one, an arbitration agreement could impose undue hardship and act as an effective bar to adjudication.  The possibility of third-party financing and the existence of a class proceeding were irrelevant, in his view.

Côté dissented.  She would have granted the stay of proceedings, but on terms.  She suggested conditions could be imposed: either that Uber advance the funds needed to initiate the arbitration proceedings or a court could apply a ‘blue-pencil’ to delete the parts of the agreement to arbitrate which were unconscionable or violated public policy.  In the latter case, she suggested all references to mediation or arbitration under the ICC Rules in Amsterdam could be cleanly excised, leaving domestic arbitration (still under Dutch law, if proven) available to Mr. Heller. While Côté viewed the agreement references as discrete and severable terms, Brown saw the illegality of the agreement as flowing from the cumulative nature of these terms which he viewed as being effectively a single provision lacking the divisibility necessary to apply the ‘blue-pencil’ solution suggested by Côté.

She differed, sometimes strongly, with the majority on almost every issue.  She concluded that the ICAA applied because the agreement was more in the nature of a commercial licensing agreement and it expressly stipulated that it did not create an employment relationship.

Even applying the Dell framework to the Ontario Arbitration Act test of invalidity, Côté concluded that the stay application could not be decided on a superficial review because the decision as to whether the ICAA or the Ontario Arbitration Act applied and Mr. Heller’s arguments for unconscionability both depended on testimonial evidence regarding Mr. Heller’s personal and financial characteristics.  She was concerned that ‘superficial review’ may in practice become a searching review of the record for the purpose of determining whether findings of fact can be made.  She observed what was, in her view and even on a thin factual record, evidence of Mr. Heller’s commercial sophistication and his ability to access and use dispute resolution procedures including class proceedings.  She also pointed to the lack of evidence about Mr. Heller’s financial circumstances at the time he entered into the agreement, the true value of his claim and the true comparative costs to Mr. Heller in pursuing this matter in the class proceedings or under the ICC Rules.  Investigation of all of these issues, and other relevant matters, to properly evaluate the accessibility of arbitration to Mr. Heller would, in her opinion, undermine the normal, systematic referral of matters to arbitration.

Côté observed that the unconscionability or public policy arguments against the choice of Dutch law in relation to the application of the ESA did not affect the evaluation of the arbitration agreement.  In her view, these were separate issues.

Further, Côté noted that articles 22(1) and (4) of the ICC Rules provide that an arbitral tribunal must make every effort to conduct the arbitration in an expeditious and cost-effective manner and must ensure that each party has a reasonable opportunity to present its case.  She therefore saw no basis for assuming that the arbitral tribunal would require Mr. Heller to travel to Amsterdam.  She also observed that Rule 38 of the ICC Rules empowers an arbitral tribunal to make decisions on costs at any time during the proceedings and to decide which party should bear the costs in the final award.

As a result, in principle, Côté described the use of unconscionability and public policy as ad hoc judicial moralism, which risk greatly increased commercial uncertainty.  They are, in her view, inimical to the widely-accepted doctrine that an arbitration agreement is only invalidated by a defect relating specifically to the arbitration agreement itself, and not by one relating merely to the underlying contract.



The Supreme Court’s previously decided case TELUS Communications Inc. v Wellman, 2019 SCC 19 recognized that while consumer class proceedings would be exempted from the effect of arbitration clauses by specific Ontario legislation, the Ontario Consumer Protection Act, that legislation did not protect TELUS’s business clients and as a result enforced the arbitration clauses in the businesses’ agreements with TELUS.  In this case, Mr. Heller was more like a consumer, but he had no specific legislative protection arising from the class proceeding he had started.  At first glance, it might be thought that in Mr. Heller’s case the Supreme Court would have continued the policy of giving effect to arbitration clauses and followed the systematic referral rule set out in the ICAA and the Ontario Arbitration Act.  In the result, the Supreme Court effectively determined that Mr. Heller’s class proceeding should be treated as if it was a consumer class proceeding.

It is possible the application of this case will be limited by the Supreme Court’s description of this as an ‘abnormal’ case, but that seems unlikely.  The case raises many questions of principle which are likely to be the subject of further judicial and practical consideration.  Some of the more important may be the following.

First, whether an arbitration agreement is ‘invalid’ or ‘null and void, inoperative or incapable of being performed’, and the basis on which courts may find exceptions to the mandatory stay, may now be more widely litigated.  This will be because of the ubiquity of arbitration agreements in standard form agreements which are now a feature of our current online economic and social lives, the variety of situations in which they are used, the apparent expansion of the court’s basis for doing so and the road map provided by this case to the drafters of arbitration clauses and those who may seek to avoid their application.

Stay applications may start to resemble forum non conveniens applications, where a variety of factors are weighed by the courts in order to determine the issues of unconscionability or public policy, including access to justice issues, prior to ordering or refusing a stay.  Creative litigators are likely to test the limits of the courts’ willingness to enforce arbitration agreements in the face of other claims of invalidity or any other possible reasons for refusing a stay including unconscionability, public policy and the denial of access to justice.

Second, this decision swings the pendulum away from commercial certainty and towards the increasing importance of the local rules of practice and procedure, including local Canadian norms of judicial tolerance for the type and nature of the evidence that can be brought to bear on these issues.  The local rules may determine the effort required to obtain a stay in favor of arbitration.

Third, creative draftspersons may re-double their efforts to try to ensure that their arbitration agreements and choice of law clauses remain enforceable so that businesses, particularly those operating across multiple jurisdictions, including within Canada, will be able to manage their affairs as efficiently as possible.

Fourth, Côté suggested the decision may undermine Canada’s growing reputation as a world-class seat for arbitration, with modern arbitration legislation.  The policy behind the Arbitration Act and the UNCITRAL Model Law was to provide only very narrow exceptions to the enforcement of arbitration clauses: ‘null and void, inoperative or incapable of being performed’ or ‘invalid’.  Neither distinguishes between arbitration agreements on the basis of a choice of law clause, the arbitral rules chosen, the cost of the arbitrator or the venue, the parties’ legal costs, any pre-arbitration mediation provision (which may add significant cost to the dispute resolution process) or the location of arbitration.  Until this case, the likelihood of being able to avoid a stay was remote.

That likelihood may have increased if courts engage in more searching investigations into the efficacy of arbitration in particular cases. Dell is perhaps more rigid in its formulation, but not out of step with other leading jurisdictions for commercial dispute resolution. For example, courts in the UK ask if there is a ‘good arguable case’, which query is determined by asking who has the better of the argument on the materials available, and which is a relative test rather than a fact-finding exercise.  Sometimes the court will engage in the determination of a question of law if it is confident it has the necessary material and the decision will not unduly prejudice a party.  For a recent example, see Airbus SAS v Generali Italia SpA, 2019 EWCA 805, in which the parties and the court agreed that the court was able to construe an arbitration agreement to determine its scope.

It is to be hoped that the courts will narrow the consequences of Uber v Heller to, at the most, consumer- or employment-related cases and not extend its application to general commercial matters.


Tristian Packwood-Greaves co-authored this article.

Learn More About Construction and Infrastructure Law

We are a preeminent Canadian construction and infrastructure law firm. Our peers and clients recognize our lawyers as the best in the construction industry.