14 Penn Plaza LLC v. Pyett: Oppression or Opportunity for U.S. Workers; Learning from Canada

I.     Introduction

Ever since its seminal Steelworkers Trilogy,1 decided in 1960, the Supreme Court has regarded arbitration under collective bargaining agreements (CBAs) as different in character from other types of arbitration. Whereas other types of arbitration serve as substitutes for litigation, arbitration under CBAs is a “substitute for industrial strife. . . . [It] is part and parcel of the collective bargaining process itself.”2

This view of labor arbitration reached its zenith in 1974 when the Supreme Court in Alexander v. Gardner-Denver Corp.3 declared labor arbitration to be separate from and to operate independently of the public legal system. In Gardner-Denver, the Court held that employees need not resort to the CBA’s grievance and arbitration procedure before bringing a lawsuit under Title VII of the Civil Rights Act of 1964, and may proceed with their lawsuits even though they have grieved and arbitrated under the CBA and lost. To the Court, the labor arbitration process was completely different from the public adjudication process. The Court observed:

As the proctor of the bargain, the arbitrator’s task is to effectuate the intent of the parties. His source of authority is the collective-bargaining agreement, and he must interpret and apply that agreement in accordance with the “industrial common law of the shop” and the various needs and desires of the parties. The arbitrator, however, has no general authority to invoke public laws that conflict with the bargain between the parties . . . .4

On April 1, 2009, in 14 Penn Plaza, LLC v. Pyett,5 the Court, in apparent disregard of a half-century of precedent, held that a “collective-bargaining agreement that clearly and unmistakably requires union members to arbitrate ADEA claims is enforceable as a matter of federal law.”6 The Court distinguished, and largely limited to their facts, Gardner-Denver and its progeny. Indeed, Justice Thomas suggested that in an appropriate case, the Court might overrule Gardner-Denver.7 Finding such a clear and unmistakable waiver of the judicial forum in the CBA that covered Pyett’s employment, the Court held that Pyett was obligated to raise his claim under the Age Discrimination in Employment Act (ADEA) through the grievance and arbitration procedure. The Court reasoned:

In this instance, the Union and the [Realty Advisory Board on Labor Relations, Inc.], negotiating on behalf of 14 Penn Plaza, collectively bargained in good faith and agreed that employment-related discrimination claims, including claims brought under the ADEA, would be resolved in arbitration. This freely negotiated term between the Union and the RAB easily qualifies as a “conditio[n] of employment” that is subject to mandatory bargaining under [the National Labor Relations Act].8

The Court continued:

Parties generally favor arbitration precisely because of the economics of dispute resolution. See Circuit City Stores, Inc. v. Adams, 532 U.S. 105, 123, 121 S.Ct. 1302, 149 L.Ed.2d 234 (2001) (“Arbitration agreements allow parties to avoid the costs of litigation, a benefit that may be of particular importance in employment litigation, which often involves smaller sums of money than disputes concerning commercial contracts”). As in any contractual negotiation, a union may agree to the inclusion of an arbitration provision in a collective-bargaining agreement in return for other concessions from the employer.9

The Court’s citation to Circuit City is significant. In Circuit City and Gilmer v. Interstate/Johnson Lane Corp.,10 cases that arose in non-union workplaces, the Court held that pre-dispute agreements to arbitrate statutory claims imposed by employers on their employees as a condition of employment are enforceable under the Federal Arbitration Act.11 Notably absent from the Court’s opinion in Pyett is any discussion of labor arbitration’s role in a private system of workplace self-governance. Also absent is a half-century of recognition that labor arbitration is a substitute for strikes and other workplace strife. Instead, the Court in Pyett regarded labor arbitration as just another substitute for litigation.12

The Court’s employment arbitration jurisprudence has been roundly criticized as advantaging employers by stripping employees of their rights to sue over discrimination and other statutory claims.13 Not surprisingly, many scholars reacted to Pyett with similar criticism.14 Others, however, suggested that compelled arbitration of statutory claims under CBAs may not be as oppressive to workers as compelled arbitration under systems unilaterally imposed by employers. Professor Sarah Cole, for example, suggested that the arbitration of statutory employment claims under CBAs will be more favorable to employees than arbitration under individual employment agreements because of the presence of the union as a repeat player to balance the employer’s influence on the process.15 Furthermore, she suggested that the grievance arbitration forum offers employees the opportunity to vindicate their statutory claims in ways that are faster than litigation and in a forum in which employees do not pay the costs of their representation.16

Professor Michael Z. Green seized the Court’s decision in Pyett as an opportunity for unions to take the lead and achieve a convergence of interests of minority employees, majority employees, unions, and employers through the arbitration of statutory discrimination claims under CBAs.17 At the 2013 Annual Meeting of the National Academy of Arbitrators, counsel for the parties to the CBA at issue in Pyett portrayed the arbitration system developed under the contract as a fair and accessible forum for resolving the discrimination claims of low-wage workers that probably would not otherwise have been litigated.18

Some attributes of employer-imposed arbitration systems are criticized for tilting the forum to favor employers, but many such attributes are not present in arbitration under CBAs. Employees covered by CBAs need not worry about securing representation because they have their union to represent them. Employees covered by CBAs also incur no forum costs and are not responsible for any portion of the arbitrators’ fee. The CBA itself is the product of active negotiation between two relatively sophisticated parties, as opposed to the arbitration system imposed on a take-it-or-leave-it basis by an employer on employees who lack bargaining power to resist. In employer-imposed arbitration systems, the employer is the only repeat player, whereas in CBA-established arbitration systems both the employer and the union are repeat players. Furthermore, there are no class action waivers in arbitration under CBAs.19 On the other hand, the arbitrator remains a privately accountable adjudicator dependent on maintaining acceptability to the parties for continued business.

Statutory discrimination claims, known as human rights claims in Canada, have been arbitrated under collective agreements in Canada for decades. Reacting to the decision in Pyett less than two months later, National Academy of Arbitrators President and noted Canadian arbitrator Michel Picher called on his colleagues south of the border to look to the Canadian experience and suggested that Pyett may increase employee access to justice.20

This Article takes up Mr. Picher’s invitation and looks to the experience on the northern side of the border. The study compares the handling of statutory human rights claims in labour arbitration in Ontario to their handling before the Human Rights Tribunal of Ontario (HRTO). The study focuses on the time period 2009–2013. This starting point was selected because amendments to the Ontario Human Rights Code effective June 30, 2008 introduced substantial changes to the operation of the HRTO.

Part II of this Article discusses the law in Canada governing the arbitration of statutory human rights claims under collective agreements. Part III details the research methodology of the study. Part IV presents and analyzes the data. Part V considers the implications of the study for the continuing debate over arbitrating statutory discrimination claims under CBAs in the United States.

II.    Arbitration of Human Rights Claims in Canada

At the same time that the U.S. Supreme Court in Gardner-Denver declared CBAs and statutes to be separate spheres,21 the law was developing differently in Canada. There are three modes by which external statutes are applied in Canadian labour arbitration. First, external law can be used as an aid to interpret the collective agreement.22 Second, any statute incorporated by reference into a collective agreement requires arbitrators to apply that statute. Finally, at least since the 1974 Supreme Court of Canada (SCC) decision in McLeod v. Egan, 23 it is clear that arbitrators are not only permitted to look to external law, but are obligated to do so. This jurisdiction has been confirmed and clarified in subsequent cases.24 In addition, provincial labour legislation, which governs labour arbitration, commonly explicitly authorises labor arbitrators to interpret and apply human rights legislation. The Ontario Labour Relations Act (OLRA),25 for instance, states:

48(12) An arbitrator or the chair of an arbitration board, as the case may be, has power, . . .

(j) to interpret and apply human rights and other employment-related statutes, despite any conflict between those statutes and the terms of the collective agreement.26

Moreover s. 54 of the OLRA provides, “A collective agreement must not discriminate against any person if the discrimination is contrary to the Human Rights Code or the Canadian Charter of Rights and Freedoms.”27

In Canada, the general approach has been to recognize concurrent jurisdiction over human rights code matters in human rights tribunals and labour arbitration, with human rights and arbitral tribunals permitted to defer an application where the same matter is being heard in another proceeding.28 In British Columbia (Workers’ Comp. Bd.) v. Figliola,29 the SCC considered whether a human rights tribunal may entertain a matter that had already been decided in arbitration. The case arose in the province of British Columbia, where the human rights legislation contains a provision analogous to the Ontario Human Rights Code’s (OHRC) direction that the Human Rights Tribunal of Ontario (HRTO) may dismiss an application where another proceeding has “appropriately dealt with the substance of the matter.”30 In Figliola, the SCC unanimously concluded that concurrent jurisdiction is not “a statutory invitation either to judicially review another tribunal’s decision, or to reconsider a legitimately decided issue in order to explore whether it might yield a different outcome.”31 As a consequence of Figliola, pursuing a grievance and arbitration under a collective agreement and filing a complaint with the human rights tribunal are two separate and mutually exclusive paths.

In drawing lessons for the United States from the Canadian experience, we must be mindful of a few potentially important differences between the two countries. First, in Canada, rather than bring their claims in court, discrimination claimants must pursue their claims before administrative tribunals. In Ontario, the province that is the subject of our study, that tribunal is the HRTO.

Second, Ontario protects a much larger range of characteristics against discrimination than does the United States. Under the OHRC, protection against discrimination is afforded race, colour, ancestry, place of origin, citizenship, ethnic origin, disability, creed (religion), sex (including sexual harassment and pregnancy), sexual solicitation or advances, gender identity, gender expression, sexual orientation, family status, marital status, age (18 or older), and record of offences.32 As in the United States, a discrimination claim can be filed using multiple bases.

Third, arbitration awards are subject to different standards of judicial review in the two countries, which could affect, for example, the level of attention that arbitrators pay to public legal authorities. In the U.S. private sector, courts enforce a labor arbitration award as long as it “draws its essence” from the CBA.33 Arbitral findings of fact are not reviewed at all by courts. “[I]mprovident, even silly, factfinding” does not provide a basis for a reviewing court to refuse to enforce an award.34

U.S. courts do recognize an exception where an award violates public policy, but hold that the public policy must be “explicit, well-defined, and dominant,” and must be “ascertained by reference to the laws and legal precedents and not from general considerations of supposed public interests.”35 Although the Supreme Court has not squarely held that an award violates public policy only when it orders a party to violate positive law, it has never upheld a refusal to enforce an award on public policy grounds, and the possibility of such a refusal in cases where the award does not command a violation of positive law appears to be mostly theoretical.36

In Canada, the standard for judicial review for particular types of agencies varies among jurisdictions. In Ontario, two standards of review are applicable to administrative tribunal decisions: the more deferential reasonableness standard and the standard of correctness.37 The reasonableness standard generally applies to questions of mixed fact and law, where “a tribunal is interpreting its own statute or statutes closely connected to its function, with which it will have a particular familiarity”38 or where a tribunal has “developed particular expertise in the application of a general common law or civil law rule in relation to a specific statutory context.”39 In contrast, correctness always applies to true questions of jurisdiction, including questions of jurisdiction between competing specialized tribunals.40 Questions of law of central importance to the legal system and outside the tribunal’s specialized area of expertise also utilize the correctness standard.41

The standard of reasonableness generally applies to the review of arbitration decisions involving the application of human rights issues. Application of human rights principles and evaluating allegations of discrimination in the context of a particular collective agreement have been found to utilize the reasonableness standard of review.42 Such questions were found to involve questions of mixed fact and law, and application of human rights principles was regarded to be within labor arbitrators’ area of expertise.43 However, arbitration awards dealing with the duty to accommodate disability are subject to the correctness standard, as this duty is regarded as a principle of general law.44

III.  The Study

The study’s initial focus was on cases decided between 2009 and 2012. This starting point was selected because amendments to the Ontario Human Rights Code effective June 30, 2008 introduced substantial changes to the operation of the HRTO. Cases were drawn from the Ontario labour arbitration and HRTO databases maintained by the Canadian Legal Information Institute (CanLII).45 Only final decisions were selected for coding. A team of six law students, three from Chicago-Kent College of Law and three from Osgoode Hall Law School, Toronto, identified potentially relevant cases using the CanLII databases. They worked in pairs of two, one Chicago-Kent student paired with an Osgoode Hall student. One pair was assigned to find all relevant arbitration awards during the specified period. Independently, each student of this pair searched the CanLII database for awards containing the term “human rights,” or any of the bases on which the Human Rights Code prohibits discrimination, as well as for “accommodation” and for “harassment.” Each student submitted his or her results to Professor Malin who reviewed them for correctness and screened out cases that did not fit our criteria.

The other two pairs were assigned to the HRTO database. The students were instructed to exclude decisions filed under the pre-July 1, 2008 statute but decided under the post-July 1, 2008 statute (known as transitional cases). Professor Slinn provided each pair with a master list of all HRTO decisions issued in a given year. Each student took his or her list and, for each tenth case in the list, reviewed that case to see whether it met the criteria for being relevant (was a case of discrimination arising out of employment) and should not be excluded (was not a transitional case, preliminary decision, interim decision, summary decision, or case resolution conference, but was a final decision on the substantive issue of discrimination). If the case met these requirements it was included in the dataset; if not, then the student picked the next case in the list until reaching a relevant one. Then, the student began the process again. The students were instructed to continue through the list until getting 40 relevant cases per year. Professor Slinn reviewed their work for correctness and screened out cases that did not fit our criteria.

The researchers trained the students on case coding and provided them with detailed coding sheets. All six students independently coded the same 20 cases: 10 arbitration awards and 10 HRTO decisions. Due to inconsistencies among the coders and errors by some of the coders, the researchers revised the coding sheet and provided additional training. The six coders independently coded another set of the same six cases: three arbitration awards and three HRTO decisions. The results were much more consistent and correct. From that point forward, the coders were paired in teams of two, one from Chicago-Kent and one from Osgoode Hall. Each team member coded a group of assigned cases independently, and then the two team members compared their results and reconciled any differences.

The study initially examined cases decided between 2009 and 2012 and a preliminary report was presented at the 2014 annual meeting of the National Academy of Arbitrators.46 After the researchers checked the coding and the data further, the size of the database, particularly the number of arbitration awards, was relatively small. The researchers decided to expand the study to examine decisions issued in 2013. By this point, four of the original six law student assistants had graduated. Professor Malin identified the relevant 2013 arbitration awards and the two remaining student assistants, one from Chicago-Kent and one from Osgoode Hall, identified the 2013 HRTO decisions, with their determinations reviewed by the researchers. The two students then coded the decisions independently of each other, compared their results, and reconciled the differences. All student coding was reviewed by the researchers for relevance and correctness. Decisions that did not meet the study’s criteria were dropped from the database and coding errors were corrected.

Professor Jon Werner, aided by two research assistants at the University of Wisconsin-Whitewater, loaded all coded cases into Excel spreadsheets. Once these were reviewed, the spreadsheets for the arbitration and HRTO cases were combined into one file that was analyzed using the Stata statistical package. Correlations between variables were initially examined; this is denoted below by the letter “r.” If two variables are positively correlated, this means that high values on one variable correspond to high values of the other variable, and low values correspond with low values. The range of r is from -1 (a perfect negative correlation) to +1 (a perfect positive correlation), with values near zero signifying no meaningful relationship between the variables. T-tests were run to compare mean differences between the arbitration and HRTO databases. A t-test addresses whether two means differ significantly from one another—that is, whether they differ beyond what would be expected by chance alone. If a difference is not statistically significant, the value is followed by the abbreviation “n.s.” Where feasible, t-tests were followed by regression analyses. Multiple regressions analyze the influence of two or more independent variables on a single dependent variable.47

IV.  The Data

A.    Case Outcomes

Do workers fare better in terms of obtaining either partial or full relief for their claims in arbitration or before the HRTO? Table 1 compares the outcomes of all cases in our database. Overall, workers fared better in arbitration than before the HRTO. Arbitration allowed a total of 52.5% of claims, while HRTO allowed 42.6%, and this difference is statistically significant (p < .05).48 There are many differences between the cases brought in arbitration and those brought before the HRTO. Accordingly, we examined which, if any, of those differences might explain the differences in case outcomes.

In Canada, just as in the U.S., union density is higher in the public sector than in the private sector.49 Not surprisingly, the cases that were handled in arbitration were significantly more likely to arise in the public sector than the private sector. Table 2 presents the data.

The difference in sector representation—that is, more public sector cases in arbitration—is significant below the .01 confidence level. However, by itself, sector is not related to case outcomes (r = .07, n.s.). This suggests that it is something other than sector that is influencing the difference in outcomes between arbitration and HRTO cases in our dataset.

Claimants, called “grievors” in arbitration and “applicants” before the HRTO, were somewhat more likely to be female before the HRTO. Table 3 presents the data. However, by itself, grievant/applicant gender was not related to case outcomes (r = -.08, n.s.), suggesting that neither male nor female grievors/applicants were either advantaged or disadvantaged in how their cases were decided before their respective tribunals. It should be noted that having multiple grievors, or having a policy grievance, is unique to arbitration, and does not occur in cases brought to the HRTO.

Table 4 shows a marked difference in the gender of the adjudicators in the two fora, and this difference is statistically significant (p < .01). That grievors have a better than 80% chance of proceeding before a male arbitrator does not appear to be an anomaly limited to Canadian labour arbitrators. For example, a 2013 survey of members of the College of Commercial Arbitrators found that 84.6% of them were men.50 However, as with previous comparisons, adjudicator gender was not related to case outcome (r = .04, n.s.).

Table 5 shows the top seven grounds claimed in the cases in each forum. Here too, there is a significant difference between the two fora, with disability claims comprising the overwhelming majority of cases in arbitration, whereas the claims before the HRTO were more diversified. All mean differences shown in Table 5 are statistically significant (p < .01).

Because of the degree to which disability claims dominated the arbitration cases, we compared case outcomes for disability claims in the two fora. Table 6 presents the results. Interestingly, disability discrimination was found by adjudicators in a somewhat larger percentage of the claims before the HRTO than in arbitration, yet this difference was not statistically significant.

Although caution must be exercised in making inferences because of the small sample size, race and sex discrimination were significantly more likely to be found before the HTRO than in arbitration. Table 7 presents the data. The differences by forum are statistically significant (p < .05).

Across both fora, claims of sex discrimination were more likely to be allowed (p < .01) and claims of race discrimination were more likely to be denied (p < .01). This appears to be driven by a high degree of self-representation among HRTO race discrimination claimants. In 14, or 30.4%, of HRTO sex claims the applicant was self-represented while the respondent was represented, compared to 22, or 52.4% of, race claims. Despite the small sample size, a t-test found the difference statistically significant at p < .05. As discussed more fully below, self-represented applicants fare significantly poorer before the HRTO. Multivariate regression analysis of sex discrimination claims found, somewhat surprisingly, that the gender of the adjudicator was not a significant influence on the outcome, but the forum was. Table 8 presents that data from regression analysis.

In every case in our arbitration database, all parties appeared at the hearing. In 13 cases before the HRTO, the respondent failed to appear. As shown in Table 9, not surprisingly, respondents who did not appear fared much worse than those who did. The results were statistically significant (p < .01) but when those cases were removed from the database, the results of subsequent regression analysis were not impacted. Thus, because the number of cases where respondents did not appear was so small, the outcomes from these 13 cases did not account for the overall difference in outcomes between arbitration and the HRTO.

There was a marked difference in the representation of parties in arbitration compared to those before the HRTO. In every arbitration case, both parties were represented. From the written records available, it was not possible to determine in every case whether the representative was an attorney, yet we are confident that in every case the representative was a professional, such as a non-attorney professional union advocate. By contrast, before the HRTO, a majority of applicants were self-represented, as were about an eighth of respondents. When a respondent was a small corporate entity that was represented by its owner, we coded the respondent as self-represented. Thirty-nine (22.2%) of the HRTO cases arose in unionized workplaces. Most HRTO cases (115 or 65.3%) did not indicate whether the workplace was unionized, and it is likely that some of those cases also arose in unionized workplaces. However, among the HRTO cases involving unionized employees, the applicant representation rate was not much different from the representation rate in HRTO cases as a whole. Table 10 presents the breakdown of representation in HRTO cases.

As shown in Table 11, not surprisingly, self-represented parties fared far worse than represented parties. Self-represented applicants had more cases denied than those who were represented (p < .01), and self-represented respondents had fewer cases denied than those who were represented (p < .01). As Table 12 shows, among HRTO applicants known to be unionized, the disparity in results between self-represented and represented parties was even greater.

As Table 13 shows, comparing outcomes only for cases where both parties are represented, the differences between arbitration and the HRTO are not statistically significant. Indeed, the rates at which claims are denied are almost identical. Therefore, from the lack of significant impact on case outcomes of the various factors by which arbitration and HRTO cases differ, which we have previously reviewed (sector, prevalence of disability claims, gender of claimants, gender of adjudicators, failures by respondents to appear), and the dramatic impact on case outcomes of the absence of representation, we conclude that it is highly likely that the significantly higher success rate for claimants in arbitration over the HRTO is mostly attributable to differences in claimant representation. All grievors in arbitration were represented, while most applicants before the HRTO were self-represented.

B.    Speed and Efficiency

A common view of arbitration is that it is a faster and more efficient dispute resolution process than litigation. A comparison of the arbitral and HRTO fora shows considerable, though not uniform, support for this commonly held view.

Table 14 shows the breakdown of adverse employment actions challenged in each forum. Where more than one action was challenged, the first adverse action was coded. Thus, if a denial of a training opportunity was followed months later by a discharge, the case was coded as a denial of training. In subsequent tables showing elapsed time, the elapsed time was calculated from the first adverse action that was challenged in the claim. For example, in a case challenging a denial of training and a subsequent discharge, the elapsed time was calculated from the training denial.

Of course, none of the grievances that were arbitrated occurred in the hiring process because employees are not subject to the collective agreement until they are hired. Table 15 excludes hiring claims and combines the remaining claims to show those that arose during the employment relationship and those that involved employment termination. The differences are not significant and this is unchanged when sector is added in—i.e., this is not a function of the differences in public- versus private-sector cases in general.

Thus, in both fora, approximately two-thirds of the claims (other than hiring claims) arose during the employment relationship. In contrast, Table 16 shows a marked difference as to when the claims are filed between arbitration and HRTO claims. The difference is highly statistically significant (p < .001). In arbitration, the 43 grievances filed after employment ended is only four more than the total number of grievances (39) challenging termination or resignation. In other words, employees grieving adverse actions that arise while they are employed do so while they are still employed. In contrast, employees complaining to the HRTO of adverse actions that arise while they are still employed generally wait until they are fired or they quit before filing their complaints. This is consistent with a recent study, which found that in non-union workplaces in Ontario, fewer than 10% of wage-hour claims are filed with the Ministry of Labour while the claimant remains employed by the employer.51

There is some indication that the presence of a union leads to employees bringing their claims more quickly than in non-union environments. Table 17 shows the stage of the employment relationship at which the claim arose in unionized HRTO cases, and Table 18 shows the stage at which the claim was filed in unionized HRTO cases. It appears that unionized employees act faster than their non-union counterparts even when they pursue redress before the HRTO rather than in arbitration.

Thus, workers not covered by collective agreements tend not to seek redress for alleged discrimination until they are terminated or they quit. Workers who grieve discrimination pursuant to procedures in their collective agreements are more likely to do so while the employment relationship continues. This difference translates into substantial differences in the amount of time it takes to resolve claims in each forum. Table 19 presents the data concerning elapsed time from various marker events in these cases, for the time period 2009–2013.

Because 2009 was the first year the HRTO was operating under its new procedures, we calculated elapsed time excluding 2009, to determine whether the HRTO might have needed a year to get up to speed under the new system. Table 20 presents the data for 2010–2013, excluding the 2009 cases.

The results are consistent with or without the 2009 cases, and several comparisons are notable. Because workers covered by collective agreements do not wait to be fired to file their claims, grievances are filed less than a month following the event that is the subject of the complaint. In contrast, HRTO applicants tend to wait until they are fired to file, resulting in claims being filed on average almost ten times later, vis-à-vis the initial event complained of than in arbitration. Grievors also get to hearings faster after filing the grievances than applicants get before the HRTO, but the difference is not significant. Overall, grievors get a decision in significantly less time after the event giving rise to the claim, and that is due mostly to grievors filing quickly after the initial event rather than waiting until they are fired.

Measured from the start of the hearing, the difference in time it takes to complete the process to the rendering of a decision is not significant. HRTO proceedings take significantly less time from the start of the hearing to the end of the hearing. This is partly due to arbitration taking about twice as many hearing days as HRTO cases. From our observations of the actual cases, multi-day arbitration hearings are much less likely to be held on consecutive days than HRTO hearings. Many arbitration hearings were spread out over a year or longer. On the other hand, arbitrators are more than twice as fast in issuing their decisions than HRTO adjudicators. We suggest that this is a positive effect of arbitrators’ need to maintain acceptability with the parties. An easy way for arbitrators to lose their acceptability is to be late with their decisions. To check this, we conducted regression analyses to see if sector or grounds claimed might affect elapsed time and therefore account for what appears to be differences based on forum. Table 21 presents the results. Unless otherwise noted, all main differences highlighted are statistically significant at p < .05.

The one consistent result is that cases proceed faster in the private sector. As noted above, cases in arbitration are more likely to arise in the public sector. These findings reinforce, rather than diminish, the speed advantages of arbitration, which we attribute to claims being filed soon after they arise and the arbitrator’s need to issue a decision quickly to maintain acceptability. We also looked at whether the parties that were represented had any impact on elapsed time, and in general, there were no statistically significant differences in elapsed time based upon representation. The one exception was that when HRTO applicants were represented, their hearings averaged 2.5 days, whereas the hearings for self-represented HRTO applicants averaged 1.7 days, and this difference was statistically significant (p < .05).

C.    Remedies

In theory, the same remedies are available in arbitration and before the HRTO. However, as shown in Tables 22 and 23, there are significant differences in the remedies requested in each forum, as well as in the remedies awarded.

Remedies in arbitration tend more to reinstatement and back pay than to compensatory damages. This is in contrast to the HRTO, where they tend more toward damages, and reinstatement is rarely requested and never ordered, even though the HRTO has the authority to award it. We believe this phenomenon may share common characteristics with the phenomenon seen earlier—HRTO claimants do not file their claims until after they are fired. A likely reason is fear of retaliation. Such fear is less likely to be present where the claimant will be protected by a union. Indeed, as seen earlier, in unionized workplaces, claimants who forgo grieving and instead file with the HRTO do so right away, i.e., they do not wait until they are fired. Because they view the work environment as hostile, terminated workers who do not have union support may not seek reinstatement because they do not want to return to the employer who discharged them. We regard the availability and use of reinstatement as a remedy in arbitration to be a positive attribute of that forum.

On the other hand, what the HRTO terms “public interest remedies,” such as requiring education or training on human rights, creating or revising human rights policies, and reporting on human rights compliance, are common before the HRTO and almost non-existent in arbitration. Table 24 singles out these differences.

Public interest remedies occur almost exclusively in HRTO; this difference by forum is statistically significant (p < .01). Follow-up analyses revealed that this difference is not impacted by adjudicator gender or applicant gender (both n.s.); that is, public interest remedies are not more likely to be awarded by female adjudicators or for female grievors/applicants. It might be asked whether the rarity of public interest remedies in arbitration resulted from the dominance of disability claims in that forum. It is possible, for example, that many disability claims could be for individual accommodations that do not lend themselves to public interest remedies. Results of follow-up regression analysis, however, show just the opposite. This is presented in Table 25.

A finding of disability is significantly positively related with the award of public interest remedies, while the arbitration forum is significantly negatively related with the award of public interest remedies. We believe that this aspect of remedies practice may reflect the negative effects of arbitrator need to maintain acceptability. Public interest remedies are the remedies that intrude the most on the respondent. An arbitrator dependent on the parties’ selection of that arbitrator for future work may be more likely to shy away from awarding such intrusive remedies, especially if the moving party is not seeking them. On the other hand, as the data show, the HRTO is not bashful about awarding these intrusive remedies, even where the applicant has not requested them.

The absence of public interest remedies in arbitration may also reflect the effects of a unionized environment. In a unionized workplace, an employer may be more likely to have human rights policies in place and may agree to needed refinements in arbitrator-initiated mediation. Such agreed-to refinements would not appear in the awards and would not appear in our database. We are unable to test for this.

A key difference between the fora is that the HRTO orders public interest remedies even when a party has not requested them. A sense of fairness internalized by many arbitrators may inhibit their awarding remedies that were not requested. It is possible that HRTO adjudicators award public interest remedies even when not requested by the applicant out of concern that self-represented applicants may not know to request them. Therefore, we looked at whether an applicant being self-represented made a difference in the awarding of public interest remedies even though the applicant had not requested them. In the 97 cases where HRTO applicants were self-represented, seven public interest remedies were requested, and 15 were awarded. However, for the 77 HRTO cases where applicants were represented, 15 public interest remedies were requested, and 39 were awarded. Thus, it cannot be said that HRTO adjudicators were awarding public interest remedies to make up for what self-represented applicants failed to request for themselves.

D.    Is Arbitration Lawless?

One concern expressed with the arbitration of statutory claims is that privately appointed and privately accountable arbitrators will be developing the public law. As Table 26 shows, in this dataset, this concern is much more theoretical than real. For arbitrators and HRTO tribunal adjudicators, deciding a novel issue of law occurs extremely rarely, and there is no meaningful difference between the fora in how often this occurs. This is likely because the substantive law had been well-developed by 2009. The role played by arbitrators, if any, in developing that law is outside the scope of this study.

However, as reflected in Tables 27 through 30, the two fora take different approaches to the law. We also added in sector, disability, and adjudicator gender, and, in general, results were unchanged when these additional variables were included in our analyses.

Although relatively rare in either forum, arbitrators are more likely to criticize or refuse to follow established authority or to identify and reconcile established authorities. They are also more likely to distinguish established authority, although that appears to be due to the higher rate of public sector cases in arbitration. Of greater concern, arbitrators are far more likely to cite themselves or other arbitrators, and far less likely to cite public bodies such as human rights tribunals and courts.52

V.    Lessons for the United States

The experience in Canada can inform the debate in the United States over Pyett. The parties to the CBA—that is, both management and the union—in Pyett portray their system as a fast, inexpensive forum for claims of low-wage workers that would not otherwise have been heard.53 To a large extent, the experience in Ontario supports that claim. The law in Ontario is sufficiently well-developed such that very few cases raise novel legal issues; both arbitrators and the HRTO are largely applying established law to the facts as they find them. Arbitration provides a forum that is accessible and can resolve claims relatively quickly after they arise. Claimants do not wait until they are fired to file. Moreover, claimants in arbitration are always represented, whereas claimants before the HRTO are mostly self-represented. That  concern applies equally in the United States for low-wage workers whose claims are frequently not of sufficient value to attract legal representation. Arbitration is the forum where discharged employees can get their jobs back and return to a workplace where they will be supported by a union.

The Ontario experience, however, does call for some notes of caution. The presence of repeat players on both sides does not completely counteract the negative effects of arbitrator need to maintain acceptability. We saw this with respect to remedies; arbitrators very rarely award public interest remedies. Although those types of remedies are generally not available in the typical individual discrimination case in the U.S., the phenomenon of adjudicators’ need for acceptability intruding on administration of the statutory scheme is something we must be vigilant about if statutory discrimination claims are to be resolved through CBA grievance and arbitration systems. That Canadian arbitrators appear to be a bit looser than public tribunals in their handling of public legal authorities suggests that if Pyett-type arbitration catches on in the United States, courts may have to rethink the level of deference they give to arbitration awards in such cases. Finally, the gender composition of Canadian arbitrators is likely comparable to the U.S. Although our data found that gender did not directly correlate with case outcomes, even in sex discrimination cases, having a forum where the adjudicators are 80% male does give an appearance of bias, even if no bias is actually present.

Parties and courts in the United States should move cautiously when considering mandating that workers covered by a CBA grieve and arbitrate their statutory discrimination claims. For example, parties pursuing this approach should consider designating a panel of arbitrators that is demographically diverse, particularly with respect to race, ethnicity, and gender. However, as claimed by the parties to the CBA at issue in Pyett, this approach can provide an accessible forum for low-wage workers whose claims might otherwise never be brought because of their low dollar value.

• 1United Steelworkers of Am. v. Am. Mfg. Co., 363 U.S. 564 (1960); United Steelworkers of Am. v. Warrior & Gulf Navigation Co., 363 U.S. 574 (1960); United Steelworkers of Am. v. Enter. Wheel & Car Corp., 363 U.S. 593 (1960).
• 2Warrior & Gulf, 363 U.S. at 578.
• 3415 U.S. 36 (1974).
• 4Id. at 53.
• 5556 U.S. 247 (2009).
• 6Id. at 274.
• 7Id. at 264 n. 8.
• 8Id. at 256.
• 9Id. at 257.
• 10500 U.S. 20 (1991).
• 119 U.S.C. § 2.
• 12See Martin H. Malin, The Evolving Schizophrenic Nature of Labor Arbitration, J. Disp. Resol. 57 (2010).
• 13See, e.g., Alexander J.S. Colvin, Mandatory Arbitration and Inequality of Justice in Employment, 35 Berkeley J. Emp. & Lab. L. 71 (2014); David S. Schwartz, Claim-Suppressing Arbitration: The New Rules, 87 Ind. L.J. 239 (2012); Katherine Van Wezel Stone, Mandatory Arbitration of Individual Employment Rights: The Yellow Dog Contract of the 1990s, 73 Denv. U.L. Rev. 1017 (1996).
• 14See, e.g., Kenneth M. Casebeer, Supreme Court Without a Clue: 14 Penn Plaza LLC v. Pyett and the System of Collective Action and Collective Bargaining Established by the National Labor Relations Act, 65 U. Miami L. Rev. 1063 (2011); Melissa Hart, Procedural Extremism: The Supreme Court’s 2008-2009 Labor and Employment Cases, 13 Emp. Rts. & Emp. Pol’y J. 253 (2009); Allan Hyde, Labor Arbitration of Discrimination Claims After 14 Penn Plaza v. Pyett: Letting Discrimination Defendants Decide Whether Plaintiffs May Sue Them, 25 Ohio St. J. on Disp. Resol. 975 (2010); Margaret L. Moses, The Pretext of Textualism: Disregarding Stare Decisis in 14 Penn Plaza v. Pyett, 14 Lewis & Clark L. Rev. 825 (2010).
• 15Sarah Rudolph Cole, Let the Grand Experiment Begin: Pyett Authorizes Arbitration of Unionized Employees’ Statutory Discrimination Claims, 14 Lewis & Clark L. Rev. 861, 862–63 (2010).
• 16Id. at 863.
• 17Michael Z. Green, Reading Ricci and Pyett to Provide Racial Justice Through Union Arbitration, 87 Ind. L.J. 367, 369 (2012).
• 18Terry Meginniss & Paul Salvatore, The Forum for Litigation of Statutory Employment Claims After Pyett: A New Approach from Management and Labor, 66 Ann. Mtg. Nat’l Acad. of Arbitrators 271 (2013).
• 19See, e.g., Federal Bureau of Prisons, 135 Lab. Arb. Rep. (bNA) 1215 (2015) (Heekin, Arb.).
• 20Michel Picher, Presidential Address: Access to Justice: The Silver Lining in Pyett, 62 Ann. Mtg. Nat’l Acad. of Arbitrators 1 (2010).
• 21See Gardner-Denver, 415 U.S. at 49–50.
• 22See, e.g., Re Sunnyside Home for the Aged and London & District Service Workers’ Union, Local 220 [1985] 21 L.A.C. (3d) 85, 96 (Can.) (P.C. Picher).
• 23[1975] 1 S.C.R. 517 (Can.).
• 24See, e.g., Parry Sound (District) Social Services Administration Board v. O.P.S.E.U., Local 324, [2003] 2 S.C.R 157 (Can.); Weber v. Ontario Hydro, [1995] 2 S.C.R. 929 (Can.); Ontario Nurses’ Assn. v. Orillia Soldiers Mem’l Hosp. [1999] 42 O.R. (3d) 692 (Can. Ont. C.A.).
• 25S.O. 1995, c 1, Sch A (Can.).
• 26Id. s. 48 (12) (j).
• 27Id. s. 54.
• 28For example, the Ontario Human Rights Code, RSO 1990, c H.19 (Can.) (OHRC) provides that the human rights tribunal may defer an application where the same matter is being heard in another proceeding, and if deferred, the applicant has 60 days to bring the application back after conclusion of the other proceeding. s. 45. However, s. 45.1 provides that the human rights tribunal may dismiss an application where another proceeding has appropriately dealt with the substance of the matter.
• 29[2011] 3 S.C.R. 422 (Can.).
• 30OHRC s. 45.1.
• 31Figliola, 3 S.C.R at 442.
• 32OHRC s. 5.
• 33United Steelworkers v. Enter. Wheel & Car Corp., 363 U.S. 593, 597 (1960).
• 34Major League Baseball Players Ass’n v. Garvey, 532 U.S. 504, 509 (2001).
• 35E. Associated Coal Co. v. United Mine Workers, 531 U.S. 57, 62–63 (2000).
• 36In the U.S. public sector, the standard of review varies among the states. Most parrot the language from the private sector but apply that language in a manner that allows for closer judicial scrutiny of awards. Nevertheless, even in the public sector, arbitration awards enjoy considerable judicial deference. See Martin H. Malin et al., Public Sector Employment: Cases and Materials 922–57 (3d ed. 2016).
• 37Dunsmuir v. New Brunswick, [2008] 1 S.C.R. 190 (Can.). Note that prior to the 2008 Dunsmuir decision, three levels of scrutiny were applicable to judicial review of decisions of administrative tribunals: reasonableness simpliciter, patent unreasonableness, and correctness. Dunsmuir combined the two reasonableness standards into a single reasonableness standard. Id. at 192.
• 38Id. at 223.
• 39Id.
• 40Id. at 225–26.
• 41Id. at 224.
• 42Ottawa Hosp. v. Ontario Pub. Serv. Emps. Union, Local 464, 2009 CanLII 9389 (Can. Ont. S.C.D.C.) at ¶¶ 8–­9.
• 43Id. at ¶ 8.
• 44Hall v. Ottawa Police Serv., 2008 CanLII 65766 (Can. Ont. S.C.D.C.) at ¶ 42.
• 45CanLII was selected because it was available free of charge to individuals in the U.S. and Canada. It was later discovered that there may be a few relevant arbitration awards in the Lexis-Nexis Canadian arbitration database that were not in CanLII, but there is no reason to believe that these awards differ in any way material to the study from those available in CanLII. U.S. law schools’ Lexis-Nexis subscriptions do not include the Canadian arbitration awards library and so the students at Chicago-Kent only had access to CanLII.
• 46Martin H. Malin et al., An Empirical Comparison of the Handling of Statutory Human Rights Claims in Labour Arbitration and Before the Human Rights Tribunal of Ontario, 67 Ann. Mtg. Nat’l Acad. of Arbitrators 342 (2014).
• 47Tradition and practice hold that a value should be considered statistically significant if it falls below the p < .05 level, and preferably below the p < .01 level, corresponding respectively to odds of 1-in-20 versus 1-in-100 of obtaining a particular value by chance alone. A signification of p < .001 indicates odds of obtaining this variable by chance as 1-in-1000. By contrast, many social scientists consider values falling between p = .05 and p = .10 to be “marginally” significant. These values may be viewed as interesting enough to present to the reader, and allows the reader to decide what practical significance should be placed upon a result obtained in a particular context.
• 48In one case, Whitby Mental Health Centre & Ontario Public Service Employees Union, Local 331, 2009 CanLII 17970 (Can. Ont. L.A.) (Stout, Arb.), the arbitrator issued a consent award that provided relief to the grievant without sustaining or denying the grievance. This case was omitted from our tally of outcomes.
• 49See Diane Galarneau & Thao Sohn, Insights on Canadian Society: Long-term trends in unionization, Statistics Canada (Nov. 26, 2013), http://www.statcan.gc.ca/pub/75-006-x/2013001/article/11878-eng.pdf, https://perma.cc/N5L9-9M7S (Table 2).
• 50Thomas J. Stipanowich & Zachary P. Ulrich, Arbitration in Evolution: Current Practices and Perspectives of Experienced Commercial Arbitrators, 25 Am. Rev. Int’l Arb. 395, 402–03 (2014). On the other hand, our database does not include cases that were settled. It is common in Ontario for arbitrators to mediate and for grievances to settle in mediation. It is possible, but outside the scope of this study, that women arbitrators may have had higher rates of settlements in mediation than men arbitrators. Our data set does not enable us to determine this.
• 51John Grundy et al., The Enforcement of Ontario’s Employment Standards Act: The Impact of Reforms 12 (June 27, 2015) (unpublished paper presented to Second Biennial Meeting, Labor Law Research Network, Amsterdam) (on file with the authors).
• 52An earlier study in Quebec had a similar finding. See Guylaine Vallée et al., Implementing Equality Rights in the Workplace: An Empirical Study, 9 Canadian Lab. & Emp. L.J. 77, 100 (2002). This may be a function of what parties cite to the adjudicators. It may be that parties cite more arbitration awards than tribunal or court decisions to arbitrators but we are unable to determine whether this is the case.
• 53See Meginniss & Salvatore, supra note 18.