Ewa Mendys-Kamphorst, CEG Europe

This is an extract from the third edition of GAR’s The Guide to Damages in International Arbitration. The whole publication is available here

Quantification of harm from cartels

The main direct effect of a successful cartel is raising the price for the goods or services sold by cartel members to their customers. If the claimant uses the cartelised goods as inputs, the cartel has – all else being equal – a direct negative impact on the claimant’s profits by increasing the price of these inputs and thereby the claimant’s costs. This price increase is known as an overcharge, which leads to overcharge harm. Less commonly pursued is compensation for damages for purchases that did not take place as a result of the direct purchaser adjusting its purchasing volumes. The source of harm is hereby the foregone profit on units no longer sold.

The negative impact on the harmed party’s profit can be reduced if, in response to the increased cost, the purchaser who uses the cartelised goods as inputs, increases the prices it charges its own customers. This is known as the pass-on effect. However, an increase of the purchaser’s own price will likely lower its sales (as some customers will reduce their purchases in response to a price increase), thereby decreasing its profits. This partly offsets the pass-on effect and is known as the volume effect.

These three main effects of a cartel on the claimant’s profits are illustrated in Figure 1. It is assumed that a cartel raises the claimant’s unit cost of production from c0 to c1, which in turn leads to an increase in the claimant’s own sales price from p0 to p1, resulting in a decrease in quantity sold from q0 to q1. The total harm is equal to the overcharge harm, minus the pass-on effect, plus the volume effect. Note that while the focus here is on cartels, the effects will be similar for any unlawful agreement that increases the prices above the counterfactual level, including some types of vertical agreements such as resale price maintenance.

Figure 1: Illustration of the main elements of the harm from cartels

In what follows, it is discussed how the three main potential effects of the cartel – the overcharge harm, the pass-on and the volume effects – can be estimated.

The overcharge harm

Estimating the harm from overcharge requires comparing the actual price paid by the purchaser during the cartel period, with the counterfactual price it would have paid absent the cartel, and multiplying the difference between the two (i.e., the estimated per-unit overcharge) by the volume of goods purchased by the claimant (volume of commerce). It is important to realise that the overcharge cannot be estimated based only on the actual or current prices; it is necessary to consider prices in the counterfactual scenario as well. Table 1 shows a hypothetical example of estimating annual harm from overcharge resulting from a cartel of apple producers, suffered by a claimant who purchases apples as an input to produce apple juice. It is assumed that the cartel was active between 2009 and 2014, and that the cartel raised the prices by 10 per cent.

Table 1: Estimating the overcharge harm

Actual price (euro/tonne) 374 385 374 374 369 330
Counterfactual price (euro/tonne) 340 350 340 340 335 300
Volume of commerce (tonnes/year) 13,000 13,500 14,000 14,500 15,000 15,000
Overcharge harm (euros) 442,000 472,500 476,000 493,000 502,500 450,000

Below, the estimation of the volume of commerce and of the counterfactual price is discussed.

Volume of commerce

Determining the volume of (affected) commerce may seem straightforward, but in reality it can be a time-consuming and demanding task. The claimant needs to be able to prove that it indeed made the purchases that it is claiming damages for. This will often require collecting the claimant’s invoice data on purchases made from the cartel members and analysing them carefully to determine which purchases were potentially affected by the cartel. This includes, among other things, identifying the time period when the affected purchases were made, product types, geographical areas or channels of distribution. Failing to delineate the scope of affected purchases properly may result in an under- or over-estimation of the harm from overcharge. For instance:

  • If the cartel only affected sales conducted through a specific distribution channel, claimants who purchased the goods through other channels may not have suffered harm, or only suffered to a lesser degree, and then using the sum of all purchases in the cartel period as the volume of commerce may overestimate the harm.
  • The cartel may have had an impact on prices even after it formally ended, for instance because prices are renegotiated infrequently and the price set during the cartel period applies until the next negotiation. In that case, limiting the volume of commerce to the purchases made during the formal cartel period will underestimate the harm.

Once the volume of commerce has been determined, the overcharge harm can be estimated by multiplying it by the estimated overcharge. While the actual price will often be known, both to the purchaser and the cartel member (at least if firms have stored the data on purchases made during the cartel period), the counterfactual price is unknown and will need to be estimated. In some cases, internal documents of cartelists may be available that provide details on the agreement, including any price increase agreed between the cartelists. However, this only reveals the attempted price increase, which the cartel may not have achieved in reality. In most cases the counterfactual will need to be estimated based on the data obtained from the claimant or on market data.

The counterfactual price

Once the relevant volume, geography and period is identified, the evaluation of the harm from overcharge involves determining what prices would have been but for the cartel. Besides using a theoretical model of oligopolistic competition, a number of useful (statistical) methods have been identified for estimating the ‘but-for’ prices. The approach chosen depends largely on the characteristics of the competition law infringement as well as the availability of data. On one end of the spectrum there are very simple methods, such as comparing simple price averages before, after and during the cartel. For these methods the data requirements are very low, but strong conditions are required to make it a credible approach. On the other end of the spectrum, structural models from industrial organisation theory are used, which are rather demanding in terms of data requirements. Finally, the more frequently used methods can be found in the middle of the spectrum and include a multivariate regression approach, the ‘difference in differences’ approach, the ‘yardstick’ or ‘benchmark’ approach, and cost-based or finance-based analysis.

Each of these methods has advantages and disadvantages and their application is highly dependent on the specifics of the case.

There are three main types of methods that are typically applied to estimate the counterfactual price: comparator-based methods, market simulations and financial analysis.

Comparator-based methods

Comparator-based methods use data on transactions of products or geographic markets unaffected by the cartel as a benchmark for the counterfactual. Common sources of comparison are: (1) historical time series of prices of the cartelised good in the same geographic market (e.g., the ‘before, during and after’ method); (2) prices of the cartelised good in a different but similar geographic market not affected by the cartel or of a related product not included in the cartel (the yardstick method); or (3) combinations of the former (e.g., the difference-in-differences method).

The before, during and after method compares the prices in the cartelised market outside the cartel period with those during the cartel period and attributes the difference to the impact of the cartel. If data from before the cartel are not available, as may be the case with long-lasting cartels, then the prices during the cartel can be compared with those after the cartel (a during and after method), although it may be more difficult to separate the impact of the cartel from the impact of other market developments.

The yardstick method compares prices in affected markets with prices in other geographic or product markets that are similar but not affected by the cartel. The advantage of the cross-sectional yardstick comparison over the temporal comparison employed in the before, during and after method is that it does not suffer from uncertainty over the actual infringement period. It also implicitly controls for common time trends (such as seasonality) that explain price changes both in the infringed and in the comparator market. However, the conditions in the comparator market must be sufficiently similar (i.e., cost structure and patterns, demand characteristics) to get valid results, and, crucially, the cartel must not have had any impact on the comparator market. In particular, in cases where cartelists reached a broad global understanding to restrict competition, there might be substantial uncertainty as to the geographic scope of the infringement.

A more sophisticated variation of the comparator-based methods is the difference-in-differences approach, which compares the development of the price in the cartelised market over time (known as the treatment group) to the price in a non-cartelised market over time (known as the control group). It therefore combines elements of the temporal before, during and after method and the cross-sectional yardstick method. The overcharge estimate is given by (1) the difference in prices in the treatment group between the pre- or post-infringement and the cartel period, less (2) the difference in prices in the control group between the two same periods. For instance, if during the cartel period prices in the cartelised market increased by 15 per cent, while in a similar non-cartelised market they increased by 5 per cent, then a very simplified version of the difference-in-differences method would lead to a conclusion that the cartel increased prices by 10 per cent.

Depending on the quantity and quality of available data, comparator-based methods can be applied by comparing arithmetic averages of prices, or by conducting econometric analysis. While the econometric method requires more data and is more demanding with respect to the analyst’s time and skills than a simple comparison of averages, it also produces more reliable results, for two main reasons. First, econometric methods can be used to test whether the difference in prices between the cartelised and non-cartelised period or market is statistically significantly different from zero relative to a certain significance level, i.e., whether there is compelling evidence that it was not caused by random price variations. Second, econometric methods are able to control for other factors that could potentially influence prices, such as increased costs or demand. This helps to isolate the impact of the cartel and could lead to conclusions that are not obvious from a casual inspection of the data, such as that the cartel had no statistically significant impact even though prices in the cartel period increased, or that the cartel resulted in a price increase that is statistically significantly different from zero even though the price level during the cartel period was not higher than in the comparator period.

Market simulations

The second type of methods used to estimate counterfactual prices is market simulations. Applying these methods involves mathematical modelling of the affected market based on assumptions regarding the type of competition that would prevail in the market absent the cartel, and on estimates of demand and cost functions derived from market data. The prices predicted by the model are then compared with the actual prices to determine the impact of the cartel. Building a model requires making a number of assumptions regarding parameters such as the price elasticity of demand, industry or firm marginal costs, and structural parameters such as the number of firms competing and the degree and nature of competition between them. This includes assumptions on whether competition takes place on prices (Bertrand) or quantity (Cournot) or on the degree of product differentiation. Through computation one can then determine equilibrium prices that would have been likely to arise absent the cartel. The parameter values such as the demand function and the degree of product differentiation can be derived from econometric analysis, or approximated through event studies and customer surveys.

A disadvantage of simulation models is that they require many assumptions on how the market works, and the results can be sensitive to relatively minor changes in these assumptions. Experts are likely to challenge the appropriateness of the underlying theoretical model, in particular where the industry is dynamic and customers are active.

Financial analysis or cost-based methods

These methods use financial analysis to estimate the counterfactual. The financial performance of either the claimant or the defendant can give indications as to the impact of the cartel. Like with the before, during and after approach, one can compare a firm’s profitability over time, under the assumption that the cartel will have had a clear and measurable impact on the financial performance of both the claimant and the defendant. To measure the financial performance, profitability measures such as the net present value or other business valuation methods can be used. For companies listed on the stock exchange, changes in stock prices can be used as indicators of how given events impacted their profitability.

Another method in this category is the cost-plus method, where the counterfactual price is derived by estimating the cost of the cartelised product based, for instance, on data from a cartel member, and then adding a profit margin that would be expected to prevail in the absence of a cartel. The estimation of the competitive profit margin can then be based on margins in comparable markets. While the advantage of this method is that cost data are often available in the companies’ records, for multi-product firms with a large common cost component it may be a challenge to identify the cost of producing only the cartelised product. In addition, it may not always be straightforward to determine what the competitive margin should be. For example, high-technology industries may be characterised by a low unit production cost, but a substantial margin to recover research and development costs would be expected, even in the absence of collusion.

Choosing the appropriate method

The choice of the most appropriate method depends largely on the availability of suitable data. For instance, if there are sufficient data on the prices before and after the cartel, and on the main cost and demand factors that may have impacted prices, the before, during and after method is likely to be most appropriate. Market simulation methods may be appropriate if no good comparator (in terms of time, geography or product) is available and if reliable assumptions can be made regarding the nature of competition, as well as the demand and cost functions. The advantage of financial analysis is that data are usually available, but it may be difficult to obtain them at the required level of disaggregation, or to make them comparable over time or across firms, which may be needed if financial performance is to be assessed against a benchmark.

Even for the same cartel, the available data will often differ depending on who is performing the analysis. For instance, a cartel member is likely to have access to data on its own costs and will therefore be in a better position to relate price increases to cost movements. On the other hand, a purchaser may have data on purchases from other suppliers that were not cartel members and be able to compare the price charged by cartel members to those charged by other suppliers. Overall, the results of the analyses of defendants and claimants are likely to differ as both parties will rely on data for the part of the market they observe.

The pass-on effect

The harm caused by the overcharge may be reduced if the customer of a cartel increases its own price in response to the increase in the price of the cartelised input. As this effect decreases the size of harm, it will typically be brought up by defendants in cartel damages cases. Depending on the jurisdiction, different rules regarding the burden of proof in demonstrating pass-on may apply. The European Commission’s Directive on antitrust damages actions states that it is the infringer who should be required to prove the passing on.Courts in different European jurisdictions differ in their approaches to the burden of proof regarding pass-on; it remains to be seen how the national case law will develop further in this respect.

The evidence for pass-on can be derived using different methods:

  • First, pass-on can be determined based on the analysis of the purchaser’s internal documents that contain information on the price-setting process. The knowledge of how the firm sets prices in practice will help to determine how an increase in the input price is likely to have impacted the price it sets for its own customers, assuming that the price-setting process itself is not affected by the increased input price.
  • Second, pass-on can be estimated using econometric methods, by examining a statistical relationship between the cartelised input price and the downstream price. Using this method requires the availability of sufficient data on both the cartelised input price and the prices charged by the claimant, as well as other factors that may have impacted the movements of the downstream price.

Economic theory is sometimes used to estimate pass-on based on a qualitative analysis of the characteristics of the market and the cartel. As an example, theory predicts that pass-on is likely to be stronger if the purchaser’s competitors were also affected by the cartel than if they were not (for instance, because they buy their inputs in markets that were not cartelised). If the purchaser’s competitors were not affected, competition from firms that do not experience a cost increase may limit the purchaser’s ability to increase its price. Theory also predicts that a stronger market competition leads to a higher pass-on, as in a strongly competitive market margins are lower and a loss of sales resulting from price increase has a less overall negative impact on the firms’ profits. However, while theory is a useful tool to structure thinking about pass-on and to provide guidance on what evidence to look for, it should not be used as sole evidence, but be supported by factual evidence – e.g., in addition to a quantitative analysis it is informative to interview relevant key personnel and review actual sales procedures.

The volume effect

Even if the increase in input prices is passed on by the purchaser into its own sales prices, it does not imply yet that harm is reduced by the full amount of the pass-on. For instance, if the cartel raises the price by €10, and the purchaser subsequently increases its own price by €10, that does not reduce the harm to zero. An increase in the claimant’s price decreases its sales and profits as some of its customers reduce their purchases in response to a higher price. The size of the volume effect depends on the degree of pass-on, the sensitivity of the demand for the claimant’s products to the price increase (described as the price elasticity of demand), and the profit margin lost on each unit of missed sales. Therefore, estimating the volume effect requires information on these variables. Depending on the market in question, the price elasticity of demand can sometimes be found in public sources, while the profit margin should be available in the purchaser’s records. To estimate the loss of volume, counterfactual estimation methods described earlier can also be used. For instance, the purchaser’s own sales volumes before, during and after the cartel could be compared, accounting for other relevant factors by which they may have been impacted.

Other effects

In individual cases, it may be necessary to take into account other effects that could either reduce or increase the harm. For example, an additional harm from a cartel can arise from what is known as an ‘umbrella effect’. This arises if an increased price charged by cartel members encourages firms that are not cartel members but active on the same market to increase their prices as well. Additional harm could also have been caused if the cartel led to a lower quality of cartelised products, although this effect may be difficult to quantify. On the other hand, harm may be reduced if other mitigating factors are present, for instance if the claimant also sells products that compete with the cartelised products and benefited from the umbrella effect.

Quantification of damages caused by exclusionary behaviour

A second example of antitrust infringements that can give rise to harm and therefore damage claims are those abuses of dominant position that are aimed at the exclusion of competitors from the market. Examples of such practices are refusals to deal, tying and bundling, predatory pricing and anticompetitive rebates. Similar harm can arise from some types of vertical restraints. An important difference to a hard-core cartel as discussed in the previous section is that a successful cartel harms customers, while it may benefit competitors who may be able to profit from the umbrella effect. By contrast, exclusionary practices directly harm competitors while the harm to customers often only arises in the long run, and may not be noticeable at the time of the claim as the abuse may not have lasted long enough to lead to the targeted competitor’s exit. Therefore, most of the claims regarding exclusionary behaviour are brought forward by competitors and not customers. Claims by customers could nevertheless arise, for instance if exclusion indeed already led to the exit of a competitor from the market and the dominant firm raised its prices afterwards.

Exclusionary behaviour harms competitors by reducing their profits. Unlike in the case of cartels, where the single most important effect is an increase of the price of the cartelised products, the mechanisms of harm caused by exclusionary behaviour are more diverse. Depending on the precise type of the infringement, its effect on competitors can be a reduction of their sales volumes, or of the price they receive, as well as an increase in their costs. Predatory pricing or tying may result in the competitor losing sales to the dominant firm, while a refusal to deal may force the claimants to use more expensive inputs and thereby increase their costs. In an extreme case, if a competitor is forced to leave the market or prevented from entry, the actual profits are equal to zero and the harm is equal to the full profits that the competitor would have earned if it were (still) active in the market.

Estimating harm resulting from the reduction of sales because of exclusionary behaviour requires estimating the volume of lost sales, and the profit that would have been earned on these sales, by using the average profit margin. To estimate counterfactual sales in the absence of the infringement, similar methods can be used as for the estimation of the counterfactual price in the case of cartels, for instance a comparison of the claimant’s performance before, during and after the infringement, or the claimant’s performance in other (geographical) markets not subject to exclusionary behaviour.

An additional complication is introduced by the fact that the impact of exclusionary behaviour may reach far into the future, if it leads to market exit or absence of entry without a possibility to enter or re-enter. Extending harm into the future makes it necessary not only to construct a counterfactual scenario for the past, but also expected counterfactual and actual scenarios for the future. This introduces an additional element of uncertainty into the estimates.

In the case of forced exit, a practical approach that can be used to predict future profits in stable markets is to assume that absent the infringement, profits from the period just before the infringement would have continued in the future. In more dynamic markets, forecasting future profits requires making adjustments for possible market developments that will have an impact on profits. If exclusionary behaviour prevented entry, it may be very difficult to determine what the profits would have been absent the infringement, which in some cases may limit the amount of damages that can be claimed to the direct costs that have already been incurred in preparing entry.

Interest on antitrust damages

Making sure that the victim of an infringement is fully compensated for the suffered harm requires that the infringer also pays an appropriate interest on the damages. This takes into account the opportunity cost of capital, i.e., the fact that absent the infringement, its victim would have been able to invest the money and earn a return on its investment. Interest may be calculated using some statutory interest rate, or it can be based on the claimant’s actual opportunity cost of capital. The latter is usually estimated as the weighted average cost of capital, which takes into account both the actual cost of debt and the cost of equity, which are then aggregated using weights based on the actual financing structure of the claimant.

Conclusion

Some arbitration cases may require the estimation of harm suffered from an infringement of competition law. If the victim of an infringement is a firm, the harm can be defined as the decrease in the claimant’s profits resulting from the infringement. To properly identify and understand the effects that need to be quantified, any quantification must begin with a thorough qualitative analysis of the infringement and the market in question. The key element of any quantification is the determination of the counterfactual – i.e., the state of the world and the claimant’s profits in a hypothetical situation without the infringement.

One type of infringement that commonly leads to harm and therefore damages claims is cartels. The most important harm from a cartel is a higher price paid by purchasers for the cartelised goods. The main element of harm is the harm from overcharge. This can be estimated based on the volume of commerce – i.e., the volume or value of purchases affected by the cartel, and the overcharge – that is, the per-unit difference between the actual prices paid by the purchaser and the prices it would have paid in the absence of the cartel. To estimate the overcharge, one needs to determine the counterfactual price. When determining the volume of commerce, attention must be paid to the proper identification of the precise scope of the cartel in terms of the affected goods, geographical territory, distribution channel and the time period. To estimate the counterfactual price, various empirical techniques can be used, which will often be based on comparisons of the cartelised market with some comparator (i.e., a market that is sufficiently similar but not affected by a cartel). The eventual harm can be lower than the harm from overcharge if the purchaser passed on all or some of the price increase to its own customers. Such a behaviour can in turn lead to the volume effect – an additional profit loss resulting from lower volumes sold.

Quantifying harm from exclusionary behaviour follows the same general principles but the focus is on the estimation of the counterfactual volumes or costs rather than the price. Additional complexity is added by the fact that harm from exclusionary behaviour extends into the future. Therefore, both the counterfactual and the actual situation must be forecast, which increases uncertainty about the actual extent of harm.

Subscribe here for related content, breaking news and market analysis from Global Arbitration Review.

Covering commercial and investment arbitration, Global Arbitration Review is the unofficial broadcaster for members of the international arbitration community, keeping them feeling up to date and informed.