Recent Developments at DG Competition: 2015/2016

The Directorate General for Competition at the European Commission enforces competition law in the areas of antitrust, merger control, and state aids. In this year’s review of the recent economic work undertaken at DG Competition we focus on a theme in each of these three areas. For merger control, we review the application of both the theoretical analysis and the empirical techniques to bidding competition in the GE/Alstom case. In terms of antitrust enforcement, we review the economic issues that were raised by an Article 101 decision with respect to information exchange in the market for container shipping. Finally, with respect to state aid we provide an overview of the sector inquiry in electricity markets that targeted capacity mechanisms, as means to ensure security of supply in the European Union.

application of both the theoretical analysis and the empirical techniques to bidding competition in the GE/Alstom case. In terms of antitrust enforcement, we review the economic issues that were raised by an Article 101 decision with respect to information exchange in the market for container shipping. Finally, with respect to state aid we provide an overview of the sector inquiry in electricity markets that targeted capacity mechanisms, as means to ensure security of supply in the European Union.
Keywords Antitrust Á Mergers Á State aid Á Auction theory Á Bidding analysis Á Exchange of information Á Electricity markets Á Capacity mechanisms 1 Introduction: Main Developments in 2015/2016 The 2015/2016 period was another busy time for the application of economic techniques to competition enforcement at the Directorate General for Competition (''DG Competition'') of the European Commission (the ''Commission'').
On the merger front, most of the advanced economic analysis carried out by the Commission was applied to in-depth merger investigations (so-called ''Phase II'' investigations). During the 18 months between January 1, 2015, and June 30, 2016, the Commission assessed 16 Phase II cases. These 16 Phase II cases included ten clearances with remedies, three abandonments, two unconditional clearances, and one prohibition. 1 A range of economic techniques was applied to the Phase II cases that were reviewed by the Commission during 2015 and the first half of 2016. By the way of illustration, this included: the application of numerical merger simulation techniques in three telecoms cases (Orange/Jazztel, TeliaSonera/Telenor and H3G UK/Telefonica); 2 the development of techniques to assess concentration in geographically differentiated markets (Cargill/ADM, and Ball/Rexam); 3 the application of bargaining theory in a vertical setting (Liberty Global/DeVijver); 4 and the use of bidding analysis (Siemens/Dresser Rand; GE/Alstom; Staples/Office Depot; and Halliburton/Baker Hughes). 5 In this article, we review in some detail the Commission's bidding analysis in GE/Alstom.
In terms of antitrust 6 the same period saw the Commission fine eight cartels as well as a dominant company for abusing its position. Furthermore, in two cases the Commission accepted commitments that meant that the companies changed the behaviour that the Commission was investigating without, however, the companies admitting that the behaviour was anticompetitive. Another such commitment decision (''Container Shipping'') that was taken in early July 2016 was particularly interesting from an economic point of view and is therefore described in some detail in this article.
Finally, for what relates to state aid, in 2016 the Commission delivered the final step in a long process that started with the launch of State Aid Modernisation by publishing the ''Notice on the Notion of Aid''. This guidance as to when public spending falls within, and outside, the scope of EU State aid control is intended in particular to facilitate public investment in the European Union by helping Member States and companies to design public funding in ways that do not distort competition.
Another priority of the Commission in the area of state aid in the last few years was how to tackle harmful tax competition. In 2015 the Commission found that Luxembourg and the Netherlands granted illegal selective tax advantages to Fiat and to Starbucks, respectively, and also concluded that the Belgian Excess Profit scheme was illegal under EU state aid rules. Most recently, in August 2016, the Commission determined that Ireland granted undue tax benefits of up to €13 billion to Apple and ordered the recovery of those tax benefits.
The energy sector represented another focal point in the Commission's enforcement with respect to the area of state aid. Particular attention has been devoted to the assessment of public measures that have been put in place to ensure capacity adequacy that is needed to ensure the security of supply in electricity markets: ''capacity mechanisms''. The work in this area encompassed both individual cases and especially a sector inquiry, which is the first of this kind in state aid.
In this article, we review in some depth three specific examples of the use of economics in DG Competition's work, each relating to the one of the three competition instruments that were applied by DG Competition: For merger control, we review the application of both the theoretical analysis and the empirical techniques to bidding competition in the GE/ Alstom case. As part of its investigation of the merger, the Commission performed a detailed assessment of competition between bidders in tenders for large industrial gas turbines that are used for electricity generation. In terms of antitrust enforcement, we review the economic issues that were raised by an Article 101 decision with respect to information exchange in the market for container shipping. This decision is important because it indicates the basic principles that firms should follow when they intend to communicate their future prices publicly. Finally, with respect to state aid we provide an overview of the contributions of the sector inquiry in electricity markets in the assessment of capacity mechanisms for security of supply under State aid rules.

Factual Background
In September 2015, the Commission cleared the acquisition of the energy businesses of Alstom by General Electric (''GE''), subject to conditions. 7 The Commission found that the transaction, as initially notified, created significant horizontal overlaps in the supply of heavy duty gas turbines (HDGTs) for gas-fired power plants. 8 HDGTs are typically supplied through lengthy and complex tenders.
In large parts of the world, including the European Economic Area (EEA), the frequency for power generation is 50 Hz, while it is 60 Hz frequency in the Americas and some parts of Asia and the Middle East. The Commission's investigation therefore focused on the markets for the supply of HDGTs (which includes sales and servicing of HDGTs) that operate at a 50 Hz frequency. 9 The Commission found that the market for 50 Hz HDGTs was a highly concentrated market with only four full technology players: the two market leaders (GE, Siemens), followed by Alstom and MHPS. 10 Power output of HDGTs is one of the main differentiating factors. In the EEA and in the 50 Hz worldwide market, most of the HDGTs sold were ''large HDGTs'' (with an output between 200 and 320 MW). Therefore, the Commission focused its analysis of tenders for the supply of large HDGTs in the market for 50 Hz frequency.

Unilateral Effects in Bidding Markets
In its review of the merger, the Commission conducted an extensive review of evidence from recent tenders for large HDGTs (this empirical evidence is summarised below). 11 In order to frame the empirical analysis properly and to interpret its results, the Commission considered it useful also to review in some depth the nature of unilateral effects in bidding markets, and the relevance of the auction design to assess the impact of a merger. 12 The Commission contrasted the nature of merger effects in auctions where bidders have good information on the characteristics of rival bidders and on demand by final customers (which have strong similarities to ''second-price auctions'', and we will thus use this phrase below in describing these auctions), relative to auctions where bidders have relatively imperfect information on some key competitive parameters when making their final offer to a customer (which have strong similarities to ''first-price auctions''). 8 Gas-fired power plants are expected to play an important role in the European energy mix in the coming decades, first as a flexible complement to renewable energies, and second in phasing out the more polluting coal-fired power plants. 9 Tenders taking place in China and Iran, while using a 50 Hz frequency, were excluded from the analysis due to specific market access barriers that lead to different competitive conditions compared to the rest of the world. For the sake of clarity, in the following, reference to tenders for 50 Hz include all countries that use a 50 Hz frequency except for China and Iran. 10 A fifth company, Ansaldo, was active in the market. However, its competitiveness and technological strength were found to be weaker compared to the other four companies. 11 Annex I of Decision in case M.7278 GE/ALSTOM, 8 September 2015 (yet to be published at the time of publication of this article). 12 See Klemperer (2005).

Unilateral Effects in Second-Price Auctions
It is well-established that in a second-price auction to purchase a product, the only constraint on the price paid by the winning bidder is determined by the willingness to pay of the runner-up bidder. This follows from the fact the price paid by the winner under this auction rule equals the bid of the second-highest bidder, and that (as a result) all bidders face incentives to reveal truthfully their willingness to pay. This outcome can be approximated in an ascending price setting, where bidders can observe the offers made by rivals and adjust their bids accordingly over time until all bidders but one drop out. The winning bidder therefore purchases the object at a price that is equal to the willingness to pay (plus a small increment) of the last bidder to drop out of the auction (i.e., the runner-up).
In a procurement setting (which is typically more relevant for merger assessments), rival bidders compete to supply a product, and may differ both in their cost of supplying the product and in the value that they offer to the buyer. If bidders have good (or perfect) information on the cost of supply of all rival bidders and on the value placed by the buyer on each offer, then the outcome of the tender has ''second-price'' features. In particular it should be expected that the winning bidder will be paid a price that allows the buyer to obtain a utility equal to the ''surplus'' (i.e., the difference between the value to the customer and the cost of supply) that is generated by the second-placed bidder. This means that the winning bidder is able to extract as profits the difference in surplus that it creates relative to the second-placed bidder (this follows from the fact that the surplus offered by the second-placed bidder is effectively the outside option for the buyer). 13 In this environment, all other competing bidders face incentives to offer their product at cost, and therefore would earn an expected margin that is close to zero if their bid was accepted by the buyer. 14 In this setting, the only constraint on the winning bidder is represented by the second-placed bidder, and all other competitors are irrelevant to the final outcome. This means that following a merger, only tenders where the merging parties were respectively the winner and the runner-up would experience a price increase (assuming that the products of the two merging parties continue to be both offered after the merger). This price increase equals the difference in the surplus associated with the second-and third-ranked bidders. One way to approximate this price effect is to examine the (average) winning margin of the second-ranked bidder in other tenders where this bidder is the winner, and the third-ranked bidder is the runnerup. 15 If one of the products of the merging parties is suppressed after the merger, 13 See Miller (2014).
14 For the sake of exposition, consider only two firms 1 and 2 with costs c1 B c2 that generate gross surplus for the customer of v1 and v2, respectively, where for simplicity it is assumed that v1 [ v2. In a second-price auction framework, firm 2 ends up bidding at cost-i.e., b2 = c2-and firm 1 wins the tender by offering a bid that makes the customer indifferent between the two competitors. This implies a winning bid of b1 = (v1 -v2) ? c2, and a profit of (v1 -v2) ? (c2 -c1). The profit of the winning bidder is therefore the difference in surplus that is generated by the two competing bidders. 15 See Shapiro (2010): ''Merging firms often claim that certain non-merging firms can and will offer an equally good alternative to customers. Customer evidence can be especially valuable in assessing this claim. There can be some tension between this claim and the presence of significant supplier differentiation. We may test this claim with evidence from procurement events in which the merging then unilateral effects should be expected in all tenders where the suppressed product was either the winner or the runner-up (thus potentially affecting a wider set of tenders).

Unilateral Effects in First-Price Auctions
Merger effects under a ''first-price'' environment manifest themselves in a different way than in a second-price auction, and resemble price effects in ''ordinary'' differentiated product settings.
In a standard first-price (or sealed-bid) auction design bidders are formally paidas-bid (that is, they are paid their final bid if successful) and make private offers to the buyer (which may or may not be updated over time, if there are multiple bidding rounds). The standard assumption in this setting is that bidders make their offers under conditions of imperfect information on the offers that are made by rival bidders and also on the value that is placed by the buyer on each bid (on the assumption that the products are differentiated). 16 Under the assumption of imperfect information, when placing its (final) offer, each bidder faces a trade-off between the probability of its bid being successful and the profit earned if the bid is successful. A higher bid will reduce the probability of being the winner, but it will increase the profit that is associated with a successful bid. The equilibrium bid will optimise this trade-off, similar to ordinary markets with differentiated products where pricing optimises the trade-off between volumes sold and profit margins on each unit sold.
In this setting a merger (by either entirely removing a competitor, or by internalising competition between rival bidders) can be expected to lead to higher bids by increasing the probability of winning faced by each bidder (ceteris paribus). This upward shift in expected sales leads to a (profit-maximizing) higher bid. The effect is larger when the ''probability diversion'' between the merging parties is greater; that is, when the impact on the probability of winning of one merging party is greater if the other merging party increases its bid (similar to a standard ''upwards pricing pressure'' logic). 17 Unilateral effects from a merger should therefore be expected in any tender where each of the two merging parties faces a reasonable probability of winning, and affects in particular the probability of winning of the other merging party (in the case where the products of both merging parties continue to be offered post-merger). 18 In a first-price auction the set of tenders that are Footnote 15 continued firms competed as finalists against these non-merging firms. If they really do offer very close substitutes, one would expect to see relatively low margins in those bidding situations''. 16 If instead bidders have very good information on these variables, then the outcome of the auction resembles a second-price environment, along the lines described above. 17 As in the logic of UPP, higher pre-merger margins are also associated with higher upward pricing pressure. See also Moresi (2010). 18 In the case of product suppression of one of the products of the two merging parties, unilateral effects would affect all tenders where the bid of the suppressed product had an impact on the probability of winning of the other bidders. affected by a merger can therefore extend well beyond the tenders where the merging parties would have been the winner and runner-up (absent the merger).

Implications for Merger Assessments
Auction theory suggests that under some conditions, a reduction in the number of bidders from n to n -1 has the same impact on expected prices under a first-and second-price environment (this result is known as the Revenue Equivalence Theorem). The difference in the auction format manifests itself in the distribution of price effects across tenders: In a second-price auction, the price effects are sharper but more concentrated (i.e., they affect only tenders where the merging parties would have been the winner and runner-up), whilst in a first-price auction the price effects are more modest and also more diffuse. 19 The indifference result that is predicted by auction theory does not imply, however, that in the context of a merger review one can safely abstract from the format and institutional setting of the tenders in the specific market at hand. 20 An understanding of the auction format is still necessary to guide the empirical analysis: for example, by determining how much weight should be placed on available data on the identity of runner-up bidders, and by guiding the analysis of the number of tenders that are likely to be affected by price effects.
The Revenue Equivalence Theorem also suggests that it would be inappropriate in the context of a merger review to ''mix & match'' the predictions of merger effects that are derived from different auction formats: for example, by assuming relatively moderate price effects (such as those typically associated with first-price auctions) whilst at the same time assuming that these price effects will apply only to relatively few bids (e.g., those where the merging parties are the winner and runnerup, as in a second-price auction). A ''mix & match'' of assumptions of this kind is likely to underestimate the overall price effects of a merger.

Qualitative Evidence on Tender Organization in GE/Alstom
In the GE/Alstom case, the Commission undertook an extensive qualitative assessment of the way in which tenders for HDGTs were typically organised, in order to inform its empirical assessment of the bidding data. This qualitative review suggested that the assumptions that are associated with first-price (or sealed-bid) auctions were more appropriate to describe tenders for HDGTs than assumptions that are related to second-price auctions.
Based on the Commission's review, the following facts emerged: Bidders had limited information about the characteristics of rival bids and about the customer valuation of each offer; information about the second-ranked bidder was frequently not available (even after the tender had been awarded) and when it was available it appeared to be subjective (with different firms often having different perceptions on the identity of the runner-up bidder); formal or informal shortlisting down to two final bidders during the course of an auction (before a best and final offer) was not the norm; and the level of ''losing'' margins (i.e., the expected level of margins for unsuccessful bids, in the event that the bid would have been accepted) was not close to zero (as predicted in a second-price environment).
As a result of this qualitative evidence, the Commission focused its core empirical analysis of the tender data on measures of closeness and competitive interactions that are more suited for first-price auctions.

Description of the Data
The Commission and the Parties with the help of their economic advisors engaged in a comprehensive data collection exercise in the GE/Alstom case. For each tender for large HDGTs in countries that use a 50 Hz frequency, data were collected for the period 2009-2014. The following information was collected for each tender: the identity of the companies at different stages (budgetary bid stage, firm bid stage, shortlist stage, runner-up, and winner); the HDGT models that were proposed by the different companies; the country where the tender took place; and the customer name. The Commission also collected information on each tender from databases that were used internally by the merging parties, including information on the expected revenue, cost, and margins that were associated with each bid.

Baseline Empirical Analysis
To assess the intensity of competition between the merging parties and the potential unilateral effects from the proposed transaction, the Commission conducted a number of empirical analyses of the bidding data. These empirical analyses, which are relevant mainly in the context of a first-price auction, are described below: Participation and Win-Loss Analyses First, the Commission analysed how often the merging parties participated against each other in tenders. The evidence indicated that submitting a binding offer for HDGTs is costly since it requires significant preparation costs for the offer and extensive discussions with potential customers. Therefore, only those firms that meet the tender specifications and expect to have a reasonable chance of winning the tender would submit a binding offer. Costly participation is, thus, indicative of product fit, and a participation analysis allows an assessment of the closeness of competition. 21 This was confirmed by the qualitative evidence that was collected by the Commission that showed that most customers invited all of the main players to participate in tenders for HDGTs, and therefore did not limit the set of possible bidders in order to save on transaction costs. The choice of participation in a given tender was therefore made by the supplier of the HDGT, and not by the buyer.
The analysis of participation data indicated that Alstom was the second bidder that GE met most often in tenders in the relevant sample, behind Siemens, but significantly ahead of MHPS and Ansaldo. This finding was also confirmed for those tenders where there was a further shortlisting of firms after the firm bid stage. This analysis indicated that GE and Alstom were close competitors for a welldefined set of bids.
The Commission also undertook a win/loss analysis, which showed that for the tenders where GE participated, it had lost to Alstom a significant number of tenders (fewer than to Siemens, but significantly more than to MHPS and Ansaldo). In other words, Alstom was successful in tenders where it met GE, achieving a significant win rate. This analysis indicated that Alstom was a significant competitive threat to GE.

Econometric Analysis on Winning Probabilities and Margins
The bidding data that were collected by the Commission also allowed a more advanced econometric analysis. As discussed above, in the context of a first-price auction, winning probability is an important element to assess the competitive strength of a bidder. Bidders with high winning probabilities are likely to affect the outcome of a bidding process and therefore the probability of winning of other bidders (and vice versa). A first analysis of the data indicated that GE achieved a significantly lower win rate for tenders where Alstom participated, compared to tenders where Alstom did not participate, which suggested the existence of a positive probability diversion ratio from Alstom to GE. A probit regression analysis, which took into account factors (other than Alstom's participation) that may have affected GE's probability to win a tender (for example tender characteristics, and the presence of other bidders), confirmed that Alstom's participation was associated with a lower winning probability for GE.
Last, the Commission analysed the (expected) margins of GE in tenders when it faced competition from Alstom. In the first analysis, the Commission found that GE's margins at the initial (firm) stage of tenders were lower for tenders where Alstom participates, and even more so at the final stage (that is, for the final bid that was submitted by GE in the context of a given tender process). In other words, the margins of GE were decreasing more from the initial stage to the final stage of the bidding process in tenders where Alstom was a competitor. This analysis indicated that, even if one were to consider conservatively that Alstom has no impact on the margins that are associated with bids at the initial stage of the bidding process (for example, because any difference in margins at that stage of the tender is entirely due to tender characteristics), the presence of Alstom was still associated with a further reduction of margins during the course of the tendering process.
In the second analysis of the margin data, the Commission analysed the expected margins of GE when it participated in tenders (considering only the final bid submitted by GE). The Commission found that GE's margins were significantly lower for tenders where it faced competition from Alstom. This was confirmed in a regression analysis after taking into account factors other than Alstom's participation (for example the tender characteristics, and the presence of other bidders) that may have affected GE's margins.
The Commission found in an additional regression analysis that GE's margins were decreasing when the number of bidders increased in tenders. In particular, GE's margins were significantly lower in tenders with four participants compared to tenders with three participants. This suggested that, even in tenders with four participants pre-merger-i.e., including GE, Siemens, Alstom, and MHPS-anticompetitive effects were likely to be present. This evidence therefore indicated that, even if one were to assume conservatively that MHPS would increase its participation post-merger in tenders where pre-merger only GE, Siemens, and Alstom were participating (thus potentially mitigating that anticompetitive effects of the merger), 22 tenders with four participants pre-merger would still suffer from significant competitive harm. Moreover, as indicated above, the fact the nonmerging parties other than Siemens participated less than Alstom in tenders that involved GE was indicative of a weaker competitive constraint exercised by these firms.
Harm from Transaction (for the Large HDGTs Segment) The quantitative analysis of bidding data indicated that GE and Alstom were close competitors by exerting significant competitive constraints on each other. The higher prices and margins that were charged by GE in tenders where Alstom was not present, combined with the large number of tenders where GE and Alstom competed against each other, supported the finding that the transaction would lead to significant competitive harm to customers.
The Commission also considered additional consumer harm for tenders where Alstom participated, but did not face GE as a competitor. This is because of the serious risk that Alstom's technology might be discontinued due to the merger. The Commission relied on its empirical analysis of bidding data to provide an illustrative estimate of the likely consumer harm from the transaction (abstracting from possible additional harm from the loss of product choice and innovation). The Commission also found that the efficiencies that were presented by the Parties would not be sufficient to offset the illustrative competitive harm to consumers.

Additional Empirical Analysis in the Context of a Second-Price Auction
One important issue that was raised in this case was the relevance of data on runnerup bidders to determine the potential loss of competition from the proposed Transaction. In particular, the Parties argued that potential price effects would be limited only to tenders where the parties were the winner and the runner-up, and the price effects were likely to be contained given the importance of a third credible player for those tenders (as would be the case in a second-price auction).
For the reasons described above, the Commission considered that a second-price auction framework was not appropriate to describe tenders for HDGTs. Nevertheless, the Commission also undertook an additional sensitivity analysis by considering the likely competitive effects in the context of a second-price auction framework. This analysis suggested that even under this alternative economic framework, the proposed Transaction would have led to significant competitive harm.
First, the Commission supplemented the runner-up data that were submitted by the Parties with data that were submitted by customers and competitors. These data revealed the presence of significant discrepancies in the perception of the identity of the runner-up firm, which cast doubt on the assumption of (ex-ante) perfect information that characterises a second-price auction (as discussed above). 23 Moreover, this exercise showed that Alstom in particular was indicated as the runner-up bidder in a significant number of tenders.
Second, as explained above, under a second-price auction format, the third choice for customers is critical to measure the potential price effects from the merger. In order to measure the competitive constraint likely to be exercised by the thirdranked bidder (typically Siemens) for tenders that were affected by the merger, the Commission examined the winning margins of each of the merging parties in tenders where each of the other non-merging parties (typically Siemens) was indicated as the runner-up. The Commission observed significant profit margins for these tenders. This finding indicated that non-merging Parties would not act as a significant constraint on price increases post-merger and therefore one could expect significant price increase for those tenders where the merging parties were the winner and the runner-up. 24 Last, the risk that some of Alstom's HDGTs models would be discontinued postmerger was important in assessing the competitive effects of the transaction in a second-price auction. With product discontinuation, anti-competitive effects spread to all tenders where the producer of the discontinued product could have expected to be the winner or the runner-up, independently of the ranking of the other merging party. The significant number of tenders where Alstom was the runner-up bidder was therefore indicative of significant competitive harm from the proposed Transaction.
Overall, while under the assumption of a second-price framework the number of tenders affected was lower than in a first-price auction framework, the potential price increases were higher. The Commission also computed an illustrative level of consumer harm under a second-price auction, finding that it was similar to the upper bound of the estimated consumer harm under the baseline first-price auction framework. These findings are broadly in line with the predictions of auction theory. 23 Ex ante information refers to information that was available to the bidders during the tender process before submitting a final binding offer. 24 If a particular bidder was an equally good alternative to GE or Alstom, we would have expected to see relatively low winning margins for GE or Alstom when they competed as finalists against this particular bidder.

Outcome of the Case
After an in-depth investigation of the merger, the Commission concluded that the acquisition of Alstom by GE would have significantly reduced competition, in particular for Large and Very Large HDGT. This conclusion partially rested on the economic and quantitative analysis of bidding data reviewed above.
The Commission ultimately approved the GE/Alstom merger, subject to the divestiture of key elements of Alstom's HDGT business to a smaller competitor (Ansaldo). The divestiture included in particular the technology for Large (GT26) and Very Large (GT36) turbines, existing upgrades and pipeline technology for future upgrades, a significant number of Alstom R&D engineers, two test facilities, and long term servicing agreements for several GT26 turbines sold by Alstom in recent years.
With the remedy, the Commission sought to replicate Alstom's role in the market as an independent firm, maintaining competition both in the short to medium run (in terms of price competition and product variety) and in the long run (in terms of the ability and incentives of the remedy taker to continue to upgrade its products and engage in R&D).

Container Shipping
In this section we report on a case 25 that involved the exchange of information that the Commission closed after the parties engaged in commitments that addressed the competition concerns that were expressed by the Commission in its preliminary assessment. 26

Factual Background
The case concerned 14 container liner shipping companies (''carriers'') 27 and their public announcements (on their websites, via the press, or other means) of intended future increases of freight prices.
Container liner shipping services-i.e. regular, scheduled services for the transport of containers by ship-are organised by trade routes between a group of ports at one end (e.g., Shanghai-Hong Kong-Singapore) and another group of ports at the other end (e.g., Rotterdam-Hamburg-Southampton). The Commission investigation concerned trade routes from Far East Asia to Northern Europe and the Mediterranean (westbound).
The Commission found that between 2009 and 2015, the shipping companies engaged in frequent public ''General Rate Increase (GRI) Announcements'': announcements that indicated the amount of the intended increase in US dollars per transported container unit (twenty-foot equivalent unit, or ''TEU''), the affected trade route and the intended date of implementation. 28 Typically, a company would make a GRI announcement about three to five weeks before the intended implementation date of the price increase, by stating its intention to increase prices on a given route by a certain amount, as of a certain date. In the following days and weeks, other parties (not necessarily all of the companies involved) would announce in turn intended price increases-often of a similar magnitude-for the same route and with a similar or identical implementation date. Such 'announcement rounds' took place some 7-12 times every year in the period that was investigated by the Commission.
It is important to note that the GRI announcements were not binding. Indeed, sometimes they were implemented by (some of) the carriers, but more often they were not: The increases could be implemented only partially, or sometimes could even be followed by a price decrease. In some (rare) cases, a company may also make a new announcement with a different GRI within the same round.
Interestingly, during the investigated period the parties never made announcements of price decreases, although the market was hit by the financial crisis and the subsequent global trade recession and prices were considered to be generally depressed (and indeed actual prices frequently fell).

Economic Analysis of Price Intentions: Private Announcements
The Commission had concerns that such GRI announcements did not provide full and binding information on new prices to customers, but merely allowed carriers to be aware of each other's pricing intentions and facilitated collusion among them.
To understand these concerns better, it may be helpful to take a step backward and consider the economic effects of exchanges of information that take the form of the disclosure of future price intentions in an oligopolistic market.
First of all, there is no doubt that if firms privately exchange information about future intended prices the consequences will be anticompetitive. This would be tantamount to rivals meeting in a room and discussing prices.
Consider the following example. A firm may want to increase prices of its product from $100 to $150, but it does not immediately apply the change to customers; rather, it reveals its intention privately to its rivals to 'test the waters'. In turn, the rivals can use unilateral announcements to signal their own preferences. For instance, they may also all think in the same way, and privately signal their intention also to increase the price to $150. In this case, it is clear that an agreement has been reached and that it can now be safely implemented: Buyers can now be told that the new price of $150 will apply. 28 Some Parties stated that they stopped publishing GRI announcements following the initiation of the Commission proceedings. If instead some of the rivals think $150 is not the optimal price, they may make private unilateral announcements to signal their preferences, and such messages may take place until there is convergence to a new price: say, $120. Only when it is manifest that this will be the new collusive equilibrium price, will the price increase from $100 to $120 be implemented with buyers.
Even though it consists of a series of individual, unilateral, and private (in the sense of not disclosed to customers) announcements, this process is effectively identical to the overtly collusive case of all of the rivals' meeting together in the proverbial 'smokefilled room', and coming out of it when the agreement has been reached. 29 In both cases, what happens is that the rivals manage to coordinate on a collusive outcome without ever experimenting with the market. If they could not talk to each other, or send messages to each other about price intentions, firms would have to operate under uncertainty, and before converging on a new price a price war may take place. For instance, suppose that after a demand shock, a firm perceives that $150 should be the new collusive price and sets the prices accordingly. If a rival responds by setting a price of $120, this may be perceived as a 'deviation' by the former firm, which may react by triggering a price war, which benefits consumers (and hurts the industry). Or perhaps, precisely because it is afraid that other firms may not want to increase prices (perhaps because they may have different perceptions about demand), the firm may not raise prices in the first place-also to the benefit of customers.
Therefore, whether the firms talk openly to each other, or arrive at an agreement through a series of unilateral private announcements of future intended prices, the effect is anticompetitive: They are able to adjust prices without the risk of losing customers. Nor can one expect any efficiency gain from such exchanges of information.

Economic Analysis of Price Intentions: Public Announcements (and Their Commitment Value)
In the container shipping case, announcements were not private but public. In principle, making prices transparent on the demand side is pro-competitive: The more precise is the information about prices, the more that buyers will be able to shop around and compare different offers without having to incur search costs. In turn, this will lead competing firms to reduce prices in order to attract customers. True, publicly announced prices mean that transparency exists also on the supply side (and we know that collusion will be enhanced when firms can observe each other's move, as they can spot deviations and promptly punish them). But, first, firms will generally try to set in place mechanisms to identify the prices that are charged by rivals, 30 so the counterfactual to publicly announced prices may well be 29 This process clearly does not guarantee that collusion will be sustained. The process described here concerns the coordination aspect of collusion, not its enforceability aspect. 30 In most markets, consumers will typically find investing to monitor market prices too costly and timeconsuming: one would need large transactions to make worthwhile the collection of such information. But for firms, the benefit from observing each other's price may be significant, and well worth the investment. And in markets where there are negotiated transactions between buyers and suppliers, the latter will try to engage with the former in order to be informed of rival's prices. Sometimes, suppliers are also themselves prices that are observed more-or-less perfectly by firms but very imperfectly observed by buyers. Second, empirical evidence shows that markets that are characterised by price transparency result in lower prices than do markets where price information is scarce on the demand side. 31 However, when price announcements are public, but do not carry commitment value (i.e., buyers cannot buy at the announced prices), then the situation becomes the same as in the case of private announcements, with firms able to exchange price signals while being sheltered from market uncertainties. In the above example, there would be little difference between firms' unilaterally and privately announcing future prices, and firms publicly announcing their intended prices if nobody can buy at those prices-at least until a new collusive equilibrium has been found. 32 Therefore, not only private announcements of intended prices, but also public announcements without commitment value should be regarded as anticompetitive.

Application of the Economic Principles to the Case
In the container liner shipping case, and in line with the economic analysis made above, ''the Commission raised the preliminary concern that GRI announcements have no committal value but are mere intentions of future pricing behaviour, upon which customers may not be able to rely. The time period between the announcement of GRIs and their actual implementation allows the shipping companies to revise the GRIs. The Commission raised the preliminary concern that whether the announced price increases are implemented or not may depend on the reactions of the other shipping companies, and that shipping companies may even occasionally issue new GRI announcements increasing their initial announcements.'' 33 The preliminary concerns that were expressed by the Commission were also in line with the European case law and the Commission Horizontal Guidelines, according to which the exchange of information about companies' intentions concerning future prices is particularly likely to lead to a collusive outcome and is considered a restriction of competition by object. 34 When the conduct is considered 'by object' the Commission does not need to analyse its actual effects, but still has to verify that it does not give rise to efficiency gains. In this case, the Commission in Footnote 30 continued purchasing (to diversify or complement their inputs when they are vertically integrated, for instance; or to satisfy an unexpected excess demand), which gives them additional insight about the prices that are charged by rivals. 31 See Motta (2004)-Section 4.2.2.2, and references therein. 32 This was the situation of the well-known Airline Tariff Publishers case, described in Borenstein (1999). 33 See Commission Decision at para. (43). 34 See references in the Horizontal Guidelines, paragraphs 73 and 74. See also the Court of Justice, which in the Chiquita case stated that ''an exchange of information which is capable of removing uncertainty between participants as regards the timing, extent and details of the modifications to be adopted by the undertakings concerned in their conduct on the market must be regarded as pursuing an anticompetitive object''. Dole v European Commission, C-286/13 P, ECLI:EU:C:2015:184, paragraph 122. its preliminary assessment discarded such efficiencies because consumers' decisions were not helped by the GRI announcements, both because they did not contain information about the final price of the freight service and because they did not have commitment value.

Economic Analysis of Price Intentions: Public Announcements with Seeming (But Not Real) ''Commitment Value''
Finally, the case at hand reveals another possible source of competition concerns: when the announced prices are public and consumers could in principle buy at those prices, but the announcement takes place at a time when no (or very few) buyers would make orders. Consider the example of airline tickets: Airlines post price on their websites for certain flights and times, and consumers can buy tickets at those publicly announced prices. Consequently, these announcements are informative and carry commitment value (note also that nothing prevents the airlines from modifying their prices as demand evolves-and indeed this is how airlines set prices); hence, we would conclude that these price postings do not raise competition concerns.
But suppose now that airlines post prices for certain flights that will take place 15 years from now. In principle, one could book seats on those flights, but few consumers are likely to do so? This would be an instance in which formally the announcement is public and contains a commitment to potential passengers; but in reality it is easy to imagine that such announcements may be used primarily to signal desired price changes to other airlines-much in the same way as a private price announcement would. And of course, in this hypothetical example, after the exchange of price signals on tickets 15 years from now has resulted in a new understanding, nothing prevents the airlines from withdrawing the 15-years-ahead price.
We are not suggesting that in the case at issue the carriers had organised such a complex scheme, but we provide the above example to explain why the Commission was worried that the GRI announcements may be made at a time when shipping orders would typically not yet be made.

The Remedies Offered by the Parties
As mentioned at the beginning of the section, the case was closed after the parties offered (and the Commission accepted) the following 'commitments' (that is, modifications in their conduct that would eliminate any competition concern). 35 First, the carriers will stop publishing and communicating General Rate Increase announcements. They will be free to make (or not make) public announcements about future prices; but if they do, they cannot limit themselves to communicating that the general price will increase in absolute or percentage terms, but they will have to announce a price that includes at least the five main elements of the total price: base rate; bunker charges; security charges; terminal handling charges; and peak season charges if applicable). These five elements represent about 90 % of the full price, and hence give customers a good basis to compare prices. 36 They should also indicate: which other charges may apply; the services to which they apply; and the period to which they relate (this can be either a fixed period or open-ended, in which case the announced prices will be valid until further notice).
Second, price announcements will be binding on the carriers as maximum prices for the announced period of validity (but carriers will remain free to offer prices below these ceilings).
Third, price announcements will not be made more than 31 days before their entry into force, which corresponds to the period when customers usually start booking in significant volumes (typically, customers plan their shipments between four weeks and one week before they need to move their consignments). This commitment reflects the risk discussed above that announcements that take place at a time when customers do not make orders may give competitors insight into each other's future prices while not being useful for customers since they are not booking yet, thus possibly leading to price coordination among the carriers.
Finally, these commitments will not apply to: a) communications with purchasers who already have an existing rate agreement in force on the route to which the communication refers; and b) communications during bilateral negotiations or communications that are tailored to the needs of specific identified purchasers.
These commitments were designed not to be too intrusive with respect to the business choices of the carriers. Indeed, the Commission leaves the carriers free to decide: whether to make public announcements or not; whether the prices are valid for a specific period or not; to lower (but not increase) prices within the period of validity of the announcement (if the period has been predetermined); to change (increase or reduce) the price if the period of application is open-ended; and to make an exception for particular customers with which there are specific agreements.
Although the commitments refer to a particular sector, with this Decision the Commission is effectively sending a signal to all companies of the basic principles that they should follow if they intend to communicate their future prices publicly: Prices should be fully informative, communicated transparently to customers, and be binding, in the sense that buyers must be able to purchase goods and services at the announced price. The last condition does not imply, of course, that firms commit once and forever to a certain price: Firms will be free to make a new announcement and change their price in the future.
In this respect, an illustrative example of a market that is reasonably transparent is the market for airline passenger services: Consumers can buy at the fares that are posted on the airlines' websites, and those fares are valid until further notice is given.
On the contrary, price announcements that are made privately among competitors, or that are public but do not carry commitment value, are likely to be considered anticompetitive and to be treated as an infringement by object of competition law.

State aid Sector Inquiry on Capacity Mechanisms in Electricity Markets
On 29 April 2015, the Commission launched the first sector inquiry under EU state aid rules. It targeted national measures in the electricity sector that are aimed at ensuring that adequate capacity is available at all times to avoid power shortages (so-called ''capacity mechanisms''). 37 Notwithstanding the importance of the security-of-supply objective that is targeted by these measures, the Commission had concerns that capacity mechanisms may unduly favour particular players in the electricity markets or create obstacles to trade across borders. The knowledge that is gathered through this exercise is expected to provide some insight as to both the Commission's enforcement of State aid rules and the Commission's Energy Union legislative proposals. The Interim Report that resulted from the inquiry was published in April 2016. 38 There is no universal definition of capacity mechanisms. Generally, they are a means to ensure the viability of electricity capacity providers by offering them public support, on top of revenues that they can obtain directly from the electricity markets, to maintain the overall capacity that is needed in the system so as to guarantee the security of electricity supplies. This public intervention may be necessary if investment signals that are produced by existing electricity markets are insufficient to ensure that the electricity generation mix and capacity is able to meet demand satisfactorily or, in other words, to ensure generation capacity adequacy.
Capacity mechanisms are not new, and they have been the object of public debate, especially with regard to the market and regulatory failures, as well as the political objective, which may justify them. Section 4.1 summarises different views on those market and regulatory failures.
Recent developments in the electricity sector-in particular, driven by market liberalisation and environmental policy interventions-have given a fresh impulse to the debate on capacity mechanisms. Given that these mechanisms may fall under EU State aid rules, the Commission has already introduced for the first time in the revised 2014 EU Guidelines on state aid for environmental protection and energy (hereinafter EEAG) a chapter that provides brief guidance for the use of public funds in ensuring electricity generation adequacy. In Sect. 4.2 we establish when capacity mechanisms are considered state aid and discuss the main principles that are followed in their assessment. 37 http://ec.europa.eu/competition/sectors/energy/state_aid_to_secure_electricity_supply_en.html. 38 A public consultation gathering opinions of stakeholders with an interest in electricity markets took place after the publication of the Interim Report. The final Report of the sector inquiry is expected to be delivered at the end of 2016. However, the recent proliferation of capacity mechanisms in the EU Member States, and the complexity and variety observed in their design, requires a more indepth examination of the arguments in favour and against these forms of support, beyond the general principles that are in the Guidelines. That was the trigger for launching the sector inquiry. Section 4.3 focuses on the contribution of the sector inquiry in providing further insight as to the assessment of capacity mechanisms for security of supply under State aid rules.

Market and Regulatory Failures that May Undermine Incentives to Invest
Liberalisation and the creation of an internal energy market have been at the heart of EU energy policy since the early 1990s. In the last decade competitive wholesale markets have developed in a large majority of Member States, and cross-border trade has intensified significantly, supported by the implementation of rules for cross border trade that allows for market coupling over most periods of operation. 39 Liberalisation has implied a transition from the central planning of investments in generation and capacity towards decentralised decision-making. On the one hand, investment decisions on generation capacity and on transmission capacity are no longer taken jointly. On the other hand, investment decisions in generation capacity are taken autonomously by private undertakings that operate in electricity generation. This process of liberalisation and regulatory change has also been accompanied by a significant change in the electricity generation mix and, over the last decade, by the impact of the economic crisis. Declining demand and increasing shares of renewables have resulted in decreasing profitability of some electricity generators, especially conventional flexible technologies such as gas and coal power plants. While the shift towards more renewable energy production is, on the one hand, a positive development that has resulted from the decarbonisation process, its impact on security of supply must be taken into consideration and managed. In this context, combined with the general ageing of existing power plants, the question of whether investments in generation capacity will be sufficient to maintain an adequate generation ''fleet'' to meet future demand has gained increasing prominence.
To the extent that low profitability reflects an excess of installed generation capacity, the resulting lower incentives to invest contribute to correct the situation of overcapacity. However, if low profitability is the consequence of market and regulatory failures, then incentives to invest may prove insufficient to maintain adequate generation capacity in the medium and long term. Given the time lags between investment decisions and actual availability of generation, current 39 Market coupling consists in automatically directing cross-border flows according to relative price differences across borders, with flows going from lower-price toward higher-price areas at any moment in time, which leads to price convergence. This is done by transmission system operators and power exchanges according to pre-established rules that govern cross-border flows, without the need for explicit acquisitions of transmission capacity. When interconnection capacity constraints cross-border flows, full price convergence across borders is not achieved, and different price levels prevail on each side of the border. overcapacity does not imply that there is no reason for concern about future security of supply. The latter will depend on whether expectations about future demand and profitability provide the right incentives for investments to be made by market participants with sufficient anticipation.
The economic literature has extensively discussed whether electricity wholesale markets can be expected to generate sufficient incentives to invest to guarantee adequate generation capacity. When this is not the case, a 'missing-money' problem arises: The market proves unable to incentivise sufficient investment because investors fear future revenues will not cover their fixed costs and appropriately remunerate their investments. 40 The missing-money problem is mainly related to the potential inability of electricity markets to deliver sufficiently high prices during periods of scarcity. Wholesale prices in perfectly competitive electricity markets should most of the times reflect the marginal cost of the most expensive generation unit that is being utilised. Instead, when demand for electricity rises sufficiently, leading to a situation of scarcity, market prices should rise above marginal cost in order to reduce demand and allow the market to clear. These transitory prices above operating costs produce margins that remunerate the fixed costs of marginal generating units. Electricity markets, absent capacity mechanisms, rely to a large extent on the rents generated during periods of scarcity to provide incentives for generators to invest in adequate generation capacity. 41 There are, however, factors that may limit the ability of electricity markets to clear spontaneously through high prices and demand contraction. The demand for electricity is often insufficiently responsive to prices, due to currently prevailing technical features of electricity delivery, which leads to situations where demand remains above available generation capacity despite high prices. In such circumstances, some kind of regulatory intervention is warranted: e.g., by rationing demand and administratively setting a price during periods of scarcity.
Economic theory indicates that, under certain assumptions, it is optimal to set a price that reflects consumers' average willingness to pay for not being disconnected when scarcity arises (the so-called value of lost load, VOLL). 42 Administrative price caps that are established by regulators, however, only rarely reflect VOLL in 40 As Joskow (2013) puts it, ''the revenue adequacy or missing money problem arises when the expected net revenues from sales of energy and ancillary services at market prices provide inadequate incentives for merchant investors in new generating capacity or equivalent demand-side resources to invest in sufficient new capacity to match administrative reliability criteria at the system and individual load serving entity levels''. 41 As Cervigni and Perekhodtsev (2013) explain, ''pricing in conditions of scarcity is a crucial element of the wholesale electricity market's design. Since the available generation capacity is far greater than demand in most hours, the competitive market-clearing price very rarely departs from the system marginal cost. Therefore the generating units with the highest variable costs rely on the extremely high prices prevailing during very few hours of scarcity to cover their fixed costs''. 42 As Cramton et al. (2013) explain: ''The market responds to VOLL by building additional capacity up to the point where a MW of capacity costs just as much as it earns from being paid VOLL during blackouts. (…) So at this point the cost of capacity equals the value of capacity to consumers, and beyond this point, consumer value per MW can only decline as the system becomes more reliable. Hence, the VOLL pricing rule causes the market to build the second-best, 'optimal' amount of capacity. This solves the adequacy problem-with help from a regulator''.
practice. It can be a challenge to estimate the VOLL accurately, which would reflect the real consumer willingness to pay for additional security of supply. 43 Regulators may also be reluctant to allow for high VOLL-based prices because it is difficult to distinguish instances of market power abuse from genuine scarcity conditions. 44 There may also be political concerns that relying on high scarcity prices may also entail higher retail prices. 45 Even in the absence of explicit price caps, penalties for imbalance at the moment of delivery can act as an implicit price cap in day-ahead and other forward markets if they are too low, because operators may prefer paying the penalty than paying high prices.
In addition to the inability of prices to rise to reflect VOLL, incentives to invest can be insufficient due to other features of real electricity markets. Generators' expectations about future returns on their investments in generation capacity can be negatively affected by several sources of uncertainty: e.g., increasing intermittency and price volatility, recurrent regulatory reforms, and boom-and-bust cycles. Moreover, the reliability of electricity systems has certain features of a public good and can have positive externalities beyond the electricity market, which lead to the suboptimal provision of capacity and reliability by the market.
Both at national and European levels, efforts are under way to implement better market designs and regulation that is aimed at improving market functioning. There is a general consensus that there exists room for improvement in the efficiency of electricity markets, most notably by: further integrating demand response and renewable generation in electricity markets; increasing cross-border integration; developing more efficient balancing and ancillary services; establishing bidding zones that better reflect underlying transmission constraints; enabling the formation of prices that reflect scarcity; and developing a more stable and harmonised regulatory environment throughout Europe.
These reforms could significantly improve the efficiency of electricity markets and also reduce concerns about the missing money problem. However, either because they may take time or because they may be insufficient to solve fully the lack of incentive to invest, regulators in many countries have implemented or planned additional measures to ensure generation adequacy. These measures typically take the form of capacity mechanisms that offer complementary remuneration to generation and demand response availability. 43 See Cramton, Ockenfels, and Stoft (2013). 44 As Joskow (2008) explains: ''Unfortunately, the supply and demand conditions which should lead to high spot market prices in a well-functioning competitive wholesale market (i.e. when there is true competitive 'scarcity') are also the conditions when market power problems are likely to be most severe (as capacity constraints are approached in the presence of inelastic demand, suppliers' unilateral incentives and ability to increase prices above competitive levels, perhaps by creating contrived scarcity, increase)''. 45 As Besser, Farr, and Tierney (2002) claim: ''In theory, energy and ancillary service markets alone can provide incentives for investment in electricity supplies. However, they can only do this by subjecting consumers to price volatility, price levels, supply shortages, and a level of risk to reliability that costumers and policymakers would find unacceptable''.

State aid Rules for Generation Adequacy
The revised 2014 EEAG provided for the first time criteria for the assessment of whether capacity mechanisms are in line with state aid rules. 46 However, the question of whether the EU state aid rules are applicable to such support measures in the first place are not addressed there.
The way to assess the existence of state aid has been established through extensive case practice and jurisprudence that are based on Article 107(1) TFEU. This Treaty provision establishes that in order for a measure to be classified as a state aid it must fulfil four cumulative conditions: (a) It must be funded by or through state resources; (b) It has to provide an economic advantage to the beneficiary, favouring certain undertakings or the production of certain goods; (c) It must have the potential to distort or threaten to distort competition; and (d) It should have an effect on trade between Member States. 47 If at least one of these conditions is not met when a capacity mechanism is established, then it falls outside the scope of the EU State aid enforcement.
If public support for ensuring generation adequacy is found to constitute state aid, the Guidelines set out six common assessment criteria that must be fulfilled in order for the measure to be declared compatible with state aid rules. Although these general criteria or principles are common in the assessment of aid measures in all sectors, individual sector Guidelines include further detailed specifications as to the particular conditions that must be fulfilled in a certain market.
First, with respect to the compatibility of state aid for security of supply, the EEAG require Member States to establish the objective of a common interest to be pursued by the public measure, by clearly defining when and where the generation adequacy problem is expected to arise, among others.
Second, it has to be demonstrated that the aid is necessary to fulfil this objective: i.e., why the market itself is not able to deliver the socially desired level of adequacy. Establishing a market failure is not an easy task and it requires a close look at all alternative market-based measures that may address the apparent market failure. For example, alternative generation options, demand-side response, as well as foreign participation through interconnection should all be taken into account when analysing the potential of the market to deliver an adequate level of security of supply.
Third, the state aid measure should be appropriate. In the particular case of generation adequacy, this translates into the requirement that the aid should remunerate exclusively the commitment of being available to deliver electricity and not the sale of electricity in the market.
The fourth and fifth conditions in the compatibility assessment refer to the socalled incentive effect and proportionality of the aid. The aid should provide the right incentives to capacity providers to act in the direction that is set by the 46 See Section 3.9. of the Guidelines (Aid for Generation Adequacy) at http://eur-lex.europa.eu/legalcontent/EN/TXT/PDF/?uri=CELEX:52014XC0628(01)&from=EN. 47 The case practice and jurisprudence regarding the issue of state aid existence have been very recently incorporated in a guidance document that was published by the Commission on ''The Notion of State Aid'' (2016). objective and furthermore, the amount of aid should not exceed what is necessary to achieve this objective.
Last (but not least), capacity mechanisms must be designed to keep potential distortions in competition and trade at a minimum. In particular, the following factors should be taken into account: openness of the measure to all technologies; participation of demand-side, storage, inter-connectors, and foreign capacity; participation of a sufficient number of players to ensure a competitive outcome; and protection of the internal market rules. A restriction in participation with respect to capacity mechanisms may be justified in cases where the objective so requires: e.g., nuclear generation may not be suitable if the capacity adequacy problem requires flexible generation to back-up a renewable ''park''.
Finally, since the EU energy policy encompasses (in addition to the security of supply) a decarbonisation objective, the EEAG allow for a preference for lowcarbon generators in case of equivalent technical and economic parameters.
The Commission has already applied these Guidelines in the 2014 Decision concerning the UK capacity market as well as in two Opening Decisions with regard to the French market in 2015. 48

Preliminary Insights from the Sector Inquiry
The Commission's Sector Inquiry has gathered information on capacity mechanisms to examine in particular whether and how they are able to ensure sufficient electricity supply without significantly distorting competition or trade in the EU Single Market. The various technological, economic and regulatory developments that were described in Sect. 4.1 have led many Member States to consider measures that are aimed at ensuring generation adequacy, by granting support for the mere availability of capacity. As part of the Sector Inquiry, the Commission requested information on a representative sample of eleven Member States that have capacity mechanisms in place or are considering them: Belgium, Croatia, Denmark, France, Germany, Ireland, Italy, Poland, Portugal, Spain, and Sweden. 49 Defining and delineating a particular capacity mechanism is not easy. The Commission has used a classification that, based on some common features observed in the existing capacity mechanisms, distinguishes six broad categories according to three criteria: (a) the exogenous variable that is chosen in the design of the mechanism to quantify the adequacy needs (volume versus price); (b) the coverage of the measure (market-wide versus targeted); and (c) the level of public intervention in setting the capacity target in market-wide mechanisms (centralised versus de-centralised).
The presentation of a few examples of capacity mechanisms in place or planned in the EU may shed some light on the differences between these types: • The UK has put in place, as approved by the Commission in 2014, 50 a ''central buyer'' mechanism, where the total required capacity is set centrally and then procured through a central bidding process in which potential capacity providers compete so that the price is determined by a competitive auction. This was set in response to a general and imminent problem of adequacy that was identified by the UK as applying from 2017/2018. The ''central buyer'' is therefore a volumebased, market wide, centralised scheme. • In 2015, France notified two capacity mechanisms to the Commission. 51 The first one is a ''tender for new capacity'' in Brittany. Typically, the beneficiary of such a tender receives some type of financing support for the construction of a new power plant with certain specifications in a certain location. France proposed this measure following indications that there was insufficient capacity in Brittany to meet increasing demand and that network constraints impede relying on electricity from other areas. A ''tender for new capacity'' is therefore a volume-based, targeted measure. • The second measure that was proposed by France is a volume-based, market wide, de-centralised scheme. The capacity mechanism that was designed by France responds to a problem that is caused by high thermo-sensitivity of the demand. Electricity consumption is very sensitive to temperature changes, especially during winter periods in France, which creates very high (albeit not very frequent) spikes in demand. France has identified the risk that market players may not have sufficient incentives to maintain or invest in electricity capacity that would run almost exclusively during such cold spells, as these winter conditions are rare and unpredictable. The ''market-wide decentralised'' mechanism in France is designed on two pillars: On the one hand, capacity providers receive capacity certificates; on the other, suppliers are obliged to procure certificates in proportion to the peak consumption of their customer portfolio. The difference with the UK market-wide mechanism is that the French capacity scheme does not rely on a central body to set the amount of capacity that will be sold in the auction, but imposes an obligation on suppliers to trade with generators bilaterally in auctions that are organised regularly. • Another relatively common category of capacity mechanisms are the ''strategic reserves''. These are capacity reserves that are kept outside the electricity market most of the time and are called to run only in specific conditions of scarcity. consumption. These types of mechanisms are usually called interruptibility schemes and the sector inquiry found that they are present in at least six Member States. They may be seen as a particular type of strategic reserves, as most of the time they do not participate in the market-only when certain scarcity conditions arise.
This set of examples illustrates the variety of approaches that are envisaged by national regulators to address concerns that are distinct regarding both their underlying causes and their geographic and temporal reach.
The Sector Inquiry has shown the importance of assessing in the first place the need for a capacity intervention through an assessment of existing market and regulatory failures and, consequently, of a potential missing money problem. The inquiry has shown that such generation adequacy assessments are not systematically conducted by Member States and when they are, methodologies vary significantly. Common criteria for the assessment of generation adequacy across the EU would give guidance to public authorities that have not yet developed their own assessments and would facilitate cross-country comparability. Assessments should identify the causes of the missing money problem, quantify its likely impact on the system's reliability, and characterise the type of capacity inadequacy that the system is expected to suffer, possibly including the amount and nature of lacking capacity, as well as its timing and location.
When identifying a missing money problem and the need for intervention, public authorities are explicitly or implicitly setting a desired level of system reliability that they want to achieve. From an economic perspective, a reliability standard that reflects consumers' willingness to pay for security of supply (i.e. consumers' VOLL) is in principle the optimal level of security of supply that regulators should want to achieve. Exceeding it would lead to higher costs without delivering value for money, unless some positive externalities of higher reliability can be objectively identified. Transparency about the estimates of VOLL, the assessment of possible externalities and the targeted reliability standard is important to be able to judge the need and the appropriateness of a capacity mechanism.
The sector inquiry has also contributed to the identification of a number of principles that can guide regulators when designing efficient capacity mechanisms. For example, the design choices of every capacity mechanism should lead to a selection of beneficiaries that are best placed to deliver the required capacity availability at minimum costs. Eligibility criteria that are as open as possible and allocation processes that are as competitive as possible can contribute to the efficacy and proportionality of the measure. Obligations and penalties that are imposed on beneficiaries are key to provide them with the right incentives to act in line with the aimed objective, once the mechanism is in place.
These relatively general principles often require complex adjustments in the mechanism design to allow for the effective participation and competition of certain types of capacity-notably demand response, newly built generation, and foreign capacity-alongside domestic existing generation. Demand response, for instance, can struggle to commit its availability with the same anticipation as generation capacity, which may justify the inclusion of shorter time frames. Uncertainty about future market conditions can inhibit investment in new capacity more than in existing generation capacity; this is something that regulators may want to compensate through the choices of mechanism design (e.g., offering longer contracts for new capacity). The participation of foreign capacity in capacity mechanisms is an important element in the process of cross-border integration of electricity systems in Europe and requires efficient rules for the relative remuneration of both interconnection and foreign capacity, to ensure that optimal incentives to invest in interconnection and capacity are given.
Finally, capacity mechanisms should be designed to minimise distortions in the functioning of electricity markets, especially to avoid increasing concerns of exercise of market power and undermining price signals for short run operation. For instance, the rules that govern the activation of a strategic reserve should ensure that it is triggered only when prices that reflect consumers' willingness to pay are not sufficient to balance the market. Otherwise, the reserve would constrain the ability of prices to reflect scarcity, which would aggravate rather than address the missing money problem.
Based on these considerations and the knowledge gathered through the sector inquiry regarding the numerous capacity mechanisms in place or planned in the EU, the Commission identified certain capacity mechanisms that are, prima facie, more likely than others to contribute to the security of supply in the most effective and least costly way. First, capacity mechanisms that are based on competitive procedures are more likely to deliver efficient outcomes than are those that are based on administrative price setting (i.e., ''capacity payments''). Second, amongst the market-based models, some mechanisms seem better placed to address more immediate or transitional problems (e.g., strategic reserves or interruptibility schemes), while others appear more suitable for long-term adequacy problems (e.g., central buyer or decentralized obligation mechanisms).
Overall, the sector inquiry has shown that capacity mechanisms cannot be a substitute for competitive electricity markets. While capacity mechanisms can be justified when a residual missing money problem is identified, they cannot replace the reforms that are pending to make electricity markets more efficient. These reforms will allow as much as possible the use of the potential of competitive markets to deliver reliable electricity efficiently to the benefit of final consumers. When needed, capacity mechanisms can contribute to provide additional investment incentives and to maintain security of supply. In order to fulfil this objective, they must address well-identified concerns and be appropriately designed.