---

 

 

IN THE COMPETITION APPEAL COURT OF SOUTH AFRICA

 

 

Case Number: 176/CAC/Jul19

 

In the matter between:

 

UNIPLATE GROUP ( PTY) LTD                Appellant

 

and

 

THE COMPETITION COMMISSION

OF SOUTH AFRICA                                 Respondent

 

 

Delivered: 25 February 2020

 

 

JUDGMENT

 

 

UNTERHALTER AJA

 

Introduction

 

[l]     The Appellant, Uniplate Group (Pty) Limited ("Uniplate"), supplies embossing machines and number plate blanks to embossers. Embossers manufacture number plates that the law requires to be fitted to all motor vehicles. Number plates are manufactured by placing a blank into an embossing machine and using a process to affix the required sequence of letters and numbers. Number plates are either acrylic or aluminium. An acrylic plate is made using an acrylic blank. An aluminium plate is made using an aluminium blank. The aluminium plate is made using one of two systems. In the type A system, the numbers are pressed into the blank and then painted. In the type B system, no painting is required. Acrylic blanks must be used in embossing machines that produce acrylic plates. So too, aluminium blanks must be used in embossing machines that produce aluminium plates.

 

[2]    New Number Plates Requisites CC ("NNPR") also supplies embossing machines and blanks to embossers. NNPR competes with Uniplate. Initially, Uniplate and NNPR supplied acrylic and type A embossing machines and blanks in competition with one another. In 2007, Uniplate developed and introduced embossing machines and blanks that utilised the type B system for the manufacture of aluminium plates. NNPR did not immediately follow suit with its own type B offering.

 

[3]    In 2010, the authorities in Gauteng required that number plates must be made of aluminium. This change led to an increase in the demand for aluminium plates in Gauteng. The type B system gained market share in 2010 - 2011. NNPR did not in this period provide a rival type B system.

 

[4]    Uniplate supplied its machines to embossers under exclusivity restraints. In essence, customers supplied with Uniplate embossing machines were required to purchase all their blanks (whether or not for Uniplate machines) and ancillaries (dies, inks, solvents, stickers) from Uniplate ("Uniplate's exclusivity restraints"). NNPR attempted to supply type B blanks to embossers who had acquired Uniplate type B embossing machines. Uniplate sought to interdict this conduct in the High Court. NNPR pleaded that Uniplate's exclusivity restraints contravened the Competition Act. And NNPR in June 2012 made a complaint to this effect to the Competition Commission ("the Commission"). In February 2013, ++ Plates, a customer of Uniplate, lodged a similar complaint with the Commission. ++  Plates complained that Uni plate imposed exclusivity restraints upon customers purchasing a Uniplate embossing machine.

 

[5]    The Commission investigated these complaints and referred the matter to the Competition Tribunal ("the Tribunal") in 2015. The Tribunal found that Uniplate's exclusive contracts foreclosed the market to Uniplate's competitors and raised barriers to entry. The Tribunal also found that customers were harmed by way of higher prices and reduced choice. As a result, the Tribunal declared that Uniplate had contravened s 8(d)(i) of the Competition Act ("the Act") and imposed a penalty ofR16 192 315.

 

[6]    Uniplate appeals these orders.

 

The principal issues

 

[7]    The Tribunal found that there was a primary market for the manufacture and supply of embossing machines and a secondary market for the manufacture and supply of number plate blanks. While acknowledging that there may be interdependence between these markets, the Tribunal found no warrant to treat these markets as one systems market, as Uniplate had contended.

 

[8]    Before this court, the Commission and Uniplate continued to differ as to how the market should be defined. But little turns on this difference, and for two reasons. First, as was the case before the Tribunal, there was agreement before us that Uniplate enjoyed a dominant position on either delineation of the market. Second, whether there is a systems market or primary and secondary markets, there is, as the Tribunal acknowledged, an interdependence between the supply of two complementary products. There may well be no independent demand for embossing machines without blanks. But there is clearly an ability to supply blanks that do not issue from the supplier of the embossing machines - why else require exclusivity?

 

[9]    What matters is whether the imposition by Uniplate of the exclusivity restraints had an anti-competitive effect upon rivals so as to foreclose the market. That may be so, as this court explained in Computicket, whether rivals are rendered less effective by reason of the aggregate effects of the exclusivity across the market or by reason of effects that occur in a part of the market, absent which the dominant firm would be more meaningfully constrained. Accordingly, whether rivalry is constrained in the supply of blanks alone or whether this occurs in the supply of machines and blanks, a foreclosure case can be made out. The supply of blanks may constitute a market or part of a market. Even if it is part of a market, the question remains whether the foreclosure of firms who would supply blanks has taken place. If it has, and no outweighing benefits are proven, that may constitute an infringement because the supply of blanks is clearly one important component of the rivalry between firms, whether the supply of blanks is a separate market or forms part of a single systems market. Accordingly, there is no reason further to engage the precise demarcation of the market.

 

[10]  The principal issues that arise for consideration in this appeal are then the following.

 

First, was the Tribunal correct in its finding that the exclusivity restraints utilised by Uniplate Computicket (Pty) Ltd v The Competition Commission [2019] ZACAC 4 at paragraphs 31 and 32 foreclosed the market to actual and potential competitors and that the Commission had discharged its burden of proof on this issue? Uniplate's submissions before this court emphasised the Commission's reliance on the likelihood of foreclosure, given the structure of the market and the nature and duration of the exclusivity. This, it is said, comes close to a revival of form-based prohibition and fails properly to consider the evidence of actual effects. Uniplate equally implicates the Tribunal in this criticism. More generally, whether the Tribunal was correct in finding that Uniplate's use and enforcement of exclusivity restraints gave rise to the foreclosure of actual and potential competitors falls for determination.

 

[11]  Second, the Tribunal found that the foreclosure of the market was significant and resulted in higher prices for blanks and a lack of choice for consumers. Uniplate contends that there was no basis for the Tribunal to have made findings as to the adverse price effect of foreclosure. And this too is a ground of appeal.

 

[12]  Third, the Tribunal rejected Uniplate's case that the exclusivity restraints were justified and that the pro-competitive gains of the restraints outweighed their anti-competitive effect. Uniplate submits on appeal that the Tribunal was in error on this score.

 

[13]  I consider these issues in this sequence. If there was no foreclosure, then the Tribunal would have attributed the higher pricing of blanks to the wrong cause. There would then be no need to consider the pricing issue, nor the case that Uniplate made to justify the exclusivity restraints. If, however, the Tribunal was correct to find that there was foreclosure, then its conclusion as to the pricing of blanks requires consideration because it contributes to the Tribunal's finding that the foreclosure was substantial. That conclusion is relevant to the burden of justification that Uniplate would then bear to show pro-competitive gains. I thus consider the issues in this sequence. Was there foreclosure? If so, did it result in higher prices for blanks? And finally, did Uniplate discharge its burden of justification, given the extent of the anti-competitive effect of its conduct ?

 

Foreclosure

 

The Tribunal's decision

 

[14]  The Tribunal considered the issue of foreclosure in this way. It first considered the contracts utilised by Uniplate to impose the exclusivity restraints. Over a long period of time, 1994 - 2014, Uniplate's agreements required that embossers procure all their blanks from Uniplate. This was so, even when a customer used machines acquired from other manufacturers, together with the embossing machine acquired from Uniplate. The agreements were of long duration, 10 years, and the majority of the cash and rental agreements provided for automatic renewal. The Tribunal considered evidence as to the number of Uniplate agreements that allowed for early termination or buy-back arrangements. It found that although Uniplate provided figures indicating that on average some 45% of agreements in the period 2010 - 2014 did allow for early termination or buy-back arrangements, these figures did not account for agreements still in force and concluded before 2010. Further, the evidence did not suggest that customers in fact were able to terminate their agreements, and where they sought to do so, customers encountered resistance from Uniplate. In sum, Uniplate imposed exclusivity restraints upon its customers of considerable scope and duration.

 

[15]  Next the Tribunal considered the extent of contestable demand in the market. Given Uniplate's acknowledged dominance, and the exclusivity restraints imposed by Uniplate, what demand was there over which competitors could compete with Uniplate? Although Uniplate recognised that contestable demand was necessarily reduced in consequence of the exclusivities in place, Uniplate nevertheless contended that significant demand remained available to competitors. The sources of that demand (even under the assumption of no early termination) were embossers entering the market, increased sales by existing customers, and embossers whose contracts with Uniplate were coming to an end.

 

[16]  Uniplate sought to substantiate its position as to contestable demand by relying on market share data that demonstrated that NNPR was able in 2012 to introduce into the market its type B system and secure a significant market share over the period 2012 - 2016 for type B blanks, while maintaining its share of the market for acrylic blanks and growing its share in type A blanks.

 

[17]  The Tribunal was little moved by this evidence. The Tribunal approved the position of the Commission's expert, Dr Mncube, that market share data are meaningless, without consideration of the counterfactual, that is to say, absent the imposition of Uniplate's exclusivity restraints. The Tribunal's own assessment of the market share data, notwithstanding these misgivings, was that Uniplate, over the period, increased its overwhelmingly dominant position for the supply of all blanks and NNPR simply retained its market share. The Tribunal found that as a result of the imposition by Uni plate of the exclusivity restraints, contestable demand was "miniscule if not absent ".

 

[18]  The Tribunal then considered the evidence of the factual witnesses called by the parties. It made the following findings. First, NNPR could only enter the blanks market if it provided embossing machine. This was so because Uniplate's rivals could not compete by winning customers to supply blanks alone. Rivals must sell machines to secure demand for blanks. And this had the effect of raising rivals costs in a market where the costs of entry were already high.

 

[19]  Second, the evidence of Mr de Lange showed that his attempts to purchase type B blanks from NNPR for his Uni plate machine were met with threats of legal action by Uniplate. This effectively precluded him from sourcing type B blanks from NNPR at a cheaper price. This, the Tribunal found, was persuasive evidence that the exclusivity restraints effectively excluded rivalry for the supply of blanks on a stand-alone basis.

 

[20]  Third, the Tribunal concluded that three international firms had explored the possibility of entering the market and that the evidence sufficed to show that they were deterred from entry or their entry was short-lived by reason of insufficient demand. This amounted to a prevention or lessening of competition because the insufficiency of demand excluded potential entry.

 

[21]  This led the Tribunal to conclude that the Commission had discharged its onus of showing that Uniplate's exclusive agreements had the likely effect of foreclosing rivals in the number plate market. The Tribunal found that the exclusivity restraints increased rivals' costs because effective competition in the blanks market required entry into the machines market. Rivals in the blanks only market remained small. New entry was discouraged because demand for blanks was tied up given the duration of the exclusivity restraints imposed by Uniplate. And finally, customers were prevented from switching to rival suppliers.

 

Likely effects and actual effects:

 

[22]  Uniplate's appeal, on the issue of foreclosure, rests m the first place upon the submission that the Commission's case before the Tribunal relied upon the likely anti­ competitive effect of Uniplate's exclusivity restraints, rather than its actual effect. The Tribunal decision paragraph [165] Commission sought to derive the likely effect from the structure of the market and the nature of the exclusivity restraints. Uniplate submits that this approach to foreclosure fails to meet the true burden that rests upon the Commission, more particularly when there is evidence of the actual effects that were felt in the market. Whether the Tribunal is said to have fallen into a like error is not made entirely clear. Certainly, the Tribunal's ultimate conclusion was that the exclusivity restraints had the likely effect of foreclosing rivals. In so doing, the Tribunal relied upon the incumbency advantages of Uni plate as a dominant firm and the salient features of the exclusivity restraints imposed by Uniplate. That, however, was not all that the Tribunal's reasoning rests upon. I nevertheless proceed to consider whether the Tribunal adopted an approach to the Commission's onus as to foreclosure that is too lenient.

 

[23]  In Computicket, this court gave an exposition of the concepts that are to be found ins 8(d). What bears emphasis is the following. First, the exclusionary act of the firm is something separate from the anti-competitive effect of that act. There must be a causal relationship between the exclusionary act and its anti-competitive effect. If the exclusionary act is taken as proof of its effect, the onus resting upon the Commission will not have been discharged. That would be a case based on conduct without regard to consequence, and does not meet the requirements for liability under s 8(d).

 

[24]  However, this does not mean that the salient features of the exclusionary act are not relevant to its likely effect. The broader the scope of the exclusionary act and the longer its duration, the more likely it is to have an anti-competitive effect. For this reason, in Computicket, the court stated that the more substantial the exclusionary conduct, the more likely it is that the impact upon the market will also be substantial. The present case illustrates this proposition. A broad restraint that requires a customer to acquire all its blanks from Uniplate for a period of 10 years and more, where Uniplate enjoys a dominant position in the market, is more likely to have an anti-competitive effect than a modest restraint of short duration. The features of the exclusionary act, once established, do not end the enquiry as to effect. That determination must still take place. But the probability as to whether an anti­ competitive effect has been caused by a particular exclusionary act will often be influenced by the relevant features of that act.

 

[25]  The second matter that warrants clarification is this. The Tribunal found that the exclusivity restraints had the likely effect of foreclosing rivals in the market. The likelihood of foreclosure should not be confused with the question as to what foreclosure has taken place. The foreclosure may be actual or potential, but to discharge the onus it must be proven on a balance of probabilities that foreclosure has taken place. The likelihood of foreclosure concerns the question as to the sufficiency of proof. The type of foreclosure that has taken place, actual or potential, marks out a different enquiry. Foreclosure may be observed when a firm leaves the market or its market share declines and these facts are attributable to the exclusionary act under scrutiny. Foreclosure may also come about because a firm that would have entered the market , or if present in the market would have expanded in the market, does not do so as a result of the exclusionary conduct. Here the firm has not done something it would have been able to do and, but for the exclusionary conduct, would have done. Foreclosure also has a temporal dimension. Foreclosure, in any of the varieties that I have referenced, may have occurred in the past. There may also be evidence that what has occurred in the past is likely to persist in the future, or that even though it has not yet happened, it is likely to happen in the future. What is observed in the market, what would have happened in the market had the exclusionary act not taken place, and what is or would have been likely to occur in the future are different aspects of foreclosure that fall under the description of actual or potential foreclosure. All are worthy of consideration and proof.

 

[26]  The Tribunal found the following: "In our view, the Commission has discharged the onus of showing that Uniplate 's exclusive agreements had the likely effect of foreclosing rivals in the number plates market." 4 I find no conceptual fault in this formulation of the onus resting upon the Commission. This conclusion was reached by the Tribunal after its consideration of the exclusivity restraints imposed and enforced by Uniplate. The Tribunal also considered and assessed the evidence as to what effect the exclusivity restraints had upon rivals in the market, whether an existing competitor, NNPR, or potential rivals who might have entered the market. The conclusory finding of the Tribunal may reasonably be understood to mean that the Commission had shown on a balance of probabilities (that is, more likely than not) that the exclusive agreements foreclosed rivals in the market. That is the meaning to be attributed to Tribunal decision [165] "the likely effect of foreclosure" and entails no diminution of the ordinary civil standard of proof.

 

[27]  I also do not consider that the Tribunal, upon a reading of its reasons, simply inferred foreclosure from the exclusionary act and the position of Uniplate as a dominant firm in the market. As I have sketched above, the Tribunal did not confine itself in this way. It certainly considered the evidence as to what effects the exclusivity restraints had upon rivals or potential rivals in the market. Nor is it an error of inferential reasoning to hold that the salient features of the exclusive restraints have a bearing on the probabilities as to whether foreclosure took place. The breadth and duration of the restraint will bear upon the contestable demand that is open to rivalry and whether that provides competitors sufficient scale so as constrain the dominant firm. This is not a case in which the Tribunal simply inferred the anti-competitive effect of foreclosure from the exclusionary conduct of Uniplate. Rather, quite permissibly, the Tribunal considered the exclusionary restraints to bear upon the probability of foreclosure. This, together with other evidence of effects, led the Tribunal to its conclusion. I find no logical fault in the Tribunal's reasoning.

 

Actual foreclosure

 

[28]  I turn next to the principal ground upon which Uniplate appeals the Tribunal's finding of foreclosure. Uniplate submits that the finding of foreclosure cannot survive the evidence that over the complaint period NNPR achieved minimum efficient scale and was able to compete effectively against Uniplate. If that is so, then there was no actual foreclosure.

 

[29]  Uniplate relies upon market share data recording the volumes of blanks and embossing machines supplied over the period 2010-2016. This data, Uniplate submits, properly analysed, show that NNPR was able to compete effectively with Uniplate over the complaint period.

 

[30]  The Tribunal considered the market share data meaningless without an appreciation of what the market would have looked like under the counterfactual that Uniplate had not imposed exclusivity restraints.5 The Tribunal also found that, even if it had regard to the data, Uniplate grew its share of the market for the supply of blanks from 72% to 76%, whereas NNPR's share was stable in the low range of 17 % - 21%, demonstrating Uniplate's enduring dominance.

 

[31]  The Tribunal's approach to the market share data is incorrect. Of course, if we had data as to market shares over the complaint period, absent the exclusionary conduct, and could compare that data with the market shares with the exclusivity restraints in place, we should have a most excellent basis to consider the effect of the restraints. But the imposition of the restraints makes this world of idealised comparison impossible of achievement. We have to interpret the data generated by the world as it is. This does not mean that there may not be data probative of the counterfactual. In some cases, there is a time-period sufficiently proximate to the complaint period when the exclusionary conduct was not in place or a range of cases to which the exclusionary conduct was not applied.

 

[32]  However, where, as in this matter, the data relevant to the counterfactual are not available or presented, there is no warrant, a priori, to reject market share data that reflects the market with the exclusionary conduct in place. This is so because empirical data, even if imperfect, may provide valuable evidence that either supports or detracts from a theory of harm. One can always imagine better data that may or may not be available. But imperfect data may not be disregarded. Its limitations may simply form part of the necessary caution with which it is interpreted. The Tribunal's position simply bears out the saying that perfection is the enemy of the good.

 

[33]  What then does the market share data show? The Tribunal found that it only confirmed Uniplate's dominance. That however is an oversimplification. In the complaint period, NNPR grew its share of type B blanks from 0 - 16%, and by 2016 to 22%. In type A blanks, NNPR achieved considerable growth in the period 2010-2015 from 28% to 52%, though the majority of this growth occurred in 2011. Nevertheless, NNPR's share of type A plates exceeded that of Uniplate in every year. In acrylic plates there was relatively little movement of shares, and both NNPR and Uniplate maintained their shares. Computing the shares for all blanks, NNPR grew its share from 17% to 21% in the complaint period, peaking at 23% in 2015. Uniplate also increased its share of all blanks from 72% to 78% in the period 2010 - 2016. lt was this growth that led the Tribunal to conclude that Uniplate had simply entrenched its dominance.

 

[34]  Uniplate's large and modestly growing share of all blanks supplied over the period 2010- 2016 is a function of two things. First, Uniplate had very large shares of the two biggest segments of blanks: acrylic and type B. Second, type A blanks over the period lost volumes, whereas type B gained volumes. And NNPR's larger share of type A blanks was thus in a somewhat diminishing sector.

 

[35]  What may fairly be concluded from this data is the following. NNPR significantly grew its position in the type A sector. It did so by taking share from Arga. Arga was a competitor of Uniplate and NNPR in the supply of acrylic and type A blanks; and, up until 2010, also supplied embossing machines. Arga's share of type A blanks greatly diminished over the period. In this, NNPR was more than able to sustain a significant competitive position against Uniplate. In the type B sector, which NNPR had initially shunned, NNPR was able to build up a significant share in a relatively short period, though nevertheless modest in comparison to that of Uniplate.In the acrylic sector, Uniplate remained considerably dominant in a static sector.

 

[36]  Uniplate contends that this data demonstrate that NNPR was able to compete effectively with Uniplate in the market. Uniplate seeks to bolster this conclusion, relying upon evidence as to the minimum efficient scale that permitted NNPR to produce plates at a cost that was competitive with Uniplate. Mr Steenkamp, the managing member of NNPR, testified that NNPR's investment in the production of type B number plates required sales of 300 000 blanks per year to recoup the investment (though this recoupment might be reached by selling different types of blanks). This, submits Uniplate, shows that Uniplate throughout the complaint period produced blanks at minimum efficient scale, given that its total sales considerably exceeded 300 000 blanks.

 

[37]  One of the principal concerns that arises from the imposition of exclusive dealing requirements by a dominant firm is that rivals will be excluded from the market or fail to achieve competitive efficiency within the market because they will not achieve minimum efficient scale. If that was not so in the case ofNNPR, and Mr Steenkamp's evidence confirms this, then this too supports the proposition that Uniplate's exclusivity restraints did not prevent NNPR  from  achieving efficient scale to compete effectively  with  Uniplate..

 

[38]  Furthermore, although the Tribunal rightly expressed scepticism that Uniplate's arrangements made significant provision for early cancellation or buybacks, whether contractually or factually, the market share data support the claim that there was sufficient contestable demand in the market to permit NNPR to secure and gain market share for the supply of blanks. NNPR did so, in the face of Uniplate having imposed exclusivity restraints since at least 1995 and did so, on Mr Steenkamp's testimony, at minimum efficient scale.

 

[39]  The more granular consideration of the market share data for blanks that I have referenced supports the position that NNPR was able to secure a competitive position in the market. In the complaint period, NNPR was able to grow its share in two of the three segments of the blanks market. Of importance, NNPR did so in the type B segment which it had previously chosen not to enter. Its entry into this segment secured a significant and growing share, if not one that that eliminated the dominant position of Uniplate. However, Uniplate had seen the opportunity of developing the type B segment given the growing demand for aluminium plates. Uniplate had a first mover advantage that it enjoyed by making an investment on risk that NNPR was not initially willing to make (NNPR wrongly predicted that the type A product would win out). In these circumstances, it could not be expected that NNPR would in a couple of years undo Uniplate's dominance in this segment.

 

[40]  One further body of evidence of some import concerns the data for embossing machine sales in the period 2010 - 2016. This data were compiled by Mr Murgatroyd of RBB Economics, the expert economist who gave evidence for Uniplate. Mr Murgatroyd's supplementary report and presentation during oral testimony compute shares based on the total number of embossing machines supplied in each year. The report also provides a table that takes account of the different dies supplied in machines supplied by NNPR (so called combination machines). Reference to the amended table (which differentiates shares for acrylic and aluminium, and presents more disaggregated data) shows that over the period NNPR was gaining significant market shares at the expense of Uniplate. NNPR's market shares steadily rose over the period from 14% to 62 % in respect of machines that make aluminium plates. NNPR's market shares in respect of machines that manufacture acrylic plates is less dramatic but rises from 36% to 63%. Some of the gain in acrylic is made at the expense of Arga. But in both categories,  Uniplate suffers significant falls over the period: in acrylic from  50%  to 37%, and in aluminium from 83% to 38%. The aggregated data shows a pattern no less significant. NNPR's share of total sales of machines rises over the period from 14% to 52%, whilst that of Uniplate declines from 83% to 48%.

 

[41]  This market share data indicate that NNPR was very successful in gaining market share in the sale of embossing machines. Once Uniplate introduced type B embossing machines, it was able substantially to increase its market share of this type of machine. Since both NNPR and Uniplate make use of exclusivity restraints, the future market for blanks is a function of machine sales. This implies that as NNPR takes ever more market share in more recent sales of new machines, its future sales of blanks tied to those machines can only grow. This in turn give rise to the likelihood that NNPR's market share for blanks will also grow.

 

[42]  Plainly new sales over the period do not reflect the shares of the existing stock of machines. Given Uniplate's dominant position in the supply of blanks and its longstanding practice of tying the supply of machines and blanks, it is likely that Uniplate machines, over the complaint period, enjoyed a dominant share of the stock of machines used by embossers. However, the sizable growth ofNNPR's share of new machines supplied into the market and its ability by 2015 to capture 50% of the sales of aluminium machines, given that it only entered the type B segment in 2012, indicate that NNPR is an effective supplier of embossing machines. This in turn is likely to support the strengthening position of NNPR in the supply of blanks. NNPR has been able to replicate Uniplate's use of tying to grow its share of the supply of type B blanks. This data thus supports the likelihood that NNPR will become an ever more significant rival ofUniplate in the supply of blanks.

 

[43]  The Tribunal considered the market share data of little evidential value. In this the Tribunal fell into error. Its assessment of this data was cursory and failed to appreciate what the trends in the data had to say about the position of NNPR and its ability to constrain Uniplate. In sum, NNPR was able to operate at minimum efficient scale to supply blanks to customers in competition with Uniplate. NNPR was able to introduce type B machines and grow market share at Uniplate's expense in supplying both acrylic and aluminium machines. In the supply of blanks, NNPR grew its market share in type A blanks, maintained its position in acrylic blanks over the complaint period, and grew its share of type B blanks from 0% to 22% by 2016. Uniplate's use of tying requirements. that had been in place since 1996. did not prevent NNPR from competing over sufficient contestable demand in the market to achieve these outcomes. The Tribunal failed to appreciate these matters and take them into account in making its assessment that the market was foreclosed by Uniplate's exclusionary conduct.

 

[44]  The question that then arises is whether the Tribunal's conclusion as to foreclosure is still supportable, notwithstanding its error in recognising the competitive strengths of NNPR. The Tribunal's conclusion on foreclosure did not rest solely upon its assessment of NNPR. The Tribunal also found that Uniplate's exclusivity restraints rendered already high barriers to entry yet higher because effective competition in the supply of blanks required an investment to produce and supply machines so as to generate demand for blanks. The Tribunal also found that Uniplate's exclusivity restraints prevented new entry into the market. Three international firms had considered entry, but the Tribunal found that there was evidence that at least two of these firms did not do so because of the dominant position of Uni plate and its imposition of exclusivity restraints. It to these issues that I now turn.

 

[45]  There can be little doubt that Uniplate's longstanding imposition of exclusivity restraints on the purchase of blanks made the ability to compete with Uniplate in supplying machines, over time, a necessary condition for securing significant demand for the supply of blanks. The robustness of this proposition may be observed by reference to the following evidence. Uniplate's success in introducing aluminium plates using type B machines required NNPR to follow suit and introduce its own type B machine. The growth ofNNPR's machine sales and its replication of Uniplate's tying arrangements were a central part of securing NNPR's competitive position and growing its sales of type B blanks

 

[46]  The relegation of Arga in the market for the supply of blanks might have been thought to found the central case for foreclosure against Uniplate. Yet neither the Commission's case before the Tribunal, nor the Tribunal's reasoning in its decision, considered this to be so. The Tribunal's consideration of firm-specific actual foreclosure was devoted to the position of NNPR. This leaves much unexplored as to why Arga came to be marginalized in the market and how far that came about by reason of Uniplate's exclusivity restraints.

 

[47]  The position of Arga is instructive. Arga was in 2010 a significant supplier of acrylic and type A blanks. In that year it was the leading supplier of type A blanks. Over the period 2010-2014 its sales drastically diminished, so much so that its supply of acrylic blanks was reduced to a 1% market share, and its share of type A blanks reduced from 52% to 10%. Arga did not supply type B blanks. In the supply of embossing machines, Arga had a market share of 14% in acrylic machines in 2010, and a 3% share in aluminium machines in that year. Thereafter, Arga appears to have exited the market for the supply of machines and never introduced a type B machine.

 

[48]  A number of issues are raised by Arga's decline. There is no evidence that either Uniplate or NNPR increased their use of exclusivity restraints in respect of type A machines over the complaint period. If anything, the shift to type B machines would have made it likely that more type A machines were coming to the end of their exclusivity period. Yet Arga suffered a drastic decline in its share of type A blanks. More generally, it is unclear whether the decline of Arga in the supply of acrylic and type A blanks was causally connected to the exclusivity restraints used by Uniplate and NNPR over the complaint period. The matter was not analysed in the evidence before the Tribunal, nor engaged by the Tribunal. So too the reasons as to why Arga no longer supplied embossing machines after 2010, and whether Arga was unable to mimic the competitive model adopted by NNPR, were also left unexamined.

 

[49]  These questions concerning Arga raise wider issues. Of the total number of embossers, what proportion was tied to Uniplate and NNPR, as a result of their use of exclusivity restraints? Of the unaffiliated remainder, what was their demand for blanks? And was that demand insufficient to support entry or sustain a blanks-only manufacturer? If entry required a manufacturer to be able to supply machines and blanks, was that possible? If so, would that have fostered sufficient rivalry to permit embossers to switch from Uni plate or NNPR? Or were the exclusivity restraints too broad and enduring to make that possible?

 

[50]  The evidence on the record is simply too fragmentary to answer these questions. Certainly, Uniplate sought to make the case before the Tribunal that Uni plate and NNPR made extensive use of exclusivity restraints in respect of type A and type B plates. It did so to support the proposition that rivalry took place in the supply of systems (machines and blanks). This attempt to make a virtue of the status quo required the Commission to enter upon the terrain that these questions mark out. But the Commission's case was focussed upon the dominance of Uni plate, its use of exclusivity restraints and their impact on NNPR. As a result, it is difficult to draw any firm conclusions as to whether blanks-only manufacturers were foreclosed from the market. Certainly, the Commission's did not make out that case.

 

[51]  The Tribunal was undoubtedly correct that Uniplate's exclusivity restraints prevented customers from switching to procure blanks from other firms. Nor can it be doubted that these restraints significantly reduced contestable demand in the market. The Tribunal's finding, however, that contestable demand was "miniscule" is not supported by the market share data because NNPR was able in the different categories to sustain or grow its share of blanks. NNPR was also able to significantly grow its share of machines, and thereby secure a source of demand for blanks that was not prevented by Uniplate's imposition of an obligation upon its customers to purchase all their blanks from Uniplate.

 

[52]  The Tribunal's found in essence that Uniplate's exclusivity restraints imposed costs upon rivals to invest in machines to supply to the market so as to compete effectively in the supply of blanks. That may well be so. But NNPR was able to make that investment, and grew its share of the market for the supply of machines. On the figures provided by Mr Murgatroyd, there is no indication that NNPR's supply of machines was not profitable. The exclusionary conduct of Uniplate required that effective competition would take place within a framework predicated upon producing and selling machines and the use of exclusivity restraints to secure demand for blanks. The facts support the conclusion that NNPR was able to meet these framework conditions and as a result engage in effective competition against Uniplate. This does not support a showing of actual foreclosure in the complaint period.

 

[53]  This gives rise to the following difficulty. The Tribunal did not recognise that NNPR was able to compete with Uniplate. NNPR did not suffer the fate of Arga, almost certainly because it adapted its business model to sell machines in the market, including type B machines, and adopted exclusivity tying arrangements of its own. This meant that the Tribunal did not consider what difference the reduction in contestable demand occasioned by Uniplate's restraints made to the effectiveness ofNNPR as a competitor. Whether NNPR would have been a more effective competitor was not a matter that the Tribunal considered. It is most likely that absent Uniplate's restraints, NNPR would have sold more plates. But what consequences this would have had for the state of competition in the market was not determined by the Tribunal.

 

[54]  Plainly, the extent and duration of Uniplate's exclusivity restraints required NNPR to compete with some vigour for the contestable demand in the market. This NNPR has done, with no small success. Whether the intensity of this competitive effort would have occurred with or without Uniplate's exclusivity restraints was not addressed by the Commission, nor decided by the Tribunal. And therefore what competitive position NNPR would have enjoyed without Uniplate's exclusivity restraints remains unknown.

 

[55]  In my estimation, the Tribunal's failure to recognise the competitive attributes that NNPR did bring to bear, with the exclusionary restraints in place, led to its failure to address the further issue as to what incremental difference to effective competition the constraint on contestable demand brought about in respect of NNPR. Nor were the consequences of Arga's position considered. Nor did the Tribunal, as a result of the case made by the Commission, answer the wider question as to what other sources of rivalry were foreclosed (save for foreign entrants to which I will come).

 

[56]  These failings give rise to the conclusion that the Tribunal's conclusion that the Commission discharged its onus to show that there was actual foreclosure in the market cannot be sustained.

 

Potential Entrants

 

[57]  This conclusion does not end the enquiry. The Tribunal also found that there was foreclosure of potential entrants into the market. Three international firms, Utsch, Utal sp and Smart had explored entry into the South African market. Utsch, a German manufacturer, had expressed an interest in buying a stake in NNPR. These negotiations failed. Utsch entered the market in 2009 and exited in 2010/201l. It appears that Utsch's exit may have been based upon regulatory uncertainty and the economic downturn. Little weight, as the Tribunal found, can be attached to the fate of Utsch.

 

[58]  The position of Utal is rather different. The evidence of Mr Steenkamp and Mr de Lange is that they were, at different times, approached by Utal to enter a business partnership. UtaL had ultimately declined to do so when it learnt that Uniplate had tied many customers by means of its exclusivity restraints. Uta confirmed in a letter to the Commission that it had not entered the market in 2014 because of Uniplate's"dominating position on the market that clearly seems to extend to a great number of embossers, thus hampering other companies to access the market"

 

[59]  Mr Steenkamp also testified that Smart had entered discussions with NNPR to set up a number plate business but decided not to pursue it on learning of the exclusivity provisions that Uniplate imposed by way of long term contracts.

 

[60]  The Tribunal found that even without the direct evidence of these international firms, there is evidence of failed entry and that shows a prevention or lessening of competition.

 

[61]  As I have already explained, foreclosure may be actual or potential. In essence, the Tribunal found that Utal and Smart did not enter the market and that the exclusionary conduct of Uniplate caused these firms not to do so.

 

[62]  Uniplate submits that the evidence relied upon by the Tribunal is hearsay and that the Commission failed properly to investigate with the international firms why they decided not to enter the market, The evidence of Messrs Steenkamp, de Lange, and Dr Mncube acknowledged that Utal supplied both machines and blanks. So too did Smart. Accordingly, both firms were in a position to enter the market to compete with Uni plate.

 

[63]  The hearsay complaint is of little moment. Messrs Steenkamp and de Lange engaged Utal in discussions concerning entry. Mr Steenkamp did so too with Smart. The fact that these firms said that Uniplate's exclusionary restraints were the reason for going no further with their discussions about a business venture is not inadmissible. Whether this was the reason for these firms to discontinue the discussions is hearsay, but the Tribunal may allow hearsay evidence. The true question is what its probative value is.

 

[64]  The difficulty lies elsewhere. When a case rests upon what was not done, it is necessary to show what would have occurred but for the exclusionary conduct. Put simply, it must be proven that the firms that did not enter would have done so, absent the exclusionary conduct, and that, had this occurred, there would have been some benefit to the state of competition. This is so because the harm to competition that s 8(d) contemplates is an anti-competitive effect. Where a firm fails to enter the market, the anti-competitive effect is co-extensive with the benefit to competition that would have occurred had the firm entered the market.

 

[65]  There are a number of reasons as to why the Tribunal's findings of the foreclosure of potential entrants cannot be sustained. First, the basis upon which a firm decides to enter a market by way of a business venture rests upon a number of factors - most commonly, whether the investment is likely to make a sufficient return,judged against other uses ofcapital and the firm's hurdle rate. In order to understand how Uniplate's position in the market and its exclusivity restraints figured in Utal's or Smart's calculus of the return on investment, considerably more would be required by way of evidence. Whether it was Uniplate's dominant position or its exclusionary conduct or other factors relevant to the market or country risk or the alternative opportunities open to these firms are matters that required evidence and consideration.

 

[66]  The reasons for and against entry into a market are seldom singular. They require explanation, and Mr Steenkamp and Mr de Lange were in no position to give this evidence. Their evidence is simply too sparse on this score to establish that these firms would have entered but for the exclusivity restraints of Uni plate. The mere fact that Utal and Smart did not enter the market plainly does not go far enough. Why they did not do so, as a matter of objective appraisal, requires rather more than the recollections of their commercial interlocutors.

 

[67]  Second, there is no assessment as to what effect the entry ofUtal and Smart would have made in the market absent the exclusivity restraint. These firms appear to have been well resourced and could supply both machines and blanks. But if they were contemplating a venture with NNPR as an existing firm in the market, what difference would this have made to the competitive constraint upon Uniplate, given that NNPR was a firm already active in the market? So too in respect of a venture with Mr de Lange: how much more effective would competition in the market have been? These matters were not addressed by the Tribunal. They required determination because the fact that a firm does not enter, even as a result of the exclusivity restraints, does not establish that the firm's entry would have made a difference to effective competition in the market. That is something that requires proof. It was not proven.

 

[68]  For these reasons, the Tribunal's findings on potential foreclosure cannot stand. Here too the Commission failed to discharge its onus to prove potential foreclosure and hence it failed to show that the exclusionary conduct of Uni plate had an anti-competitive effect.

 

Pricine

 

[69]  There remains one further matter to consider. The Commission, in its complaint referral, contended that Uniplate's exclusivity restraints resulted in Uniplate charging higher prices for blanks. The Commission's expert report compared Uniplate's prices for blanks in Gauteng, the Western Cape and KZN with those ofNNPR. The Commission found there to be significant differences over a range of types and sizes.

 

[70]  The RBB report by Mr Murgatroyd criticized this finding. *U complained that the comparison was not informative because the Commission had compared the list prices of Uniplate with NNPR's actual volume-weighted average selling prices, which were likely to reflect discounts.

 

[71]  The Commission's expert did not ultimately seek to defend the comparison of prices made in his report, but testified that his analysis was simply offered as some support for the factual evidence of Mr de Lange that the plates bought from Uniplate's distributor were 20% more expensive than alternative suppliers. Dr Mncube also acknowledged that a pricing analysis may not be straightforward or useful, absent a counterfactual without exclusive contracts.

 

[72]  The Tribunal found that there was significant foreclosure, resulting in higher prices for blanks. This finding is expressed to be a consequence of foreclosure. *U is not clear that the Tribunal intended to treat the higher pricing of blanks as an independent basis for concluding that there were anti-competitive effects. Rather, the Tribunal appears to have simply drawn the conclusion that if, as it found, there was significant foreclosure, then higher price would result because of the lack of effective competition.

 

[73]  This finding cannot stand, whether it is a consequential conclusion, or an independent basis for finding an anti-competitive effect. As a consequential conclusion it cannot stand because I have found that the Tribunal should not have found that the Commission had met its onus to show foreclosure. If there was no showing of foreclosure, the factual premise for the consequential conclusion is lacking.

 

[74]  As an independent basis for finding an anti-competitive effect, the finding also cannot stand because there was no comprehensive analysis of pricing upon which reliance could be placed that was available to the Tribunal. Dr Mncube's analysis was not defended by him as such a study. It was a makeweight of limited utility, as he acknowledged. For the rest, the evidence is slight and impressionistic.

 

[75]  The Tribunal's finding may well have been something of an afterthought, given that the Tribunal provided no analysis of the evidence that supported its finding. The finding however is challenged on appeal, and that challenge is well founded.

 

Conclusion

 

[76]  I hold that the Tribunal's findings as to foreclosure, both actual and potential, cannot be sustained. So too, the Tribunal's finding as to Uniplate's pricing. *G the finding is a consequence of foreclosure, then it fails with the case of foreclosure. *G the finding is put forward as a direct consequence of Uniplate's exclusionary conduct, it lacks a proper evidential foundation and cannot be allowed to stand. In view of these conclusions, it is unnecessary to consider whether Uni plate made good its defence of pro-competitive gains.

 

[77]  In the result the appeal is upheld.

 

The following order is made:

 

1      The appeal is upheld, with costs.

 

2.     The Tribunal's order is set aside.

 

3.     In its place the following order is made: The complaint is dismissed.

 

 

David Unterhalter

Acting Judge of Appeal

 

Davis JP and Rogers JA concurred in the judgment of Unterhalter AJA

 

APPEARANCES

 

For the appellant:            Advocate M. A Wesley

Instructed by:                  Cliffe Dekker Hofmeyer Inc.

 

For the respondent:        Layne Quilliam

Instructed by:                  The Competition Commission of South Africa

 

Heard on:                           10 December2019

 

Judgment delivered on:      25 February 2020

---