‘24. Where, a claimant has established a breach of statutory duty (as in this case the claimants have by their reliance on the Decision) it is entitled to recover damages to be assessed applying the measure of loss that is usually applied in English law to the assessment of loss caused by a tort – that is “ … that sum of money which will put the party who has been injured, or who has suffered, in the same position as he would have been in if he had not sustained the wrong for which he is now getting his compensation or reparation” – see Livingstone v. Rawyards Coal Co(1880) 5 App Cas 25 per Lord Blackburn at page 39; and, specifically in relation to competition claims, Sainsbury’s Supermarkets Ltd v Mastercard Inc and others[2020] UKSC 24 ;[2020] 4 All ER 807 , where it was held that in a competition infringement case, the primary claim will be for the claimant’s direct loss measured by the overcharge attributable to the relevant cartel activity less any part of that overcharge passed down stream to the claimant’s customers – see paragraphs 182 and 194. 25. There is no requirement to enquire into which defendant sold what products to each claimant. The claimants plead … that each defendant is jointly and severally liable for any losses caused by the cartel and the defendants admit that averment … It has not been suggested … that this was wrong or that some other principle should apply and to the extent necessary I accept the claimants’ submission that each of the defendants is jointly and severally liable for any loss caused to the claimants by the cartel. 26. In assessing the overcharge to be attributed to the relevant unlawful cartel activity it is usually necessary (and it is necessary in this case) to assess what would have happened had the infringement not occurred. This is so because in order to identify the Overcharge, it is necessary to identify and then exclude from the assessment of loss all price movements during the Relevant Period down to the date when the claimants respectively entered administration that would have occurred had there been no infringement. 27. In this case, very complex economic evidence involving statistical modelling at various levels of complexity and sophistication was deployed by both parties but in particular by the defendant in an attempt to identify what part of the price increases in LCD panels over the Relevant Period was attributable to the cartel’s infringing activity. The level of mathematical sophistication that this evidence involved does not however lead to the conclusion that the counter factual elements that have to be excluded can be quantified with precision – at best such sophistication should narrow the range of uncertainty that arises in most counter factual assessments.’
‘65. Furthermore, it is in our view unfortunate that the judge in the present case should have found assistance in what Popplewell J said in Asda at [307], when the anti-competitive conduct in that case was not remotely comparable to the concerted and dishonest worldwide cartel in which ABB participated. Any suggestion, in a case of the present type, that the court should “err on the side of under-compensation” is liable to give entirely the wrong impression, quite apart from the obvious point that the aim of the court should always be to give the right amount of compensation, without erring in either direction. All that said, however, we remain of the view that, when paragraph [12(9)] of the judgment is read as a whole, it does not betray any fundamental error of approach which vitiated the judge’s performance of his task. The most that can be said, in our view, is that in considering the judge’s approach to, and assessment of, the evidence before him, we should be alert to the possibility that he may have been unduly prone to give ABB the benefit of the doubt, or to err on the side of under-compensation, when (of necessity) wielding the broad axe or broad brush. (We observe, in passing, that the two metaphors appear to have become interchangeable in the authorities, although the images they conjure up are very different. For our part, we prefer to guide ourselves by reference to Lord Shaw’s time-hallowed “exercise of a sound imagination and the practice of the broad axe”, while reminding ourselves of the dangers of using any vivid metaphor to express a legal doctrine). … 221. Thirdly, when in the supplementary judgment the judge returned to where we think he should have started, and he sought to assess an appropriate discount to reflect the risk of over-compensation arising from the possible future operation of the Cap, we think it particularly unfortunate that he should have referred to the “rule” of erring on the side of under-compensation where the court is compelled to use a broad brush, and the “need” to give ABB the benefit of any doubt in the calculation of damages: see [15(3)(d)]. As we have explained, there is in our view no such rule and no such need.’
‘Following delivery of this judgment in draft to the parties, the claimants drew my attention to subsequent authorities that decide that the cautionary principle referred to above should no longer be followed. I had not consciously made any adjustments applying this principle because in the circumstances I considered it unnecessary to do so. When I referred to paragraph 28 that was generally shorthand for the broad brush/broad axe approach. Avoiding over compensation is different from erring on the side of under compensation or giving the benefit of any doubts to the defendant. Had I made any such adjustments I would have identified them specifically. However, I have since reviewed each of the conclusions I arrived at where it was necessary for me to apply the broad axe/ broad brush approach consciously applying that principle but without qualification. I have concluded that no adjustments to the figures arrived at are required.’
‘106. Mr Parker’s opinion remained that it was preferable to use a market specific demand measure. I agree. I have not been persuaded that endogeneity undermined use of the semi conductor metric. It is a market that is enormous when compared to that for LCD screens and services multiple different commercial and consumer sectors. Dr Van der Laan’s reasons for thinking there was an endogeneity risk posed by the use of the semi conductor data was speculative, untested and was one that Dr Van der Laan was unable to say was material. I agree with Mr Mussa’s submission that it would be wrong to prefer a less economically relevant metric in the absence of any evidence of a material risk that using a more economically relevant metric would produce statistically biased results. 107. In the end broadly I prefer Mr Parker’s evidence that the semi-conductor billing data is the best available proxy variable because it is very largely exogenous and closely correlated with monitor demand and no other proxy has been identified that more satisfactorily satisfies these requirements. That said, plainly there is a risk that to a limited extent the semi conductor proxy may under estimate overcharge by virtue of a limited endogenous element. Allowing for that impact requires the application of the principles summarised at paragraph 28. I return to that issue below.’
‘114. Whether a cost/price lag has occurred is by definition highly fact sensitive. There is no evidence of such a phenomenon in the LCD screen industry. If and to the extent the question is one of argument I consider that the assumptions made by Dr Van der Laan are likely counterbalanced by the fact that the reason for the cartel was to control prices by price fixing and controlling increases in supply. Such motivation suggests that it is highly unlikely that there would be any significant delay in changing prices by reference to changes in cost any more than in respect of changes in demand. Failing to do so would adversely affect profit on one side of the equation or market share on the other (depending on whether the costs and demand changes were up or down).’
‘143. Applying the principles summarised in paragraph 28 above and taking account of the endogeneity risk posed by use of the semi-conductor proxy as the demand variable in combination with the uncertainty already noted concerning the effect of post cartel price persistence leads me to conclude that if I was to adopt Mr Parker’s figure that would under estimate to a limited degree the level of overcharge. In my view a modest adjustment upwards of Mr Parker’s figures is required to eliminate the possibility of under recovery by reference to these sources of uncertainty but limited so as to take account of what I consider is likely to be the modest effect of both factors and the need to be cautious to avoid over compensation. Taking into account all that I have so far considered, I conclude that the probable overcharge rate suffered by the claimant that is attributable to the cartel’s infringing activity is 8% for monitors, 4% for notebooks and 14% for TVs.’
‘111. Mr Parker did not accept this reasoning. Why this is so involves a significant amount of very technical statistical maths. However it starts with the point that the adjustments carried out by Dr Van der Laan mean that the only explanatory factor that affects the price level in a given period is the price level in the two previous periods, which Mr Parker contends makes no sense when it is agreed (as it is here) that prices will be affected by costs, supply and demand. This is described by Mr Parker as “… not credible…”. He added that he had tested his model using a lagged price variable and had concluded that it should be rejected because “ … while there was a statistically significant effect of the lagged price variable, this had arisen because the lagged price variable was correlated with costs, rather than because last period’s price had any causal effect on this period’s price …” Mr Parker’s opinion was also that by omitting lag to both the demand and capacity variables that resulted in a substantially increased overcharge that does not occur when they are included.’
‘Agree. Dr Van der Laan presents evidence that there was price persistence, i.e. that the price of an LCD panel in a month is partly determined by the price of the LCD panel in the previous month. There is a statistically significant effect up to 9 months – see Van der Laan 2, 4.3. Both Dr Van der Laan and Mr Parker conclude that there was an overcharge (the estimate for Notebook panels is not significant in Mr Parker’s baseline model but with the small changes suggested by Dr Van der Laan there is a statistically significant overcharge on both Notebook and Monitor panels). An overcharge during the Cartel Period in combination with price persistence implies that prices during the immediate Post-Cartel Period were affected by the Cartel: it would have taken time for prices to return to the competitive level. Additionally, costs were likely to be higher due to Cartel as a result of reduced learning effect and delayed investment in fabs. This implies that prices in the Post-Cartel Period could have been affected by delayed investment for around two years (the time it takes for a new fab to become operational).’
‘Disagree. Dr Van der Laan presents no evidence that this is the case. I do not consider the Infringement to have reduced investment in production capacity (rather, there was evidence of capacity growth throughout the period) or to have led to permanently higher production costs. In any event, even if there had been capacity restraint, there is no reason to think that this would have been maintained beyond the Infringement. See Parker 2, Section 5.3. I do not consider that Dr Van der Laan has demonstrated price persistence for 9 months in his analysis in Van der Laan 2. His analysis shows effects in months 1 and 2, but no significant effects for monitors in months 3, 4, 5, 6, or 8, with month 7 being significant but incorrectly signed (that is, a higher price 7 months previously actually reduced current month prices). For notebook panels, Dr Van der Laan finds significant effects in months 1, 2 and 3, but after that largely a mix of insignificant and incorrectly signed lags. It is not economically plausible that prices 4 months previously have no impact, but prices 9 months previously do have an impact, as I would expect the impact to reduce over time. I therefore consider that Dr Van der Laan’s analysis is consistent with price persistence for at most 2-3 months.’
‘I conclude that the inherent probabilities are as Dr Van der Laan says and that it would be both unusual and highly unlikely that the effects of a worldwide price-fixing cartel could be eliminated instantaneously on the coming to an end of the cartel.’
‘46. It is necessary next to consider two other concepts that took up a significant part of the expert oral evidence at trial. Dr Van der Laan maintained that the soundness of the conclusions reached by Mr Parker depended on the application of statistical tests and that applying those tests to Mr Parker’s regression model demonstrated that it could not safely be relied on. Although a number were mentioned in the course of cross examination, the two that the Claimants placed most reliance on were “R Squared” and “Ramsey RESET”. Although Dr Van der Laan’s position was that the statistical tests that I have referred to were the primary determinant of whether the model had been correctly formulated, Mr Parker considered that these were of subsidiary importance because he considered the most important thing to do when preparing a model was to ensure that all the right variables (and proxy data sources) were being used in order to ensure that the exercise was as Mr Parker put it “economically sensible”. 47. In principle I should make clear even at this stage that I prefer Mr Parker’s approach to that adopted by Dr Van der Laan on this issue. It reflects what has been said in other cases where these issues have arisen - see by way of example paragraph 302 of Marcus Smith J’s judgment in Britned Developments Ltd (ibid.), where he recorded that both experts in that case were agreed and he accepted that in order to be reliable a regression analysis had to be well specified – that is to “ … take account of the main drivers of project prices and not take account of factors which were irrelevant.” - and in academic writing and other guidance – see (a) the principles set out in Carter-Hill: Principles of Econometrics (4th Ed.) at 233-234 and (b) the comment of the Commission at paragraph 81 of its Practical Guide that carrying out a regression analysis requires: “… a good understanding of the industry concerned, in the first place, to formulate the right hypotheses when constructing the regression equation and to make the right choice as to the factors that are likely to have significantly influenced the variable of interest (and which should therefore be included in the analysis)”.’ “… a good understanding of the industry concerned, in the first place, to formulate the right hypotheses when constructing the regression equation and to make the right choice as to the factors that are likely to have significantly influenced the variable of interest (and which should therefore be included in the analysis)”.’
‘143. Applying the principles summarised in paragraph 28 above and taking account of the endogeneity risk posed by use of the semi-conductor proxy as the demand variable in combination with the uncertainty already noted concerning the effect of post cartel price persistence leads me to conclude that if I was to adopt Mr Parker’s figure that would under estimate to a limited degree the level of overcharge. In my view a modest adjustment upwards of Mr Parker’s figures is required to eliminate the possibility of under recovery by reference to these sources of uncertainty but limited so as to take account of what I consider is likely to be the modest effect of both factors and the need to be cautious to avoid over compensation. Taking into account all that I have so far considered, I conclude that the probable overcharge rate suffered by the claimant that is attributable to the cartel’s infringing activity is 8% for monitors, 4% for notebooks and 14% for TVs.’
‘215. … But the question of legal causation is straightforward in the context of the retail business in which the merchant seeks to recover its costs in its annual or other regular budgeting. The relevant question is a factual question: has the claimant in the course of its business recovered from others the costs of the MSC [merchant service charge], including the overcharge contained therein? …’
‘225. … In accordance with the compensatory principle and the principle of proportionality, the law does not require unreasonable precision in the proof of the amount of the prima facie loss which the merchants have passed on to suppliers and customers.’
‘228. By way of summary on the legal test for causation in relation to a pass-on form of mitigation defence, we respectfully conclude that DAF must prove a direct and proximate causative link between the Overcharge and any increase in prices by the Claimants. It is not enough for DAF to say that all costs, including increases in costs, are fed into the Claimants’ or their regulators’ business planning and budgetary processes. There must be something more specific than that …’
‘(1) Knowledge of the Overcharge or the specific increase in the cost in question; (2) The relative size of the Overcharge against the Claimants’ overall costs and revenue; (3) The relationship or association between what the Overcharge is incurred on and the product whose prices have been increased; and/or (4) Whether there are identifiable claims by identifiable purchasers from the Claimants in respect of losses caused by the Overcharge.’
‘150. … Factual causation involves consideration of whether the effect of the mitigating conduct was in fact to reduce or eliminate the claimant’s loss, whereas legal causation concerns whether, even if the effect of the mitigating conduct was in fact to reduce or eliminate the claimant’s loss, as a matter of legal policy, it should serve to reduce or eliminate the damages payable by the defendant to the claimant.’
‘151. In terms of factual causation, DAF could only succeed in its argument on SPO [Supply Pass On] if it could establish that the prices charged by Royal Mail and BT to their customers were higher because of the overcharge, in other words if it could establish (and the burden of proof is on DAF) that the overcharge had been passed on to those customers. The CAT was unanimous as to this requirement at [223] of its judgment where it said: “we consider that DAF must prove that there was a direct and proximate causative link between the Overcharge and any increase in prices by the Claimants. That means that there must be something more than reliance on the usual planning and budgetary process, into which the Overcharge was input and at some point prices increased.” I agree with Mr Ward KC that the CAT was applying the correct legal test, as recently restated by this Court in Stellantis (as cited at [23] above).’
‘154. The CAT concluded that none of the four factors was present in this case, a conclusion which was not challenged on appeal. However, as set out at [85] above, Mr Beard KC did submit that the majority had been wrong to conclude at [573] that: “we think that in a situation where none [of the four factors] are present, the evidence of factual causation needs to be that much stronger so that the requisite proximity can be established.” In my judgment, that submission is misconceived. In circumstances where none of the four factors which might establish the requisite degree of proximity to establish a direct causative link between the Overcharge and the prices charged by the Claimants is present, it is both logical and common sense to conclude that there would need to be some other evidence of factual causation to establish that requisite degree of proximity. Far from that majority conclusion being confused and incorrect as Mr Beard KC submitted, its analysis is clear and correct, disclosing no identifiable error of law.’
‘194. In relation to the second of these points (the role of LCDs as inputs to downstream production) Mr Parker makes two points, each of which he maintains support the proposition of a high degree of downstream pass on. Both are based on the proposition that as a matter of economic theory, pass on is more likely to occur where a cartel affects the variable costs of those who are being supplied with the cartelised product, because variable costs directly influence (or are at least likely to influence) short term pricing decisions. This is significant in this case because LCD panels are a key component for producing LCD monitors and notebooks and the cost of LCD monitors and notebooks is a variable cost in the claimants’ business of producing and/or selling to end users desktop “bundles” – that is a computer, monitor and peripherals such as a keyboard and mouse – and notebooks. Mr Parker’s evidence is that this is all the more significant in the downstream market in this case because variable cost contribution of the LCD element was substantial as a proportion of the total cost of the units being sold. Mr Parker’s conclusion derived from this factor was that “…(t)he nature of the increases in cost to LCD monitor manufacturers and to downstream PC retailers caused by an Overcharge for LCD Panels – being both substantial and impacting variable costs – increases the likelihood of high pass-on of any Overcharge by Granville downstream…”’
‘226. Given the nature of the evidence relevant to the downstream pass on issue (and the conclusions I have reached so far), I consider that the approach summarised in paragraph 28 above is one that I must apply to the downstream pass on issue in the circumstances of this case. Had this claim involved companies currently trading, I would have expected much more relevant material to have been available supported by relevant oral evidence from officers and managers (or retired officers and managers) as to how sector wide variable market wide costs increases were managed. As is apparent from what I said above, there is no such evidence.’
‘243. This leads me back to the approach to be adopted in assessing damages in follow on infringement cases. The evidence available satisfies me first that as a matter of economic theory it is more probable than not that the claimants would have passed on at least some of any market wide variable cost shock. Further, the evidence summarised in this section of the judgment satisfies me that it is more likely than not that the probable level of pass on would be between 50% and 100%. This leads me to conclude that I should reject the claimants’ case that the defendants have failed to prove any pass on and that I should therefore conclude that there was 0% pass on. That is with respect entirely unreal. 244. By the same token I reject the notion that I should conclude that pass on was 100%. That is equally unreal in the circumstances of this case for the reasons that I have summarised above including because pass on was less likely in relation to products where the starting margin was higher, it is likely that any pass on would be delayed or staggered over a short period either to allow a “while stocks last” campaign or to await the launch of a new or replacement product (which appears to have happened frequently and which would limit the effect of Add On sales of items such as extended warranties) and/or because of the incidence of psychological pricing policies. However, it is probable that when viewed as a whole the claimants’ consumer retail business was a low margin business that was undoubtedly operating in a highly competitive market. In my judgment therefore it is improbable that industry wide cost increases could be ignored or action delayed for long. All this leads me to conclude that across the claimants’ retail consumer businesses it was probable that industry wide variable cost increases would be passed on at a level of somewhere between 50% and 100%.’
‘246. As I have set out in paragraph 28 above, the assessment of damages in competition infringement cases is likely to involve estimation and approximation to arrive at an appropriate figure within the bounds of uncertainty established by the evidence. It involves the application of a broad brush or a broad axe depending on what element in the assessment is being examined. As always there are two factors in play – the need on the one hand to ensure that a claimant is compensated for infringement of its or their rights but to err on the side of under compensation where there is uncertainty in order to avoid the claimant recovering more by way of damages than it would be entitled to applying conventional principles. 247. The degree of uncertainty depends on the evidence available. In relation to the issue I am now considering this has allowed the narrowing of the range of uncertainty concerning downstream pass on to that mentioned above – that is between 50-100%. In my judgment, applying the principles summarised in paragraph 28 above, taking account of the whole of the expert and other evidence relevant to this issue, that pass on by the claimants is likely to have altered over time and applied in different ways in relation to different products and balancing the need to ensure the claimants are properly compensated against the requirement to exercise caution where there is uncertainty leads me to conclude that I should assess pass on at a blended rate of 65%.’
‘246. As I have set out in paragraph 28 above, the assessment of damages in competition infringement cases is likely to involve estimation and approximation to arrive at an appropriate figure within the bounds of uncertainty established by the evidence. It involves the application of a broad brush or a broad axe depending on what element in the assessment is being examined. As always there are two factors in play – the need on the one hand to ensure that a claimant is compensated for infringement of its or their rights but to err on the side of under compensation where there is uncertainty in order to avoid the claimant recovering more by way of damages than it would be entitled to applying conventional principles. 247. The degree of uncertainty depends on the evidence available. In relation to the issue I am now considering this has allowed the narrowing of the range of uncertainty concerning downstream pass on to that mentioned above – that is between 50-100%. In my judgment, applying the principles summarised in paragraph 28 above, taking account of the whole of the expert and other evidence relevant to this issue, that pass on by the claimants is likely to have altered over time and applied in different ways in relation to different products and balancing the need to ensure the claimants are properly compensated against the requirement to exercise caution where there is uncertainty leads me to conclude that I should assess pass on at a blended rate of 65%.’