In its December 2021 decision in Aranda Textile Mills (Pty) Ltd v. The Competition Commission of South Africa, the Competition Appeal Court of South Africa overturned a Competition Tribunal ruling after finding a lack of evidence of anti-competitive practices. The case dealt with a bidding process involving a sole manufacturer who provided varying prices to different bidders who offered its product, while also directly submitting its own bid. That manufacturer, along with one of the other bidders, was charged with Competition Act offences by South Africa’s Competition Bureau and was found guilty of bid-rigging by the Competition Tribunal.
However, on appeal, the Court ruled that it was not automatically anti-competitive for a manufacturer to directly submit a bid when its resellers were also bidding on that same solicitation. Further, the Court ruled that in the absence of direct evidence, expert evidence was required to prove the elements of the collusion offence.
As the Court summarized, the bidding rules required bidders to “declare that there was no consultation, communication, agreement, or arrangement with any competitor regarding prices or formulas to calculate prices with the aim of winning or not winning the bid”. Further, the Court noted that the government’s internal audit committee found that the defendant had “tendered with the intention not to win the bid”, which put in it in conflict with the required non-collusion attestations.
However, the Court found no direct evidence that the parties had colluded prior to submitting their bids and that the Tribunal’s ruling relied on inferential reasoning to arrive at a guilty verdict. The Court noted the evidentiary challenges associated with bid-ridding prosecutions given their typical reliance on circumstantial evidence:
In this case, the direct evidence of Aranda and Mzansi was largely ignored by the Tribunal save that it cherry picked bits of their direct evidence which it adopted to support its mosaic hypothesis. A cautious and fact sensitive approach must be adopted when relying wholly on circumstantial evidence in order to conclude that there has been a contravention. This approach may also be warranted in the context that we have here, where the Commission was unable to lead direct evidence of the contravening conduct.
In cartel cases, it may well be difficult to find direct evidence which means a combination of direct and circumstantial may be necessary or only circumstantial evidence. As stated, often cartels are clever and avoid traces of evidence thereby concealing their conduct. Arrangements are made on a nod and nothing is noted on paper; leaving the court to rely on the best evidence which may be circumstantial.
While the Court found that “communication would of necessity have taken place” between the manufacturer and distributor, that communication did not necessarily prove price-fixing or collusion. The Court cautioned that the evidence should be considered as a whole and not in a vacuum:
Conduct evidence is evidence that competitors behaved consistently with the existence of the alleged cartel agreement. Conduct evidence will be most persuasive if it cannot be explained by ordinary market forces or competitive business behaviour. A judge should consider whether particular behaviour would have occurred in the absence of a cartel, having regard to unilateral commercial and economic interests of the competitors. Conduct evidence can include evidence of bidding patterns, information exchanges between competitors, abnormally high sustainable profits and past violations of competition laws. It could also include structural evidence which is evidence that explains why certain structural features make a particular market more susceptible to cartel conduct. In fact, the cross examination of Mrs Paruk on profitability, showed that Mzansi’s profits were very modest.
Evidence has to be assessed holistically and not in a vacuum. No one piece of circumstantial evidence would provide a sufficient basis on its own to infer an agreement or collusion. Where a piece of circumstantial evidence may be capable of supporting a number of inferences, it has to be the most plausible. It must be assessed contextually. The reasoning should not be aimed at using the inferences to answer a question which the adjudicator has an inclination for.
The process of inferential reasoning calls for an evaluation of all the evidence and not merely selected parts. The inference that is sought to be drawn must be consistent with all the proven facts. If it is not, then the inference cannot be drawn or it must be the ‘more natural, or plausible, conclusion from amongst several conceivable ones when measured against the probabilities.
The Court also noted that no expert evidence was introduced by the Commission to support the Tribunal’s ruling:
Whilst the existence of the alleged agreement or collusion may be inferred from circumstantial communication and conduct evidence, proving an agreement or concerted practice with indirect or circumstantial evidence remains problematic. Of the two, communication evidence is considered to be the more important as economic evidence is invariably ambiguous unless supported by expert evidence. As much is clear from this case, where the Tribunal found that the economic evidence regarding pricing was a compelling factor in support of a contravention, yet no expert evidence of the economic structure was presented by the Commission.
The Court ultimately concluded that the Commission over-reached in declaring that pre-bid communications between a manufacturer and distributor automatically constituted bid-rigging:
Commerce is robust in our modern era and many discussions would routinely take place between manufacturer and distributor and even where they are competitors such as in this tender. Caution must be exercised when drawing inferences from communication that take place simply because one is a manufacturer and the other a distributor of the product.
In light of the lack of evidence, that is of proven facts, the Tribunal had no basis to draw the inferences it did in concluding that Aranda and Mzansi directly or indirectly fixed a price for the tender or agreed any other trading condition or that there were collusive dealings on a balance of probabilities between Aranda and Mzansi.
The Court also found that the varying prices offered by the manufacturer to different distributors did not automatically constitute evidence of price-fixing:
The Tribunal in essence found that the 56% price difference which Aranda gave to Mzansi to be staggeringly high and that it put a tenderer like Vilankosi at a significant disadvantage. There was no evidence to suggest that despite Aranda giving Mzansi a good price it was left at a loss. Price discrimination cannot of itself indicate collusion. Insufficient economic evidence was led on this. There are too many variables involved in pricing. In the absence of a proper analysis this fact of alleged price discrimination found by the Tribunal is wrong.
The Court found that the Commission erred in finding that the vertical relationship between the manufacturer and distributor automatically became an illegal horizontal relationship when both parties bid on the same opportunity without proving how this was improper. The Court concluded that “there was no significant attempt to illustrate how the effects of this alleged relationship harms competition.” The Court ultimately ruled that “in the absence of proven facts and the exclusion of another plausible explanation, an infringement cannot be made by way of inferential reasoning.” The appeal was therefore upheld.
As this case illustrates, bid-rigging cases are subject to complex evidentiary considerations. While bidding practices may superficially appear to be a form of illegal collusion, the totality of relevant evidence should be considered in order to make a proper determination. Public institutions should therefore be careful about rushing to judgment to sanction suppliers for the appearance of bidding improprieties and should establish due process procedures to deal with these issues when they arise.
