Cartels distort the market in several ways. Perhaps most obviously they allow firms to raise prices directly to generate higher profit margins. However, they can also lead to higher costs overall by letting inefficient firms retain market share and by dampening incentives to innovate. As Nobel prize-winning economist Sir John Hicks put it: “the best of all monopoly profits is a quiet life”.1
If cartels disrupt normal competitive mechanisms that promote cost efficiency, simply looking at whether the cartel led to higher profit margins will not reveal the full extent of harm to customers.
The question of cost inefficiency effects loomed large in BritNed v ABB – the first cartel damages claim to reach final judgment in the English courts.2
BritNed, the operator of the interconnector connecting the Dutch and UK electricity grids, claimed damages due to its purchase of the submarine power cable from ABB. ABB was found by the European Commission (EC) to have been party to a global cartel relating to high voltage submarine and underground power cables.3 The decision established that the cartelists had agreed to allocate the BritNed project to ABB by either declining to participate in the tender or by submitting uncompetitive bids.4
In his judgment, which has since been appealed by both parties, the judge found no evidence of inflated margins, but did award modest damages for the cartel’s impact on costs under two discrete heads of loss:
- ‘Baked-in inefficiencies’ that were deemed to have been passed-on to BritNed (€7.5 million); and
- A share of ‘cartel-related cost savings’ that ABB accessed from reduced bidding costs and its improved ability to manage future capacity (€5.5 million).
We do not expect the second head to survive the Court of Appeal and do not focus on this aspect of the judgment in this article.5
The baked-in inefficiency damages represent under 3% of the final price of €265 million agreed between BritNed and ABB for the cartelised project. This compares with the approximately €58 million derived from the approximately 22% overcharge produced by the claimant’s econometric analysis – which the judge considered “insufficiently reliable to be used in any way at all.”6
In our first article on the BritNed judgment, we set out our views on the court’s detailed evaluation of the econometric evidence of overcharge.7 In this article, we focus on the treatment of cost inefficiencies that underpinned the damages award. We compare this treatment with consulting industry best practice for assessing efficiency in the context of advising business on performance improvement. Building on our own experience of combining these skillsets in cartel litigation, we then conclude by considering the implications for addressing efficiency in cartel damages claims going forward.
What does economic theory say about cartel impacts on efficiency?
Economic theory identifies three separate types of efficiency that competition promotes and which cartels risk impairing:
- Allocative efficiency: competition puts pressure on firms to price close to their costs resulting in lower profit margins and better prices for customers. The closer price is to cost, the more efficiently the economy can allocate scarce resources between competing needs. Prices above cost mean that customers who value the product more than the cost of producing it are nonetheless deterred from buying it. This leads to inefficient levels of consumption and output and a sub-optimal allocation of society’s resources.8
- Productive efficiency: competition puts pressure on firms to reduce costs, operate efficiently and ensure – through a Darwinian 'survival of the fittest' selection process – that inefficient firms lose market share to more efficient rivals. Market allocation cartels disrupt this process and thereby risk allowing survival of not just the fittest, but also the fattest. Wrongly allocating contracts to inefficient firms leads to higher costs and ultimately higher prices.9
- Dynamic efficiency: competition also puts pressure on firms to innovate. In the long run, this may have the most powerful impact on consumer welfare if it leads to paradigm shifts in the quality of products and/or radically improved production processes. Cartels could amount to pressing a 'pause button' on innovation as again firms feel less pressure to work hard to stay ahead of the competition.10
Broadly we can refer to the first type of harm as direct price effects and the other two as cost inefficiency effects.11 There is substantial economic literature on the materiality of such adverse effects. For example:
- Direct price effects: famously, the European Commission’s practical guide on antitrust damages quotes a meta-study showing average overcharges of 20%. The study, which as we observed in our first BritNed article should be treated with caution, primarily captures price effects but could also include estimates of cost inefficiency effects.12
- Cost inefficiency effects: Disney, Haskel and Heden (2003) analyzed the impact of competition on productivity and found that market competition increases both the level and growth of productivity.13 Günster, Carree and van Dijk (2011) undertook an analysis of 141 firms that participated in 49 European cartels and found that innovation (modelled as R&D investments) was lower during the cartel period.14
Hence, a complete cartel damages assessment may need to consider all these potential impacts. Whether they apply in practice in a specific case will of course depend on the evidence.
How were these effects assessed in BritNed v ABB?
As discussed above, this case involved a market allocation cartel that resulted in ABB winning the BritNed contract. The EC found that other rivals agreed either not to bid or to submit a phoney bid much higher than ABB's.
We first summarise what the judge did and then consider the expert evidence and the judge’s assessment of it, before presenting our own views.
THE JUDGE'S APPROACH
To determine the extent to which BritNed may have overpaid, the judge sided with the claimant to define “the overcharge […] as the difference between (i) the price agreed between ABB and BritNed and (ii) the price that would have been agreed – whether with ABB or another provider – had the cartel not operated.”15 (Emphasis added).
This clearly goes beyond price effects to also capture cost effects – in particular any productive inefficiency from the cartel blocking BritNed's access to more efficient suppliers than ABB. It also seems to capture dynamic inefficiencies to the extent that suppliers may have innovated more absent the cartel. As the judge observed: "baked-in inefficiencies might arise because of an absence of internal pressure to produce a competitive price or (anterior to this) an absence of internal drive within ABB to improve the products it was selling."16 (Emphasis added).
Unfortunately, data on rivals’ costs and prices was apparently unavailable as ABB was the only defendant involved in the litigation. Only data on rivals’ costs during the cartel can comprehensively inform the productive inefficiency question.17 Similarly, only data on all suppliers’ costs after the cartel ended can comprehensively inform the dynamic inefficiency question and a long period after the cartel may be needed to capture the benefits of an innovation.
Accordingly, the judge recognised that he “cannot realistically assess what a rival bid would have been.”18 Instead, he attempted to calculate ABB’s counterfactual bid, where ABB faces competition from efficient rivals. This comprised evaluating the evidence on direct price effects and further considering whether there was prima facie evidence of inefficiency in ABB's costs. In short, the judge found no evidence of direct price effects, but did conclude ABB's costs were subject to "baked-in inefficiencies".
To quantify these inefficiencies, the judge emphasised that the evidence was "exiguous" and clarified he could not just assume that they existed.19 However, he did ultimately find an inefficiency based on internal documents suggesting that rival cables are 20% thinner than ABB’s. The judge concluded that this led to excessive copper costs and that, absent the cartel, ABB would either have lost the contract or would have had to absorb the extra costs of its less efficient technology.20
To support this reasoning, the judge pointed to the fact that after the cartel ended, ABB lost 9 out of the 10 most similar projects to BritNed (compared to winning 14 out of 14 during the cartel), and on the project it did win post-cartel, the client had specifically requested a thicker copper cable.21
Using what he described as a ‘broad brush’ approach, the judge estimated the overcharge resulting from this inefficiency at €7.5 million, equivalent to assuming that ABB would have had to lower its bid by some 15% of the copper-related costs.22
It can be seen that this approach tries to capture the productive inefficiency associated with a market allocation cartel. It does not capture any dynamic inefficiency effects.
THE EXPERTS' APPROACHES AND THE JUDGE’S VIEWS ON THESE
Both claimant and defendant economic experts focused exclusively on trying to estimate direct price effects and no expert evidence on cost inefficiencies was presented.23
Both experts developed models that compared ABB’s bids during and after the cartel. The defendant’s expert used a simple margin comparison taking ABB’s accounting costs as read, and also used these costs in econometric models (alongside other relevant factors) that sought to explain ABB's prices.
The claimant expert raised the prospect of the cartel causing cost inefficiencies to argue against using ABB’s cost data in the econometric analysis. Instead of using ABB’s actual costs, the claimant expert used proxies for efficient costs based on industry-wide cost-indices for key cost inputs (namely copper and aluminium) and tried to control for other 'legitimate' factors that should affect competitive prices.24
The judge found the defendant's margin analysis useful to support the finding of no direct price effects, but recognised that margins could not address the question of whether the cartel lead to cost inefficiencies. The judge was thoroughly unpersuaded by the defendant expert’s view that he had not found evidence of any material cost inefficiencies.25
However, the judge disagreed with the claimant expert’s blanket dismissal of the cost data. In essence, he found that the possibility of the cartel causing cost inefficiencies did not mean that the ABB cost data was insufficiently reliable for the examination of direct price effects.26 He also found that the claimant expert had relied on speculation to justify abandoning ABB cost data in the econometric model and her “attacks on the reliability of the direct costs recorded in the project pricing models (PPMs) to be misconceived”.27
Although the judge did ultimately reject the claimant’s econometric evidence of price effects, this was for other reasons other than the use of cost-proxies per se.28
Overall the judge’s approach to cost inefficiencies seems sensible in principle. He made the distinction between direct price effects and cost inefficiency effects – and determined that the prospect of the latter should not preclude the use of actual costs to determine the former. He also found that the examination of the latter requires a specific and separate focus. Given the paucity of expert evidence and relevant data available, he used a 'broad brush' to base his findings on evidence suggestive of some (modest) productive inefficiencies and stopped there in the absence of evidence on dynamic inefficiency.
One may ask whether the experts could have produced further evidence to help the judge assess the productive inefficiency question in a more robust way. To quantify the resulting harm, the judge would have required an estimate for counterfactual efficient costs. Various methodologies are available to derive such estimates.
The most obvious solution is to benchmark ABB’s costs against those of its rivals. However, as noted above, ABB’s rivals were not part of the proceedings and information on their costs was apparently not available.
An alternative possible approach is the use of cost-proxy models that substitute actual or reported costs with cost-indices (for key input costs, such as copper costs) and product characteristics that drive costs (such as cable length). In theory, this can isolate cartel-related cost inefficiencies from legitimate cost factors that drive prices, as cost-indices are unaffected by the cartel. If the cartel led ABB (and indeed all cartelists) to incur higher costs during the cartel, then an index may illuminate that by only allowing for changes in the price of key input costs between the during and after cartel periods.
Continue reading "Our views" and sections on a cost-proxy shortcut & illustration and what good looks like when assessing firm efficiency by downloading the full PDF
1. J. R. Hicks, "Annual survey of economic theory: The theory of monopoly," Econometrica, Volume 3, Number 1, January 1935, p. 8
2. BritNed v ABB (EWHC 2616, 2018). Judgment available at: https://www.judiciary.uk/judgments/britned-v-abb-another/
3. Decision of the European Commission dated 2 April 2014 in Case AT.39610 – Power Cables
4. See Decision of the European Commission, paragraph 395 and BritNed v ABB, paragraph 141
5. The cartel savings element – a surprise to many – does not make sense in a compensation framework (there is no mention of exemplary damages) as the claimant could not have been harmed by these cost savings to ABB. Indeed, the savings relate to ABB not having to compete to win projects, yet the judge’s main finding (to rule out an explicit overcharge) was that ABB did actually compete for the BritNed project specifically, if not for others. There is also the tension of assuming the inefficiencies were passed-on but the savings were not –otherwise the savings would have to be offset against, not added to, the inefficiency damages
6. BritNed v ABB, paragraph 417
8. This allocative inefficiency is not to be confused with inefficiencies that arise from market allocation cartels (see next bullet). See also Competition: Theory and Practice, Massimo Motta, Chapter 2, Section 2.2. Note that reduced allocative efficiency is not necessarily revealed by a comparison of margin as the cartel may have improved the ability of all firms to pass-on cost increases during the cartel. See Quantifying cartel damages and cost pass–through, by Burak Darbaz, Mat Hughes & David Vincent of AlixPartners (forthcoming)
9. The Competition and Markets Authority (CMA) describes these two effects as (i) a ‘market sorting’ effect, i.e. the process of ensuring inefficient firms leave the market, and (ii) an ‘x-inefficiency’ effect, i.e. the difference between the most efficient behaviour that the firm is capable of and its actual behaviour. Productivity and Competition (2015, CMA45 see paragraph 3.17). Confusingly, in paragraph 3.18, the CMA also describes the market-sorting effect in terms of allocative inefficiency
10. See for example Barnett (2008) https://www.justice.gov/atr/speech/maximizing-welfare-through-technological-innovation
11. Innovations that lead to a step-change in improving quality, as opposed to reducing production costs, can nonetheless be thought of as leading to a step reduction in "quality adjusted" costs
12. http://ec.europa.eu/competition/antitrust/actionsdamages/quantification_guide_en.pdf (paragraph 143). The numbers derive from a meta-study prepared for the European Commission by the consultancy Oxera, which built on a previous study by academics Connor and Lande (http://ec.europa.eu/competition/antitrust/actionsdamages/quantification_study.pdf, page 90). A detailed description of the methodologies used in the individual studies is not available in the Oxera study
13. Disney, R., Haskel, J. and Heden, Y. (2003). Restructuring and Productivity Growth in UK Manufacturing. The Economic Journal, 113(489), pp.666-694. The authors find a significant relationship between market competitiveness – measured using various proxies such as market concentration – and the level and growth of productivity. They also show that 90% of productivity growth can be explained by the exit of inefficient firms and the entry of efficient ones (as opposed to internal cost efficiency programmes)
14. Günster, A., Carree, M. and van Dijk, M.(2011). Do Cartels Undermine Economic Efficiency? Working paper. The authors show that R&D expenses, when expressed as a percentage of sales, fall by an average of 0.2 percentage points during the cartel period.
15. BritNed v ABB, paragraph 18
16. BritNed v ABB, paragraph 367
17. Note that rivals' costs during the cartel may of course also be affected by dynamic inefficiency. This means that data on rivals’ costs after the cartel may be required to consider both dynamic and productive inefficiencies.
18. BritNed v ABB, paragraph 451(3)
19. BritNed v ABB, paragraph 446 and 447
20. BritNed v ABB, paragraph 446 and 449
21. BritNed v ABB, paragraph 448(4)
22. BritNed v ABB, paragraph 451(3) and (4). The inefficiency adjustment was limited to the copper element of the cable. The suggestion from the claimant’s economist that the approach should be extended to other raw material and production costs was dismissed due to a lack of evidence. BritNed v ABB, paragraph 451(5)
23. In the next section, we consider a possible argument that the claimant’s model was also implicitly seeking to capture productive inefficiencies through the use of proxies
24. BritNed v ABB, paragraph 290(2) & 318. One of the control variables was the total volume of copper – the importance of this is explained in the next section.
25. “Mr Biro acknowledged the risk that direct costs might be inflated for this reason but considered that there was no reason to believe it was a material factor. … “No reason to believe” is a peculiarly weak formulation… I do not consider that Mr Biro – an expert economist – would be able to identify such inefficiencies, as I am sure Mr Biro would himself accept.” (BritNed v ABB, paragraphs 366/7)
26. This assessment differs from that of a Dutch court in a recent case, also involving ABB. In that case the court emphasised the implications of the bidridding process on the cartelists’ incentives to reduce costs and thus rejected ABB’s margin analysis as irrelevant. District Court of Gelderland, Mar. 29, 2017 (TenneT TSO BV, Saranne BV / ABB BV, ABB Ltd.), available at http://deeplink.rechtspraak.nl/uitspraak?id=ECLI:NL:RBGEL:2017:1724 (in Dutch), see paragraph 4.6
27. BritNed v ABB, paragraph 265
28. BritNed v ABB, paragraphs 417 and 419