Communications, Energy and Paperworkers Union of Canada, Local 30 v. Irving Pulp & Paper, Ltd., 2013 SCC 34, [2013] 2 SCR 458

 

SUPREME COURT OF CANADA

 

Citation: Communications, Energy and Paperworkers Union of Canada, Local 30 v. Irving Pulp & Paper, Ltd., 2013 SCC 34, [2013] 2 S.C.R. 458

Date: 20130614 Docket: 34473

 

Between:

Communications, Energy and Paperworkers Union of Canada, Local 30

Appellant

and

Irving Pulp & Paper, Limited

Respondent

- and -

Construction Owners Association of Alberta, Construction Labour Relations — an Alberta Association, Enform, Canadian National Railway Company, Canadian Pacific Railway Company, Via Rail Canada Inc., Alberta Federation of Labour, Communications, Energy and Paperworkers Union of Canada, Local 707, Canadian Civil Liberties Association, Alliance of Manufacturers & Exporters of Canada, carrying on business as Canadian Manufacturers & Exporters, Canadian Mining Association, Mining Association of British Columbia, Mining Association of Manitoba Inc., Québec Mining Association, Ontario Mining Association, Saskatchewan Mining Association and Power Workers’ Union

Interveners

 

 

 

Coram: McLachlin C.J. and LeBel, Fish, Abella, Rothstein, Cromwell, Moldaver, Karakatsanis and Wagner JJ.

 

Reasons for Judgment: (paras. 1 to 55)   Joint Dissenting Reasons: (paras. 56 to 118)

Abella J. (LeBel, Fish, Cromwell, Karakatsanis and Wagner JJ. concurring)   Rothstein and Moldaver JJ.(McLachlin C.J. concurring)

 

 

 

Communications, Energy and Paperworkers Union of Canada, Local 30 v. Irving Pulp & Paper, Ltd., 2013 SCC 34, [2013] 2 S.C.R. 458

Communications, Energy and Paperworkers

Union of Canada, Local 30                                                                            Appellant

v.

Irving Pulp & Paper, Limited                                                                    Respondent

and

Construction Owners Association of Alberta,

Construction Labour Relations — an Alberta Association, Enform,

Canadian National Railway Company,

Canadian Pacific Railway Company, Via Rail Canada Inc.,

Alberta Federation of Labour,

Communications, Energy and

Paperworkers Union of Canada, Local 707,

Canadian Civil Liberties Association,

Alliance of Manufacturers &

Exporters of Canada, carrying on business as

Canadian Manufacturers & Exporters,

Canadian Mining Association,

Mining Association of British Columbia,

Mining Association of Manitoba Inc.,

Québec Mining Association, Ontario Mining Association,

Saskatchewan Mining Association and

Power Workers’ Union                                                                                Interveners

Indexed as:  Communications, Energy and Paperworkers Union of Canada, Local 30 v. Irving Pulp & Paper, Ltd.

2013 SCC 34

File No.:  34473.

2012:  December 7; 2013:  June 14.

Present:  McLachlin C.J. and LeBel, Fish, Abella, Rothstein, Cromwell, Moldaver, Karakatsanis and Wagner JJ.

on appeal from the court of appeal for new brunswick

                    Labour relations — Arbitration — Collective agreements — Management rights — Privacy — Employer unilaterally imposing mandatory random alcohol testing policy for employees — Whether unilaterally implementing random testing policy a valid exercise of employer’s management rights under collective agreement — Whether employer could unilaterally implement policy absent reasonable cause or evidence of workplace alcohol abuse.

                    Administrative law — Judicial review — Standard of review of labour arbitration board’s decision — Employer unilaterally imposing mandatory random alcohol testing policy for employees holding safety‑sensitive positions — Whether arbitration board’s decision that harm to employees’ privacy outweighed policy’s benefits to employer was reasonable.

                    The Union brought a grievance challenging the mandatory random alcohol testing aspect of a policy on alcohol and drug use that the employer, Irving, unilaterally implemented at a paper mill.  Under the policy, 10% of employees in safety sensitive positions were to be randomly selected for unannounced breathalyzer testing over the course of a year.  A positive test for alcohol attracted significant disciplinary action, including dismissal. 

                    The arbitration board allowed the grievance.  Weighing the employer’s interest in random alcohol testing as a workplace safety measure against the harm to the privacy interests of the employees, a majority of the board concluded that the random testing policy was unjustified because of the absence of evidence of an existing problem with alcohol use in the workplace.  On judicial review, the board’s award was set aside as unreasonable.  The New Brunswick Court of Appeal dismissed the appeal. 

                    Held (McLachlin C.J. and Rothstein and Moldaver JJ. dissenting):  The appeal should be allowed.

                    Per LeBel, Fish, Abella, Cromwell, Karakatsanis and Wagner JJ.:  The legal issue at the heart of this case is the interpretation of the management rights clause of a collective agreement.  The scope of management’s unilateral rule‑making authority under a collective agreement is that any rule or policy unilaterally imposed by an employer and not subsequently agreed to by the union must be consistent with the collective agreement and be reasonable.

                    A substantial body of arbitral jurisprudence has developed around the unilateral exercise of management rights in a safety context resulting in a carefully calibrated “balancing of interests” proportionality approach.  Under it, and built around the hallmark collective bargaining tenet that an employee can only be disciplined for reasonable cause, an employer can impose a rule with disciplinary consequences only if the need for the rule outweighs the harmful impact on employees’ privacy rights.  This approach has resulted in a consistent arbitral jurisprudence whereby arbitrators have found that when a workplace is dangerous, an employer can test an individual employee if there is reasonable cause to believe that the employee was impaired while on duty, was involved in a workplace accident or incident, or was returning to work after treatment for substance abuse.

                    A unilaterally imposed policy of mandatory random testing for employees in a dangerous workplace has been overwhelmingly rejected by arbitrators as an unjustified affront to the dignity and privacy of employees unless there is evidence of enhanced safety risks, such as evidence of a general problem with substance abuse in the workplace.  The dangerousness of a workplace is clearly relevant, but this does not shut down the inquiry, it begins the proportionality exercise.  It has never been found to be an automatic justification for the unilateral imposition of unfettered random testing with disciplinary consequences.

                    In this case, the expected safety gains to the employer were found by the board to range from uncertain to minimal, while the impact on employee privacy was severe.  The board concluded that eight alcohol‑related incidents at the Irving mill over a 15‑year period did not reflect the requisite problem with workplace alcohol use.  Consequently, the employer had not demonstrated the requisite safety concerns that would justify universal random testing.  As a result, the employer exceeded the scope of its rights under the collective agreement.

                    The applicable standard for reviewing the decision of the labour arbitrator is reasonableness.  The board’s decision must be approached as an organic whole, not as a line‑by‑line treasure hunt for error.  In this case, based on the findings of fact and the relevant jurisprudence, the decision was a reasonable one.

                    Per McLachlin C.J. and Rothstein and Moldaver JJ. (dissenting):  There is agreement with the majority that the standard of review is reasonableness.  However, there is disagreement as to the application of that standard.  In striking down the policy, the board departed from an arbitral consensus that has attempted to strike a balance between competing interests in privacy and safety in the workplace.  In so doing, it came to an unreasonable decision. 

                    This case asks whether management’s exercise of its unilateral rule‑making power can be justified under a collective agreement.  That question is one the New Brunswick legislature has delegated to labour arbitrators.  Their decisions are entitled to a protected zone of deference in which the courts should not willingly enter as competing arbiters of labour policy.  Reasonable people may well differ about the wisdom of the legislative choice to entrust labour arbitrators with a policy‑making function that potentially carries serious repercussions for public safety.  Indeed, the fact that the public interest — not merely that of employer and employee — is relevant in cases such as this one may counsel a reassessment of that choice.  But that decision is one for the New Brunswick legislature, not this Court. 

                    Nevertheless, under reasonableness review there is a difference between judicial abdication and judicial restraint.  To the extent a particular arbitral award is unreasonable — as this one is — it remains liable to being set aside on judicial review.

                    Where arbitral consensus exists, it raises a presumption — for the parties, labour arbitrators, and the courts — that subsequent arbitral decisions will follow those precedents.  Consistent rules and decisions are fundamental to the rule of law.  Therefore, arbitral precedents in previous cases concerning management’s unilateral adoption of a random alcohol testing policy shape the contours of what qualifies as a reasonable decision in this case.

                    The arbitral jurisprudence does not recognize an unqualified right of employers to unilaterally impose workplace rules on their employees outside of the collective bargaining process.  Rather, the onus is on the employer to justify such rules based on compliance with standards established by the arbitral jurisprudence.  In this case, the only standard in dispute was the reasonableness of the policy.  The key question is the threshold of evidence that the employer was required to introduce in order to meet its burden to demonstrate reasonableness and thereby justify its random alcohol testing policy.

                    From a review of the relevant arbitral decisions, what emerges is an arbitral consensus that an employer must demonstrate evidence of an alcohol problem in the workplace in order to justify a random alcohol testing policy.  That is the evidentiary threshold accepted by arbitrators who have upheld such policies and those who have struck them down.  Thus, barring some explanation, whether implicit or explicit, for its basis for departing from it, that is the evidentiary threshold the board in this case should have applied.  That is not, however, what the board did.

                    Though purporting to apply the test emerging from the arbitral consensus, the board elevated the threshold of evidence that Irving was required to lead in order to justify its random alcohol testing policy and offered no reason for doing so.  The board required evidence of a “significant” or “serious” problem at the Irving mill.  The standard reflected in the arbitral consensus, however, is evidence of “a” problem.  The difference between the two approaches is obviously a marked one and it cannot be ignored.  The board then required that the evidence of alcohol use be tied or causally linked to the accident, injury or near miss history at the plant.  Again, there is no support for such a requirement in the arbitral jurisprudence.  An employer does not have to wait for a serious incident of loss, damage, injury or death to occur before taking action.  To require such a causal connection is not only unreasonable, it is patently absurd.  It is the application of this higher evidentiary standard which dictated the board’s decision to strike down the policy.  The evidence was the decisive factor. 

                    To be clear, it was open for the board in this case to depart from the arbitral consensus in reaching its conclusion, provided it had a reasonable basis for doing so.  In so departing, it was thus incumbent upon the board to provide some explanation for its reasoning.  Here, the board provided no explanation whatsoever — whether implicit or explicit, reasonable or unreasonable — for the new evidentiary standard that it applied.  In the absence of a reasonable explanation for its novel test, the board must be taken as having misapplied the existing test, which in the circumstances of this case rendered its decision unreasonable.

                    The reasonableness of the board’s reasoning is further undermined by its inference that the risk at the Irving mill was not high based on the fact that only 10% of mill employees in safety‑sensitive positions were tested in any given year.  The inference was unreasonable because it failed to recognize that: even low testing percentages can be highly effective; testing a higher percentage of employees in order to establish the reasonableness of a workplace testing policy would perversely incentivize employers and lead to a greater intrusion into the privacy of employees; and the threshold set by Irving is not out of the mainstream for random alcohol testing.

                    In sum, the board departed from the legal test emerging from the arbitral consensus by elevating the threshold of evidence Irving was required to introduce in order to justify a policy of random alcohol testing.  In the absence of any explanation whatsoever, it is impossible to understand why the board thought it reasonable to do what it did.  In the circumstances of this case, its decision thus fell outside the range of reasonable outcomes defensible in respect of the facts and law.

Cases Cited

By Abella J.

                    Referred to:  Dunsmuir v. New Brunswick, 2008 SCC 9, [2008] 1 S.C.R. 190; Nor‑Man Regional Health Authority Inc. v. Manitoba Association of Health Care Professionals, 2011 SCC 59, [2011] 3 S.C.R. 616; Newfoundland and Labrador Nurses’ Union v. Newfoundland and Labrador (Treasury Board), 2011 SCC 62, [2011] 3 S.C.R. 708; Entrop v. Imperial Oil Ltd. (2000), 50 O.R. (3d) 18; Re United Steelworkers, Local 4487 & John Inglis Co. Ltd. (1957), 7 L.A.C. 240; Re United Brewery Workers, Local 232, & Carling Breweries Ltd. (1959), 10 L.A.C. 25; Re Public Utilities Commission of the Borough of Scarborough and International Brotherhood of Electrical Workers, Local 636 (1974), 5 L.A.C. (2d) 285; United Electrical, Radio, and Machine Workers of America, Local 524, in re Canadian General Electric Co. Ltd. (Peterborough) (1951), 2 L.A.C. 688; Re Hamilton Street Railway Co. and Amalgamated Transit Union, Division 107 (1977), 16 L.A.C. (2d) 402; Re Lumber & Sawmill Workers’ Union, Local 2537, and KVP Co. (1965), 16 L.A.C. 73; Metropolitan Toronto (Municipality) v. C.U.P.E. (1990), 74 O.R. (2d) 239, leave to appeal refused, [1990] 2 S.C.R. ix; Charlottetown (City) v. Charlottetown Police Association (1997), 151 Nfld. & P.E.I.R. 69; N.A.P.E. v. Western Avalon Roman Catholic School Board, 2000 NFCA 39, 190 D.L.R. (4th) 146; St. James‑Assiniboia Teachers’ Assn. No. 2 v. St. James‑Assiniboia School Division No. 2, 2002 MBCA 158, 222 D.L.R. (4th) 636; Esso Petroleum Canada and C.E.P., Loc. 614, Re (1994), 56 L.A.C. (4th) 440; Canadian National Railway Co. and C.A.W.‑Canada (Re) (2000), 95 L.A.C. (4th) 341; Weyerhaeuser Co. and I.W.A. (Re) (2004), 127 L.A.C. (4th) 73; Navistar Canada, Inc. and C.A.W., Local 504 (Re) (2010), 195 L.A.C. (4th) 144; Rio Tinto Alcan Primary Metal and C.A.W.‑Canada, Local 2301 (Drug and Alcohol Policy) (Re) (2011), 204 L.A.C. (4th) 265; Imperial Oil Ltd. and C.E.P., Loc. 900 (Re) (2006), 157 L.A.C. (4th) 225; Imperial Oil Ltd. v. Communications, Energy & Paperworkers Union of Canada, Local 900, 2009 ONCA 420, 96 O.R. (3d) 668; Metropol Security, a division of Barnes Security Services Ltd. and U.S.W.A., Loc. 5296 (Drug and Alcohol testing) (Re) (1998), 69 L.A.C. (4th) 399; Trimac Transportation Services — Bulk Systems and T.C.U. (Re) (1999), 88 L.A.C. (4th) 237; Fording Coal Ltd. v. United Steelworkers of America, Local 7884, [2002] B.C.C.A.A.A. No. 9 (QL); ADM Agri‑Industries Ltd. v. National Automobile, Aerospace, Transportation and General Workers’ Union of Canada (CAW‑Canada), Local 195 (Substance Abuse Policy Grievance), [2004] C.L.A.D. No. 610 (QL); Petro‑Canada Lubricants Centre (Mississauga) and Oakville Terminal and C.E.P., Local 593 (Re) (2009), 186 L.A.C. (4th) 424; Communications, Energy and Paperworkers Union, Local 777 v. Imperial Oil Ltd., May 27, 2000 (unreported); Greater Toronto Airports Authority v. Public Service Alliance of Canada, Local 0004, [2007] C.L.A.D. No. 243 (QL); DuPont Canada Inc. and C.E.P., Loc. 28‑O (Re) (2002), 105 L.A.C. (4th) 399; R. v. Dyment, [1988] 2 S.C.R. 417; R. v. Shoker, 2006 SCC 44, [2006] 2 S.C.R. 399.

By Rothstein and Moldaver JJ. (dissenting)

                    Esso Petroleum Canada v. Communications, Energy & Paperworkers’ Union, Local 614, [1994] B.C.C.A.A.A. No. 244 (QL); Entrop v. Imperial Oil Ltd. (2000), 50 O.R. (3d) 18; Canadian National Railway Co. and C.A.W.‑Canada (Re) (2000), 95 L.A.C. (4th) 341; Greater Toronto Airports Authority v. Public Service Alliance of Canada, Local 0004, [2007] C.L.A.D. No. 243 (QL); Rio Tinto Alcan Primary Metal and C.A.W.‑Canada, Local 2301 (Drug and Alcohol Policy) (Re) (2011), 204 L.A.C. (4th) 265; Nor‑Man Regional Health Authority Inc. v. Manitoba Association of Health Care Professionals, 2011 SCC 59, [2011] 3 S.C.R. 616; Alberta (Information and Privacy Commissioner) v. Alberta Teachers’ Association, 2011 SCC 61, [2011] 3 S.C.R. 654; Canada (Canadian Human Rights Commission) v. Canada (Attorney General), 2011 SCC 53, [2011] 3 S.C.R. 471; Dunsmuir v. New Brunswick, 2008 SCC 9, [2008] 1 S.C.R. 190; Toronto (City) Board of Education v. O.S.S.T.F., District 15, [1997] 1 S.C.R. 487; St. Anne Nackawic Pulp & Paper Co. v. Canadian Paper Workers Union, Local 219, [1986] 1 S.C.R. 704; CAIMAW v. Paccar of Canada Ltd., [1989] 2 S.C.R. 983; Canada (Citizenship and Immigration) v. Khosa, 2009 SCC 12, [2009] 1 S.C.R. 339; Prestressed Systems Inc. and L.I.U.N.A., Loc. 625 (Roberts) (Re) (2005), 137 L.A.C. (4th) 193; Halifax (Regional Municipality) and N.S.U.P.E., Local 2 (Re) (2008), 171 L.A.C. (4th) 257; Re Monarch Fine Foods Co. and Milk and Bread Drivers, Dairy Employees, Caterers and Allied Employees, Local 647 (1978), 20 L.A.C. (2d) 419; Trimac Transportation Services — Bulk Systems and T.C.U. (Re) (1999), 88 L.A.C. (4th) 237; Re United Steelworkers and Triangle Conduit & Cable Canada (1968) Ltd. (1970), 21 L.A.C. 332; Newfoundland and Labrador Nurses’ Union v. Newfoundland and Labrador (Treasury Board), 2011 SCC 62, [2011] 3 S.C.R. 708; Re Lumber & Sawmill Workers’ Union, Local 2537, and KVP Co. (1965), 16 L.A.C. 73; British Columbia (Public Service Employee Relations Commission) v. BCGSEU, [1999] 3 S.C.R. 3; Imperial Oil Ltd. v. Communications, Energy & Paperworkers Union of Canada, Local 900, 2009 ONCA 420, 96 O.R. (3d) 668; Imperial Oil Ltd. and C.E.P., Loc. 900 (Re) (2006), 157 L.A.C. (4th) 225; Fording Coal Ltd. v. United Steelworkers of America, Local 7884, [2002] B.C.C.A.A.A. No. 9 (QL); Continental Lime Ltd. and B.B.F., Loc. D575 (Re) (2002), 105 L.A.C. (4th) 263; Weyerhaeuser Co. and I.W.A. (Re) (2004), 127 L.A.C. (4th) 73; ADM Agri‑Industries Ltd. v. National Automobile, Aerospace, Transportation and General Workers’ Union of Canada (CAW‑Canada), Local 195 (Substance Abuse Policy Grievance), [2004] C.L.A.D. No. 610 (QL); Communications, Energy and Paperworkers Union, Local 777 v. Imperial Oil Ltd., May 27, 2000 (unreported); Provincial‑American Truck Transporters and Teamsters Union, Loc. 880, Re (1991), 18 L.A.C. (4th) 412; Alberta (Education) v. Canadian Copyright Licensing Agency (Access Copyright), 2012 SCC 37, [2012] 2 S.C.R. 345.

Statutes and Regulations Cited

49 C.F.R. Part 382 (2001) (U.S.).

Canadian Charter of Rights and Freedoms .

Industrial Relations Act, R.S.N.B. 1973, c. I‑4, ss. 32(1), 55(1), 77(1).

Labour Relations Act, 1995, S.O. 1995, c. 1, Sch. A, s. 48(12)(j).

Rail Safety (Adoption of National Law) Regulation 2012, No. 662 (N.S.W.).

Railway Safety Act 2005 (I.).

Authors Cited

Borack, Jules I.  “Costs and Benefits of Alternative Drug Testing Programs”.  San Diego:  Navy Personnel Research and Development Center, 1998.

Brown, Donald J. M., and David M. Beatty, with the assistance of Christine E. Deacon.  Canadian Labour Arbitration, 4th ed., vol. 1.  Toronto:  Canada Law Book, 2013 (loose‑leaf updated March 2013, release 29).

Canada.  House of Commons.  Minutes of Proceedings and Evidence of the Standing Committee on National Health and Welfare, including “‘Booze, Pills & Dope’:  Reducing Substance Abuse in Canada”, No. 28, 2nd Sess., 33rd Parl., 1987, p. 25.

Canada.  Minister of National Health and Welfare.  “Government Response to the Report of the Standing Committee on ‘Booze, Pills & Dope’:  Reducing Substance Abuse in Canada”.  Ottawa:  Minister of Supply and Services, 1988.

Keith, Norm, and Ailsa Jane Wiggins.  Alcohol and Drugs in the Canadian Workplace:  An Employer’s Guide to the Law, Prevention and Management of Substance Abuse.  Markham, Ont.:  LexisNexis, 2008.

Mitchnick, Morton, and Brian Etherington.  Labour Arbitration in Canada, 2nd ed.  Toronto:  Lancaster House, 2012.

Snyder, Ronald M.  Collective Agreement Arbitration in Canada, 4th ed.  Markham, Ont.:  LexisNexis, 2009.

United States.  Department of Transportation.  “Current Random Testing Rates” (2013) (online:  http://www.dot.gov).

                    APPEAL from a judgment of the New Brunswick Court of Appeal (Drapeau C.J. and Turnbull and Robertson JJ.A.), 2011 NBCA 58, 375 N.B.R. (2d) 92, 348 D.L.R. (4th) 105, 30 Admin. L.R. (5th) 269, 216 L.A.C. (4th) 418, 969 A.P.R. 92, [2011] N.B.J. No. 230 (QL), 2011 CarswellNB 356, affirming a decision of Grant J., 2010 NBQB 294, 367 N.B.R. (2d) 234, 199 L.A.C. (4th) 321, 946 A.P.R. 234, [2010] N.B.J. No. 331 (QL), 2010 CarswellNB 494, setting aside an award of an arbitration board, [2009] N.B.L.A.A. No. 28 (QL).  Appeal allowed, McLachlin C.J. and Rothstein and Moldaver JJ. dissenting.

                    Daniel Leger, David Mombourquette and Joël Michaud, for the appellant.

                    Neil Finkelstein, Steven Mason, Brandon Kain, Byron Shaw and William Goss, Q.C., for the respondent.

                    Barbara B. Johnston and April Kosten, for the interveners the Construction Owners Association of Alberta, Construction Labour Relations — an Alberta Association and Enform.

                    Robert Dupont, Simon‑Pierre Paquette and Johanne Cavé, for the interveners the Canadian National Railway Company, the Canadian Pacific Railway Company and Via Rail Canada Inc.

                    Written submissions only by Ritu Khullar and John Carpenter, for the intervener the Alberta Federation of Labour.

                    Ritu Khullar, for the intervener the Communications, Energy and Paperworkers Union of Canada, Local 707.

                    Joshua S. Phillips and Karen Ensslen, for the intervener the Canadian Civil Liberties Association.

                    Norman A. Keith, Ailsa Jane Wiggins and Anna Abbott, for the intervener the Alliance of Manufacturers & Exporters of Canada, carrying on business as Canadian Manufacturers & Exporters.

                    Peter A. Gall, Q.C., Andrea Zwack and Melanie Vipond, for the interveners the Canadian Mining Association, the Mining Association of British Columbia, the Mining Association of Manitoba Inc., the Québec Mining Association, the Ontario Mining Association and the Saskatchewan Mining Association.

                    Andrew K. Lokan, Emily Lawrence and Christopher M. Dassios, for the intervener the Power Workers’ Union.

                    The judgment of LeBel, Fish, Abella, Cromwell, Karakatsanis and Wagner JJ. was delivered by

[1]                              Abella J. — Privacy and safety are highly sensitive and significant workplace interests. They are also occasionally in conflict. This is particularly the case when the workplace is a dangerous one.

[2]                              In a unionized workplace, these issues are usually dealt with in the course of collective bargaining. If an employer, however, decides not to negotiate safety measures before implementing them, and if those measures have disciplinary consequences for employees, the employer must bring itself within the scope of the management rights clause of the collective agreement.

[3]                              The legal issue at the heart of this case is the interpretation of the management rights clause of a collective agreement. This is a labour law issue with clear precedents and a history of respectful recognition of the ability of collective bargaining to responsibly address the safety concerns of the workplace — and the public.

[4]                              A substantial body of arbitral jurisprudence has developed around the unilateral exercise of management rights in a safety context, resulting in a carefully calibrated “balancing of interests” proportionality approach. Under it, and built around the hallmark collective bargaining tenet that an employee can only be disciplined for reasonable cause, an employer can impose a rule with disciplinary consequences only if the need for the rule outweighs the harmful impact on employees’ privacy rights.  The dangerousness of a workplace is clearly relevant, but this does not shut down the inquiry, it begins the proportionality exercise.

[5]                              This approach has resulted in a consistent arbitral jurisprudence whereby arbitrators have found that when a workplace is dangerous, an employer can test an individual employee if there is reasonable cause to believe that the employee was impaired while on duty, was involved in a workplace accident or incident, or was returning to work after treatment for substance abuse. In the latter circumstance, the employee may be subject to a random drug or alcohol testing regime on terms negotiated with the union.

[6]                              But a unilaterally imposed policy of mandatory, random and unannounced testing for all employees in a dangerous workplace has been overwhelmingly rejected by arbitrators as an unjustified affront to the dignity and privacy of employees unless there is reasonable cause, such as a general problem of substance abuse in the workplace. This body of arbitral jurisprudence is of course not binding on this Court, but it is nevertheless a valuable benchmark against which to assess the arbitration board’s decision in this case.

[7]                              It cannot be seriously challenged, particularly since Dunsmuir v. New Brunswick, 2008 SCC 9, [2008] 1 S.C.R. 190, that the applicable standard for reviewing the decision of a labour arbitrator is reasonableness (Dunsmuir, at para. 68; Nor-Man Regional Health Authority Inc. v. Manitoba Association of Health Care Professionals, 2011 SCC 59, [2011] 3 S.C.R. 616, at paras. 31 and 42; Newfoundland and Labrador Nurses’ Union v. Newfoundland and Labrador (Treasury Board), 2011 SCC 62, [2011] 3 S.C.R. 708).

[8]                              In a thoughtful and meticulous decision of almost 80 pages, a majority of the arbitration board in this case, applying the arbitral consensus, concluded that the employer, Irving Pulp & Paper, Limited, exceeded the scope of its management rights under a collective agreement by imposing random alcohol testing in the absence of evidence of a workplace problem with alcohol use.  In my view, based on the board’s findings of fact and its reliance on the arbitral consensus for determining the scope of the employer’s rights under the collective agreement in such circumstances, the decision was a reasonable one.

Background

[9]                              Irving operates a kraft paper mill in Saint John, New Brunswick. Between 1991 and 2006, Irving had no formal policy with respect to alcohol and drug use at the mill. In 2006, it unilaterally adopted a “Policy on Alcohol and Other Drug Use” under the management rights clause of the collective agreement without any negotiations with the union.  The policy imposed drug or alcohol testing for employees holding positions designated by Irving as “safety sensitive”. 

[10]                          The policy contained a universal random alcohol testing component, whereby 10% of the employees in safety sensitive positions were to be randomly selected for unannounced breathalyzer testing over the course of a year. A positive test for alcohol, that is, one showing a blood alcohol concentration greater than 0.04%, attracted significant disciplinary action, including dismissal. Failure to submit to testing was grounds for immediate dismissal.

[11]                          Among the employees randomly tested under this aspect of the policy was Perley Day, a member of the Communications, Energy and Paperworkers Union of Canada, Local 30.  Mr. Day was a teetotaller who had not had a drink since 1979.  His breathalyzer test revealed a blood alcohol level of zero. The Union filed a grievance on his behalf challenging only the random alcohol testing aspect of the policy. 

[12]                          The rest of the testing policy was not challenged. Under it, employees were subject to mandatory testing if there was reasonable cause to suspect the employee of alcohol or other drug use in the workplace, after direct involvement in a work-related accident or incident, or as part of a monitoring program for any employee returning to work following voluntary treatment for substance abuse.

[13]                          Mr. Day’s inclusion in the class of employees occupying safety sensitive positions was undisputed, as was the fact that the workplace represented a dangerous work environment.  However there were only eight documented incidents of alcohol consumption or impairment at the workplace over a period of 15 years from April 1991 to January 2006. Nor were there any accidents, injuries or near misses connected to alcohol use. By December 2008, when the arbitration was heard, the testing policy had been in effect for 22 months, during which not a single employee had tested positive on either a random test or a test for reasonable cause.

[14]                          The absence of evidence of any real risk related to alcohol led a majority of the board to conclude that there was little benefit to the employer in maintaining the random testing policy.  Weighing the employer’s interest in random alcohol testing as a workplace safety measure against the harm to the privacy interests of employees, the board therefore allowed the grievance and concluded that the random testing policy was unjustified:

        The question is now one of proportionality. What needs to be measured are the benefits that will accrue to the employer through the application of the random alcohol testing policy against the harm that will be done to the employee’s right to privacy. If the random alcohol testing policy is to be justified, these must be in proportion. Here the employer’s scheme gets into heavier weather.

 

        In a word, on the evidence I heard, I do not conclude that any significant degree of incremental safety risk attributable to employee alcohol use has been demonstrated to exist in this workplace. Taken with the low testing percentages, I believe it is likely that the employer’s policy will seldom, if ever, identify any employee with a blood alcohol concentration over the 0.04% Policy cut-off limit. I therefore see little or no concrete advantage to the employer to be gained through the random alcohol testing policy.

 

     On the other side of the balance scale, I have to consider the employee’s right to privacy.  Rights to privacy and to the related right of security of the person are important and prized incidents of Canadian citizenship. Reactions to invasions of them tend to be prompt, visceral, instinctive and uniformly negative. When the testing is random — that is, without articulable cause — as it is here, an already high bar is raised even higher. This considerably increases the burden of justification on the employer.

 

        The invasion of that privacy by the random alcohol testing policy is not a trifle. It effects a significant inroad. Specifically, it involves a bodily intrusion and the surrender of bodily substances. It involves coercion and restriction on movement. Upon pain of significant punishment, the employee must go promptly to the breathalyzer station and must co-operate in the provision of breath samples. As we saw with Mr. Day, there can be an element of public embarrassment. Taking its results together, the scheme effects a loss of liberty and personal autonomy. These are at the heart of the right to privacy.

 

        On the evidence, the gains likely to result to the employer from random alcohol testing rule[s] run from uncertain to exist at all to minimal at best. The inroads into employee privacy are significant and out of proportion to any benefit, actual or reasonably to be expected to be had by the employer and disclosed by the evidence. The employer has not been able to tilt the balance in its favour and therefore justify the imposition of random alcohol testing as a proportionate response to a demonstrated incremental risk caused by the attendance of employees at work with alcohol in their bodies. I therefore find that the random alcohol testing provisions of the Policy do not meet the KVP reasonableness test, and for that reason are unenforceable. That portion of the Policy therefore must be, and hereby is set aside. [Emphasis added; A.R., vol. IV, at pp. 71-73.]

[15]                          On judicial review, the board’s award was set aside as unreasonable because of the dangerousness of the workplace.  The New Brunswick Court of Appeal dismissed the appeal. The court applied a bifurcated standard of review. It applied a correctness standard to the board’s analytical framework for determining the validity of the employer’s random alcohol testing policy and a reasonableness standard to the board’s factual findings. Using this segmented approach, the Court of Appeal substituted its own legal framework and concluded that no balancing of interests was required in a dangerous workplace, whether or not it was unionized. As a result, it held that employers can unilaterally impose random alcohol testing in any dangerous workplace, unionized or non-unionized, without having to show reasonable cause, such as evidence of an existing problem with alcohol use. It also found the board’s findings regarding the degree of dangerousness at the workplace to be unreasonable.

[16]                          In my respectful view, the Court of Appeal erred in disregarding this Court’s direction that decisions of labour arbitrators be reviewed for reasonableness and that deference be paid to their legal and factual findings when they are interpreting collective agreements. This misapplication of the standard of review led the Court of Appeal away from its required task of determining whether the board’s decision fell within a range of reasonable outcomes, and towards a substitution of its own views as to the proper legal framework and factual findings. It also led the court essentially to disregard the remarkably consistent arbitral jurisprudence for balancing safety and privacy in a dangerous workplace, and to impose instead a novel, unfettered and automatic remedy outside the existing consensus and expectations in the labour relations community about how these issues are to be approached under a collective agreement.

Analysis

[17]                          At the outset, it is important to note that since we are dealing with a workplace governed by a collective agreement, that means that the analytical framework for determining whether an employer can unilaterally impose random testing is determined by the arbitral jurisprudence. Cases dealing with random alcohol or drug testing in non-unionized workplaces under human rights statutes are, as a result, of little conceptual assistance  (Entrop v. Imperial Oil Ltd. (2000), 50 O.R. (3d) 18 (C.A.)).

[18]                          It may be tempting to suggest that dangerous unionized workplaces should be beyond the reach of the collective bargaining regime, freeing an employer both from the duty to negotiate with the union and from the terms of the collective agreement.  This suggests, Cassandra-like and evidence-free, that collective bargaining is the altar on which public and workplace safety is sacrificed and that only employers have the capacity to address these concerns.

[19]                           But the reality is that the task of negotiating workplace conditions, both on the part of unions and management, as well as the arbitrators who interpret the resulting collective agreement, has historically — and successfully — included the delicate, case-by-case balancing required to preserve public safety concerns while protecting privacy. Far from leaving the public at risk, protecting employees — who are on the front line of any danger — necessarily also protects the surrounding public. To suggest otherwise is a counter-intuitive dichotomy. 

[20]                          And this without any evidence that dangerous workplaces that are unionized have experienced any, let alone a disproportionate number of, accidents resulting from collectively bargaining safety measures.  It also assumes that no balancing is required at all once a finding is made that a workplace is dangerous.  This not only negates any recognition of the significant privacy interests at play, it wrongly assumes that when there is no collective agreement, an employer is free to exercise its own discretion about worker safety.  All provinces have legislation protecting worker safety, thereby restricting an employer’s wishes.  And, as we saw in Entrop, even in a non-unionized workplace, an employer must justify the intrusion on privacy resulting from random testing by reference to the particular risks in a particular workplace.  There are different analytic steps involved, but both essentially require attentive consideration and balancing of the safety and privacy interests.

[21]                          As the board recognized, the only possible source of the employer’s asserted right to impose random alcohol testing unilaterally was the management rights clause in the collective agreement:

                    4.01.    The Union recognizes and acknowledges that it is the right of the Company to operate and manage its business subject to the terms and provisions of this agreement.

The legal issue, as a result, is whether implementing a random alcohol testing policy was a valid exercise of the employer’s management rights under the collective agreement.

[22]                          When employers in a unionized workplace unilaterally enact workplace rules and policies, they are not permitted to “promulgate unreasonable rules and then punish employees who infringe them” (Re United Steelworkers, Local 4487 & John Inglis Co. Ltd. (1957), 7 L.A.C. 240 (Laskin), at p. 247; see also Re United Brewery Workers, Local 232, & Carling Breweries Ltd. (1959), 10 L.A.C. 25 (Cross)).      

[23]                          This constraint arises because an employer may only discharge or discipline an employee for “just cause” or “reasonable cause” — a central protection for employees. As a result, rules enacted by an employer as a vehicle for discipline must meet the requirement of reasonable cause (Re Public Utilities Commission of the Borough of Scarborough and International Brotherhood of Electrical Workers, Local 636 (1974), 5 L.A.C. (2d) 285 (Rayner), at pp. 288-89; see also United Electrical, Radio, and Machine Workers of America, Local 524, in re Canadian General Electric Co. Ltd. (Peterborough) (1951), 2 L.A.C. 688 (Laskin), at p. 690; Re Hamilton Street Railway Co. and Amalgamated Transit Union, Division 107 (1977), 16 L.A.C. (2d) 402 (Burkett), at paras. 9-10; Ronald M. Snyder, Collective Agreement Arbitration in Canada (4th ed. 2009), at paras. 10.1 and 10.96).

[24]                          The scope of management’s unilateral rule-making authority under a collective agreement is persuasively set out in Re Lumber & Sawmill Workers’ Union, Local 2537, and KVP Co. (1965), 16 L.A.C. 73 (Robinson). The heart of the “KVP test”, which is generally applied by arbitrators, is that any rule or policy unilaterally imposed by an employer and not subsequently agreed to by the union, must be consistent with the collective agreement and be reasonable (Donald J. M. Brown and David M. Beatty, Canadian Labour Arbitration (4th ed. (loose-leaf)), vol. 1, at topic 4:1520).

[25]                          The KVP test has also been applied by the courts. Tarnopolsky J.A. launched the judicial endorsement of KVP in Metropolitan Toronto (Municipality) v. C.U.P.E. (1990), 74 O.R. (2d) 239 (C.A.), leave to appeal refused, [1990] 2 S.C.R. ix, concluding that the “weight of authority and common sense” supported the principle that “all company rules with disciplinary consequences must be reasonable” (pp. 257-58 (emphasis in original)).  In other words:

The Employer cannot, by exercising its management functions, issue unreasonable rules and then discipline employees for failure to follow them. Such discipline would simply be without reasonable cause. To permit such action would be to invite subversion of the reasonable cause clause. [p. 257]

[26]                          Subsequent appellate decisions have accepted that rules unilaterally made in the exercise of management discretion under a collective agreement must not only be consistent with the agreement, but must also be reasonable if the breach of the rule results in disciplinary action (Charlottetown (City) v. Charlottetown Police Association (1997), 151 Nfld. & P.E.I.R. 69 (P.E.I.S.C. (App. Div.)), at para. 17; see also N.A.P.E. v. Western Avalon Roman Catholic School Board, 2000 NFCA 39, 190 D.L.R. (4th) 146, at para. 34; St. James-Assiniboia Teachers’ Assn. No. 2  v. St. James-Assiniboia School Division No. 2, 2002 MBCA 158, 222 D.L.R. (4th) 636, at paras. 19-28).

[27]                          In assessing KVP reasonableness in the case of unilaterally imposed employer rules or policies affecting employee privacy, arbitrators have used a “balancing of interests” approach. As the intervener the Alberta Federation of Labour noted:

Determining reasonableness requires labour arbitrators to apply their labour relations expertise, consider all of the surrounding circumstances, and determine whether the employer’s policy strikes a reasonable balance. Assessing the reasonableness of an employer’s policy can include assessing such things as the nature of the employer’s interests, any less intrusive means available to address the employer’s concerns, and the policy’s impact on employees.  [I.F., at para. 4]

[28]                          In the earliest privacy cases using a balancing of interests approach, arbitrators generally found that employers could only exercise a unilateral management right to search an individual employee’s personal effects if there was a reasonable suspicion that the employee had committed theft. Universal random searches — that is, random searches of the entire workforce — were rejected as unreasonable unless there was a workplace problem with theft and the employer had exhausted less intrusive alternative measures for addressing the problem (Morton Mitchnick and Brian Etherington, Labour Arbitration in Canada (2nd ed. 2012), at pp. 308-9; Brown and Beatty, at topic 7:3625).

[29]                          The balancing of interests approach was subsequently applied in assessing the reasonableness of unilaterally imposed employer policies calling for universal random drug or alcohol testing of all employees performing safety sensitive work. Universal random testing refers to the testing of individual employees randomly selected from all or some portion of the workforce. As in the search cases, arbitrators rejected unilaterally imposed universal random testing policies as unreasonable unless there had been a workplace problem with substance abuse and the employer had exhausted alternative means for dealing with the abuse.

[30]                          In a workplace that is dangerous, employers are generally entitled to test individual employees who occupy safety sensitive positions without having to show that alternative measures have been exhausted if there is “reasonable cause” to believe that the employee is impaired while on duty, where the employee has been directly involved in a workplace accident or significant incident, or where the employee is returning to work after treatment for substance abuse. (See Esso Petroleum Canada and C.E.P., Loc. 614, Re (1994), 56 L.A.C. (4th) 440 (McAlpine); Canadian National Railway Co. and C.A.W.-Canada (Re) (2000), 95 L.A.C. (4th) 341 (M. Picher), at pp. 377-78; Weyerhaeuser Co. and I.W.A. (Re) (2004), 127 L.A.C. (4th) 73 (Taylor), at p. 109; Navistar Canada, Inc. and C.A.W., Local 504 (Re) (2010), 195 L.A.C. (4th) 144 (Newman), at pp. 170 and 177; Rio Tinto Alcan Primary Metal and C.A.W.-Canada, Local 2301 (Drug and Alcohol Policy) (Re) (2011), 204 L.A.C. (4th) 265 (Steeves), at para. 37(b)-(d).)

[31]                          But the dangerousness of a workplace — whether described as dangerous, inherently dangerous, or highly safety sensitive — is, while clearly and highly relevant, only the beginning of the inquiry.  It has never been found to be an automatic justification for the unilateral imposition of unfettered random testing with disciplinary consequences. What has been additionally required is evidence of enhanced safety risks, such as evidence of a general problem with substance abuse in the workplace.

[32]                          The blueprint for dealing with dangerous workplaces is found in Imperial Oil Ltd. and C.E.P., Loc. 900 (Re) (2006), 157 L.A.C. (4th) 225 (“Nanticoke”), a case involving a grievance of the employer’s random drug testing policy at an oil refinery, which the parties acknowledged was highly safety sensitive. Arbitrator Michel Picher summarized the principles emerging from 20 years of arbitral jurisprudence under the KVP test for both drug and alcohol testing:

•         No employee can be subjected to random, unannounced alcohol or drug testing, save as part of an agreed rehabilitative program.

 

•         An employer may require alcohol or drug testing of an individual where the facts give the employer reasonable cause to do so.

 

•         It is within the prerogatives of management’s rights under a collective agreement to also require alcohol or drug testing following a significant incident, accident or near miss, where it may be important to identify the root cause of what occurred.

 

•         Drug and alcohol testing is a legitimate part of continuing contracts of employment for individuals found to have a problem of alcohol or drug use. As part of an employee’s program of rehabilitation, such agreements or policies requiring such agreements may properly involve random, unannounced alcohol or drug testing generally for a limited period of time, most commonly two years. In a unionized workplace the Union must be involved in the agreement which establishes the terms of a recovering employee’s ongoing employment, including random, unannounced testing. This is the only exceptional circumstance in which the otherwise protected employee interest in privacy and dignity of the person must yield to the interests of safety and rehabilitation, to allow for random and unannounced alcohol or drug testing. [Emphasis added; para. 100.]

[33]                          There can, in other words, be testing of an individual employee who has an alcohol or drug problem.  Universal, random testing, however, is far from automatic.  The reason is explained by Arbitrator Picher in Nanticoke as follows:

     . . . a key feature of the jurisprudence in the area of alcohol or drug testing in Canada is that arbitrators have overwhelmingly rejected mandatory, random and unannounced drug testing for all employees in a safety sensitive workplace as being an implied right of management under the terms of a collective agreement. Arbitrators have concluded that to subject employees to an alcohol or drug test when there is no reasonable cause to do so, or in the absence of an accident or near miss and outside the context of a rehabilitation plan for an employee with an acknowledged problem is an unjustified affront to the dignity and privacy of employees which falls beyond the balancing of any legitimate employer interest, including deterrence and the enforcement of safe practices. In a unionized workplace, such an extraordinary incursion into the rights of employees must be expressly and clearly negotiated. It is not to be inferred solely from general language describing management rights or from language in a collective agreement which enshrines safety and safe practices. [Emphasis added; para. 101.]

[34]                          Significantly, Arbitrator Picher acknowledged that the application of the balancing of interests approach could permit general random testing “in some extreme circumstances”:

        It may well be that the balancing of interests approach . . . would allow for general random, unannounced drug testing in some extreme circumstances. If, for example, an employer could marshal evidence which compellingly demonstrates an out-of-control drug culture taking hold in a safety sensitive workplace, such a measure might well be shown to be necessary for a time to ensure workplace safety. That might well constitute a form of “for cause” justification.

 

(Nanticoke, at para. 127)

[35]                          In the case before him, however, since there was no evidence of a substance abuse problem at the oil refinery, the random drug testing component of the policy was found to be unjustified (Nanticoke, at para. 127). His decision was upheld as reasonable by the Ontario Court of Appeal (Imperial Oil Ltd. v. Communications, Energy & Paperworkers Union of Canada, Local 900, 2009 ONCA 420, 96 O.R. (3d) 668).

[36]                          The balancing of interests approach has not kept employers from enacting comprehensive drug and alcohol policies, which can include rules about drugs and alcohol in the workplace, discipline for employees who break those rules, education and awareness training for employees and supervisors, access to treatment for substance dependence, and after-care programs for employees returning to work following treatment.

[37]                          But I have been unable to find any cases, either before or since Nanticoke, in which an arbitrator has concluded that an employer could unilaterally implement random alcohol or drug testing, even in a highly dangerous workplace, absent a demonstrated workplace problem (Esso Petroleum, at pp. 447-48; Metropol Security, a division of Barnes Security Services Ltd. and U.S.W.A., Loc. 5296 (Drug and Alcohol testing) (Re) (1998), 69 L.A.C. (4th) 399; Trimac Transportation Services — Bulk Systems and T.C.U. (Re) (1999), 88 L.A.C. (4th) 237; Canadian National, at pp. 385 and 394; Fording Coal Ltd. v. United Steelworkers of America, Local 7884, [2002] B.C.C.A.A.A. No. 9 (QL), at para. 30; ADM Agri-Industries Ltd. v. National Automobile, Aerospace, Transportation and General Workers’ Union of Canada (CAW-Canada), Local 195 (Substance Abuse Policy Grievance), [2004] C.L.A.D. No. 610 (QL), at para. 77; Petro-Canada Lubricants Centre (Mississauga) and Oakville Terminal and C.E.P., Local 593 (Re) (2009), 186 L.A.C. (4th) 424 (Kaplan), at pp. 434-37; Rio Tinto, at para. 37(a) and (d)).

[38]                          In the only two arbitration decisions that have upheld random alcohol testing, the employers were found to be justified in implementing random alcohol testing for employees in safety sensitive positions because there was a demonstrated general problem with alcohol use in a dangerous workplace (Communications, Energy and Paperworkers Union, Local 777 v. Imperial Oil Ltd., T. J. Christian, Chair, May 27, 2000, unreported (“Strathcona”); Greater Toronto Airports Authority v. Public Service Alliance of Canada, Local 0004, [2007] C.L.A.D. No. 243 (QL) (Devlin) (“GTAA”)).

[39]                          In Strathcona, the arbitrator upheld the termination of an employee in a safety sensitive position at an oil refinery who tested positive on a random alcohol test. Imperial Oil Limited had implemented the random testing policy after surveying employees across all its facilities about alcohol-related incidents and near misses. According to the survey, the plant operations group that included the grievor’s position had a disproportionately high rate of accidents due to substance abuse, with 2.7% of employees reporting that they had personally had near misses due to substance use in the previous 12 months. The arbitrator accepted the survey results as a “rational and sufficient foundation for the random testing Policy” (p. 73).  He concluded that “there is evidence of a problem with alcohol use by employees at the Strathcona Refinery” (p. 60).  On that basis, he upheld the reasonableness of the random testing policy and the consequential discipline.

[40]                          In GTAA, the employer had a random alcohol and drug testing policy for individuals occupying safety sensitive positions at Pearson International Airport in Toronto. The arbitrator acknowledged that “the safety-sensitive nature of a particular industry [is] not, in itself, sufficient to outweigh the privacy interests of individual employees and to support a regime of random testing” (para. 251) and that “[a]rbitrators have required evidence of a drug and/or alcohol problem in the workplace which cannot be addressed by less invasive means” (para. 254).

[41]                          The evidence showed a “pervasive problem” with alcohol (GTAA, at para. 262).  Both employer and union witnesses testified about numerous occasions when they had seen employees drinking on the job or storing alcohol at work, smelled alcohol on other employees’ breath, or found empty liquor containers on site. There were also concerns that alcohol abuse at work often went unreported. Based on this evidence, the arbitrator concluded that random alcohol testing was a reasonable employer policy. Because there was little evidence of on-the-job drug use, however, the random drug testing aspect of the policy was found not to be justified.

[42]                          This arbitral consensus, which was carefully applied by the board, helps inform why its decision was reasonable on the facts of this case.

[43]                           The board framed the question using the accepted KVP balancing of interests approach: Was the benefit to the employer from the random alcohol testing policy in this dangerous workplace proportional to the harm to employee privacy?

[44]                          To assess the employer’s side of the balance, the board canvassed the risks that the employer intended to address by random alcohol testing. It examined both the risk associated with the particular grievor’s position as a millwright and the risk associated with the particular workplace. After reviewing the employer’s risk assessments of different safety sensitive positions, the board found that the workplace was “one in which great care must be taken with safe work practices”. There were “risks and dangers in the operations performed both to the incumbent, and to others, as well as to the environment and to property”. The board therefore concluded that “the mill in normal operation is a dangerous work environment”. These conclusions have not been challenged.

[45]                          But, as previously noted, the fact that a workplace is found to be dangerous does not automatically give the employer the right to impose random testing unilaterally. The dangerousness of the workplace has only justified the testing of particular employees in certain circumstances: where there are reasonable grounds to believe that the employee was impaired while on duty, where the employee was directly involved in a workplace accident or significant incident, or where the employee returns to work after treatment for substance abuse. It has never, to my knowledge, been held to justify random testing, even in the case of “highly safety sensitive” or “inherently dangerous” workplaces like railways (Canadian National) and chemical plants (DuPont Canada Inc. and C.E.P., Loc. 28-O (Re) (2002), 105 L.A.C. (4th) 399), or even in workplaces that pose a risk of explosion (ADM Agri-Industries), in the absence of a demonstrated problem with alcohol use in that workplace. That is not to say that it is beyond the realm of possibility in extreme circumstances, but we need not decide that in this case.

[46]                          This obliged the board to consider whether there was evidence of an alcohol-related problem in the workplace. There were eight documented alcohol-related incidents at the mill from April 29, 1991, to January 11, 2006. Only one witness, a former employee, gave any evidence about alcohol use in the workplace, but the board found his evidence to be “dated” and “not persuasive”.

[47]                          The board concluded that these eight incidents over a 15-year period did not reflect a significant problem with workplace alcohol use. As a result, the board concluded that there was a “very low incremental risk of safety concerns based on alcohol-related impaired performance of job tasks at the site”.

[48]                          While the employer had argued that deterrence was a major benefit of random alcohol testing, the board was not satisfied that there was any evidence of a deterrent effect at the mill. The only evidence supporting the employer’s view was that of its expert witness, who described deterrence as the main theoretical goal of random alcohol testing policies, but had no information about this particular workplace. In the board’s view, the lack of any positive test results in almost two years of random alcohol testing was equally consistent with the opposite conclusion: that there was no workplace alcohol abuse to deter.

[49]                          On the other side of the balance was the employee right to privacy. The board accepted that breathalyzer testing “effects a significant inroad” on privacy, involving

coercion and restriction on movement. Upon pain of significant punishment, the employee must go promptly to the breathalyzer station and must co-operate in the provision of breath samples. . . . Taking its results together, the scheme effects a loss of liberty and personal autonomy. These are at the heart of the right to privacy.

[50]                          That conclusion is unassailable. Early in the life of the Canadian   Charter of Rights and Freedoms , this Court recognized that “the use of a person’s body without his consent to obtain information about him, invades an area of personal privacy essential to the maintenance of his human dignity” (R. v. Dyment, [1988] 2 S.C.R. 417, at pp. 431-32).   And in R. v. Shoker, 2006 SCC 44, [2006] 2 S.C.R. 399, it notably drew no distinction between drug and alcohol testing by urine, blood or breath sample, concluding that the “seizure of bodily samples is highly intrusive and, as this Court has often reaffirmed, it is subject to stringent standards and safeguards to meet constitutional requirements” (para. 23).

[51]                          In the end, the expected safety gains to the employer in this case were found by the board to range “from uncertain . . .  to minimal at best”, while the impact on employee privacy was found to be much more severe. Consequently, the board concluded that the employer had not demonstrated the requisite problems with dangerousness or increased safety concerns such as workplace alcohol use that would justify universal random testing. Random alcohol testing was therefore held to be an unreasonable exercise of management rights under the collective agreement.  I agree.

[52]                          This is not to say that an employer can never impose random testing in a dangerous workplace. If it represents a proportionate response in light of both legitimate safety concerns and privacy interests, it may well be justified.  

[53]                          Moreover, the employer is not only always free to negotiate drug and alcohol testing policies with the union, as was said in Nanticoke, “such an extraordinary incursion into the rights of employees must be expressly and clearly negotiated” (para. 101 (emphasis added)).  But where, as here, the employer proceeds unilaterally without negotiating with the union, it must comply with the time-honoured requirement of showing reasonable cause before subjecting employees to potential disciplinary consequences. Given the arbitral consensus, an employer would be justifiably pessimistic that a policy unilaterally imposing random alcohol testing in the absence of reasonable cause would survive arbitral scrutiny.

[54]                          The board’s decision should be approached as an organic whole, without a line-by-line treasure hunt for error (Newfoundland Nurses, at para. 14).  In the absence of finding that the decision, based on the record, is outside the range of reasonable outcomes, the decision should not be disturbed.  In this case, the board’s conclusion was reasonable and ought not to have been disturbed by the reviewing courts.

[55]                          I would therefore allow the appeal with costs throughout.

                    The reasons of McLachlin C.J. and Rothstein and Moldaver JJ. were delivered by

[56]                          Rothstein and Moldaver JJ. (dissenting) — Where labour and management fail to agree on the introduction of a new workplace policy, legislatures have delegated the task of adjudicating their dispute to labour arbitrators.  In this case, a union challenged management’s proactive adoption of a random alcohol testing policy at a paper mill, which the union accepts is inherently dangerous, carrying risks that go beyond the mill’s four corners.  An arbitral board struck down the policy. 

[57]                          In striking down the policy, we conclude that the board departed from an arbitral consensus that has attempted to strike a balance between competing interests in privacy and safety in the workplace.  The board put its thumb on the scales and upset the careful balance established in the arbitral jurisprudence.  In so doing, it came to an unreasonable decision.  Accordingly, we respectfully dissent from the majority opinion upholding the board’s decision.

I.       Overview

[58]                          Irving Pulp & Paper, Limited (“Irving”) operates a paper mill in Saint John, New Brunswick.  The mill is located along the banks of the Saint John River, near the point where the river empties into the Bay of Fundy.  The mill’s operations involve hazardous chemicals and gases, heavy machinery and equipment, high-pressure boilers and steam lines, and high-voltage electric lines.  It is uncontroversial that the mill, in normal operation, is a dangerous environment that presents risks not only to the employees of the mill, but also to the public, to property, and to the environment.  The evidence discloses that malfunctions at the mill carry “a potential for ‘catastrophic failures’” (board’s reasons, [2009] N.B.L.A.A. No. 28 (QL), at para. 101).

[59]                          On February 1, 2006, Irving adopted a comprehensive policy concerning employee drug and alcohol use at the mill.  The company was not unique in this regard.  After the Exxon Valdez ran aground in Prince William Sound in 1989, spilling hundreds of thousands of barrels of crude oil into the ocean, Imperial Oil Ltd., the Canadian subsidiary of Exxon Mobil Corporation, began implementing similar policies at its Canadian oil refineries (Esso Petroleum Canada v. Communications, Energy & Paperworkers’ Union, Local 614, [1994] B.C.C.A.A.A. No. 244 (QL) (McAlpine), at para. 5; Entrop v. Imperial Oil Ltd. (2000), 50 O.R. (3d) 18 (C.A.), at para. 5).  The operators of dangerous facilities in a variety of other industries have taken similar steps.  See, e.g., Canadian National Railway Co. and C.A.W.-Canada (Re) (2000), 95 L.A.C. (4th) 341 (M. Picher) (“C.N.R.”) (rail operations); Greater Toronto Airports Authority v. Public Service Alliance of Canada, Local 0004, [2007] C.L.A.D. No. 243 (QL) (Devlin) (“GTAA”) (airport ground operations); Rio Tinto Alcan Primary Metal and C.A.W.-Canada, Local 2301 (Drug and Alcohol Policy) (Re) (2011), 204 L.A.C. (4th) 265 (Steeves) (aluminum smelter).

[60]                          In explaining its decision to employees, Irving said that “the implementation of this policy is one more component of our overall safety program, which minimizes the risks associated with our operations in order to ensure a safe, healthy and productive workplace” (A.R., vol. II, at p. 70).  The policy included both proactive and reactive components and addressed various issues from voluntary assessment and rehabilitation to drug and alcohol testing in defined circumstances.  One aspect — the one that gave rise to this case — required that employees in designated “Safety Sensitive Positions . . .  be subjected to unannounced random tests for alcohol” using a breathalyser (A.R., vol. II, at p. 76 (emphasis added)).  A positive test showing a blood alcohol concentration greater than 0.04 percent would lead to disciplinary action, determined on a case-by-case basis.  The policy did not provide for random drug testing.  It did require testing of employees in safety-sensitive positions for drug or alcohol use after an accident in the workplace (“post-incident” testing) and where there was a reasonable basis to suspect alcohol or drug use or possession (“reasonable cause” testing).

[61]                          On March 13, 2006, Perley Day, who worked in the mill’s maintenance department in a safety-sensitive position, was informed by his supervisor that he had been randomly selected for a breathalyser by a computer program managed off-site by an independent third party.  This upset Mr. Day, who has been a teetotaler since 1979.  He nonetheless went along with the test, because failing to do so could have led to disciplinary action.  He tested negative.  On April 12, Mr. Day’s union, Communications, Energy and Paperworkers Union of Canada, Local 30 (“Union”), filed a grievance with Irving on his behalf.  Mr. Day grieved that “there was no reasonable grounds to test or a significant accident or incident which would justify such a measure” (A.R., vol. II, at p. 62).  At bottom, Mr. Day objected to the random alcohol testing component of the policy; he had no quibble with those aspects concerned with so-called reasonable cause or post-incident testing.

[62]                          Mr. Day’s grievance ultimately went before a labour arbitration board, where a majority of the board set aside the random alcohol testing portion of the company’s policy.  The board applied what it understood to be the existing test in the arbitral jurisprudence for review of employer rules concerning drug and alcohol testing and concluded that “[t]he inroads into employee privacy are significant and out of proportion to any benefit, actual or reasonably to be expected to be had by the employer and disclosed by the evidence” (para. 123).  On judicial review, the court did not take issue with the board’s articulation of the legal test, but it quashed the board’s decision because it was “unreasonable in that it is not an outcome which is defensible in the context of their earlier findings regarding the dangerous nature of the workplace and the minimally intrusive nature of the testing” (2010 NBQB 294, 367 N.B.R. (2d) 234, at para. 70).  On appeal, the New Brunswick Court of Appeal applied a standard of correctness to the board’s decision and concluded that the test in the arbitral jurisprudence was flawed because “[e]vidence of an existing alcohol problem in the workplace is unnecessary once the employer’s work environment is classified as inherently dangerous” (2011 NBCA 58, 375 N.B.R. (2d) 92, at para. 52).  On the strength of its new test, the Court of Appeal found the board’s decision incorrect and thus dismissed the appeal.

[63]                          We would affirm the decisions of the two courts below quashing the board’s decision, but do so for different reasons.  We agree with the majority that the appropriate standard of review is reasonableness.  In our view, however, the board made two findings that are fatal to the reasonableness of its decision.  First, though purporting to apply the test emerging from the arbitral consensus, the board misstated an element of the test that was essential to its ultimate decision.   More specifically, the board elevated the threshold of evidence that Irving was required to lead in order to justify its random alcohol testing policy, but it offered no reason for doing so.  Second, in applying the evidentiary element of the test, the board supported its conclusion by making an unreasonable inference from the factual record.  Because these findings rendered the board’s decision unreasonable, we would dismiss the appeal and affirm the order of the court below quashing the board’s decision.

II.     Analysis

[64]                          At the heart of the dispute between Irving and the Union is the quantum of evidence that the operator of a dangerous workplace is required to introduce before it can exercise its management rights under the parties’ collective agreement to adopt a proactive (that is, random) as opposed to a reactive (that is, a reasonable cause or post-incident) alcohol testing policy.  In our view, the consensus reflected in the arbitral jurisprudence provides an answer to that question.  Before turning to that jurisprudence and the board’s departure from it, we begin our analysis with the standard of review, which occupied much attention at the Court of Appeal and before this Court. 

A.     The Standard of Review for Labour Arbitration Awards Is Reasonableness

[65]                          There is no question in this case about the appropriate standard of review:   it is reasonableness.  As Fish J. emphasized for a unanimous Court only two years ago, “[p]revailing case law clearly establishes that arbitral awards under a collective agreement are subject, as a general rule, to the reasonableness standard of review” (Nor-Man Regional Health Authority Inc. v. Manitoba Association of Health Care Professionals, 2011 SCC 59, [2011] 3 S.C.R. 616, at para. 31).  This case is no exception.

[66]                          The Court of Appeal’s conclusion that a standard of correctness is warranted in this case rests, at bottom, on its assertion that “at its core this appeal is of importance to the public at large” (para. 56).  With respect, the prospect that this dispute may be of wider public concern because of the risks posed by the mill cannot, on its own, transform the legal question here into a “questio[n] of law that [is] of central importance to the legal system as a whole and that [is] outside the adjudicator’s expertise” (Alberta (Information and Privacy Commissioner) v. Alberta Teachers’ Association, 2011 SCC 61, [2011] 3 S.C.R. 654, at para. 30, citing Canada (Canadian Human Rights Commission) v. Canada (Attorney General), 2011 SCC 53, [2011] 3 S.C.R. 471, at para. 18; see also Dunsmuir v. New Brunswick, 2008 SCC 9, [2008] 1 S.C.R. 190, at paras. 55 and 60).  On the contrary, this case asks whether management’s exercise of its unilateral rule-making power can be justified under a collective agreement.  That question is plainly part of labour arbitrators’ bread and butter.  This dispute has little legal consequence outside the sphere of labour law and that, not its potential real-world consequences, determines the applicable standard of review. 

[67]                          The privileged position of labour arbitrators is a product of “their distinctive role in fostering peace in industrial relations” (Nor-Man, at para. 47), which “is important . . . to society as a whole” (Toronto (City) Board of Education v. O.S.S.T.F., District 15, [1997] 1 S.C.R. 487, at para. 36).  Since at least St. Anne Nackawic Pulp & Paper Co. v. Canadian Paper Workers Union, Local 219, [1986] 1 S.C.R. 704, this Court has been unequivocal in emphasizing the respect that courts must show for the legislative choice to delegate such decisions to labour arbitration boards.  As Estey J. observed:

The more modern approach is to consider that labour relations legislation provides a code governing all aspects of labour relations . . . .

 

. . .

 

What is left is an attitude of judicial deference to the arbitration process. . . . It is based on the idea that if the courts are available to the parties as an alternative forum, violence is done to a comprehensive statutory scheme designed to govern all aspects of the relationship of the parties in a labour relations setting. Arbitration . . .  is an integral part of that scheme, and is clearly the forum preferred by the legislature for resolution of disputes arising under collective agreements. [Emphasis added; pp. 718-21.]

[68]                          As a consequence of this legislative choice, labour arbitrators are entitled to a “protected zone of deference” (Nor-Man, at para. 43) in which the courts should not willingly enter as competing “arbiters of labour policy” (CAIMAW v. Paccar of Canada Ltd., [1989] 2 S.C.R. 983, at p. 1005, per La Forest J.).  That is especially so because unwarranted judicial intervention risks short-circuiting negotiations between management and labour by delivering through judicial fiat what the legislature has said should be subject to collective bargaining between the parties.

[69]                          The Court of Appeal was of the view that “[a]s matter of policy, this Court must decide whether an employer is under an obligation to demonstrate sufficient evidence of an alcohol problem in the workplace before adopting a policy requiring mandatory random alcohol testing” (para. 52 (emphasis added; emphasis in original deleted)).  We respectfully disagree.  That policy choice is one that the Legislative Assembly of New Brunswick has delegated to the collective bargaining process and, where disputes emerge, to labour arbitrators, whose decisions the legislature has shielded with a privative clause.  See Industrial Relations Act, R.S.N.B. 1973, c. I-4, ss. 32(1), 55(1) and 77(1).[1] 

[70]                          Reasonable people may well differ about the wisdom of the legislative choice to entrust labour arbitrators — or courts, for that matter — with a policy-making function that potentially carries serious repercussions for public safety and the environment.  One leading arbitrator has suggested that the “sensitive treatment” of drug and alcohol testing policies by arbitral boards and human rights tribunals “has given a sufficiently fair and balanced protection to the interests of both employees and employers, so as to avoid the need for the more blunt and draconian alternative of legislative regulation” (C.N.R., at p. 366 (emphasis added)).  That may well be the case. 

[71]                          But the fact that the public interest — not merely that of employer and employee — is relevant in cases such as this one may counsel a reassessment of the legislative choice to delegate policy-making for drug and alcohol testing to the collective bargaining process and to labour arbitrators.  It is one thing for employers and employees to negotiate a balance as they see fit with respect to their own privacy and safety.  It is a different matter, however, to leave the public interest to the vicissitudes of the bargaining table.  Of course, it would be counterintuitive to suggest that employees do not care for their own safety or, indeed, the safety of their neighbours.  The point is simply that employees, employers, and the public may each strike the balance between privacy and safety differently.  And where disputes between employers and employees emerge, it is not immediately apparent to us why an adjudicative body that is expert in the resolution of private labour disputes, but not in weighing broader considerations concerning the safety and environmental interests of the public at large, is best positioned to serve as the guardian of the public interest.  Indeed, nothing in the relevant legislation even requires, let alone suggests, that labour arbitrators should assume this role.

[72]                          The New Brunswick legislature has within the scope of its legislative authority the power to take drug and alcohol testing outside the purview of the collective bargaining process, as some other legislative bodies have done in certain contexts.  See, e.g., Code of Federal Regulations, 49 C.F.R. Part 382 (United States); Rail Safety (Adoption of National Law) Regulation 2012, No. 662 (New South Wales); Railway Safety Act 2005 (Ireland).  Indeed, some experts have suggested there is an “overwhelming argument” in this country for “legislative direction and definition that would add consistency, uniformity of meaning, and predictability for all workplace stakeholders” (N. Keith and A. J. Wiggins, Alcohol and Drugs in the Canadian Workplace: An Employer’s Guide to the Law, Prevention and Management of Substance Abuse (2008), at p. 240).  That decision, however, is one for the New Brunswick legislature and not for this Court — no matter how strongly we might favour such a step.

[73]                          Nevertheless, under reasonableness review there is a difference between judicial abdication and judicial restraint.  We reiterate that “the domain reserved to arbitral discretion is by no means boundless” (Nor-Man, at para. 52).  To the extent a particular arbitral award is unreasonable — as we would hold the award here is — it remains liable to being set aside on judicial review. 

B.     The Role of Arbitral Consensus in Defining the “Range of Reasonable Outcomes”

[74]                          In recent years, this Court has emphasized that reasonableness is “a single standard that takes its colour from the context” (Canada (Citizenship and Immigration) v. Khosa, 2009 SCC 12, [2009] 1 S.C.R. 339, at para. 59; see also Alberta Teachers, at para. 47).  The factual and legal context in which a decision is made is critical to assessing its reasonableness for the simple reason that “[r]easonableness is not a quality that exists in isolation” (Paccar, at p. 1018, per Sopinka J.).  Rather, when a reviewing court brands a decision as “reasonable” or “unreasonable”, it is necessarily making a conclusion about the relationship between the ultimate decision and the facts and law that underlie it.  The context of a decision thus shapes the “range of possible, acceptable outcomes which are defensible in respect of the facts and law” (Dunsmuir, at para. 47) or, more simply, the “range of reasonable outcomes” (Khosa, at para. 4). 

[75]                          The context of this case is informed in no small part by the wealth of arbitral jurisprudence concerning the unilateral exercise of management rights arising under a collective agreement in the interests of workplace safety.  We will say more about the “balancing of interests” test that has emerged from that jurisprudence in a moment, but for now the salient point is that arbitral precedents in previous cases shape the contours of what qualifies as a reasonable decision in this case.  In that regard, we agree with our colleague, Abella J., who describes this “remarkably consistent arbitral jurisprudence” as “a valuable benchmark against which to assess the arbitration board’s decision in this case” (paras. 16 and 6).

[76]                          The arbitral cases themselves stress the importance of arbitral consensus in shaping subsequent awards.  For example, in Prestressed Systems Inc. and L.I.U.N.A., Loc. 625 (Roberts) (Re) (2005), 137 L.A.C. (4th) 193, Arbitrator Lynk spoke of a “‘common law’ of the unionized workplace” and observed:

While statutes and collective agreements form the foundation for the law of the unionized workplace in Ontario today, as well as providing the source for arbitral authority, any statement on the scope of labour arbitration law would be defici[en]t and incomplete without also including the interpretative function that arbitration awards play in building upon and adding to the law on workplace relations. When an arbitral rule or principle has emerged through industrial relations practice and become broadly accepted in a series of arbitration awards, then, even though the governing statute, the broader common law and the collective agreement may be silent on the matter, this principle at some point crystallizes and becomes part of the law of the unionized workplace. The duty of management to act fairly and reasonably, the estoppel doctrine, the KVP principle on company rules and the doctrine of the culminating incident, to name but only a few, have all become part of the legal regime of the workplace through the arbitral “common law”. [Emphasis added; pp. 206-7.]

[77]                          Thus no arbitral board is an island unto itself.  As it is with the common law, which matures with the benefit of experience acquired one case at a time, so it is with the arbitral jurisprudence.  Indeed, in this case, the arbitral board cited multiple prior arbitral awards for the proposition that Mr. Day had a right to privacy in his workplace (para. 19, citing Halifax (Regional Municipality) and N.S.U.P.E., Local 2 (Re) (2008), 171 L.A.C. (4th) 257 (Veniot), which referred to Prestressed Systems; Re Monarch Fine Foods Co. and Milk and Bread Drivers, Dairy Employees, Caterers and Allied Employees, Local 647 (1978), 20 L.A.C. (2d) 419 (M. Picher); Trimac Transportation Services — Bulk Systems and T.C.U. (Re) (1999), 88 L.A.C. (4th) 237 (Burkett)).

[78]                          Respect for prior arbitral decisions is not simply a nicety to be observed when convenient.  On the contrary, where arbitral consensus exists, it raises a presumption — for the parties, labour arbitrators, and the courts — that subsequent arbitral decisions will follow those precedents.  Consistent rules and decisions are fundamental to the rule of law.  As Professor Weiler, a leading authority in this area, observed in Re United Steelworkers and Triangle Conduit & Cable Canada (1968) Ltd. (1970), 21 L.A.C. 332:

This board is not bound by any strict rule of stare decisis to follow a decision of another board in a different bargaining relationship.  Yet the demand of predictability, objectivity, and impersonality in arbitration require that rules which are established in earlier cases be followed unless they can be fairly distinguished or unless they appear to be unreasonable. [Emphasis added; p. 344.]

See, also D. J. M. Brown and D. M. Beatty, Canadian Labour Arbitration (4th ed. (loose-leaf)), at topic 1:3200 (including discussion of the “Presumption Resulting From Arbitral Consensus”); R. M. Snyder, Collective Agreement Arbitration in Canada (4th ed. 2009), at p. 51 (identifying Professor Weiler’s view as “typical”). 

[79]                          Thus, while arbitrators are free to depart from relevant arbitral consensus and march to a different tune, it is incumbent on them to explain their basis for doing so.  As this Court has stressed, “reasonableness is concerned mostly with the existence of justification, transparency and intelligibility within the decision-making process” (Dunsmuir, at para. 47).  Because judges are not mind readers, without some explanation, whether implicit or explicit, for a board’s departure from the arbitral consensus, it is difficult to see how a “reviewing court [could] understand why the [board] made its decision” (Newfoundland and Labrador Nurses’ Union v. Newfoundland and Labrador (Treasury Board), 2011 SCC 62, [2011] 3 S.C.R. 708, at para. 16).  Reasonableness review includes the ability of courts to question for consistency where, in cases like this one, there is no apparent basis for implying a rationale for an inconsistency.

[80]                          In this case, as we will explain, the board departed from the legal test emerging from the arbitral consensus by elevating the threshold of evidence Irving was required to introduce in order to justify a policy of random alcohol testing.  The board, however, offered no explanation — whether implicit or explicit, reasonable or unreasonable — for doing so.  In the absence of any explanation whatsoever, we are unable to understand why the board thought it reasonable to do what it did.  In the circumstances of this case, its decision thus fell outside the range of reasonable outcomes defensible in respect of the facts and law. 

C.     The Arbitral Jurisprudence Reveals a Consensus on Assessing Workplace Drug and Alcohol Policies

[81]                          The arbitral jurisprudence does not recognize an unqualified right of employers to unilaterally impose workplace rules on their employees outside of the collective bargaining process.  Rather, the onus is on the employer to justify such rules based on compliance with standards first articulated in the seminal arbitral decision of Re Lumber & Sawmill Workers’ Union, Local 2537, and KVP Co. (1965), 16 L.A.C. 73 (Robinson).  The “KVP test” has six distinct elements, the primary one being that the rule must be reasonable.  In this case, the only question was the reasonableness of the rule (board’s reasons, at para. 30).  Before this Court, neither party challenges the applicability or reasonableness of the KVP test and we therefore accept it as establishing the guiding framework for analysis for the purposes of the present appeal. 

[82]                          The rather abstract concept of KVP reasonableness has been given shape in various contexts, including drug and alcohol testing policies, by a further “balancing of interests” test (see, e.g., Esso Petroleum, at para. 73; C.N.R., at pp. 367-69).  The test recognizes that an employee’s right to privacy is “a core workplace value, albeit one that is not absolute” (Trimac, at p. 260).  Accordingly, the test seeks to determine “the extent to which mandatory random drug [or alcohol] testing furthers the objective of a safe and productive workplace and a corresponding assessment of the extent to which it invades individual privacy” (Trimac, at p. 259).  Again, before this Court, neither party challenges the applicability or reasonableness of the balancing of interests test.  They do, however, have divergent understandings as to what it actually requires in the circumstances of this case.  Accordingly, in what follows, we review the relevant jurisprudence in some detail.

                    (1)      What Is the Appropriate Test in These Circumstances?

[83]                          A measure of precision is required when discussing the relevant arbitral jurisprudence in the area of drug and alcohol testing because there are different testing scenarios, with different tests applying depending on the rule an employer seeks to justify.  First, one must distinguish between testing for drugs from that for alcohol.  Second, one must distinguish reasonable cause or post-incident testing from random testing.  Taking both distinctions together, the matrix of possible options reveals four distinct testing scenarios: reasonable cause/post-incident drug testing, reasonable cause/post-incident alcohol testing, random drug testing, and random alcohol testing. 

[84]                          The Irving policy that spawned the grievance in this case addressed reasonable cause and post-incident testing for both drugs and alcohol, as follows:

Post-incident: Employees employed in Safety Sensitive Positions will be subject to post-incident tests for alcohol and drugs. After a work-related accident or other incident (an “Incident”) the decision to refer an Employee(s) for a test will be made by an on-site Supervisor investigating the Incident, in conjunction with a second person (a health professional, another Supervisor, or Company Security) wherever practicable.

. . .

Reasonable Cause: Employees employed in Safety Sensitive Positions will be subject to reasonable cause tests for alcohol and drugs. Where the Company determines there is reasonable cause to suspect alcohol or other drug use or possession in violation of this policy, testing will be performed. The decision to test shall be made by a Supervisor, in conjunction with a second person (e.g. another Supervisor or Company Security) wherever practicable. The decision will be based on specific, personal and documented observations resulting from, but not limited to:

 

-          observed use or evidence of use of a substance (e.g. smell of alcohol);

-          erratic or atypical behaviour of the Employee;

-          changes in the physical appearance of the Employee;

-          changes in behaviour of the Employee; or

-          changes in the speech patterns of the Employee. [Emphasis added; A.R., vol. II, at pp. 75-76.]

[85]                          In contrast, the random portion of the policy was concerned only with alcohol testing and it is that portion — and only that portion — of the policy that is subject to the instant grievance:

Random Testing: Employees employed in Safety Sensitive Positions will be subjected to unannounced random tests for alcohol. In addition, applicants to a Safety Sensitive Position must pass an alcohol and/or drug test before entry to the position or re-entry to the position where they have participated in a treatment program. [Emphasis added; A.R., vol. II, at p. 76.]

It bears noting the language for each of these three provisions is similar to those used in other drug and alcohol testing policies.

[86]                          Turning to the first distinction mentioned above, between drug and alcohol testing, the cases recognize that testing for alcohol “stand[s] on a different footing” from testing for drugs (Entrop, at para. 106).[2]  For example, alcohol tests are usually conducted with a breathalyser, which provides an immediate result concerning present alcohol impairment in a minimally invasive manner.  Though drug testing technology has advanced, it does not provide an immediate detection of drug impairment, which may affect the determination of whether it is reasonably necessary to ensure safety in the workplace (Imperial Oil Ltd. v. Communications, Energy & Paperworkers Union of Canada, Local 900, 2009 ONCA 420, 96 O.R. (3d) 668, at para. 61). 

[87]                          In light of the distinction found in the arbitral cases between drug and alcohol testing, we do not view the decision in Imperial Oil Ltd. and C.E.P., Loc. 900 (Re) (2006), 157 L.A.C. (4th) 225 (M. Picher) (“Nanticoke”), as conclusively shaping the range of reasonable outcomes in this case (board’s reasons, at paras. 30-33).  Nanticoke decided only the issue of a random drug testing policy and must be seen in that context.  Both the reasons of the arbitral board and the Court of Appeal for Ontario in Nanticoke make that abundantly clear.  See Nanticoke, at paras. 112-13:  “The Company reasons that the oral fluid drug test which it now administers is fully analogous to the breathalyser test . . . [but the board concludes that] the buccal swab [drug] test does not equate to the breathalyser” (emphasis added); Imperial Oil, at para. 47: “. . . both sides placed considerable reliance on . . . an established body of arbitral case law that directly concerned the subject matter of the Board hearing — random drug testing in the workplace” (emphasis added).  Indeed, Arbitrator Picher’s comments in Nanticoke are properly read as being confined to random drug testing:

It may well be that the balancing of interests approach, which we favour, would allow for general random, unannounced drug testing in some extreme circumstances. If, for example, an employer could marshal evidence which compellingly demonstrates an out-of-control drug culture taking hold in a safety sensitive workplace, such a measure might well be shown to be necessary for a time to ensure workplace safety. [Emphasis added; para. 127.]

[88]                          More problematically, the Nanticoke arbitral reasons, as the board in this very case noted, are self-contradictory and, further still, are out of step with the more recent arbitral jurisprudence to the extent they speak to random alcohol testing.  The board here provided a reasonable — indeed, a convincing — explanation for declining to follow Nanticoke to the extent it discussed random alcohol testing.  See, e.g., board’s reasons, at para. 55: “. . . I cannot accept [Nanticoke] as correct [because] I believe it to be mistaken in principle”; at para. 57: “I also have to note that Nanticoke itself is not thoroughgoing in following its own model”; at para. 61: “. . . I would question [the Nanticoke model’s] value as an explanatory mechanism with respect to this board’s issue”; and, at para. 69: “. . . I would not accept the first element of the [Nanticoke] model [that unannounced random testing is prohibited, save as part of an agreed rehabilitative program] when it comes to random alcohol tests” (emphasis added).  Thus, like the board in this case, we think that Nanticoke is of limited utility.

[89]                          Turning then to the second distinction mentioned above, between reasonable cause and post-incident testing, on the one hand, and random testing, on the other, the arbitral jurisprudence recognizes “significant differences between the principles” applicable to these two types of testing, (Fording Coal Ltd. v. United Steelworkers of America, Local 7884, [2002] B.C.C.A.A.A. No. 9 (QL) (Hope), at para. 36).  In the context of certain safety-sensitive positions, for example, arbitrators have required some evidence of drug or alcohol use in the workplace in order to justify a random testing policy, but have not required such evidence where testing was based on reasonable cause or a workplace incident.  See, e.g., Continental Lime Ltd. and B.B.F., Loc. D575 (Re) (2002), 105 L.A.C. (4th) 263 (Freedman): “Evidence of a problem may be necessary to support a policy of random testing, but I do not think it is necessary to support a [reasonable cause] policy such as here” (p. 284); Weyerhaeuser Co. and I.W.A. (Re) (2004), 127 L.A.C. (4th) 73 (Taylor): “. . . where safety is clearly a justifiable concern . . . [an employer] does not have to prove the existence of a drug and alcohol problem as a precondition to the introduction [of a reasonable cause or post-incident testing policy]” (p. 108);  GTAA: “While a different approach has been adopted in cases involving random testing, . . . in respect of reasonable cause and post-accident/incident testing, . . . an employer need not demonstrate a history of substance abuse in the workplace . . .” (para. 221).

[90]                          In sum, care must be taken to identify the appropriate test in the arbitral jurisprudence.  The cases illustrate that there is a difference between how arbitral boards have assessed a random testing policy and one based on reasonable cause, and a difference between testing for drugs and testing for alcohol.  We thus avoid reliance on cases not directly applicable in the context of a challenge to a random alcohol testing policy.  Fording Coal “[did] not involve random testing” at all (para. 40), and the same is true for the decisions in Rio Tinto and ADM Agri-Industries Ltd. v. National Automobile, Aerospace, Transportation and General Workers’ Union of Canada (CAW-Canada), Local 195 (Substance Abuse Policy Grievance), [2004] C.L.A.D. No. 610 (QL) (Springate).  The grievance in Trimac concerned only “mandatory random drug testing”, not random alcohol testing (p. 276 (emphasis added)).  These cases must be put in their proper context — and that is not the context of this case.

                    (2)     The Arbitral Consensus on Random Alcohol Testing

[91]                          Having established the importance of identifying the relevant arbitral consensus, we turn now to an examination of the cases concerning the legal test for random alcohol testing policies.  While the general principles emerging from the broader arbitral jurisprudence may assist in situating random alcohol testing in the wider context, it is these cases that shape the contours of what is a reasonable outcome in the context of this case.

[92]                          We are aware of two arbitral decisions, Communications, Energy and Paperworkers Union, Local 777 v. Imperial Oil Ltd., May 27, 2000, unreported (Christian) (“Strathcona”), and GTAA, in which an arbitrator found a random alcohol testing policy to satisfy the demands of KVP reasonableness.  In both cases, the arbitrators accepted that the policies applied to what were legitimately safety-sensitive positions.  In both cases, the employer used breathalyser tests, with a 0.04 percent blood alcohol concentration level, which the arbitrators accepted as the least intrusive means of identifying present intoxication.  The key question for our purposes is the threshold of evidence that the employer was required to introduce in order to meet its burden to demonstrate KVP reasonableness.

[93]                          Before reviewing the two random alcohol testing awards, however, a preliminary observation is warranted.  The standard applied in both Strathcona and GTAA is the progeny of earlier jurisprudence in Provincial-American Truck Transporters and Teamsters Union, Loc. 880, Re (1991), 18 L.A.C. (4th) 412 (Brent) (“Truck Transporters”), that drew a distinction between testing where there was “reason to demand a test” (which we would understand today as including reasonable cause or post-incident testing) and otherwise “mandatory universal testing” (which we would understand as including random testing) (p. 425).  In Truck Transporters, the company sought to justify a policy of mandatory drug and alcohol tests of its drivers.  The board was of the view that

[i]f mandatory universal testing is to be justified, absent a specific term allowing it, then there should at least be evidence of a drug and/or alcohol problem in the workplace which cannot be combated in some less invasive way. [Emphasis added; p. 425.]

[94]                          The first random alcohol testing case to adopt the Truck Transporters standard came three years later in Esso Petroleum.  There, Arbitrator McAlpine, borrowing the precise language of Truck Transporters, was of the view that the employer had to establish “evidence of a drug and/or alcohol problem in the workplace” as part of what was then emerging as the balancing of interests test (para. 104 (emphasis added)).  On the facts presented to the board, Arbitrator McAlpine noted that “there have been no reported incidents at [the facility] involving drugs and alcohol”, “no report of employees reporting for work impaired”, “no safety violations involving drugs or alcohol”, and “no accidents where drug[s] and alcohol were suspected” (para. 151).  In short, the company had provided no evidence whatsoever, and its random alcohol testing policy was thus unreasonable.

[95]                          Turning then to the cases where a random alcohol testing policy was upheld as reasonable, in Strathcona, the earlier of the two decisions, the testing policy concerned employees at an Imperial Oil refining facility, much like Esso Petroleum.  After a thorough review of the arbitral cases, including quotation from the Truck Transporters language mentioned above (at p. 69), the board concluded that the appropriate evidentiary threshold was as follows:

The question is whether there is evidence upon which the Employer could rationally conclude that alcohol and drug use might cause catastrophic damage at the Strathcona Refinery. [Emphasis added; p. 73.]

The board relied on two sets of evidence.  First, it looked to a national employee survey conducted by an independent party on behalf of the employer.  Actual workplace accidents in which an employee’s own use of “alcohol, medications or street drugs” was thought to be a contributing factor were reported by 0.5 percent of employees and a further 1.7 percent reported it being a factor in “near misses”, in each case in the 12 months prior to the survey (p. 56).  The board inferred alcohol use at the Strathcona facility based on the representativeness of the survey (p. 59).  That data, in the board’s view, “provide[d] a rational and sufficient foundation” of “a problem or potential problem which justifies the Employer in implementing a [random alcohol testing] process” (p. 73).  Second, the board noted that there was evidence of one worker who was at the workplace while intoxicated.  The board observed that such a “real case” might even be said to “provid[e] the best evidence of a problem at the Strathcona Refinery, and the need for, and effectiveness of, the Policy” (p. 74 (emphasis added)).

[96]                          In GTAA, the more recent case, the random alcohol testing policy concerned employees involved with the ground operations at Pearson International Airport in Toronto.  After surveying prior arbitral decisions, including Truck Transporters (at para. 251) and Esso Petroleum (at para. 252), Arbitrator Devlin concluded that, “in cases involving random testing, Arbitrators have required evidence of a drug and/or alcohol problem in the workplace” (para. 254 (emphasis added)), thus echoing the precise language of the earlier decisions.  He heard testimony that employees “consumed alcohol at work or during meal breaks”, that management had “frequently found empty beer or alcohol bottles in vehicles or in the garbage”, and that “on a few occasions, employees took beer into snow-clearing equipment” (para. 256).  The arbitrator ultimately concluded the testimony of alcohol use represented evidence of “a far more pervasive problem” (para. 262).

[97]                          Taking these cases together, what emerges is an arbitral consensus that an employer must demonstrate evidence of an alcohol problem in the workplace in order to justify a random alcohol testing policy.  That is the evidentiary threshold accepted by arbitrators who have upheld such policies (Strathcona, GTAA) and those who have struck them down (Esso Petroleum).  Thus, barring some explanation, whether implicit or explicit, for its basis for departing from it, that is the evidentiary threshold the Union, management, and this Court should be able to presume the board in this case applied.  But as we explain next, that is not what the board did.

D.     The Arbitral Board’s Departure From the Arbitral Consensus

[98]                          The board in this case was well aware of the relevant arbitral jurisprudence.  As we noted earlier, it reviewed the decision in Nanticoke and reasoned — compellingly in our view — that it was not helpful in the present case (para. 61).  The board then proceeded to review both Strathcona and GTAA, including quoting from the same passages we cite above, and concluded:

[These cases] demonstrate a fact finding process centred on risk in the particular workplace and the means adopted to address it, and, balancing the interests involved, move to a conclusion. Where the evidence supports the need for such a policy and the balancing of interests warrants it, the employer’s policy prevails; where it doesn’t, the employee’s right to privacy carries the day. [Emphasis added; para. 69.]

The board also noted, referring to the earlier cases:

There are numerous statements in the cases to the effect that an employer, to be successful, must lead evidence of a problem existing in its own workplace, but as a general statement I think this is somewhat overbroad. Evidence of risk may be available from the nature of the industry itself. The cases recognize a lighter burden of justification on an employer engaged in the operation of an ultra-hazardous endeavour. [Emphasis added; para. 75.]

[99]                          As a preliminary matter, we note the board’s reference to an “ultra-hazardous endeavour” comes from remarks in Arbitrator Picher’s decision in C.N.R., a leading case that concerned reasonable cause and post-incident testing (see C.N.R., at pp. 377-78).  The board here appeared willing to depart from the arbitral consensus such that the lower evidentiary requirement could be applied to random testing cases, but ultimately concluded that the Irving mill did not fit into that category of facilities recognized in C.N.R. (para. 103).  According to the jurisprudence, Irving thus had to lead some evidence of an alcohol problem at the mill in order to establish the reasonableness of its policy. 

[100]                      We would not impugn the board’s finding as to the level of dangerousness of the mill and the requirement flowing from that finding that Irving had to lead evidence of alcohol use.  While the respondent made much of the board’s conclusion in this regard, in our view, nothing in this case turns on whether the Irving mill was “ultra-hazardous” (whatever that may mean) or not because, as we will explain, the board departed from the consensus evidentiary standard after it concluded that some evidence was required.  That unreasonable finding is sufficient to decide this case.

[101]                      In any event, as we explained earlier in these reasons, we know of no case in which an arbitral board has approved of random alcohol testing absent some evidence of alcohol use (see, e.g., Strathcona and GTAA).  That fact, while not dispositive, at minimum shapes the range of reasonable outcomes in this case.  Of course, an arbitral board in some future case may think it reasonable to adopt the principles in C.N.R. in order to conclude that no evidence is required to justify random alcohol testing in the context of a particular dangerous environment.  We, however, have no occasion to go that far in this case.  Our focus is not the rule emerging from the arbitral jurisprudence, which we accept as reasonable for present purposes, but rather an outcome that unreasonably departs from it.

[102]                      Turning then to evidence introduced in this case, Irving relied on the testimony of Leo Moorehouse, who served as industrial relations superintendent at the mill from 1987 to 2008, and an exhibit provided by the company listing specific incidents of employee intoxication or alcohol consumption at the mill.  The exhibit listed seven instances, dating from April 29, 1991, through January 11, 2006 — the last, we hasten to add, being a mere three weeks before the policy’s implementation — where employees identified by name were variously “under the influence of alcohol, consuming and in possession of alcohol on company premises”, “under the influence of alcohol while at work”, and “consuming alcohol on company premises” (A.R., vol. II, at p. 121).  Mr. Moorehouse testified that the exhibit was “by no means an exhaustive list” and that he had “witnessed [alcohol use at the mill] on a lot of occasions” (board’s reasons, at para. 107).

[103]                      The board found Mr. Moorehouse’s testimony “not persuasive”, but it did think that the exhibit was “more helpful” in assessing the evidence of alcohol use at the mill (para. 108).  The board then concluded:

This evidence is not to be dismissed, and I do not do so, but it cannot be said to be indicative of a significant problem with alcohol-related impaired performance at the plant. As well, such as it is, it is not tied in with what the actual experience has been in this plant, with accident, injury and near-miss history, and with what group or groups of employees. I therefore have no idea of what the elements of any such record are; still less whether any lapses have been causally linked to the abuse of alcohol. [Emphasis added; para. 109.]

We note that though the phrase “significant problem with alcohol-related impaired performance” was used in the board’s final reasons, the draft reasons employ the phrase “serious problem with alcohol abuse” in the very same paragraph (A.R., vol. I, at p. 68 (emphasis added)).  The language in the paragraph is otherwise identical between the draft and final reasons, and we do not know what led to the revision.  Both the draft and final reasons are included in the appellant’s record before this Court and both versions are signed and dated.

[104]                      Two issues become immediately apparent from the board’s conclusion as to the evidence.  First, the standard it applied was one of “significant problem” (based on the final version of the reasons) or a “serious problem” (based on the draft).  In either case, as we have just discussed, that is not the standard reflecting the arbitral consensus for justification of a random alcohol testing policy.  In none of the cases of which we are aware, whether those that upheld such policies or those that set them aside, have we seen language requiring evidence of a “significant” or “serious” problem.  Rather, the standard has been that of evidence of a problem.  The difference between the two approaches is obviously a marked one and it cannot be ignored.

[105]                      Second, the board required that the evidence of alcohol use be “tied” or “causally linked” to “accident, injury and near-miss history” at the plant.  Again, there is no support for such a requirement in the arbitral jurisprudence.  While it is true that the board in Strathcona relied on survey data that indicated alcohol use “was thought to be a contributing factor” in workplace incidents (p. 56), there is no support in that case for the conclusion that the employer must establish cause and effect between alcohol use and a workplace incident.  Indeed, the reasons in Strathcona say exactly the opposite:

. . . an Employer does not have to wait for “a serious incident of loss, damage, injury or death” to occur before taking action. Likewise, given the inherent risks at the Refinery, the Employer is not bound to bide its time, patiently building a case in favour of random testing, one incident after another. [Emphasis added; p. 73.]

In any case, to require that an employer tie alcohol use to actual incidents at the mill, as the board in this case did, is not only unreasonable, it is patently absurd.  The arbitral cases recognize that evidence of alcohol use at an inherently dangerous facility such as the Irving mill — where the impact of a catastrophic failure could extend well beyond the safety of workers — is “a problem” enough.

[106]                      Taking these two points together, it is beyond question that the board in this case applied an evidentiary standard unknown to the arbitral jurisprudence.  And it is the application of that higher standard which, in our view, dictated the board’s conclusion in this case.  As such, this is not a matter of quibbling with a few arguable statements or intermediate findings in the board’s reasons — the higher evidentiary standard is the basis for the board’s ultimate conclusion.

[107]                      It is clear from the board’s reasons that it accepted that “the Irving plant is one in which great care must be taken with safe work practices” and that “the mill in normal operation is a dangerous work environment” (paras. 98 and 102).  There was also no dispute that Mr. Day’s job was properly characterized as safety-sensitive, and his job role was “noted [in the evidence] as showing one of the highest risks in the plant” (para. 90).  Furthermore, the board also accepted that the use of a breathalyser was “minimally intrusive”, “among the alternatives, ha[d] the lowest impact on the privacy right”, and was “a reasonable choice for this employer” (paras. 117-18).  Those conclusions are not challenged by either party on appeal. 

[108]                      However, when it came to balancing the interests as part of KVP reasonableness, the board concluded that “[Irving’s] scheme [got] into heavier weather” (para. 119) because the company failed to demonstrate a “significant degree of incremental safety risk attributable to employee alcohol use” at the plant (para. 120).  In other words, the board concluded that the company’s policy was unreasonable because the evidence of alcohol use that Irving introduced fell short of the higher standard the board applied.  The evidence, in short, was the decisive factor. 

[109]                      To be clear, and as we observed earlier, it was open for the board in this case to depart from the arbitral consensus in reaching its conclusion, provided it had a reasonable basis for doing so.  In so departing, it was thus incumbent upon the board to provide some explanation for its reasoning.  Here, the board provided no explanation whatsoever — whether implicit or explicit, reasonable or unreasonable — for the new evidentiary standard that it applied.

[110]                      The board’s departure from the arbitral consensus resulted in a decision that fell outside the range of reasonable outcomes defensible in the facts and law.  In the absence of a reasonable explanation for its novel test, the board must be taken as having misapplied the existing test, which in the circumstances of this case rendered its decision unreasonable.  See Alberta (Education) v. Canadian Copyright Licensing Agency (Access Copyright), 2012 SCC 37, [2012] 2 S.C.R. 345: “Because the Board’s finding of unfairness was based on . . .  a misapplication of the CCH factors, its outcome was rendered unreasonable” (para. 37, per Abella J.). 

[111]                      Whether Irving in fact introduced evidence sufficient to meet the evidentiary standard reflected in the arbitral consensus is not the issue before us.  Our concern in this appeal is whether the appropriate standard was applied by the board.  Though we take no position on the sufficiency of the evidence brought to bear by Irving, we do note that the documented history of alcohol use by individual employees would appear to be similar to the evidence accepted as sufficient in other cases as reasonably justifying similar policies in similar contexts.  The actual determination of that matter, however, is the responsibility of a future arbitral board, should the policy be challenged again.

E.      Further Unreasonableness in the Board’s Decision

[112]                      We turn finally to a further aspect of the board’s reasoning that undermined the reasonableness of its ultimate conclusion.  The board drew an adverse inference as to the reasonableness of the random alcohol testing policy based on the fact that only 10 percent of mill employees in safety-sensitive positions were tested in any given year.  Irving’s choice of that figure was characterized as “indirect evidence from which the inference can be drawn that plant management does not regard the incremental safety risk posed by alcohol in this plant as being high” (para. 110).  In other words, it was used to support the board’s conclusion that there was insufficient evidence of an alcohol problem at the mill.

[113]                      The board’s inference was unreasonable for three reasons.  First, as the board itself recognized, “[b]ecause the random alcohol testing policy is based upon deterrence, the percentage chosen for testing represents its estimate of what is required to achieve that goal” (para. 110 (emphasis added)).  In other words, the figure should be understood not as an indication of the level of the problem, but what it will take to solve the problem.  Indeed, the value of a random alcohol testing program comes not from what it detects, but from what it deters.  Academic literature — not to mention common sense — teaches that even low testing percentages can be highly effective in deterring the relevant conduct.  See, e.g., J. I. Borack, “Costs and Benefits of Alternative Drug Testing Programs”, U.S. Navy Personnel Research and Development Center (March 1998) (explaining that a 20 percent random test rate “achiev[ed] significant benefits” in deterring drug use among service members while tripling the test rate to 58 percent would provide “modest increases [in deterrence] . . . but at significantly higher cost” (p. 15)).

[114]                      Second, the board’s reasoning would perversely incentivize employers to test a higher percentage of their employees in order to establish the reasonableness of their workplace drug and alcohol testing policies.  Manifestly, testing a greater number of employees leads to a greater intrusion into the privacy of those employees.  Indeed, to the extent a testing threshold were higher than reasonably necessary to achieve the desired deterrent effect, it may well fail to satisfy the minimal impairment analysis arbitrators have conducted as part of the balancing of interests. 

[115]                      Third and finally, the threshold set by Irving is hardly out of the mainstream for random alcohol testing.  For example, the U.S. Department of Transportation, a leading policy-maker in this area, sets a 10 percent threshold for employers subject to its regulations, as the board in this case recognized (para. 111; see also U.S. Department of Transportation, “Current Random Testing Rates” (online)). 

III.    Conclusion

[116]                      The decision of the board in this case cannot be said to fall within the range of reasonable outcomes defensible in respect of the facts and law.  Though purporting to apply the accepted test from the arbitral jurisprudence, the board unreasonably departed from it.  And in applying its own novel test, the board compounded the unreasonableness of its finding by reasoning in a manner that was again unreasonable. 

[117]                      To be sure, the decisions of labour arbitration boards command judicial deference.  But, in our respectful view, “deference ends where unreasonableness begins” (Khosa, at para. 160, per Fish J.). 

[118]                      For these reasons, we respectfully dissent.

                    Appeal allowed with costs throughout, McLachlin C.J. and Rothstein and Moldaver JJ. dissenting.

                    Solicitors for the appellant:  Pink, Larkin, Fredericton.

                    Solicitors for the respondent:  McCarthy Tétrault, Toronto.

                    Solicitors for the interveners the Construction Owners Association of Alberta, Construction Labour Relations — an Alberta Association and Enform:  Dentons Canada, Calgary.

                    Solicitors for the interveners the Canadian National Railway Company, the Canadian Pacific Railway Company and Via Rail Canada Inc.:  Heenan Blaikie, Montréal; Canadian National Railway Company, Montréal.

                    Solicitors for the interveners the Alberta Federation of Labour and the Communications, Energy and Paperworkers Union of Canada, Local 707:  Chivers Carpenter, Edmonton.

                    Solicitors for the intervener the Canadian Civil Liberties Association:  Ursel Phillips Fellows Hopkinson, Toronto.

                    Solicitors for the intervener the Alliance of Manufacturers & Exporters of Canada, carrying on business as Canadian Manufacturers & Exporters:  Gowling Lafleur Henderson, Toronto.

                    Solicitors for the interveners the Canadian Mining Association, the Mining Association of British Columbia, the Mining Association of Manitoba Inc., the Québec Mining Association, the Ontario Mining Association and the Saskatchewan Mining Association:  Heenan Blaikie, Vancouver.

                    Solicitors for the intervener the Power Workers’ Union:  Paliare Roland Rosenberg Rothstein, Toronto; Power Workers’ Union, Toronto.