Solidarity obo Van Zyl v KPMG Services (Pty) Ltd and Others (LC) (unreported case no JR960/12, 10-10-2013) (Fourie AJ)

By Moksha Naidoo

The question posed by the court in this application was to what extent an arbitrator’s error in accessing the credibility of witnesses when faced with two mutually destructive versions, renders such award reviewable.

The applicant was dismissed for allegedly uttering a racial slur while in the presence of three colleagues. One colleague reported the incident to the employer, KPMG. At an internal inquiry, the colleague testified to what she had heard and, despite the applicant denying this allegation, he was found guilty and dismissed.

In arbitration proceedings, under the auspices of the Commission for Conciliation, Mediation and Arbitration (CCMA), KPMG led the evidence of the employee who laid the complaint. Not only did the applicant continue to deny such wrongdoing, he further led, as witnesses, the evidence of the remaining two colleagues who were present when the incident allegedly occurred. Both colleagues denied having heard the applicant make these remarks.

Faced with these mutually destructive versions the arbitrator accepted KPMG’s version over that of the applicant’s. The arbitrator based his findings on the sole ground that no reason could be advanced as to why KPMG’s witness would fabricate her version, especially in light of the fact that there was no ‘bad blood’ between the applicant and KPMG’s witness.

At the core of this application was the fact that the arbitrator failed to assess the credibility of any of the witnesses when arriving at his decision.

On review, the court referred to the judgment of Stellenbosch Farmers’ Winery Group Ltd and Another v Martell Et Cie and Others (2003) 1 SA (11) SCA wherein the Supreme Court of Appeal laid out the accepted test applicable to both a trail court and an arbitrator when faced with a factual dispute. According to the judgment (at para 5) the court had to come to a conclusion on the disputed issues by making findings on –

• the credibility of the various factual witnesses;
• their reliability; and
• the probabilities.

The court’s finding on the credibility of a particular witness will depend on its impression about the veracity of the witness. This finding will, in turn, depend on a variety of subsidiary factors, such as –

• the witness’ candour and demeanour in the witness-box;
• his or her bias, latent and blatant;
• internal contradictions in his or her evidence;
• external contradictions with what was pleaded or put on his or her behalf, or with established fact or with his or her own extra-curial statements or actions;
• the probability or improbability of particular aspects of his or her version; and
• the calibre and cogency of his or her performance compared to that of other witnesses testifying about the same incident or events.

A witness’ reliability will depend, apart from some of the factors above, on –

• the opportunities he or she had to experience or observe the event in question; and
• the quality, integrity and independence of his or her recall thereof.

Finally, an analysis and evaluation of the probabilities and improbabilities of each party’s version on each of the disputed issues are necessary components in coming to a conclusion. In the light of its assessment of all of the above factors the court will then, as a final step, determine whether the party burdened with the onus of proof has succeeded in discharging it.

While acknowledging the above approach to be sound in law, the court pointed out that this approach needs to be exercised with some caution in order to maintain the distinction between reviews and appeals. To apply this onerous test, applicable when appealing credibility findings of a trial court, to review proceedings under the Labour Relations Act 66 of 1995 (LRA), could blur the distinction between appeals and reviews, especially in light of the recent decision in Herholdt v Nedbank Ltd (Cosatu as Amicus Curiae) 2013 (6) SA 224 (SCA), wherein the SCA, in setting out the proper approach to reviews under the LRA held:

‘In summary, the position regarding the review of CCMA awards is this: A review of a CCMA award is permissible if the defect in the proceedings falls within one of the grounds in s 145(2)(a) of the LRA. For a defect in the conduct of the proceedings to amount to a gross irregularity as contemplated by s 145(2)(a)(ii), the arbitrator must have misconceived the nature of the inquiry or arrived at an unreasonable result. A result will only be unreasonable if it is one that a reasonable arbitrator could not reach on all the material that was before the arbitrator. Material errors of fact, as well as the weight and relevance to be attached to particular facts, are not in and of themselves sufficient for an award to be set aside, but are only of any consequence if their effect is to render the outcome unreasonable’ (at para 25).

Against this backdrop, the court held that it is not axiomatic that an award wherein the arbitrator failed to properly apply the aforementioned test was reviewable for this reason alone.

On this point, Fourie AJ, at para 17 held: ‘While arbitrators should always aspire to meet the exacting standard set by the Supreme Court of Appeal in Stellenbosch Farmers’ Winery for the proper assessment of conflicting versions by a finder of fact, an arbitration award that does not live up to this standard will not automatically be subject to review. Arbitrators are empowered to deal with the dispute with a minimum of legal formalities, their decisions are immune from appeal, and the legislature has set a high bar for reviewing arbitration awards. Errors committed by an arbitrator in the assessment thereof will not necessarily vitiate an award.’

In adopting this approach to the merits before it the court found that, by failing to assess the credibility of any of the witnesses, the arbitrator in casu fell short of the standard aspired to in the Stellenbosch case. However, this alone did not render the award reviewable without first considering whether his decision fell within the band of reasonableness. Further to this it was not necessary for the arbitrator to have found the applicant and his or her witnesses unreliable for him or her to find their version improbable (see Transnet Ltd v Gouws and Others (LC) (unreported case no JR206/09, 25-4-2012), at paras 11 to 20).

Factors, as recorded in the proceedings that supported a finding that it was improbable for KPMG’s witness to have fabricated her version were:

• The complainant did not initially mention the applicant’s name in her complaint and intended for KPMG to send a general instruction for employees to divest from such behaviour.
• KPMG disciplined the applicant after an investigation and not at the behest of the complainant.
• The complainant was reluctant to participate in both the inquiry and at arbitration and did not intend for the applicant to be dismissed.

In light of the above, the court held that in finding KPMG’s version more probable, the arbitrator’s decision was not one that a reasonable person could not come to given the evidence before him or her.

The application was dismissed with no order as to costs.

Moksha Naidoo BA (Wits) LLB (UKZN) is an advocate at the Johannesburg Bar.

This article was first published in De Rebus in 2013 (Dec) DR 46.