Many arbitrations involve competition law claims. In arbitration proceedings that involve contracts, one party may claim that a particular contractual provision is anticompetitive, or that market power or abusive conduct has distorted the market in a manner that is relevant to the question facing the tribunal. We have seen competition issues raised in disputes over the review of prices in major long-term contracts for the sale or purchase of energy. Competition claims can also arise in investor–state disputes, as when the state’s actions or omissions are alleged to distort competition. Below we address two key issues that are relevant to the calculation of damages from competition claims: the determination of market shares, and calculating overcharges in price-fixing and monopolisation cases.
Determining market shares
To estimate damages from a competition law claim, the expert must compare the financial position of the claimant under two alternative scenarios. One scenario is the actual scenario that reflects the distortions of the alleged anticompetitive conduct. The other scenario is inherently hypothetical: one in which the alleged anticompetitive conduct had never occurred. Market shares are relevant because the respondent’s alleged conduct may well have suppressed the market share of the claimant, so the damages analysis must assess the market share that the claimant could have reasonably expected to obtain in the absence of the disputed conduct.
Assessing a hypothetical scenario is not unique to competition law claims; it follows directly from the broader principle of making the claimant whole, which governs the estimate of damages from a broad spectrum of legal claims. When someone breaches a contract, the estimate of damages must consider what would have happened, but never did, had the respondent continued to honour the contract. At times, the analysis can lead to the conclusion that the breach of a contract prevented a company from increasing its market share. In such a case, estimating damages from a contractual claim can become, in principle, similar to the analysis of a competition claim, as both cases require careful assessment of potentially market-wide repercussions from the disputed conduct.
However, claims for damages from competition claims are particularly susceptible to challenges of speculation, because it can often seem relatively straightforward to reconstruct a hypothetical scenario of uninterrupted contract performance, as opposed to reconstructing a scenario in which the market shares of various competitors differ significantly from reality.
For claimants, there are several responses to allegations of speculation over market shares or market prices. A general logical response is that the respondent is itself responsible for the need to speculate; the respondent’s conduct has distorted the actual performance of the market, requiring the very inquiry that the respondent declares speculative. A more specific evidentiary response involves a careful review of the records retained in the ordinary course of business by the claimant. Most claimants retain business plans and financial projections that they developed in the ordinary course of business prior to the commencement of the disputed conduct. Those plans often contain projections of the market shares and prices that the claimants anticipated, and a sound analysis of damages should investigate their reasonableness. Their reasonableness should command particular weight if the claimant demonstrated a willingness to invest, or was able to attract financing, in reliance on the specific financial projections.
To check the reasonableness of a claimant’s projections there are two common techniques: an analysis of time series data, and comparisons to other geographic or product markets in which the claimant is also present. The time series approach looks at the claimant’s performance prior to the initiation of the disputed conduct, or after its termination, and may also entail extrapolations from growth that occurred beforehand. Comparison to other geographic and product markets can test the ability of the claimant to succeed in other areas where the conduct in dispute never occurred. The field of economics has many formal tools and its own technical language, but it is often useful to recall the prominence of the two relatively simple concepts discussed above: ‘before-and-after’ analyses and extrapolations from comparable situations elsewhere.
Often the key challenge to estimating damages involves the need for reasonable adjustments. That is, two geographic markets are rarely perfectly comparable. One market may provide a reasonable benchmark for a competitive price, which can provide insight into how another market would have performed in the absence of a cartel or other disputed conduct. However, if the producers in either the benchmark market or the market in dispute face unique taxes or costs, then a reasonable approach can require adjustments to the competitive benchmark price available elsewhere.
Similarly, before-and-after analyses often confront the need for adjustments. If the price was low in a market before the disputed conduct and high afterwards, a logical inference is that the difference will reflect the impact of the disputed conduct. However, the key challenge is often to isolate the effects of other factors that may have coincided with the disputed conduct. If the market was naturally moving towards a period of relative scarcity, then one can argue that the price would have increased regardless of the disputed conduct. Changes in input costs could also have driven up prices, regardless of the disputed conduct. Assessing damages from the competition claim can, therefore, entail a modification of a simple before-and-after comparison: the change in market price over time can provide the logical starting point, to which the analysis should either add or deduct to account for other simultaneous changes that occurred.
At times it is only feasible to perform one of the two general approaches: if the claimant is a new entrant then the absence of any prior track record may render infeasible any before-and-after analysis. Alternatively, data may be available for a before-and-after analysis while other markets elsewhere are either not comparable or not sufficiently transparent to yield reliable data. However, at times both approaches are feasible, and a comprehensive analysis should compare their results. If the two approaches provide similar results, then each one reinforces the reliability of the other. If the results differ significantly, then the damages expert should inquire into the underlying reasons. The inquiry can lead either to the refinement of one approach, until it yields similar results to the other, or the abandonment of one approach as less reliable. The debate among experts can in fact boil down to disagreements over the choice between a before-and-after approach and the reliance on extrapolations from comparable markets elsewhere.
The economic literature has developed several useful tools for assessing market shares under alternative hypotheses. Many tools are variations of a basic ‘Cournot’ model, which assumes logical producer decisions in response to changes in prices by their competitors and changes in the market. A key ingredient for these models is the elasticity of demand, which measures the sensitivity of demand to changes in price. Techniques have developed for estimating demand elasticities, and it is often possible to find independent papers that estimate elasticities for products ranging from electricity to alcohol. Competition authorities often use variants of Cournot models to assess the potential effects of proposed mergers on prices, so a logical instinct is to apply the same models to estimate damages from claims of anticompetitive conduct in arbitration proceedings. Nevertheless, our experience is that the models may not work well for a variety of reasons, which relate to the differences between arbitrations and proposed mergers.
First, a specific claim of anticompetitive conduct may involve a shock to the market that is exceptionally strong in magnitude. If so, it could be inappropriate to assess the impact of disputed conduct by applying a standard competition model that contains general estimates of the elasticity of demand. The empirical literature often derives elasticities from statistical analyses of time periods that do not involve significant market disruptions. The resulting elasticity estimates can have little predictive power if extrapolated to situations that the market has not in fact witnessed before. For example, an arbitration claim may involve allegations of anticompetitive conduct in the aluminium market. A natural temptation would be to quantify damages to a purchaser of aluminium using the same sort of Cournot model that one would use in a proposed merger between two aluminium companies. A key input to the model would be a demand elasticity, which may come from a statistical study that considered a period of time where aluminium prices were relatively stable. The elasticity study may suggest that each 1 per cent reduction in prices would stimulate consumption by 0.5 per cent. The estimate can be reliable if extrapolated to situations where aluminium prices change modestly, but we would urge caution in the use of such a model if the arbitration involved a claim that a cartel of aluminium companies overcharged by 30 per cent. A more reliable approach may be to model the economics of the different uses of aluminium, and to assess whether a 30 per cent price reduction would fundamentally change the economics of the existing uses, or perhaps could even stimulate new uses or attract a significant supply-side response.
Another concern is that the assessment of damages often requires a focus on a specific company, while the most common use of Cournot models and their variants is to assess the typical market-wide parameters of concern in merger proceedings. Economic models certainly have the tools to estimate company-specific impacts; the concept of elasticity extends beyond markets to individual companies as well. Market elasticity reflects the intuition that higher prices in a market reduce total consumption. Similarly, the elasticity of an individual company’s product reflects the notion that, if an individual company acts alone in raising prices, then it will lose market share to its competitors. When populated with elasticities for specific companies, sophisticated models can supplement estimates of market-wide impact with estimates for specific companies.
Formal economic models often draw upon the elasticity estimates available in independent published academic papers. Many papers look at market-wide effects, but few papers estimate the specific elasticities confronting individual companies. In any given arbitration the expert will not likely be able to draw upon an independent published paper to derive elasticity estimates for a particular company. Moreover, differences in quality and brand image across companies can complicate the measurement of elasticities for specific producers. It can be impractical to develop an elasticity estimate from scratch for a specific company, or the data may not be readily available. Practical considerations can, therefore, counsel for a different approach, in which the expert does not construct and run a large formal market model, but assesses the sales of a specific company in the absence of disputed conduct by considering multiple sources of information, and studying separately the likely responses of competitors and customers.
A third concern is that certain economic models have a built-in bias to distribute impacts proportionately among specific companies. An economic model may offer to estimate a market-wide solution, supplemented by predictions for specific market participants. However, some models contain algorithms that embody an internal inertia: for the sake of simplicity the model may decide that the company with the largest market share would gain the most sales as aggregate demand for the product increases, while smaller companies would grow proportionately less. At first glance, it may seem reasonable to presume that effects on companies will be proportional to their existing market shares, but such a presumption can actually be perverse. If the disputed conduct caused a decline in a particular claimant’s market share, then the claimant’s small size could unwittingly trick a model into thinking that the claimant would have continued to remain small in the absence of the disputed conduct.
Given the limitations of formal economic models, the best approach to damages will often require a balance between two approaches. The expert should have familiarity with the standard tools and concepts of competition economics, which come predominantly from competition proceedings involving proposed mergers and investigations of disputed conduct. However, the expert should also maintain a keen awareness of the limitations to the standard tools and concepts. A challenge is to determine when standard approaches face limitations related to the differences between typical competition cases and the particular arbitration at hand. If necessary, the expert should be prepared to depart from formal economic models in favour of a more business-oriented approach, which looks heavily at accounts, internal financial projections and other factors to consider closely the claimant’s costs and competitive advantages or disadvantages.
When determining market shares, it is important to adopt a dynamic focus. A static analysis would simply impute a value to lost market share, as if the claimant had lost the share permanently. However, if the arbitration itself puts an end to the disputed conduct, then the claimant may eventually catch up with the market share that it could have reasonably anticipated in the absence of the disputed conduct. If so, then an appropriate damage analysis should look forward over time, and consider that the lost market share will diminish with the passage of time, and eventually cease as the claimant fully regains the same position it would occupy in the hypothetical competitive scenario. It would, therefore, be reasonable to measure future damages that diminish over time; the challenge is to determine the pace of the claimant’s catch-up.
With respect to catch-up, the tables reverse with respect to allegations of speculation. A respondent may ordinarily feel inclined to levy accusations of speculation, but the respondent has a natural incentive to support a damage estimate that looks into the future to project a rapid catch-up rate that reduces the total damage estimate. A respondent can legitimately complain that the greater speculation lies in a fully static analysis that refuses to contemplate future changes. If the claimant simply imputes a value to lost market share, as if it had lost the share in perpetuity, then the calculation contains an internal contradiction: the analysis presumes that disputed conduct has caused a loss in market share, without considering that the cessation of the disputed conduct could lead to a restoration of market share. We can imagine reasons why a claimant’s lost market share may in fact be irreversible, but a sound analysis of damages should be able to identify and articulate the reasons.
Calculating overcharges in price-fixing and monopolisation cases
The specific claim confronting an arbitration tribunal may be that the respondent has charged an excessive price to the claimant, either as a result of the respondent’s participation in a cartel or as an abuse of its dominant position. As with the discussion of market shares above, the calculation of damages requires the analysis of an alternative scenario that never occurred: one in which the respondent charged a reasonable price.
If the members of an arbitration tribunal are not themselves experts in competition law, they may look to prior decisions by competition authorities for guidance. However, tribunals should recognise limitations to the precedent. Competition commissions often have access to detailed data concerning the costs of companies involved in price-fixing and monopolisation. A competition authority may feel compelled to mandate the provision of information concerning costs, and to measure the extent to which a dominant company’s prices exceeded its costs. In the well-known United Brands case, the court found the European Commission deficient for failing to assess the dominant company’s underlying costs. However, it would be a mistake to conclude that arbitration tribunals must make any damage award for excessive pricing contingent upon a demonstration that the respondent’s prices exceeded its underlying costs.
The court in the United Brands case rejected the European Commission’s analysis because an analysis of costs would have been useful in that particular case, and because the Commission had the power to compel the production of costs, but did not do so. The situation in an arbitration can differ significantly. First, a tribunal might not feel that it has the power to compel the production of the respondent’s underlying cost data. Second, depending on the particular case, it may be possible to reach robust conclusions concerning the excessive nature of an incumbent’s prices without knowing its underlying costs precisely.
To illustrate, we discuss a hypothetical example in which the respondent’s costs would not actually provide useful insight into excessive pricing. Consider a city whose market for commercial real estate has vast excess capacity, defined as significantly more office and retail space than demanded. During a phase of excess capacity, a competitive market would generate market prices that are significantly lower than the rental rates that would recover an efficient company’s costs plus a return on the construction of buildings. In fact, a normal competitive market would send a rational economic signal to building developers, to stop constructing new buildings. The rational economic signal is by definition a market price that does not recover costs plus a reasonable return. The failure to recover costs should lead the market to halt new construction until demand recovers and occupancy rates bounce back. After eliminating excess capacity, a competitive market would finally justify increases in rental rates to levels that recover costs and a reasonable return. However, if a monopolist controls all of a city’s commercial real estate, then the monopolist can abuse its position to bypass the normal cycle of price declines during periods of vast excess capacity. During such a period it would be absurd to absolve the monopolist of excessive pricing based on a determination that its prices were in line with the costs of constructing buildings. If one could identify a comparable city with a competitive market structure and the same amount of excess capacity, then the rental rates in that city would provide better insight into the question of excess capacity. In contrast, looking at the monopolist’s costs would only cloud the picture and risk inaccurate conclusions. Similarly, comparing a monopolist’s prices to its costs could also risk inaccurate conclusions during a situation of tight capacity in the commercial real estate market. During such periods, even a competitive market could charge prices that significantly exceed underlying costs.
Given that underlying costs may at times only cloud the picture, it is useful to reflect on the reasons why it has become common for European competition authorities to use underlying costs as a benchmark for excessive pricing. As explained above, part of the reason is that competition authorities have the ability to compel defendants to provide their cost data, and there is a natural temptation to analyse all available data for the sake of rigour. However, another reason may be that in many cases of dominance, the competition authorities do not have any obvious benchmark that would make them comfortable bypassing a cost analysis. They do not have the luxury in the example above of identifying another city with a competitive market structure that is sufficiently comparable.
However, private arbitrations can present significantly different types of cases compared to those that often confront a competition authority. A large number of international commercial arbitrations involve commodities. Commodities naturally lend themselves to arbitration owing to the prevalence of long-term contracts, which provide useful tools for obtaining committed supplies. The supplier often lies in a different country to the purchaser. Coming from different countries, the supplier and the producer may each hesitate to accept the jurisdiction of the counterparty’s courts for fear of bias, so they may agree to an arbitration clause in search of a neutral forum. At the same time, long-term commodity contracts often raise complex commercial and legal issues that lend themselves to specialisation, and arbitration offers an opportunity for the parties to submit disputes to more specialist adjudicators than often found in national courts.
Since the early 2000s, many different commodities have witnessed significant improvements in the development of liquid trading. Trading now generates natural competitive price benchmarks for many products, ranging from metals to energy. At the same time, international commodity markets often rely heavily on large infrastructure investments for production, transportation and delivery. Infrastructure investments are often large and long-lasting, and take significant time to develop. The supply of many commodities, therefore, reacts more slowly than the demand, so commodities are particularly prone to cycles of excess capacity and scarcity, which make underlying costs a poor benchmark for assessing competitive price levels, as discussed above by reference to the illustrative example of commercial real estate.
Infrastructure constraints can simultaneously permit the emergence of regional pockets of market power. The European natural gas industry provides a notable example. Parts of continental Europe have multiple buyers and sellers that participate in liquid traded markets for natural gas, which generate competitive price signals. Other markets in Eastern Europe still face questions regarding dominance by a sole natural gas producer, as evidenced by the European Commission’s recent investigation of Gazprom. Pipelines link Eastern and Western Europe, but the constraints and high costs of pipeline infrastructure can in fact continue to segregate markets into distinct geographic areas. The European natural gas industry, therefore, presents a situation that rarely arises in the competition cases confronted by European authorities: to the west there are competitive markets with liquid trading that provide natural competitive benchmarks, while to the east there are some markets that still face questions of dominance. The liquid trading to the west likely provides more insight into competitive price levels than the underlying costs of a producer in Eastern Europe. If the issue of excessive pricing arises in arbitration proceedings over a long-term European natural gas contract, it can, therefore, make sense to rely on the competitive benchmarks as opposed to examining the underlying costs of the dominant gas supplier. Similar issues are bound to arise with respect to other commodities that have liquid trading hubs in particular geographic areas, with constraints that can still lead to market power problems in other areas.
The existence of a competitive benchmark does not immediately remove all challenges to measuring the extent of monopolistic pricing. The price generated by a competitive commodities exchange tends to involve a separate geographic market than the one in dispute – the exchange reflects effective competition by buyers and sellers, while one company’s dominance or the existence of a cartel reduces competition in a distinct geographic market. Since the exchange will not reflect trading in the same geographic area as the disputed conduct, the analysis of damages should determine whether and how to adjust the traded prices. Depending on the circumstances, the correct answer may involve adding or deducting transportation costs, or considering other economic variables that can distinguish geographically separate markets.
A second challenge is whether the respondent sells the product subject to other terms and conditions that render it fundamentally distinct from the contracts traded in a commodity exchange. Long-term commodities contracts tend to be relatively complex and tailor-made to specific situations. In contrast, the trading on commodities exchanges tends to involve standardised contracts. The seller pursuant to a complex long-term contract may claim that its tailor-made provisions render the product more valuable than indicated by prices on a commodities exchange. A reasonable damage analysis will consider the possible merits of such claims, and whether an appropriate adjustment would be to raise or lower the exchange traded price when deriving a competitive benchmark for the calculation of overcharges.
We conclude that to measure damages from competition claims in arbitration proceedings can require approaches that differ from the typical assessment of market shares and competitive prices in merger proceedings, and that differ from the assessment of excessive pricing by competition authorities. Analytical tools and approaches have developed over time to provide useful insight in competition proceedings, and an understanding of existing practice contributes to an expert’s assessment of damages from competition claims in arbitrations. However, equally essential is an awareness of the limitations to the common tools and approaches, given differences in the issues that can arise in arbitrations and those confronted by competition authorities.
 Carlos Lapuerta is an expert in economic analysis and financial valuation and Richard Caldwell is an economics and financial expert at the Brattle Group.