Jump to Content Jump to Main Navigation

8 Design of Remedies

Gunnar Niels, Helen Jenkins, James Kavanagh

From: Economics for Competition Lawyers (2nd Edition)

Dr Gunnar Niels, Dr Helen Jenkins, James Kavanagh

From: Oxford Competition Law (http://oxcat.ouplaw.com). (c) Oxford University Press, 2015. All Rights Reserved. Subscriber: null; date: 27 October 2020

Fining Guidelines — Remedies in merger cases

(p. 359) Design of Remedies

8.1  After the Diagnosis, What’s the Cure?

8.1.1  Catchers versus cleaners

8.01  Most of the scholarly literature on competition law, and most of this book so far, has focused on the identification and analysis of competition problems. Much less attention has been paid to the design of remedies for these problems. As one commentator noted, ‘Everybody likes to catch them, but nobody wants to clean them.’1 This gap is increasingly (p. 360) recognized among competition officials and practitioners. Remedies matter a great deal for the effectiveness of competition law enforcement. Spending a substantial amount of resources on an investigation and concluding that there is a competition problem is of little value if a suitable remedy cannot be found. As stated by the DOJ, ‘Without a proper remedy, winning a judgment of a [Sherman Act] section 2 violation is similar to winning a battle but losing the war.’2 It quotes the former FTC Chairman William E Kovacic (1999): ‘Responsible prosecutorial practice dictates that government agencies begin an abuse of dominance case only after they first have defined their remedial aims clearly and devised a convincing strategy for achieving them if the defendant’s liability is established.’

8.02  We use the term remedy here in a wide sense. It includes not only remedial actions to alter the structure of the market or the behaviour of companies, but also the imposition of fines on the perpetrators and the award of financial damages to the victims. Jurisdictions around the world use different combinations of these types of remedy. The appropriate remedy will depend on the type of case. In mergers the simplest remedy is a prohibition of the merger. Not much else may be required, as competition in the market will continue as was. However, there are many merger cases in which the authority does not wish to prohibit the merger outright, instead approving it in exchange for certain commitments by the parties aimed at preserving competition. These commitments can be structural—divesting activities or assets—or behavioural, where the merged entity promises to engage or not engage in certain behaviour. The picture is more varied for cases involving restrictive agreements and abuse of dominance. A simple prohibition may not be enough. Fines are often imposed, serving the dual purpose of punishing the perpetrators and discouraging future infringements. Some jurisdictions take punishment and deterrence a step further and impose prison sentences for hardcore cartel agreements. Many jurisdictions have endorsed the objective of compensation for the victims of the infringements, be they competitors or customers, through private damages actions before national courts (a topic discussed in Chapter 9). Where competition has been harmed by restrictive agreements or abusive conduct, remedies should aim at restoring it.

8.03  This chapter discusses the economic principles that can assist in the design of remedies in merger and conduct cases. Merger control is the area in competition law where the design of structural remedies has received most attention. Competition authorities around the world now have extensive experience with conditional merger clearances, and there is an increasing number of studies evaluating past merger remedies.3 The market investigations regime in the United Kingdom offers useful lessons on both structural and behavioural remedies. These investigations deal with markets in which competition is not functioning effectively, but for reasons other than abuse of dominance or restrictive agreements, and therefore the remedies are often aimed at changing structural or behavioural features of the market.

(p. 361) 8.1.2  Structural versus behavioural remedies: Courts and competition authorities as regulators?

8.04  Competition authorities often prefer structural over behavioural remedies, especially in a merger context.4 Structural remedies can be horizontal or vertical. Horizontal separation involves the sale of assets that compete with each other in the same product and geographic market. Vertical separation means divesting assets in different layers of the supply chain—it is the opposite of vertical integration. Structural remedies can be used to restore competition in the market, or indeed to create more competition than there was before the intervention. Their main attraction in theory is that the authority needs to strike only once: after dissolving a position of market power (or removing a bottleneck problem in a vertical supply chain), it can leave the newly created competitive structure to its own devices without having to worry about ongoing supervision. On the other hand, structural remedies are by nature rather intrusive—the authority is forcing a company to cut off some of its limbs. This may create inefficiencies—we saw in Chapters 6 and 7 how vertical integration can be an efficient solution to co-ordination problems. Moreover, structural remedies may be complex to execute in practice. The authority needs to monitor whether a viable buyer is found within the established time period, and indeed whether the new buyer raises merger concerns itself. For example, in 2010 the European Commission allowed Unilever to buy Sara Lee’s body care business on the condition that it divest the Sanex deodorant business (including all trade mark and other IP rights related to Sanex, customer contracts, production equipment, and key personnel). The acquisition of Sanex by Colgate-Palmolive Group in 2011 was then itself reviewed by the Commission, and ultimately cleared.5

8.05  Behavioural remedies require ongoing or periodic monitoring. This makes them less attractive to competition authorities. A commonly held view in US antitrust law is that ‘judges and juries (and antitrust enforcers) are ill-equipped to act as industry regulators deciding the terms on which a firm should be required to sell its products or services’.6 The US Supreme Court has warned against remedies that require courts to ‘assume the day-to-day controls characteristic of a regulatory agency’.7 There is some logic in this. Sector regulators specialize in the ongoing supervision of an industry, and have specific skills and expertise to do so. Most competition authorities do not. Yet over the years this has not prevented competition authorities from applying behavioural remedies in merger and conduct cases, and taking on the supervisory burden that comes with them. Why is this? First, structural remedies may not always be proportionate. Second, appropriate sector regulations may not be in place in the market in question (or the market has recently been deregulated), which means competition law is then seen as the main tool to keep dominant companies in check. The neat separation of competition law and sector regulation as (p. 362) envisaged in US antitrust law is not always feasible in practice. Competition authorities often end up doing supervisory work akin to what regulators do. Third, many competition authorities seem confident these days to impose behavioural remedies, having gained experience over the years and learned useful lessons from the regulation of networks and utilities. Some competition authorities also have formal sector regulator responsibilities. The Australian Competition and Consumer Commission and New Zealand Commerce Commission have traditionally combined the functions of competition authority and sector regulator. The ACM in the Netherlands was formed in 2013 by merging the competition authority with the telecommunications regulator (with energy and transport regulation already having resided within the former). Most of the sector regulators in the United Kingdom have been given powers to apply competition law to their sector, as has the telecommunications regulator in Mexico that was formed in 2013. The field of economic regulation has developed extensive insight and sophisticated tools that can be of direct use in the design of competition law remedies, both behavioural (access and price regulation) and structural (vertical separation). Perhaps the main lesson from regulation is that effective monitoring of behaviour with regard to pricing and access is by no means impossible—the tools and practical experience exist—but the supervisory structure can be costly to implement and maintain.

8.06  Finally, we note that competition law (especially at EU level) is sometimes used as a stick to force through market changes and thereby support other regulatory policies, such as liberalization, structural separation, and cross-border integration. This chapter contains several examples of merger and conduct cases in the financial services, energy, and telecommunications industries where the remedies imposed sought to contribute to these wider policy objectives, sometimes perhaps even going beyond addressing the identified competition concerns.

8.1.3  The remainder of this chapter

8.07  Sections 8.2 and 8.3 deal with structural remedies in merger and conduct cases, respectively. The latter section also describes regulatory insights on vertical separation. Section 8.4 discusses behavioural remedies in merger and conduct cases and sets out the regulatory principles of access and price regulation. Section 8.5 covers the concept of fair, reasonable, and non-discriminatory (FRAND), which is used in a number of different contexts in access and licensing cases, and is of increasing relevance to competition law. In section 8.6 we explore the lessons from behavioural economics for designing remedies. The Microsoft and Google cases are useful illustrations. Section 8.7 deals with the principles of setting fines, with reference to the theory of incentives (including the incentives of rational criminals) and behavioural economics. The section also covers the use of economics (or possible lack of it) in the European Commission’s Fining Guidelines, including in assessing companies’ ability to pay fines.8 Finally, section 8.8 covers another area where competition law can learn from regulation and public policy in general, namely that of cost–benefit analysis (CBA). In recent years, competition policy has seen an increasing emphasis on the measurement of costs and benefits, not only of specific interventions and remedies, but also of having a competition regime in the first place.

(p. 363) 8.2  Structural Remedies in Mergers

8.2.1  Divestment as the most direct means to restore competition

8.08  Divestment remedies are common in merger cases. Short of prohibiting a merger outright, divestments are the most direct means of restoring or maintaining competition. They are relatively straightforward where existing business units are sold off in their entirety, or the assets can be run with a high degree of independence. Going back to some of the examples of mergers we saw in Chapters 2 and 7, individual cinemas, holiday parks, and supermarkets are not difficult to separate from the rest of the chain they were part of. By the same token, they can be easily integrated into a competing chain which acquires them (sometimes not much more than a change in livery is required). The same often holds for the divestment of products and brands. Many spirit brands have changed hands following conditional merger clearances in the beverages industry, including sixteen different brands when Pernod Ricard acquired Allied Domeqc in 2005 (among which were Sauza tequila, Courvoisier cognac, and Kuemmerling bitter), and Whyte & Mackay Scotch whisky in 2014 when Diageo acquired United Spirits.9 Several pharmaceutical mergers have been approved following divestment of products or treatments in markets where the parties had overlapping activities. In 2015, Novartis agreed to divest two of its cancer treatments when it acquired GSK’s oncology business, and Mylan divested four products in five countries when it acquired part of Abbott Laboratories.10

8.09  In other industries, separating assets or products is less straightforward. Two merging hospitals may find it difficult to divest activities that take place under the same roof. In 2013 the CC prohibited the merger between two hospitals in Bournemouth in the United Kingdom, considering divestment remedies to be infeasible:

We found SLCs in 55 different clinical areas. Each of these would need to be addressed by any remedy. The services affected by the SLCs are not easily divisible from the rest of the merged parties’ operations and it would be impossible to separate all of these services from the rest in order, for example, to divest them while allowing the remaining services to be merged.

We therefore did not find that there were remedies other than prohibition that would be effective.11

8.10  Bank branches, like supermarkets and cinemas, can be sold off relatively easily as a matter of divesting property. However, this may not solve the competition concerns in question if customers keep their current account with the original bank and use other branches or channels (internet, cash machines) to access their account. Italy and Switzerland had experiences with structural remedies for bank mergers before the financial crisis that began in 2007–08.12 In Italy these included branch divestments, but also (p. 364) share divestments and behavioural remedies to deal with the high degree of inter-locking directorates (directors sitting on each other’s boards) in Italy’s banking sector. The 1998 merger between Union Bank of Switzerland and Swiss Bank Corporation created the largest Swiss bank at the time, UBS. The Swiss Competition Commission imposed structural remedies in the form of divestments of a number of subsidiaries and of bank branches in certain Swiss regions. In an evaluation of the remedies seven years later, the authority found that the attempt to attract new competitors in regional markets had failed, mainly because many of the divested branches were in economically unattractive regions, and because shifting customers of the branches to the new buyer worked only to a limited extent. The financial crisis created a wave of bank bail-outs and restructurings, and the European Commission imposed extensive branch divestment remedies on banks across Europe as a condition for approving these rescue measures. In the United Kingdom the Lloyds Banking Group and the Royal Bank of Scotland (RBS) announced plans in 2009 to divest hundreds of bank branches in order to win Commission state aid approval.13 The process of completing the Lloyds divestments took more than five years. The RBS divestments have not been completed yet.

8.2.2  How much divestment, and to whom?

8.11  Basic questions when determining the appropriate divestments are: how many assets must be sold off, and to whom? As regards the first question, the answer is the more the original market structure can be emulated the better. But even if the divested assets represent a smaller market share than that of the smaller of the two merging parties before the transaction, the remedy can be effective if the assets are particularly attractive and are scooped up by a rival with ambitions to grow in the market. This to some extent answers the second question: who should be allowed to buy the assets? The market is generally best placed to decide on this (the keenest buyer with plans to use the assets most productively will usually step forward in a sales process), though authorities will want to avoid the assets being purchased by a competitor which already has a large market share.

8.12  The Fortis/ABN AMRO merger case in 2007 is one where the structural remedy went a step further than seeking to restore competition to pre-merger levels.14 A consortium of the Royal Bank of Scotland, Banco Santander Central Hispano, and Fortis acquired ABN AMRO for €72 billion (the largest ever bank merger in Europe). This raised competition concerns in the Netherlands where ABN AMRO was one of the traditional top three banks, and Fortis (from Belgium) one of its more rapidly growing competitors. The main problem was in the provision of banking and factoring services to commercial customers (factoring is a form of financing where the bank takes over its client’s outstanding invoices to third parties). The European Commission required the divestment of a significant commercial banking division that existed within ABN AMRO, together with a number of business centres and a factoring division. Altogether this divestment package was nearly twice the size of Fortis’s activities in commercial banking before the merger, and it removed the overlap (p. 365) in factoring. The Commission was satisfied that this remedy would solve the competition concerns and that the divested assets were commercially viable if sold to the right purchaser. In addition, however, the Commission determined that the purchaser had to be a bank that was able to serve international corporate customers and that had a presence in the major world and European financial centres. This requirement effectively meant that the assets could not be acquired by one of the smaller Dutch banks, some of which had demonstrated ambitions to grow in the commercial banking market but lacked such international presence. The Commission probably saw the remedy negotiations as an opportunity to promote cross-border integration of European banking markets, a policy objective it had long sought to achieve through various means. In the event, both the deal and the execution of the remedies were delayed, and changed somewhat in nature, when the two banks had to be rescued from collapse in 2008.

8.13  The Ryanair/Aer Lingus case (2013) is an example where arguably far-reaching structural remedies were offered but ultimately rejected by the European Commission. Ryanair tried several times to acquire the other major Irish airline, Aer Lingus. Following a prohibition by the Commission in 2007 and an aborted attempt in 2009, in 2013 Ryanair offered significant structural remedies:15 divestment of Aer Lingus’s operations on forty-three of the forty-six short-haul routes where the two airlines overlapped, and of take-off and landing slots to International Airlines Group (IAG, formed in 2011 after the merger between British Airways and Iberia) such that the latter could operate three major routes between London and Irish airports. Flybe, a British regional airline, was identified as an upfront buyer of the operations on the forty-three routes (and would receive access to the required aircraft, as well as support with publicity and brand awareness). Both Flybe and IAG committed to operating the newly acquired routes for a minimum of three years. The Commission welcomed the ‘fix-it-first’ aspects of the proposed remedies, where upfront buyers are identified and then enter into a legally binding agreement (conditional on the Commission accepting the remedy in question).16 However, it questioned whether the forty-three routes would be a viable business (for example, they would no longer benefit from the connecting passengers on Aer Lingus’s long-haul routes). The Commission also did not consider Flybe to be capable of competing effectively with Ryanair, as it was seen to have limited financial resources (despite Ryanair offering it €100 million in cash as part of the remedy) and little experience in the Irish market. The Commission found that IAG would not place a strong competitive constraint on Ryanair, and (like Flybe) would have little incentive to stay on the new routes after the three years. It doubted that the divestment remedies could be implemented in a timely manner, and ultimately prohibited the merger again.

8.14  As it happened, Ryanair was subsequently required by the UK competition authorities to sell its minority stake in Aer Lingus which it had built up as part of the acquisition strategy (we discussed this in Chapter 7), and in 2015 it was IAG who acquired Aer Lingus—a deal approved by the Commission subject to a number of remedies, including the release (p. 366) of five daily slot pairs at London Gatwick Airport to facilitate entry on routes to Dublin and Belfast.17

8.2.3  Completed mergers: Cleaning up the mess afterwards

8.15  In many jurisdictions there have been policy debates about merger notification thresholds, and indeed whether merger notification should be mandatory or voluntary (the United Kingdom is one of the few voluntary notification regimes). Mandatory notification with relatively low thresholds can be burdensome to both businesses and the competition authority. A voluntary regime relies on self-assessment by merging parties, and in practice most relevant mergers will be notified if the parties wish to have legal certainty and there is a credible threat that the competition authority will investigate the merger even after completion. The CMA (like the OFT before it) has indeed reviewed several completed mergers, as have competition authorities elsewhere. Yet unwinding a completed transaction can be costly and difficult where the two businesses have already been integrated, overlapping staff have been made redundant, and commercially sensitive information has been exchanged.

8.16  The Twin America case is an extreme example where a merger was unwound six years after it took place.18 In March 2009, Coach USA and City Sights formed Twin America, a joint venture (JV) that combined their ‘hop-on, hop-off’ bus tour services in New York City. These services use open-top double-decker buses, and take visitors past the city’s leading tourist attractions and neighbourhoods. Coach USA, using the Gray Line brand, had been the long-standing market leader in New York. City Sights entered in 2005, resulting in vigorous head-to-head competition between the two operators, with tourists benefitting from fare discounts, improved service, and novel ticket packages (e.g. tickets combined with boat tours). However, in 2009 the two companies combined their operations under Twin America, basically creating a monopoly, although continuing to operate both the Gray Line and City Sights brands. The formation of Twin America fell outside the merger notification criteria of the Hart-Scott-Rodino Act, but the DOJ and the New York State Attorney General challenged it in July 2009. The matter was put on hold for a few years because of jurisdictional questions—immediately after the challenge the parties applied to the Surface Transport Board, the federal regulator, for approval of the JV, and only when this application was rejected in 2012 could the DOJ and Attorney General pursue the case.

8.17  Sightseeing in New York through hop-on, hop-off bus tours was considered a separate relevant market. Each year, approximately 2 million visitors use these services, spending more than $100 million. As stated by the prosecutors, it was clear that a hypothetical monopolist would find it profitable to impose a SSNIP because the actual monopolist—Twin America—did in fact raise its prices significantly: the JV’s stated aim was to improve profitability through raising fares by 10 per cent, and the price of Gray Line’s most popular service, the All Loops Tour, increased from $49 to $54. The market was characterized by significant entry barriers, in particular the requirement to obtain authorization from the New York City Department of Transportation for each bus stop location. Gray Line and City Sights each held large portfolios of bus stop authorizations near all of the city’s top (p. 367) attractions. Recent entrants faced persistent difficulties in obtaining authorization for new bus stops from the department (this is an example of the absolute entry barriers we described in Chapter 3). The remedy agreed in March 2015 required Twin America to relinquish to the department the complete set of City Sights bus stop authorizations (around fifty in total). This was expected to create opportunities for new entrants to obtain bus stops at key locations. Twin America could keep the existing Gray Line bus stops, but would not be allowed to apply for new bus stop authorizations for five years.

8.18  In addition to this divestment remedy, Twin America was required to ‘disgorge’ $7.5 million in extra profits it had made through its unlawful formation (the DOJ and New York State shared the proceeds evenly). In US law, disgorgement is meant to prevent unjust enrichment and deter future infringements. In determining this amount, the prosecutors took into account the fact that Twin America had already agreed to pay $19 million to settle a private class action lawsuit that was filed after the DOJ and the Attorney General brought their case. The class on whose behalf the action was brought includes all persons who purchased a hop-on, hop-off ticket between February 2009 and June 2014 (they could claim back up to $20 per ticket).19 In broad terms, if Twin America carried 2 million passengers per year, and each paid $5 extra due to the illegal merger, the total payment of $26.5 million to the prosecutors and the customer class would correspond to just over two-and-a-half years of the extra profits that the JV made. In all, remedying this completed merger took six years, and involved a combination of divestments, a prohibition on future expansion, repayment of extra profits made, and payment of damages to customers via a class action.

8.3  Structural Remedies in Conduct Cases

8.3.1  The classics: Standard Oil and AT&T

8.19  In conduct cases the use of structural remedies has been rare. The two most famous corporate break-ups in the United States—Standard Oil (1911) and AT&T (1982)—took place in very different times and circumstances, but they usefully illustrate some of the economic principles that are of relevance when designing a structural remedy.

8.20  Following a lawsuit by the DOJ, the Supreme Court ordered in 1911 that Standard Oil Company of New Jersey be dissolved and split into thirty-four companies.20 For decades, Standard Oil—co-founded and majority-owned by John D Rockefeller—had dominated the refinement and shipment of oil in the United States, with a market share of 80–90 per cent (Scherer and Ross, 1990, pp. 450–1). It had achieved this position through a combination of superior efficiency (nothing wrong with that of course), the acquisition of more than 120 rival companies (merger control did not exist at the time), and a variety of notorious business practices. The latter included obtaining preferential treatment from railroad companies, unfair practices against competing pipelines, local price cutting at points where competition was to be suppressed, industrial espionage, and the operation of bogus independent companies. It was these practices that the DOJ condemned, and for which the (p. 368) Supreme Court accepted dissolution of the company as the remedy. Reference was also made to the ‘enormous and unreasonable profits’ earned by Standard Oil because of its monopoly power.

8.21  Interestingly, Standard Oil argued in its defence that the companies it controlled in different parts of the United States operated quite independently and competitively (it also seemed to argue that its control of the industry was ‘the result of lawful competitive methods, guided by economic genius of the highest order’). While this argument was not accepted, it did perhaps give the authorities a clue that making those companies truly independent by splitting Standard Oil was a feasible option. Indeed, rather than a single company, Standard Oil was a ‘combination’ or ‘trust’, controlling a host of other companies through shareholding arrangements—Standard Oil and its contemporaneous brethren gave rise to the term ‘antitrust’ when the Sherman Antitrust Act was passed in 1890. It was this ‘combination’ that the Supreme Court ordered to be dissolved so as to ‘neutralize’ its monopoly power. The court considered it insufficient for Standard Oil merely to cease the above-mentioned practices, since the market was already effectively monopolized. The position of the DOJ had been that the existence of Standard Oil, ‘with the vast accumulation of property which it owns or controls, because of the infinite potency for harm and the dangerous example which its continued existence affords, is an open and enduring menace to all freedom of trade, and is a byword and reproach to modern economic methods’. The Supreme Court agreed in the end. While there have been heated debates ever since the ruling about whether the structural remedy was appropriate, the break-up did, with some delay, result in greater competition among the newly created oil companies—these included the likes of (predecessors of) ExxonMobil, Chevron, and ConocoPhillips. As for John D Rockefeller, he was said to have done rather well out of the break-up, as the minority shares he had kept in the resultant companies all increased significantly in value.

8.22  The break-up of AT&T was agreed in a settlement with the DOJ in 1982 and became effective from 1 January 1984.21 AT&T operated the ‘Bell System’, which had a telephone monopoly in the United States from the early twentieth century, and consisted of entities delivering long-distance and local calls, equipment manufacturing, and R&D. In the 1960s and 1970s, new communications technologies began to compete with AT&T at the ‘edges’ of the Bell System. Rival telephony equipment and wireless intercity call services sought to establish a market presence, but were heavily reliant on AT&T for interconnection and compatibility with the rest of the system. AT&T responded to these competitive threats through a range of hostile actions, including refusals to interconnect and accept ‘foreign’ equipment, or to do so only at a high price and with delays. AT&T also heavily lobbied the sector regulator, the FCC, to help protect its monopoly status and thereby preserve the integrity of the telephone system. The DOJ filed a lawsuit against AT&T in 1974 and, after eight years of legal battles, a settlement was reached. The Bell System was to be broken up into a long-distance telephony arm—the new AT&T, which also kept the equipment-manufacturing and R&D entities—and seven separate incumbent local exchange carriers (known as the regional Bell operating companies, or ‘Baby Bells’). This was therefore a combination of vertical and horizontal separation. It may seem paradoxical that such a radical intervention was made at a time when antitrust law in the United States tended more (p. 369) towards the laissez-faire (see Chapter 1). However, the DOJ’s position was that a structural break-up would be the most effective way of introducing competition once and for all, bringing an end to the persistent disputes and need for oversight of AT&T’s behaviour. The court that reviewed the case also questioned whether the FCC was realistically capable of regulating a company the size of AT&T (Scherer and Ross, 1990, pp 462–4).

8.23  What happened next? Competition certainly developed in the long-distance call market (Sprint and MCI were major challengers). Long-distance call rates fell sharply (local call rates increased, because they had previously been subsidized by the long-distance calls). The equipment and R&D entities were not as successful as the new AT&T had hoped, again due to external rivals. As competition developed further in telephony markets over the years, consolidation took place among the Baby Bells. In 2005 AT&T was acquired by SBC Communications (keeping the name AT&T), reuniting it with three of the former Baby Bells. However, this partial reintegration was not seen as problematic given that competitive dynamics in the market had changed significantly since 1982. One thing that did not work so well after the break-up was that extensive legal and regulatory disputes continued. These have related in particular to the behaviour of the Baby Bells, which inherited local monopolies and have frequently been accused of imposing excessive access charges. In Chapter 4 we discussed the Trinko case (2004), which involved a refusal to supply by Verizon Communications, one of Baby Bells (formerly known as Bell Atlantic), and where AT&T was one of the affected competitors.22 The need for extensive oversight by the courts and the FCC has therefore remained ever since the break-up.

8.3.2  More recent examples of structural remedies in conduct cases

8.24  Relatively few structural remedies have been imposed under competition law since the AT&T case, although there are some examples. At one stage during the DOJ’s investigation into Microsoft, in 2000, a district court ordered the company to be broken up, separating its operating system from the production of other software23—commentators spoke of the ‘Baby Bills’ (after Microsoft founder Bill Gates, then the richest man in the world). However, this was overturned on appeal, and the DOJ eventually agreed to a settlement through which Microsoft would share its application programming interfaces with third parties.24 A break-up was off the table. In 2014 it was Google’s turn to be threatened with structural separation. The European Commission opened an investigation into Google’s practices in the online search market in 2010. The parties came close to a settlement (see section 8.6 below) but failed. In November 2014 the European Parliament adopted a resolution calling for the unbundling of search engines from other commercial services (though the resolution is non-binding).25

8.25  Energy is another sector where competition law tools have been used to enforce further liberalization. In 2013 the European Commission closed an abuse of dominance investigation into CEZ, the Czech electricity incumbent, after the company offered to divest significant generation capacity. CEZ had a market share of more than 60 per cent in electricity generation and was vertically integrated. The Commission had been concerned about a number of practices that hindered new entry, such as pre-emptive booking of transmission capacity. It considered that: ‘Transfer of some of CEZ’s generation capacity to a competitor represents a clear-cut solution to the identified competition concerns. Transfer of (p. 370) generation capacity is necessary in this case as no other type of remedy can address the effects of CEZ’s conduct.’26

8.26  Given the variety of remedy mechanisms discussed in this chapter, this last statement by the Commission (‘no other type of remedy can address’) may be somewhat pessimistic. Yet the divestment remedy brought a long inquiry to an end, and contributed to a lower level of concentration in electricity generation in the Czech Republic, which had been a long-standing regulatory objective. Abuse of dominance actions by the European Commission also resulted in structural remedies in the German and Italian domestic gas and electricity markets.27

8.27  In the United Kingdom there have been some structural remedies under the market investigations regime. In Chapter 6 we discussed how in 1989 the authorities ordered vertical separation between brewery companies and pubs.28 A more recent example is the forced divestment of airports by BAA, the company that used to own the main airports in London and Scotland. In 2009 the CC required BAA to sell both Gatwick and Stansted airports (to different purchasers), leaving it with Heathrow and Southampton airports in the south-east of England, and to sell either Edinburgh or Glasgow airport, leaving it with two airports in Scotland instead of three.29 In its market investigation into private healthcare in 2014, the CMA ordered the divestment of one of the two major London private hospitals operated by HCA, a US-based private healthcare group.30 This divestment was aimed at introducing greater rivalry in this area, and thus better outcomes on price, quality, and range of services. The London private hospital market was seen as concentrated and difficult to enter, and HCA was the largest operator, with eight hospitals and a market share of more than 45 per cent by number of admissions and 55 per cent by revenue. To support the remedy, the CMA determined that the relevant insurers were required to roll over their existing contract terms with the divested hospital for at least eighteen months following the divestment.

8.3.3  Economic insights from network regulation: Vertical separation

8.28  Vertical separation has been introduced in a number of network industries in Europe and elsewhere since the early 1990s—among the first were the gas, electricity, and rail industries in the United Kingdom.31 The main aim is to facilitate entry in the potentially competitive layers of the supply chain (i.e. where there is no natural monopoly), and to restrict the incumbents’ incentives to discriminate against non-integrated competitors in the provision (p. 371) of network access. In other network industries where the incumbents have remained vertically integrated, regulators often impose accounting separation, a much milder form of separation aimed at monitoring the pricing and financial performance of the network facility. In the last decade, vertical separation seems to have come back into fashion in Europe as a tool to address persistent discrimination problems where lighter forms of access regulation have failed. Due in part to pressure from their national regulators, the telecoms incumbents in the United Kingdom (BT), Italy (Telecom Italia), and Sweden (TeliaSonera) each introduced a type of ‘functional’ vertical separation. This is not full vertical separation since ownership remains the same, but the network business is set up as an independent unit that offers the same terms and conditions for access to all service providers downstream, including the incumbent’s own retail business—a concept referred to as ‘equivalence’. In the energy sector, the European Commission and national regulators have been promoting structural ‘unbundling’ of the large energy companies, in particular splitting the transmission networks for electricity and gas, which are natural monopolies, from the more competitive activities (generation/production and distribution). We saw earlier how competition law investigations have been instrumental in further driving structural separation in several EU Member States. Debates about vertical separation have also been raging in the financial services sector, particularly as regards the vertical separation or integration of securities exchanges and clearing and settlement systems, and of payment system infrastructure and payment system providers.32

8.29  Over the years, a large body of economic literature has explored the relative merits of separated and integrated structures in network industries.33 Experience has shown that it remains tempting for vertically integrated companies to engage in the vertical leveraging of market power and foreclosure of entrants downstream. Hence, the overarching question faced by regulators has been whether the efficiency benefits of integration (lower transaction costs, the elimination of double marginalization, better co-ordination) outweigh these vertical foreclosure effects. In industries that have not been separated, foreclosure has typically been dealt with through access regulation—by mandating the incumbent to provide third parties with access to its bottleneck facilities on non-discriminatory terms. However, discrimination and margin squeeze disputes have been common in this model (we saw examples in Chapter 4), and it has frequently proved difficult for regulators to monitor cases of discrimination, particularly in relation to non-price terms. For this reason vertical separation between the upstream bottleneck and downstream competitive businesses has again come to be seen as a suitable remedy, just as it was at the time of the AT&T break-up in the United States.

8.30  The main downside of vertical separation is simply that it eliminates the benefits of vertical integration. In theory, an integrated company with market power upstream and downstream will sell to more consumers at a lower price than its separated equivalent—this is the double-marginalization problem of separation that we discussed in Chapter 6. Vertical separation may also reduce incentives to invest or to innovate upstream, because the separated network business would no longer derive any profits from such efforts in the downstream activities. A further argument against separation is that it hinders the co-ordination of investment and production decisions, given that the upstream company no longer has (p. 372) direct contact with end-user demand. Moreover, the process of separation itself comes at a cost. Separation entails a one-off direct cost, as well as the ongoing costs of maintaining the separated structure. These include the reorganization of the existing integrated company and, where ownership is still held in common, the prohibition of certain flows of information between the separated businesses.

8.31  The experience of the rail sector in Great Britain demonstrates the complexity of the co-ordination issues that can arise as a result of vertical separation. The split between the network operator and the train operators (which also exists in several other EU Member States) has proved to be an effective way of ensuring non-discrimination, but it has also led to a lack of both investment co-ordination and of focus on user requirements (and plenty of scope to blame the other entity when things go wrong).34 The energy sector also has features that make vertical separation less attractive. Integrated companies may have greater incentives to construct interconnectors between national energy markets, and may be more able to co-ordinate the operation and scheduling of the various layers in the supply chain (balancing supply and demand at all times is an overriding imperative in energy markets). There are indications that vertically integrated energy companies tend to spend more on R&D investment—they can apply new innovations to a variety of activities, and thus have a better chance of internalizing the benefits of this investment (Markard et al., 2004). However, there are also advantages to vertical separation in the energy sector, and the introduction of competition in certain layers of the supply chain has been successful in many countries.

8.32  Given the potential upsides and downsides of vertical separation, the policy decision of whether to impose such a remedy is inherently complex. A proper CBA is needed. The outcome will depend on the specific features of the sector and the country in question. Sectors such as rail and energy tend to place greater emphasis on investment as a regulatory objective in its own right, whereas in telecoms regulation the promotion of entry and competition is considered of central importance. Another question is where in the supply chain to separate. The main criterion is to split the naturally monopolistic from the potentially competitive activities. Technological change may render a given form of separation obsolete. For example, as the telecoms industry migrates from traditional copper networks to fibre networks, local-loop unbundling, a structural remedy heavily promoted in the early 2000s, has lost some of its relevance. In the ‘next-generation’ networks the local exchanges are no longer the critical interconnection points in the network.

8.4  Behavioural Remedies: Price and Access

8.4.1  Behavioural remedies in competition cases

8.33  Behavioural remedies are frequently applied in both conduct and merger cases, despite competition authorities’ usual preference for structural remedies. In mergers these remedies are often aimed at mitigating the effects of potential bottlenecks. Two recent examples are the acquisitions in the Netherlands and Belgium by Liberty Global, a cable operator.35 In 2014 the company merged the two major cable networks in the Netherlands. One of the behavioural remedies imposed by the European Commission was a commitment to provide sufficient internet interconnection capacity to providers of innovative over-the-top (OTT) services, such as Netflix and Amazon Instant Video, which increasingly compete with TV platforms, for a period of eight years. The Commission also required Liberty Global to remove certain restrictions from its carriage agreements with TV broadcasters, thus allowing these broadcasters to offer their content through OTT services.36 In 2015 Liberty Global was permitted to purchase two major TV channels in Belgium, Vier and Vijf, on condition that it license these channels to other TV distributors on FRAND terms for seven years.

(p. 373) 8.34  In conduct cases the behavioural remedy often goes hand in hand with the prohibition of the anti-competitive practice itself. A company that is found to have abused its dominant position is effectively told to stop and not do it again. A fine may help to reinforce the message and also to deter others from engaging in the same type of conduct. The judgment itself, if published and explained clearly, can become a useful part of the case law. Sometimes not much else may be required from the court or competition authority. In 2014 the English High Court ruled that Luton Airport had abused its dominant position by granting National Express exclusive access to the airport bus station for bus services to and from London for a period of seven years (the bus station being seen as an essential facility for operators of these services).37 This finding of abuse was sufficient for the airport to agree to an access deal with the excluded operator, Arriva, without the court prescribing any of the terms for such access (Granatstein and Niels, 2015). However, in other cases the court or authority will not only want to put an end to the anti-competitive practice but also to ensure that the dominant company behaves differently in future—for example, supplying a product on FRAND terms. In 2011 the European Commission agreed a commitment with Standard & Poor’s in relation to the pricing of US International Securities Identification Numbers.38 These securities identifiers are essential for interbank communication, clearing and settlement, and for reporting to authorities. Standard & Poor’s had been designated by the American Bankers Association as the monopoly numbering agency. The European Commission opened an investigation into what it considered to be unfairly high prices, and Standard & Poor’s agreed to set a cost-reflective price ($15,000 per year, adjusted for inflation) for five years, and to submit a confidential report to the Commission every year on the implementation of the commitment.

8.35  Such behavioural remedies (in the United States they are often referred to as affirmative-obligation remedies) not only require ongoing monitoring, but also careful design to create the right incentives and market outcomes. This is where competition law can learn from the economic principles of price and access regulation, a theme we turn to now. Price regulation arises where there are concerns about excessive prices resulting from market power. Access regulation may also be designed to remedy high prices, but is usually concerned with remedying exclusionary practices by vertically integrated companies.

(p. 374) 8.4.2  Economic insights from network regulation: Price regulation

8.36  The field of network and utility regulation has developed extensive theoretical and practical knowledge of price regulation.39 Here we summarize some of the basic principles that are relevant when considering a pricing remedy in competition law. The two main types of price regulation are rate-of-return regulation and price-cap regulation.

8.37  Under rate-of-return regulation (also referred to as cost-plus regulation), prices are set at a level that reflects the cost of service, and the supplier is allowed a fixed return on its asset base. This asset base is equal to the historical investment in fixed assets, minus what has been deducted for depreciation. For an agreed base period, often twelve months, the supplier calculates its operating costs, depreciation, capital employed, and cost of capital. The regulator audits these calculations and determines the fair return—that is, revenues minus operating costs and depreciation, generating a profit that covers the cost of capital. The certainty of earning a particular return provides incentives for suppliers to invest sufficiently in the service and thus guarantee security of supply. A drawback of simple rate-of-return regulation is that it provides little incentive for efficiency. If a company’s costs of service increase, it can simply raise its price and continue making the same rate of return. Companies may also have the incentive to over-invest in their asset base (referred to as ‘gold-plating’).

8.38  Price-cap regulation involves setting either price or revenue caps for a fixed time period (often three to five years), thereby providing an incentive for the company to reduce costs during this period. The cap itself is set with reference to a reasonable profit level similar to that in rate-of-return regulation, but because the cap doesn’t change over the period, the company retains the benefits of any cost reductions for the remainder of the period. The regulator reviews the cap at the end of each period. If the company has been making high returns, the regulator will reduce the allowed prices or revenues for the next period. If returns are low (and if this is attributable to external cost or demand factors, rather than to inefficiency), prices are permitted to rise. A common form of price cap is the formula RPI – X. RPI is the retail price index, a measure of consumer price inflation. The X captures efficiency targets derived from reasonable capital and labour productivity gains for the regulated industry in question. Under this form of regulation, prices can increase with inflation but are required to decrease by X per cent in real terms per year. This mechanism incentivizes suppliers to improve their efficiency during the price-cap period as they get to keep any efficiency gains greater than X per cent. Incentives to over-invest (or gold-plate) do not arise as this would reduce the profitability of the company. This feature of the price cap may also mean, however, that suppliers have incentives to reduce investments or quality of service below optimal levels in order to save costs.

8.39  A variant of the price cap is the revenue cap. Under the former, a company’s revenues will change in proportion to volumes (every unit is sold at the same price). It will therefore be incentivized to increase volumes in order to raise revenues through the course of a price control period. The downside is that the company is exposed to the risk of reduced profits due to an unanticipated decline in volumes. This risk will be exacerbated when there are high fixed costs, since these costs will be incurred irrespective of the volumes sold. In contrast, (p. 375) under a revenue cap, the revenue that a company is permitted to generate remains constant irrespective of the volume of output. The company has no particular incentive to increase volumes (since price per unit would have to come down if the revenue cap is reached), but nor is it exposed to uncontrollable downward shocks in volume.

8.40  The regulatory burden is high under any of the above mechanisms. Detailed expertise and analysis are required to assess the companies’ costs and asset base, compare these with efficient benchmarks, and determine the cost of capital or reasonable return. Experience with rate-of-return regulation in the United States suggests that extensive supervision is required to avoid the gold-plating of assets. In turn, price-cap regulation often requires additional quality regulation to ensure that investments are not reduced beyond the point where quality is affected. An additional complexity arises where the market is not a natural monopoly but rather one where competition exists, or is being promoted, at the same time as prices of the dominant supplier are regulated. In a natural monopoly, reasonable profits are determined with reference to what capital providers actually invested in the asset base. In potentially competitive markets, the assets of the dominant supplier may have to be valued based on what an entrant would have to invest to replace them. This way, prices are still regulated but not at such low levels as to discourage new entry. Sector-specific regulators have been created to address these institutional and practical challenges of price regulation. Competition authorities and courts have not. The additional cost of expertise and analysis that is required by the authority or court when implementing a price remedy should be considered in the CBA of such a remedy.

8.4.3  Economic insights from network regulation: Access regulation

8.41  Access regulation has been of central importance in network industries. The first step is to decide whether to impose an access obligation. The economic principles behind this are not very different from the assessment of essential facilities under Article 102 (see Chapter 4). In the field of regulation, the focus has mostly been on access to bottleneck and natural monopoly facilities such as rail infrastructure, electricity transmission networks, and last-mile telephone lines. With an access obligation usually comes the need to regulate the access price as well (otherwise access can be notionally offered but made prohibitively expensive). Many of the general principles of price regulation discussed above apply to access regulation as well—capping prices at competitive levels should aim to ensure that the network can earn a reasonable return and has the right incentives to invest and provide high quality. But there are additional policy objectives that are specific to access pricing: to promote efficient levels of entry in downstream markets that are reliant on the network, and to prevent vertically integrated network operators from leveraging their market power.

8.42  A general principle to achieve these objectives is that access terms should be non-discriminatory between integrated and independent competitors. But the price level and structure also matter. Making competition work effectively in the downstream market means stimulating entry, but only by companies that are efficient (note the similarity with the as-efficient competitor test for abuse of dominance cases). Aligning efficient entry and profitable entry is generally achieved by ensuring that access prices are cost-reflective. Some of the main access pricing mechanisms are discussed below. We note here that competition law sometimes, but not always, prescribes cost-reflective access pricing—this is often a matter of judgement (and discretion), not just economics. In Chapter 4 we saw the example of Attheraces, where a cost-plus approach to access pricing was rejected because the input (p. 376) (horseracing data) had significant ‘economic value’ to the purchaser’s downstream operations (a broadcast service) that was greater than the supplier’s cost of producing the input.40

8.43  A first set of access-pricing mechanisms uses marginal costs as the basis. As we noted in Chapter 1, optimal pricing theory states that prices should equal marginal or incremental cost if efficient resource allocation is to be achieved. Incremental cost pricing also often means that prices are neither excessive nor predatory. The question is what to take as the relevant increment—we addressed this in Chapter 4 when discussing cost benchmarks for abuse of dominance cases. Short-run marginal cost is one option. If train operators face the actual marginal cost that their services impose on the rail network, they will be incentivized to utilize network capacity efficiently. A train service will be operated wherever the benefit of that service to the operator (in revenues) exceeds the cost that the service imposes on the network. This holds true up to the point where network capacity is fully utilized and marginal costs become much larger—that is, at that point marginal costs include the costs of network expansion required to supply the next unit. If there is effective competition in the downstream market (train services), retail prices will reflect the network access charges, thereby providing efficient incentives for usage by end-consumers. The downside is that short-run marginal costs do not include fixed costs. In most network industries fixed costs are substantial, and therefore the network operator would not recover its total costs from the access charge. This could negatively affect investment incentives and ultimately lead to sub-optimal network capacity.

8.44  A variant used frequently in regulation, and recognized in competition law as well, is long-run incremental cost (LRIC) or LRAIC pricing, which we explained in section 4. An example of a competition case in which this variant was used is Albion Water, concerning the access price for non-potable water to supply a paper factory in the United Kingdom.41 LRIC pricing allows for recovery of the fixed costs of providing additional network capacity. As the relevant increment becomes larger or is considered over a longer time period, more costs become incremental. Although it deviates from short-run efficient price signals (if there is excess capacity, pricing at short-run marginal costs gives the right signals for usage of the capacity), LRIC pricing can be more efficient over time since it takes into account both short-run marginal costs and the contribution to capital costs of incremental use of the network, thus preserving incentives to invest in efficient network provision. It also allows incremental investments to be directed to those parts of the network where there is an efficient need for them.

8.45  Fully allocated cost pricing is a top-down approach whereby total costs are identified and, for a given volume projection, the access charge is set such that total revenue is sufficient to cover total costs. The European Commission inquiry into excessive charges for access to Belgacom’s subscriber data is one (albeit somewhat old) example where fully allocated pricing was used in an abuse of dominance remedy.42 This pricing mechanism is fundamentally different from marginal cost pricing. Total costs will include an appropriate return on capital and allowances for depreciation, so that this mechanism can provide for appropriate (p. 377) investment incentives. However, since charging is based on average rather than marginal cost, it does not provide efficient usage incentives. One particular problem arises where the network capacity is under-utilized and hence fixed costs are spread over fewer users. This increases the average costs and thereby the access price, thus perversely reducing the number of users even further.

8.46  The efficient component pricing rule is the last access pricing mechanism we discuss here. It is also known as the Baumol–Willig rule (Baumol, 1983; Willig, 1979). Two competition cases where the rule was discussed at length (one in negative and the other in positive terms) in the context of an alleged abuse of dominance are Albion Water and Telecom New Zealand.43 The efficient component pricing rule is applied to situations in which there is a vertically integrated network provider. Its main objective is ensuring efficient entry downstream. The access price is set such that only companies that are as efficient as the incumbent enter. It takes as a starting point the downstream (retail) price charged by the incumbent. Say this is €1.00 per unit. The access price is then equal to that retail price minus the incumbent’s downstream marginal cost per unit. If this downstream marginal cost is €0.20, the access price is €0.80. At this access price, only entrants that have downstream marginal costs of €0.20 or lower get sufficient ‘headroom’ to make a profit (assuming that they cannot raise the retail price above the €1.00 charged by the integrated company). This mechanism is therefore also known as ‘retail-minus’. In general it is relatively straightforward to implement as only downstream costs need to be analysed in detail. It is less strict on the network operator than marginal cost pricing since the retail price is taken as given—that is, there is no scrutiny of whether the €1.00 retail price reflects competitive or monopoly levels. This does not provide for any means to control end-consumer prices if those are set inefficiently high to start with. For example, if the true incremental cost of network access is €0.70, with the downstream marginal cost being €0.20, the competitive retail price would be €0.90 not €1.00. This is sometimes seen as a drawback of the efficient component pricing rule.

8.4.4  Behavioural remedies with side effects: Online hotel bookings

8.47  Competition authorities increasingly seek to resolve investigations into anti-competitive conduct by agreeing a settlement or commitments with the parties, rather than taking the matter to a final decision. We have seen a number of examples in this chapter. This has clear advantages from the perspective of procedural efficiency, and may save significant enforcement costs. The downside is that the merits of the case may not be fully tested or made public. The parties (and sometimes the authorities) may be so keen to end the investigation that they agree to remedies even if they do not agree with the assessment of the problem. A worse situation is where a settlement agreement between the authority and the investigated parties does not sufficiently account for negative side-effects on third parties, such as competitors or customers. An example of such a situation is the online hotel booking case in the United Kingdom, which the CAT ruled on in 2014.44

(p. 378) 8.48  In 2010 the OFT had opened an investigation into the online sale of room-only hotel bookings by online travel agents (OTAs). It was targeted at the two major OTAs (Expedia and Booking.com) and a large hotel chain (InterContinental Hotels Group) but the practices of concern seemed to be widespread in the market. Distribution agreements established that OTAs would offer hotel accommodation at the prices set by the hotels themselves, and would not lower these prices through discounts. The OFT considered this to be a restriction by object akin to resale price maintenance. The restrictions on discounting limited ‘intra-brand’ price competition for the same hotel rooms among OTAs, and between OTAs and the hotels’ direct online sales channels. They also created barriers to entry for new OTAs wishing to grow by offering discounts. In January 2014 (shortly before being merged into the CMA), the OFT closed the investigation through a commitment decision. The parties agreed to remove the restrictions on discounting the headline hotel room rates. However, importantly, OTAs would offer discounts only to customers who had actively signed up to a ‘closed group’, and could not publicize information about the level or extent of discounts outside the closed group.

8.49  Skyscanner, a meta-search company, appealed against this commitment decision, mainly on the basis that such discounting to closed groups would reduce overall price transparency in the market and threaten the business model of meta-search and price-comparison websites. The CAT agreed, and annulled the decision. It held that the OFT should have considered the potentially detrimental side-effects of the commitments. The OFT itself had acknowledged the consumer and efficiency benefits of price transparency in online markets, and the growing importance of meta-search and price-comparison websites in online hotel bookings. The CAT noted that because the OFT had approached the matter as an object infringement it had not gathered sufficient information to properly assess the effects of the commitments. It found the concerns about the negative side-effects on price transparency and the search business to be plausible, even if no empirical evidence of such effects was available at that stage (given that the commitments had not yet been put in place). This case is a useful reminder that the costs and benefits of any remedies must be analysed carefully and in the round, including the effects on third parties.

8.5  Behavioural Remedies: FRAND

8.5.1  FRAND remedies in competition law

8.50  FRAND is a generic formulation for the terms that can be offered under a pricing or access remedy. It is increasingly used in a variety of contexts, both in regulation and competition law. Earlier we saw the example of the Liberty Global/De Vijver merger in 2015 where one of the remedies was to license TV channels to rival distributors on FRAND terms. Other FRAND remedies in competition cases include BSkyB’s obligation to supply access to its technical platform services in the UK pay-TV market, and Microsoft’s obligation to make interface information available in order to allow rival servers to achieve full interoperability with its servers and PCs running on Windows.45

(p. 379) 8.51  The ‘non-discriminatory’ condition is usually interpreted in the same manner as the general criteria for anti-competitive price discrimination under the abuse of dominance rules (see Chapter 4): the access terms should not distort competition between downstream buyers, whether vertically integrated or independent. It does not necessarily mean that all buyers get exactly the same price; there may be justifications for setting differential terms, and if buyers do not compete with each other in downstream markets there is limited risk of a competitive distortion.

8.52  The ‘fair’ and ‘reasonable’ conditions are more difficult to interpret. Some commentators have given separate meanings to what is ‘fair’ and what is ‘reasonable’ (in the United States the concept used to be called RAND, without the ‘fair’, but this is mostly a matter of semantics and US commentators now also commonly refer to FRAND). However, the terms ‘fair’ and ‘reasonable’ are in essence the same, as they both seek to capture the principle that access terms must strike the right balance between facilitating efficient competition and product development in the downstream market, and providing the upstream provider with sufficient incentives to invest in the facility or input. The price and access regulation principles discussed previously in this chapter are of direct relevance here. Like in other contexts, what is fair and reasonable is usually a matter of degree and judgement, reflecting particular trade-offs that must be made—for example, between short- and long-term efficiency, and between cost- and value-based access prices.

8.5.2  FRAND remedies: IP licensing and abuse of dominance

8.53  One context in which the concept of FRAND has gained particular prominence in the last ten years is in the licensing of IP. While FRAND terms for access to physical networks or inputs are commonly set with reference to costs, this is less straightforward for IP licences. Many products in the high-tech and telecoms industries—mobile handsets, memory chips, and game consoles, for example—are based on common technology standards, which are often set by standard-setting organizations (SSOs). Standards usually incorporate a large number of patented technologies. These patents are called standard-essential patents (SEPs) and are valuable—all users of the standard have to pay royalties to the patent holder. Patents that are excluded from the standard tend to be less valuable. Before making a particular patent part of the standard, SSOs often require the holder to commit to license it on FRAND terms. SSOs began to incorporate FRAND principles in the 1970s and 1980s, when industries were becoming increasingly aware of the potential commercial value of technology standards and had to devise practical solutions to agreeing a standard and avoiding problems of hold-up. Hold-up in this context occurs when the manufacturers in the SSO must select a particular technology for the standard, but are concerned that they will subsequently be charged exploitative royalty rates for that technology and will be unable to switch because it has become part of the standard. FRAND is designed to give IP holders a reasonable return so as to preserve their incentives to innovate, but at the same time prevent them from unreasonably benefiting from the additional value that accrues to them from being included in the standard. The use of the FRAND principle has facilitated the development and acceptance of new standards. However, SSOs that require licensing on FRAND terms, such as the European Telecommunications Standards Institute (ETSI) and the Joint Electronic Device Engineering Council (JEDEC) in the United States, have never really come up with a precise definition of the concept. For a long time this lack of definition was perhaps not a major problem because licensing disputes were relatively rare. (p. 380) Yet in the last ten years the FRAND rules have led to numerous legal challenges—most notably in what has been dubbed the ‘smartphone wars’ since 2009, with the likes of Apple, Google, HTC, Microsoft, Nokia, and Samsung suing each other in jurisdictions across the world. These FRAND disputes under IP law increasingly end up in front of competition authorities as well.

8.54  How the competition rules apply to FRAND in IP licensing is still an undeveloped area. One question is whether not complying with FRAND obligations amounts to an abuse of dominance. Another is how FRAND should be determined. As regards the first question, there is a spectrum between laissez-faire and more interventionist approaches, just as there is in other areas of competition law. An example of the former is the Rambus case in the United States.46 In 2008 a court overturned a finding by the FTC that Rambus, a technology licensing company, had breached section 2 of the Sherman Act. The FTC considered that Rambus had unfairly ‘ambushed’ an SSO—the JEDEC, which worked with industry participants to develop standards for random access memory chips—by not declaring its intention to acquire and subsequently enforce patent rights covering technology that was essential to implement two particular standards. The FTC also concluded that as part of this ‘ambush’ Rambus had charged prices above the (F)RAND levels that the SSO would have sought. Assessing this behaviour under competition law, and taking FRAND as the benchmark, was a previously untested approach. The court disagreed with the FTC and stated that:

the Commission failed to sustain its allegation of monopolization. Its factual conclusion was that Rambus’s alleged deception enabled it either to acquire a monopoly through the standardization of its patented technologies rather than possible alternatives, or to avoid limits on its patent licensing fees that the SSO would have imposed as part of its normal process of standardizing patented technologies. But the latter—deceit merely enabling a monopolist to charge higher prices than it otherwise could have charged—would not in itself constitute monopolization.47

8.55  The European Commission also opened an abuse of dominance investigation into Rambus, and reached a settlement in 2009 whereby Rambus would place a worldwide cap on its royalty rates for products under the JEDEC standards (and zero rates for a technology used in a particular standard that had been adopted when Rambus was a JEDEC member).48 Another early investigation under EU competition law was into Qualcomm, following complaints by various mobile phone and chipset manufacturers (including Ericsson, Nokia, Texas Instruments, Broadcom, NEC, and Panasonic).49 The alleged abuse concerned the terms under which Qualcomm licensed its SEPs under the WCDMA standard, which forms part of the 3G standard for European mobile phone technology (also referred to as UMTS). The investigation was closed in 2009 after the complaints were withdrawn, and hence did not shed much light on how FRAND would be treated under Article 102.

8.56  Two subsequent European Commission investigations, into Motorola and Samsung, reached a conclusion in 2014 and set out the circumstances in which SEP holders can abuse (p. 381) a dominant position by filing injunctions against users of the SEP—that is, these cases did not clarify how FRAND should be interpreted, but they do circumscribe the conduct that owners of SEPs can engage in.50 The Commission stated that where the holder of a SEP has committed to license it on FRAND terms, seeking injunctions to exclude competitors from the downstream market is anti-competitive if those competitors are ‘willing’ to take a licence on FRAND terms. In the case of Motorola (whose patents have been owned by Google since 2011), the Commission found that an injunction against Apple in a German court led to competitive harm: a temporary ban on Apple’s sales of iPhones and iPads in the German market, and Apple having to enter into an onerous settlement agreement whereby it had to give up its claims against the validity of the Motorola patents. Apple was considered to have been a ‘willing’ licensee since it had agreed to a court setting the FRAND rate in the event of a dispute. In a parallel case, Samsung had sought injunctions against Apple in a number of jurisdictions. In exchange for closure of the investigation, Samsung committed not to seek such injunctions against ‘willing’ licensees for a period of five years. Following these decisions there is still much debate about what constitutes a ‘willing’ licensee, in addition to the ongoing debate about the meaning of FRAND.

8.5.3  FRAND: Economic criteria

8.57  Economics can shed some light on how to think about determining FRAND, even if the meaning of FRAND is not yet settled. The nature of IP rights means that static cost-based pricing rules do not provide useful guidance for setting price ceilings (in contrast with essential physical facilities where both capital and operating costs can be more easily identified and related to usage of the facility). The typically low marginal cost of distributing IP means that setting prices based on marginal costs—which is normally considered the efficient price level—would not generate sufficient incentives for innovation. It is therefore necessary to look for alternatives. One is to carry out a profitability analysis of the investments and returns over the lifetime of the patent. This is a more comprehensive analysis than the static cost-based pricing rules and uses the financial tools discussed in Chapter 3. With the right data, the profitability of innovative activities can be estimated, making appropriate adjustments for intangible assets, unsuccessful as well as successful innovation efforts, and risk. The price would be set such that the innovator can earn returns that cover its cost of capital. This approach still leaves open the policy question of whether it is sufficient to allow patent owners a fair return on their investments in innovation (which is often the approach for owners of essential physical facilities), or whether patent owners should be entitled to a share of the profits (‘economic value’) that downstream companies make by using the patented technologies.

8.58  Another set of economic approaches tries to capture the notion that the FRAND terms should reflect the added value of the patent itself, but not the value derived from the patent’s inclusion in the standard. One of these is the Swanson–Baumol approach, whereby the price that the IP holder would be willing to charge prior to the acceptance of the IP into the standard is taken as the fair price, as it reflects the value of the IP independently of the value of the standard (Swanson and Baumol, 2005). Another is the Shapley value approach, (p. 382) resulting in a FRAND price that awards each IP holder the value that represents its relative contribution to the standard.51 We discuss these two approaches below.

8.59  To alleviate concerns about exploitation of ex post market power by an IP holder that has been accepted into a standard, the Swanson–Baumol approach is based on prospective licensees negotiating licence terms before their acceptance into the standard, when there is still active competition between technologies. Licence terms set at that point do not reflect the value of incorporation into the standard. This is broadly the model that two SSOs—the Institute of Electrical and Electronics Engineers and the VITA Standards Organization—have been exploring: in their applications to the SSO, IP holders reveal the maximum royalty rate that would apply for the lifetime of the standard, were they to be accepted (Treacy and Kostenko, 2007). The Swanson–Baumol approach is auction-based, so the resulting licensing rates reflect the value that each patent brings to the standard over and above the value that the next-best solution could provide. While this approach is appealing in relation to the future pricing of technology that is yet to be incorporated into a standard, its usefulness as a benchmark for making ex post assessments is more limited. One would have to verify the extent to which, at the time that the particular patent was included in the standard, alternative technological solutions were available that could have been substituted for the patent in question.

8.60  The Shapley value approach to FRAND is based on a theoretical form of patent selection, and can be defined as follows (Layne-Farrar et al., 2007, p. 24):

Suppose that there are n patent-owners, one for each patent involved . . . Suppose the patent-owners arrive at the SSO in random order each with her patent in her pocket, with all possible arrival sequences equally likely. Now suppose that in each sequence, each patent-owner receives the amount by which her patent increases the value of the best standard that can be built from the patents that are already at the SSO when she arrives . . . The Shapley value gives the average of such contributions over all possible arrival sequences—each patent thus receives the average (over arrival sequences) of its marginal contribution.

8.61  To illustrate, consider a standard with two complementary technologies: one to enable the characters in a computer game to move left, and one to enable them to move right. The end-product (the game) needs both in order to be of any use. While for most users these technologies will be of equal value, the payouts to the patent holders may not be the same under the Shapley value approach. If just one company develops the technology to move left, while two companies work out alternative technologies to move right, the payoffs would be two-thirds to the former, and one-sixth to each of the latter—this is shown in Table 8.1, which has all six arrival sequences and shows for each which technology (R1, R2, or L) renders the product valuable upon arrival, that is, once there is a left and right technology. Note that in this analysis both inventors of moving right receive a payoff, even if only one of them is ultimately included in the standard.

Table 8.1  Illustration of FRAND pay-offs based on the Shapley value approach

Arrival sequence of the technologies

Which technology adds the value?

Final pay-off to each technology after six sequences

R1, R2, L


R1: 1/6th

R1, L, R2


R2: 1/6th

R2, R1, L


L: 4/6ths

R2, L, R1


L, R1, R2


L, R2, R1


8.62  Despite their different approaches, the Swanson–Baumol and Shapley value methodologies share similarities in their outcomes. An IP holder that faces no competition from alternative technologies vying to be included in the standard will always earn greater returns than an IP holder for a technology for which there is competition. In practice, this could lead to (p. 383) the paradoxical outcome of a monopoly provider of a peripheral technology earning greater returns than a provider of a more fundamental element of the standard for which alternatives exist. However, economically, this outcome might still be considered efficient—scarce goods are priced more highly; this is what you get in well-functioning markets, even if it may not seem ‘fair’.

8.63  In any event, it has proved difficult to translate these two theoretical approaches into practical tests. In addition, some courts have questioned the appropriateness of the theoretical result that a patent that adds significant value to a standard (like the moving right technology in the computer game example above) should get a low royalty just because it had a competitor before inclusion into the standard. In Innovatio (2013), an IP case in the United States involving wireless internet technology, the court accepted that the FRAND rate should reflect the value of the technology and not the hold-up value from inclusion in the standard. However, it rejected the argument that when two alternative patents compete for incorporation into the standard the FRAND rate could be driven down to zero, as ‘it is implausible that in the real world, patent holders would accept effectively nothing to license their technology’.52

8.5.4  FRAND: A bargain?

8.64  A framework for determining FRAND (and reasonable royalties more broadly) used in IP law in the United States is that of the hypothetical negotiation between a willing licensor and a willing licensee. The reference for this is the Georgia-Pacific case of 1970 (concerning patents for particular types of plywood):

The amount that a licensor (such as the patentee) and a licensee (such as the infringer) would have agreed upon (at the time the infringement began) if both had been reasonably and voluntarily trying to reach an agreement; that is, the amount which a prudent licensee—who desired, as a business proposition, to obtain a license to manufacture and sell a particular article embodying the patented innovation—would have been willing to pay as a royalty and yet be able to make a reasonable profit and which amount would have been acceptable by a prudent patentee who was willing to grant a license.53

(p. 384) 8.65  This negotiation framework is very broad—there are many possible outcomes—and gives great weight to what ‘reasonable’, ‘prudent’, and ‘willing’ buyers and sellers agree with each other in the market. The framework does have some grounding in economics, namely in the famous bargaining problem as formalized by John Nash (1950b). In its simplest general form the bargaining problem involves two parties negotiating the distribution of a fixed sum (or pie) between them. This can be a seller and a buyer (or licensor and licensee, as here). The main parameters of the bargaining are the buyer’s maximum willingness to pay and the minimum price the seller is willing to accept. If the former is larger than the latter, there is scope to reach an agreement (there is a pie to be shared). Another key parameter is the ‘disagreement point’, which reflects the value to each player if no agreement is reached. For example, if neither party gets any value when no deal is struck, the disagreement point would be (0,0). If one of the parties has a ‘walk-away’ option, it would get a positive value at the disagreement point (you can see how having an alternative improves the party’s bargaining position). Any resulting distribution can be an equilibrium outcome. From an economic perspective what matters most for welfare is that a deal is struck at all—any additional transaction where the buyer’s willingness to pay exceeds the seller’s minimum desired price enhances economic welfare (as we saw in Chapter 1).

8.66  Evidently the bargaining problem doesn’t say much about distributive justice. At one extreme, the buyer may be powerful or cunning enough to appropriate the entire surplus—that is, the price ends up at the minimum level the supplier was willing to accept. At the other extreme the seller extracts the entire surplus from the buyer (in the FRAND context, this would mean that the patent holder obtains the entire downstream profit from the licensee). One type of outcome is the Nash bargaining solution, which maximizes the joint surplus made by the two parties. In the case where the two parties are symmetric (e.g. they have the same payoff at the disagreement point) the solution is to share the pie equally. This is intuitive, but also somewhat simplistic. In the Oracle v Google case of 2011, concerning patents related to the Java software platform, a US court rejected the 50/50 Nash bargaining solution as a basis for setting reasonable royalty rates:

It is no wonder that a patent plaintiff would love the Nash bargaining solution because it awards fully half of the surplus to the patent owner, which in most cases will amount to half of the infringer’s profit, which will be many times the amount of real-world royalty rates. There is no anchor for this fifty-percent assumption in the record of actual transactions.54

8.67  Quoting some formulae from an economics paper on the Nash bargaining solution, the court added that: ‘No jury could follow this Greek or testimony trying to explain it. The Nash bargaining solution would invite a miscarriage of justice by clothing a fifty-percent assumption in an impenetrable facade of mathematics.’55

8.68  In Microsoft v Motorola (2013) another US court sought to adjust the Georgia-Pacific criteria to the specific context of FRAND for SEPs.56 Motorola, the licensor, had offered a rate of 2.25 per cent of the end price of each Microsoft product using its SEPs (in particular the Xbox), amounting to around $4 billion in cash. The court ruled that (p. 385) Motorola’s offer was not in line with FRAND, and determined a much lower range of FRAND rates (expressed as per-unit prices rather than as a percentage of Microsoft’s product price). The court did this with reference to various comparable licensing rates in the market that had been set in a FRAND commitment context. Such benchmarking against other rates in the market is a common approach in FRAND disputes. The court referred to the hypothetical negotiation framework, but added a number of specific factors to be considered when assessing FRAND for SEPs. First, the SEP holder has committed to licensing on FRAND terms, so cannot walk away from the negotiation. Second, bargaining parties would take into account the fact that the licensee implementing the standard must deal with more than one SEP holder, and must avoid excessive ‘royalty stacking’ (i.e. where the SEP holders in aggregate charge such high royalties that the licensee cannot make any profit downstream). While this case has narrowed the Georgia-Pacific negotiation framework somewhat, it still leaves plenty of room for many different bargaining outcomes. A relatively wide range of royalty rates could be considered FRAND under the hypothetical bargaining and market benchmarking approaches.

8.6  Behavioural Remedies: Insights from Behavioural Economics

8.6.1  Nudges and liberal paternalism

8.69  In Chapter 3 we discussed behavioural economics and showed how the presence of consumer biases and bounded rationality on the demand side can have a negative impact on market outcomes. In this section we explore how behavioural economics can help in the design of remedies targeted at the demand side. These are different from the remedies discussed so far, which have focused mostly on the supply side—remedies targeted at market structure and at the behaviour of companies. Within behavioural economics there are different schools of thought about remedies (and policy interventions more broadly). One of these is ‘liberal paternalism’, a phrase coined by Richard Thaler and Cass Sunstein (2003). Liberal paternalist interventions try to influence consumers’ choices in a way that will make them better off, and without restricting their options. These interventions have become known as ‘nudges’ (Thaler and Sunstein, 2008). Liberal paternalism recognizes that it is not completely understood why people behave the way they do. In addition, it acknowledges that a heavy-handed approach by policy-makers—for example, banning certain products, or subsidizing others—is not always desirable or cost-effective, and may have unintended consequences. Therefore, liberal paternalism involves designing remedies that, to some extent, work with consumers’ biases and limited decision-making abilities, rather than seeking to correct them. This does not necessarily mean more intervention, but rather ‘smarter’ intervention. Given the influence of framing and bounded rationality in consumer decisions, nudging seeks to alter the choice architecture within which people make decisions. This can be of use in markets where competition is perceived to be ineffective due to a lack of customer searching and switching.

8.70  One such remedy is to simplify information disclosure to salient points—this is aimed at overcoming framing, information overload, and inertia, and hence at improving decision-making. Timing also matters, as information is most relevant at the point when a decision is made. To give an example, Cass Sunstein, retained by the US government, produced a memorandum for all executive departments and agencies entitled ‘Disclosure and (p. 386) simplification as a regulatory tool’.57 This stated that departments and agencies should provide ‘clear, salient information at or near the time that relevant decisions are made’. This is often at the point of selection or purchase, in which ‘agencies highlight the most relevant information in order to increase the likelihood that people will see it, understand it, and act in accordance with what they have learned’. In a competition context, the UK Financial Conduct Authority (FCA), the financial regulator, proposed such information-based remedies in its market study into general insurance add-ons.58 An experiment was carried out on insurance add-ons to be purchased with a number of primary products: a home boiler, a tablet computer, a laptop, a luxury holiday for two, and a 12-day car hire in Spain. It turned out that 65 per cent of consumers bought the first insurance offer they saw without further search if the add-on was first introduced at the point of sale, whereas only 16 per cent purchased the first insurance product they saw if the add-on was introduced up front (i.e. at the start of the process of purchasing the primary product). In addition, the framing of add-on prices resulted in consumers seemingly not understanding the ‘real’ price of add-ons. Where monthly prices were displayed rather than the annual price, 30 per cent of consumers changed their mind about purchase once they saw the annual cost.

8.71  Another type of liberal paternalist remedy focuses on activating consumers to make a choice—the ‘forced choice’, which aims to prevent inertia or simply going with the default option. This could be achieved by requiring consumers to confirm pro-actively (e.g. by telephone or online) whether they wish to renew their service contract or insurance policy with their current provider, rather than the contract being renewed automatically. This may overcome the poor choices made as a result of consumer inertia. A related remedy is to use default opt-ins or opt-outs. Where a particular outcome is clearly superior from the consumers’ (as opposed to the sellers’) point of view, the policy might be to set that outcome as the default, without restricting consumers’ ability to choose an alternative. For example, it may be that people are defaulted into an employee pension plan, but can then opt out. This can help overcome inertia and loss aversion (the fear of regret of opting into the wrong policy), and ultimately achieves an arguably better policy outcome with more people having a pension plan.

8.72  An example of such a remedy in a competition context is Ofcom’s prohibition in 2011 of automatically renewable contracts (also known as rollover contracts) in retail telephony and broadband markets.59 The UK telecoms regulator recognized the benefits of such contracts (e.g. convenience), but found that these benefits were outweighed by the negative effects on consumer switching and hence effective competition. An econometric analysis had shown that customers who were on such contracts tended to switch less frequently than those who were not. Ofcom considered a remedy aimed at providing more information to consumers about the automatic renewals. However, it found that the consumer harm resulted from the opt-out nature of these contracts, not from a lack of information as such. The remedy therefore focused on making the renewal process an opt-in rather than (p. 387) an opt-out process—that is, customers would actively have to renew their contract upon expiry. A more heavy-handed remedy that Ofcom also considered was to abolish minimum contract periods altogether and allow customers to switch at any point in time (as some other EU countries have done via legislation). However, the regulator considered this remedy to be disproportionate, as it would deprive customers of the discounts and other benefits they may receive when signing up to a contract for a particular period. This example does not mean that automatic renewals are always problematic. In its 2014 investigation into the UK motor insurance market, the CMA did not find automatic renewals of annual policies to be a major barrier to switching, as switching rates in this market were relatively high.60 This shows the importance of careful empirical testing of the competition problems and potential remedies on a case-by-case basis.

8.73  Liberal paternalist interventions tend to come at a lower cost than more intrusive interventions, such as subsidies or education programmes. Implementing them does not require a complete understanding of consumer preferences or decision-making processes. If the intervention does not work effectively, there are often few unintended negative consequences. Liberal paternalist interventions can also be aimed at ensuring that affected consumers (e.g. naive consumers) are better off without making others (e.g. sophisticated consumers) worse off, as both types of consumer still have all choices available to them. Yet there is a fine line between liberal paternalism and simple paternalism. The CC investigation into personal current accounts in Northern Ireland perhaps came close to it. One of the competition concerns identified was that, in general, consumers do not actively search for alternative personal current accounts. However, the CC also found that ‘customers are generally not particularly interested in personal current accounts’, and that 80 per cent of those surveyed who had not switched bank gave as a reason that they had been with their current provider for a long time.61 If consumers do not particularly care about a product, should competition authorities? This is a policy question that merits consideration before intervening in a market.

8.6.2  More intrusive interventions to restore consumer sovereignty

8.74  Sometimes more intrusive remedies are preferred which do restrict consumer choice sets (e.g. banning certain products), but are still aimed at increasing the power that consumers have over those choices that are available to them. These remedies are still consistent with the aim of reinforcing ‘consumer sovereignty’ (i.e. whereby consumer preferences determine which goods and services are produced), but go beyond just nudging consumers towards the desired behaviour. By restricting the choice set, these interventions may make some consumers better off but also make others worse off. They therefore require a great deal of caution—policymakers ultimately have to assign weights to the welfare impacts on the different groups of producers and consumers.

8.75  A far-reaching remedy proposed by the CC in the payment protection insurance (PPI) market investigation (which we also discussed in Chapter 3) was a prohibition on selling PPI at (p. 388) the credit point of sale.62 PPI was sold as an add-on insurance to consumers who take out credit (the primary product), and provided cover against events that may prevent them from keeping up with their repayments. The remedy prohibiting PPI at point of sale would give consumers time to shop around and encourage them to consider more carefully the price and quality features of PPI before purchasing it. It would also give stand-alone providers of PPI the opportunity to offer their products more effectively, without having the disadvantage of not being available at the point of sale. This was expected to increase competition in the PPI market. The CC concluded that other remedies, such as the provision of more information, would not be sufficient to address the competition issues. On appeal, the CAT found that the CC had failed to take into account the loss of convenience to consumers that would follow from the imposition of the point-of-sale prohibition.63 The case was remitted to the CC for reconsideration, which then conducted further research—in particular, in the form of a number of conjoint analyses among consumers—to assess the costs and benefits of the prohibition.64

8.76  This consumer research raised some interesting distributional issues. The survey evidence suggested that 60 per cent of personal loan PPI customers preferred the convenience of purchasing PPI at the point of sale. Hence for this majority the prohibition remedy would remove the current preferred choice. However, the CC determined that the other 40 per cent of consumers who preferred to purchase away from the point of sale valued this option more than those who preferred to buy at the point of sale (notwithstanding the fact that those who wished to buy away from the point of sale could already do so). It therefore placed greater weight on the 40 per cent group, and confirmed its previous conclusion that, on balance, consumers as a whole would benefit from a point-of-sale prohibition. This example illustrates that it is far from straightforward to design remedies that will benefit everyone, and that a CBA of remedies often requires authorities to value the relative preferences of different groups of consumers. In this particular case, the prohibition remedy was estimated to result in an inconvenience to 60 per cent of customers and a potential benefit to the remaining 40 per cent.

8.6.3  The Microsoft remedies

8.77  The European Commission Microsoft cases are useful examples of how behavioural economics can add to the assessment of remedies in abuse of dominance investigations. Consumer biases played a role in the Media Player case, which focused on Microsoft’s practice of bundling (by default and free of charge) Windows Media Player with its Windows operating system.65 Microsoft was dominant in the operating system market, but competitors were vying to gain market share in the emerging media player market. By bundling Media Player as the default application through which all media would play when a consumer bought a PC, Microsoft was considered to have leveraged its dominance in the operating system market into the market for media players.

(p. 389) 8.78  Traditional approaches would suggest that this default does not matter hugely if consumers are rational. After all, with a few clicks they could download an alternative media player free of charge. However, when viewed from the perspective of behavioural economics, it follows that many consumers may not switch away from the preloaded Microsoft software that comes as the default with the PC. The European Commission found that the pre-installation of Media Player on Windows created the potential for leveraging since, on the demand side, ‘users who find WMP pre-installed on their client PCs are in general less likely to use alternative media players as they already have an application which delivers media streaming and playback functionality’.66 Equally, the Commission found that: ‘A supply-side aspect to consider is that, while downloading is in itself a technically inexpensive way of distributing media players, vendors must expend resources to overcome end-users’ inertia and persuade them to ignore the pre-installation of WMP.’67

8.79  Hence the finding of abuse took account of the behavioural biases that consumers face. The remedies to this competition problem also had behavioural economics aspects. Yet one remedy that the Commission put forward turned out to be ineffective, and could perhaps have been improved if it had been based more on insights from behavioural economics. This was the remedy that Microsoft should make available versions of Windows with and without Media Player installed. This came with a strict warning that Microsoft must refrain from using any technological, commercial or contractual means which would have the equivalent effect of tying Media Player to Windows. The unbundled version of Media Player had to be of the same quality as the bundled version. As it happened, few copies of the version without Media Player were sold (Ahlborn and Evans, 2009). This is perhaps not surprising since both versions were offered at the same price. When presented with the two options, consumers would be most likely to choose the version with Media Player already installed. The bundled version remained an attractive default option. A more effective remedy might have been to include a CD containing a random selection of media players for consumers to choose from. By forcing consumers to make a conscious choice in this way, and by not including a default option, consumers might have been better placed to select a media player most suited to their needs.

8.80  In a subsequent case in 2009 the Commission investigated the bundling of the Microsoft Internet Explorer web browser with Windows (akin to the US investigation into Microsoft a decade earlier, when the Netscape browser was the victim of such bundling).68 The remedy in this case was more in line with behavioural economics insights. Users of Windows-based PCs with Internet Explorer as their default web browser, and who received Windows Update (which updates their operating system), would be taken to what was referred to as the choice or ballot screen. This screen would give users an opportunity to choose whether to install a competing web browser and, if so, which one. It would display, at random, the main browsers that users could click on, including Internet Explorer. The Commission reasoned as follows:

Displaying five web browsers in a prominent manner, and seven more when the user scrolls sideways, strikes an appropriate balance between the need to have a workable choice screen (p. 390) that users are likely to make use of and making the choice screen as accessible as possible to web browser vendors. If the choice screen presented too many web browsers, users could be overwhelmed and as a consequence would be more likely not to exercise a choice at all, but rather to dismiss the entire choice screen.69

8.81  The design of the choice screen was carefully considered as part of this remedy. It had to avoid any framing biases in favour of Internet Explorer or another browser—the screen was made to look different from the Explorer environment, and the initial proposal to list the various browsers alphabetically was changed to listing them randomly. The process also had to be as simple as possible for users—only one or a few clicks were required, and the new icon would be automatically pinned to the task bar, replacing the Internet Explorer icon.

8.82  This remedy in effect eliminated the default option of having Internet Explorer, instead forcing users to make an active choice of their preferred browser. The forced-choice approach seemed more effective in cutting the tie between Windows and Internet Explorer than the unbundling remedy in the Media Player case. It should be noted that in 2013 the European Commission imposed a €561 million fine on Microsoft for not implementing the browser remedy correctly (reportedly there had been a technical error in some versions of Windows which meant that customers did not see the choice screen and thus were deprived of the choice of browser).70 This illustrates the need for ongoing supervision when behavioural remedies are applied.

8.6.4  The Google remedies

8.83  As internet usage grew worldwide in the early 1990s, so did the prominence of search engines: Excite, Infoseek, AltaVista, Yahoo!, Lycos, and others. Google entered the fray only in 1998, but it did so with a new and highly innovative search technology. Existing search engines mainly ranked results by how many times the search terms appeared on a page, but had begun to struggle with helping users navigate through huge amounts of web content, a lot of it spam. Google’s method revolved around links between pages as a primary metric for page relevance, and significantly improved search functionality. By 2000, Google was the largest search engine in the world, and still is today. It made it into the top five global companies by market capitalization, and ‘to google’ has become a widely used verb. This is a case study of successful innovation in a competitive market resulting in significant benefits to consumers and huge financial rewards and high market shares for the innovator.

8.84  With high market shares have come competition concerns, especially as Google has moved into other online activities on the back of its powerful position as a search engine—including online shopping, advertising, news, travel services, and mapping. Among a range of competition law actions against Google (and actions under other areas of law, including copyright, patents, and privacy), the European Commission opened an investigation in 2010 into abuse of dominance by Google in the online search market.71 This followed complaints by a number of competitors in search, including Foundem, ejustice.fr, and Microsoft, who were (p. 391) subsequently joined by several others. One of the Commission’s concerns was that Google’s general search services, which in Europe had a market share above 90 per cent, systematically favoured the company’s own specialized or vertical search services, now called Google Shopping. In search results for particular products and services Google was said to display its own services more prominently, at the expense of other specialized search providers. In line with the findings of behavioural economics, salience in search results can have a decisive influence over consumers’ choices, even if alternative options can be found lower down the list. Consumers persistently tend to select the more prominently listed options, and do not tend to make much effort looking further. In competition law terms, the Commission considered that Google used its dominant position in the market for general search to the detriment of competitors, innovation, and consumers in the market for comparison shopping (specialized or vertical search).

8.85  At one stage of the investigation, in 2014, the Commission came close to agreeing a remedy whereby Google would present search results such that competing offerings would appear as prominently as Google’s own.72 In behavioural economics terms this remedy would seem to go a long way towards mitigating concerns about biases from the way search results are framed. It would be made clear on the screen which are the Google Shopping results and which are the alternatives the user could click on. Both sets would be presented in the same manner; if the Google results had photos or a picture then the alternatives would as well. In presenting the remedies, the Commission gave illustrations of a search for gas grills and one for cafés in Paris. The first of these is shown in Figure 8.1. The picture at the top shows the search results without the remedy. The most prominent listings are for five vendors of gas grills who have paid Google for the advertising. The picture at the bottom shows what would happen with the remedy. The first search results shown are now three options from Google Shopping (by vendors who paid Google directly), and right next to it, and equally salient, three options from rival vertical search providers (Supaprice, Kelkoo, and Shopzilla). The results page for cafés in Paris after the remedy (not shown here) similarly had a prominent listing of three alternative search providers that could be used to find such cafés (pagesjaunes.fr, viamichelin.fr and Yelp).

Figure 8.1  Before and after: Proposed (but rejected) remedies for Google

Source: European Commission (2014), ‘Statement on the Google investigation’, press conference, 5 February.

8.86  The Commission at that stage of the investigation rejected calls for a more heavy-handed remedy preventing Google from displaying its own specialized search results:

The objective of the Commission is not to interfere in Google’s search algorithm. It is to ensure that Google’s rivals can compete fairly with Google’s own services. Google should not be prevented from trying to provide users with what they’re looking for. What Google should do is also give rivals a prominent space on Google’s search results, in a visual format which will attract users.73

8.87  One of the sticking points in discussing the remedy displaying alternatives was how to determine which three competitors would appear in the prominent alternative slots. The suggestion was to auction these slots, which was meant to strike a fair balance by giving competitors access to a valuable position in the search results, but not without payment. In the end, the remedy was not accepted, and the investigation is ongoing at the time of writing.(p. 392)

(p. 393) 8.7  Setting Fines

8.7.1  Headlines and main principles

8.88  Hefty fines imposed by competition authorities now regularly grab the headlines. This has raised awareness of competition law among business communities and even the general public. In the ten years from 2005 to 2014 the European Commission imposed cartel fines totaling €18 billion (in the decade of the 1990s total cartel fines were only €615 million).74 The Commission fined Microsoft €899 million in 2008 and €561 million in 2013, in both instances for failing to comply with abuse of dominance remedies (the first of these fines was reduced to €860 million by the General Court in 2012).75 Intel was fined €1.1 billion in 2009 for abuse of dominance, a decision upheld by the General Court in 2014.76 The Bundeskartellamt in Germany imposed a record total of more than €1 billion in fines in 2014, large parts of which were imposed on the members of the sausages and sugar cartels.77 The French competition authority has also been active on this front—on a single day in 2014 it imposed its highest-ever fines of €950 million on cartels in personal care and cleaning products (involving L’Oréal, Procter & Gamble, Sara Lee, Reckitt Benckiser, and Unilever among other companies).78

8.89  These high fines raise a number of questions: how do competition authorities determine the level of a fine? Is there any economic basis for the fine? Can companies afford these fines? We address these questions in this section. Many competition authorities have published guidelines on how they set fines, such as the European Commission’s 2006 Fining Guidelines.79 Fines function as a punishment, but more importantly they are also aimed at deterring infringements. Indeed, the main objective stated in the Fining Guidelines is to ensure that the fine has the necessary deterrent effect. We start here by exploring the economic principles of fines and their effect on deterrence.

8.90  Why do you park your car neatly between the white lines of a parking bay and put enough money in the parking meter? Economists have two answers to this. The first is: it’s because you know that otherwise you’ll get a fine. You respond to financial incentives. The second answer is more akin to the principles of behavioural economics: you may have an intrinsic motivation to be law-abiding, so you would pay the car park charge even if the chances of a fine were low. Economists have joined scholars of law and psychology in studying the mindset of those tempted to break the law. The traditional economic framework in essence sees (p. 394) would-be offenders making a rational trade-off between the rewards of the illegal activity and the risk of being caught.80 Behavioural economics has introduced some further subtleties to this framework.

8.91  Two conditions must hold in this framework in order to achieve deterrence: the likelihood of being caught must be sufficiently high (in some places the chance of getting a parking ticket is rather higher than in others); and the fine must be sufficiently high. Economic theory identifies two possible reference points to determine the level of fine: the harm to society caused by the crime, and the illicit gains made by the perpetrator. On the first basis, if the cost to society of a particular crime is €1,000, and the offender is caught with 100 per cent certainty, the fine should be set at €1,000. The harm to society can then be repaired (provided that the authorities redistribute the collected fines to those who have suffered—this does not always happen). In reality, very few crimes are punished with 100 per cent certainty. If the probability of detection and enforcement is only, say, 20 per cent, a fine of €1,000 is too low—only €200 is recovered on average for every crime costing €1,000. Instead, a more appropriate fine would be €5,000—one in five criminals is caught, and a total of €5,000 is collected in fines, covering the cost to society of the five crimes. This very simple framework can be expanded with additional features—for example, the cost of enforcement. Having a police force (or a competition authority) and a court comes at a cost, and this cost must be weighed against the benefits of fighting crime. As a result, the socially optimal degree of law enforcement is usually not to catch 100 per cent of criminals but some smaller proportion. It can be optimal to let some criminals get away with it. In this framework there is to some extent a trade-off between the probability of detection on the one hand, and the level of fine on the other. In theory, the same result can be achieved either through very active enforcement (leading to a high proportion of criminals being caught) and low fines, or through more limited enforcement and higher fines. The disadvantage of the former approach is that enforcement costs are high. The disadvantage of the latter is that the fines may be disproportionate for those few criminals who do get caught (in the above example, a justice system may frown upon a criminal having to pay a fine of €5,000 for a crime that cost society only €1,000).

8.7.2  Fines and deterrence

8.92  A number of additional economic principles must be considered when applying the above framework for setting fines aimed at deterrence. One problem with setting fines with reference to the cost of the crime to society is that this may not have a deterrent effect if the fine is lower than the benefit obtained by the criminal. If you are in a real hurry to get to a client meeting on time, and the fine for speeding is €30, you may well consider that a risk worth taking. However, if the fine is €300, your rational calculation may lead you to behave differently. Whether you decide to break the law to get to your meeting on time will depend on many factors—such as how much the client is worth to you, or your hourly charge-out rate—but the point is that the fine effectively becomes like a price.

8.93  A much-discussed example of where deterrence does not work if the fine becomes a price is that of a number of private day-care centres in Haifa, Israel.81 In order to address the problem of late pick-ups by parents, the centres began to charge parents a penalty of 10 shekels per child every time they arrived more than ten minutes late for pick-up (p. 395) (adding to their monthly bill of 1,400 shekels). The result was the opposite of what was intended: instead of having a deterrent effect, the new fining system resulted in an increase in the number of late pick-ups. Why? The reason was that parents were more than willing to pay the fine in exchange for the extra time their child was in day care. The fine effectively became like any other price in the consumers’ rational calculation (to put the 10 shekel fine into context, a babysitter cost around 15–20 shekels per hour). Another way of explaining the outcome—and here behavioural biases are at work—is that the parents felt less guilty about arriving late and taking advantage of the teachers’ goodwill because they were now paying for it.

8.94  These examples show that deterrence can best be achieved through fines that negate the illicit benefits obtained by offenders. If the cost to society of you speeding is €15, and the chance of being caught is 50 per cent, proponents of fines based on the harm to society would say that a fine of €30 results in the socially optimal level of crime. The fact that at this level of fine there are some well-off motorists who can afford to ‘pay off’ the authorities for the right to speed is simply part of the optimum in this framework. However, you can also see that deterrence is not achieved, as the propensity to break the law will be income-related. Some commentators therefore do not regard this as the optimal situation in the context of competition law, and would rather set fines with reference to illicit gains so as to achieve greater deterrence (e.g. Wils, 2006). The calculation is similar to the one above. If the extra profit from entering a cartel is €1,000 and the cartel will be punished with 100 per cent certainty, a fine of €1,000 achieves effective deterrence. If the probability of punishment is only 20 per cent, a fine of €1,000 is insufficient as the would-be cartelist will take the 80 per cent chance of receiving a positive pay-off from the crime—instead, the deterrent fine is €5,000. Behavioural economics identifies two additional reasons why fines may have to be set even higher than that. One is the ‘availability bias’—people tend to forget past fines after a while and hence may not be sufficiently deterred; regularly grabbing the headlines with high fines is one way competition authorities can avoid this cognitive bias. The other reason is optimism bias; criminals tend to underestimate the probability of something bad happening to them (i.e. getting caught), and hence an uplift in the fine may be required to make deterrence effective.

8.95  A final economic principle we discuss here relates to predictability—to achieve deterrence, should authorities make the level of fines predictable for would-be offenders? According to the CFI’s ruling in the plasterboard cartel case in 2008 the answer is no:

Moreover, it is important to ensure that fines are not easily foreseeable by economic operators. If the Commission were required to indicate in its decision the figures relating to the method of calculating the amount of fines, the deterrent effect of those fines would be undermined. If the amount of the fine were the result of a calculation which followed a simple arithmetical formula, undertakings would be able to predict the possible penalty and to compare it with the profit that they would derive from the infringement of the competition rules.82

8.96  Is the court’s reasoning economically sound? Isn’t the ‘simple arithmetical formula’ precisely the one we saw earlier capturing the rational calculation made by the would-be criminal, and doesn’t this calculation allow authorities to determine the fine with optimal deterrence? There is no clear-cut answer. Leaving would-be infringers in the dark about the eventual (p. 396) penalty, as the CFI proposed, can have a deterrent effect if companies wish to avoid the risk of arbitrarily high fines. However, some companies, or individuals within those companies, may well be willing to take that risk, in which case deterrence is not achieved. In most policy contexts some predictability of penalties is usually considered desirable to achieve effective and fair law enforcement.83 In competition law there is an additional advantage of having predictable fines: companies that are already in a cartel can make a rational calculation of the benefit of applying for leniency in exchange for blowing the whistle (and hence avoid the fine), which can have the effect of more cartels being reported voluntarily.

8.7.3  The European Commission fining guidelines: Where is the economics?

8.97  The Fining Guidelines set out the steps that the Commission follows in its calculations. They explain how the basic amount of a fine is arrived at, and how adjustments are made for factors such as whether the company has engaged in anti-competitive behaviour before (fine adjusted upwards if this is the case) or whether it was co-operative during the investigation (fine adjusted downwards, or even cancelled in the case of leniency). The Fining Guidelines are to some extent based on the economic principles discussed earlier. The basic amount is set with reference to the value of the company’s sales in the relevant market and the duration of the infringement, two factors that are seen as ‘an appropriate proxy to reflect the economic importance of the infringement’.84 The term ‘economic importance’ can be interpreted as the importance either to the economy as a whole or to the perpetrators. Both are positively correlated with the value of sales in the relevant market in question—larger companies do more harm, and gain more by doing so, than smaller companies. But the Guidelines stop short of actually considering the harm to the economy or the illicit gains. Nor are the various steps in the Fining Guidelines very mechanistic; there is a high degree of judgement and discretion involved at each step.

8.98  To illustrate how the Guidelines work in practice, consider the example of the chloroprene rubber cartel decision of 2007 against Bayer, DuPont/Dow, ENI, and a number of other companies.85 The starting point for the basic amount of the fine is the value of sales in the relevant market or markets. The cartel covered chloroprene rubber sales worldwide, but for the fine the relevant turnover was limited to the EEA (no detailed market definition was required here; in other cases economists play a role in delineating the relevant market and calculating the relevant sales that form the basis for the fine). For example, ENI’s relevant turnover in the last full year of the infringement was €26 million. To this the Commission then applied a ‘gravity factor’. This can be up to 30 per cent of the sales. There has been some debate about whether the Commission must analyse the actual effects of the cartel to determine the gravity factor,86 but this does not normally seem to be a requirement. In the chloroprene rubber case the Commission noted that the infringement concerned a worldwide hardcore cartel, its members having a 100 per cent market share. It set the gravity factor at 21 per cent, resulting in an amount of €5.46 million (21 per cent of €26 million), and multiplied this by the duration of the cartel of nine years, so nine times €5.46 million equals €49.1 million. The Commission next added an ‘entrance fee’ of 20 per cent. The stated aim of this fee is to deter companies from entering into price-fixing agreements (it seems somewhat odd, if not to say arbitrary, to include such an additional deterrence (p. 397) factor at this stage of the calculation). This gave the basic amount for ENI of €59 million (€49.1 million plus 20 per cent).

8.99  The next step in the Fining Guidelines is to adjust the basic amount for aggravating and mitigating circumstances. In the chloroprene rubber case there was none of the latter (the Commission rejected arguments that ENI subsidiaries did not fully implement the agreements, ceased to take part in the agreements at an earlier stage, or participated only passively). The aggravating circumstance for ENI and Bayer was recidivism. Both companies had been fined for cartel behaviour before. It did not matter to the Commission that this was for different subsidiaries and different cartel products (polypropylene and citric acid, respectively), nor that many years had passed since the last cartel. ENI’s fine was increased by 60 per cent to €94.4 million (Bayer’s punishment for recidivism was a 50 per cent uplift of the fine). Next, for good measure, the Commission imposed a further specific increase of 40 per cent for (again) deterrence, arguing that this was appropriate for ENI, whose overall turnover was much larger than that for chloroprene rubber alone (the other members got either no or a smaller uplift for this). The result was a total of €132.16 million (€94.4 million plus 40 per cent). The final steps were to consider this amount in light of the maximum fine of 10 per cent of total turnover (not exceeded in the case of ENI), any leniency reduction (not applicable for ENI; Bayer got a 100 per cent reduction for being the first to blow the whistle, Tosoh got 50 per cent, and DuPont/Dow 25 per cent), and a possible inability to pay the fine (not applicable for any of the cartel members in this case). Hence the Commission fined ENI €132.16 million. In 2015 the ECJ confirmed an earlier ruling by the General Court which cut the recidivism uplift for ENI from 60 per cent to 50 per cent and the specific deterrence uplift from 40 per cent to 20 per cent, thus reducing ENI’s fine to €106.2 million.87

8.100  You can see from this example that fines have some economic grounding, in that the basic amount relates to the relevant turnover made by the infringer (and economists can play a role defining the relevant market and calculating the relevant turnover). Most of the steps set out in the Fining Guidelines have some logic to them. Yet the actual setting of the fine is not a very scientific process, and it is not always clear how specific numbers are picked at each stage, either by the Commission or, subsequently, by the courts.

8.7.4  Inability to pay fines

8.101  An infringer’s ability or inability to pay can be relevant in determining the fine. In countries such as Finland and Switzerland, speeding tickets are set with reference not only to the speed recorded but also to the income of the offender (headline-grabbing fines of several hundreds of thousands of euros have been imposed under these rules—some people are hard to deter). In competition law, the inability to pay has become a topical issue given the (p. 398) high fines imposed by some authorities. EU case law has actually established that the financial situation of infringers should not in principle influence the level of the fine. In its 2002 judgment on the district heating pipes cartel, the CFI stated as follows:

As regards the applicant’s ability to pay the fine, it is sufficient to observe that, according to a consistent line of decisions, the Commission is not required, when determining the amount of the fine, to take into account the poor financial situation of an undertaking concerned, since recognition of such an obligation would be tantamount to giving an unjustified competitive advantage to undertakings least well adapted to the market conditions.88

8.102  Yet it has also been recognized that there are specific situations in which the inability to pay a fine is a relevant factor to consider. Causing companies to go bankrupt by imposing fines they cannot afford may harm competition. The Fining Guidelines state as follows:

In exceptional cases, the Commission may, upon request, take account of the undertaking’s ability to pay in a specific social and economic context. It will not base any reduction granted for this reason in the fine on the mere finding of an adverse or loss-making financial situation. A reduction could be granted solely on the basis of objective evidence that imposition of the fine as provided for in these Guidelines would irretrievably jeopardise the economic viability of the undertaking concerned and cause its assets to lose all their value.89

8.103  These are stringent conditions. The company must go beyond showing that it is loss-making. It must demonstrate a strong possibility of financial failure as a direct result of the fine. There is currently relatively little guidance or case law on how these conditions should be interpreted.90 How can financial economics be used to meet the Commission’s requirement of ‘objective evidence’?

8.104  The most direct route to demonstrating an inability to pay is to show that the fine will leave the company insolvent (solvency refers to a company having enough assets to cover its liabilities). In general, the most serious problems in paying will emerge if the fine is greater than the market value of shareholders’ equity. Since a company’s liabilities must equal its assets, not only would shareholder equity be wiped out, but the company would not have enough assets to pay all its debts, leading to balance sheet insolvency. For companies listed on a stock exchange, the insolvency condition could be reflected in the fine being greater than market capitalization, which in general reflects the NPV of the future payments to shareholders. Solvency constraints may mean that companies with highly leveraged capital structures (i.e. a lot of debt relative to equity) may have less capacity to pay fines than companies with lower leverage.

8.105  The Fining Guidelines emphasize issues of solvency. However, liquidity is also relevant as it may threaten the survival of a company in difficulty (liquidity refers to the extent to which a company has the cash, or assets that can be quickly converted into cash, to meet its short-term obligations). There are a number of ways in which liquidity crunches can threaten survival. Companies often require a certain amount of working capital in order to operate (p. 399) effectively, and liquidity will be required to cover these needs. Similarly, many sectors have cyclical demand, and companies need to maintain a buffer of available cash or credit lines in order to be able to meet costs during downturns. In a solvent company, all of these points are essentially about timing. In theory, if a company is solvent but has insufficient liquidity to meet both a fine and its other obligations, financial markets should be willing to provide funding to the company so that it can meet its expenditure needs. However, in practice, companies may face capital rationing, which means that they cannot finance existing or new assets. It is this capital rationing that can make even solvent companies unable to raise funds (something you can expect to happen more frequently in a financial crisis). There are therefore a number of reasons why, even if sufficient funds can be raised by a company to pay a fine, the result may be a liquidity crunch which imperils its survival.

8.106  Demonstrating liquidity constraints is more complex than assessing solvency. Determining the maximum fine that the company would be able to pay in the short run (defined here as the period over which the company cannot access financial markets) involves comparing, on one hand, the amount of cash and credit lines it has with, on the other hand, the combined total of expected short-term losses, requirements for short-term debt repayment and taxes, and working capital requirements. This would then represent the maximum fine that could be paid without the company having to access the financial markets.

8.107  The Commission has taken into account, and sometimes accepted, arguments regarding inability to pay in a number of cases in recent years. Its published decisions provide little detail on the analysis performed (possibly for confidentiality reasons). In the bathroom fittings cartel decision of 2010—involving seventeen different companies, including the likes of Hansgrohe, Ideal Standard, and Villeroy & Boch—the Commission assessed the ability to pay as follows.

In assessing the undertakings’ financial situation, the Commission examined the companies’ recent and current financial statements as well as their projections for subsequent years. The Commission considered a number of financial ratios measuring the companies’ solidity, profitability, solvency and liquidity as well as its equity and cash flow situation. In addition, the Commission took into account relations with outside financial partners such as banks and relations with shareholders. The analysis also took into account restructuring plans.

The Commission assessed the specific social and economic context for each undertaking whose financial situation was found to be sufficiently critical. In this context, the impact of the global economic and financial crisis on the bathroom fitting sector was taken into account. The Commission also concluded for the five undertakings concerned that the fine would cause their assets to lose significant value.

As a result of the Commission’s analysis, the fines of three companies were reduced by 50% and those of another two by 25% given their difficult financial situation.91

8.108  There will continue to be much debate about the level of fines under competition law and the capacity of companies to pay them. Existing economic and financial criteria can be applied to produce the required ‘objective evidence’ of a company’s inability to pay. More case law and explanation of the Commission’s assessments would also be useful in providing clarity.

(p. 400) 8.8  Measuring the Costs and Benefits of Remedies, and of Competition Law

8.8.1  Why measure costs and benefits?

8.109  We have mentioned a number of times that any remedy that is being considered requires a careful CBA. Economics has a role to play in this kind of analysis. In the last ten years, competition authorities around the world have been placing great emphasis on measuring the effects of competition law enforcement, including remedies. Carrying out a CBA is generally good policy practice, and is increasingly required from policy-makers and regulatory authorities around the world. For example, the European Commission has issued Impact Assessment Guidelines to ensure that Commission initiatives and EU legislation are prepared on the basis of transparent, comprehensive, and balanced evidence.92 In competition law, CBA can be applied to specific interventions and remedies, or to the competition regime itself. A thorough ex ante CBA enables better policy decisions (e.g. more effective remedy design). Ex post CBA allows for the evaluation of past remedies and lessons to be learned for future actions.

8.110  There is an additional reason why measuring costs and benefits is important for competition policy. We have seen a spectacular proliferation of competition law globally. Awareness of competition policy has probably never been so widespread among businesses and the public at large. Proponents of competition policy would consider this growing awareness to be a positive development. It makes life easier for competition authorities, and probably means that more businesses refrain from engaging in illegal anti-competitive behaviour in the first place (the deterrence effect). However, this proliferation, in combination with high fines and an increasingly proactive stance taken by competition authorities, is bound to lead to someone asking at some point: what is it all good for? Competition policy needs a robust answer when it is held to account in this way. In recent years competition authorities have, quite successfully, played the ‘consumer welfare’ card in justifying their actions. It is a popular message to the public that competition law is there to protect consumers against anti-competitive practices by companies. However, this consumer welfare justification has some downsides as well, and should be used with care. For one, it can create false expectations and hence place competition authorities under pressure to intervene whenever a company is seen to be ‘ripping off’ its customers. Competition policy requires more than the consumer welfare argument to sustain political legitimacy. This is where the measurement of the effects of competition law enforcement comes in.

8.8.2  A remedy in dispute: CBA in the grocery inquiry

8.111  An example of where competition law expressly requires a form of CBA of remedies by the authority is the market investigation regime in the United Kingdom. In its 2008 investigation into grocery retailing, the CC found the market to be too concentrated and hindered by unnecessary barriers to entry.93 As part of a package of remedies it recommended that a (p. 401) ‘competition test’ be applied to retail planning applications to local government authorities for large grocery stores. The CC thus envisaged a change to local authority regulations that had previously been more or less indifferent as to who developed a particular piece of land. The competition test implied that planning permission would be granted if, in addition to satisfying the normal planning restrictions, the retailer was a new entrant in the local area; or the total number of fascias (i.e. grocery retailer brands) in the area was four or more; or there were three or fewer fascias but the retailer would have less than 60 per cent of sales within the local area. Under this test, it would be difficult for a retailer with more than a 60 per cent local market share to expand an existing store or to build a new one, as it would not obtain planning permission.

8.112  However, on appeal the CAT found that there was insufficient evidence that the costs of applying the competition test would be justified:

[T]here is a significant gap in the Commission’s analysis in relation to the ‘costs’ of the competition test. The Report does not fully and properly assess and take account of the risk that the application of the test might have adverse effects for consumers as a result of their being denied the benefit of developments which would enhance their welfare, including by leaving demand ‘unmet’.94

8.113  The CAT also criticized the CC’s assessment of the likely benefits of such a test:

[T]he Commission seems simply to have based its proportionality assessment on an assumption that the whole of the estimated customer detriment would be remedied by the test, in combination with the other remedies . . . There is in the Report no recognition or weighing of the non-acknowledged possibility that the existing AEC [adverse effects on competition] might not be satisfactorily remedied or mitigated for many years.95

8.114  The CAT’s ruling implies that advancing a remedy such as the competition test and establishing its effectiveness and proportionality requires a comprehensive CBA. Following the appeal, the CC duly undertook further analysis on the likely effects of the competition test. In a subsequent decision in 2009, it concluded that the test would deliver positive value to consumers on the basis that any reduction in consumer welfare in the short run would be offset by the longer-term benefit of increased competition.96 However, the CC also introduced a de minimis provision such that the test would be passed in cases where store extensions were small (less than 300 square metres), provided that the store had not been extended in the previous five years. The CC concluded that this new provision would not have a material impact on the effectiveness of the remedy, but would be a more proportionate response to the competition problems it had identified. This case illustrates the importance of undertaking the CBA as an integral part of the competition analysis, not as an afterthought.

8.8.3  What to measure: Identifying the counterfactual

8.115  CBA can be undertaken at distinct levels, depending on the policy question at hand. An important initial step is to identify the objective of the CBA and to be clear about the counterfactual against which costs and benefits must be measured. CBAs can be applied to (p. 402) remedies in individual competition cases, or to the competition regime more broadly. One distinction is between the costs and benefits of competition legislation and those of the competition authority. Both are relevant but distinct policy questions. To take the example of the Netherlands, a CBA could be undertaken for the Competition Law 1998. The relevant counterfactual for this analysis would be a situation in which the Competition Law was not enacted and the NMa (or its successor, the ACM) had not been created. This would basically be the situation pre-1998 when the Economic Competition Law 1956 was in place, enforced by the Ministry of Economic Affairs—a regime that was generally considered inactive (the Netherlands used to be regarded as a ‘cartel paradise’). In this counterfactual some reliance might be placed on interventions by the European Commission under the EU competition rules, which would come into play in the counterfactual without the 1998 law (in other words, the CBA of a national competition authority should consider only those cases that the European Commission would not cover, or cover less well). A separate CBA can be conducted for the competition authority itself. The counterfactual is one with a competition law in place but without a competition authority enforcing it.97 Thus, the costs and benefits of, say, the ACM would be assessed against a situation in which the Competition Law 1998 was not enforced by the authority but rather through private litigation. A relevant question to ask in such an analysis would be: which types of anti-competitive conduct can effectively be addressed through private actions? Many business-to-business disputes involving restrictive agreements or abuse of dominance probably can. The costs and benefits of these should then not be ascribed to the competition authority.

8.116  A further distinction—of most relevance for this chapter—can be made between the costs and benefits of the competition authority and the costs and benefits of specific decisions and remedies. For the former, the absence of the authority is the counterfactual. For the latter, the authority is assumed to be in place, and the incremental analysis refers to the costs and benefits that are attributable to the enforcement action or remedy in question. The analysis can be applied either to individual decisions or to a cumulative set of decisions (e.g. all merger decisions, or all abuse of dominance decisions).

8.8.4  How to measure: Categories of costs and benefits

8.117  The next question that arises is what costs and benefits should be included in the analysis, and how they should be weighed. At a superficial level you could do a very simple calculation: the substantial fines imposed by competition authorities in recent years are orders of magnitude higher than their annual budgets. Take the European Commission’s figures for 2013: it imposed fines of €1.7 billion on banks for fixing Euro and Yen interest rate derivatives, and €561 million on Microsoft.98 These fines alone far outweigh DG Competition’s operating expenditure for 2013, which was just short of €400 million.99 In one sense, therefore, it might be claimed that through its cartel and Microsoft actions (p. 403) alone DG Competition already provided substantially greater benefits to EU taxpayers than it has cost them.100 However, this comparison is not quite right: an important guiding principle for CBA is that it should be performed from a total economic welfare perspective. This means that costs and benefits to all the various participants in the economy—consumers, producers, government, taxpayers—should be included in the calculations. Money transfers between different participants—such as a fine paid by a company to the competition authority (or state treasurer)—are not a net benefit to the economy. As noted in Chapter 1, economists generally have little to say about distributive effects from transfers. However, if considered appropriate from a policy perspective, different weights can be given to different groups—for example, consumer welfare may be given greater weight than producer welfare, or poor consumers may be given greater weight than rich consumers.

8.118  Table 8.2 provides an overview of the main categories of costs and benefits in CBA. These same categories are of relevance to any policy question—that is, whether the CBA refers to competition legislation, the competition authority, or specific enforcement actions and remedies. The main difference will lie in the counterfactual against which the categories of costs and benefits are measured. For example, to assess the costs and benefits of the competition authority, the category of ‘direct costs of the authority’ needs to cover its entire budget. To assess a specific remedy imposed by that authority, the category covers only the costs incurred by the authority in relation to that decision (note that the table does not include income from fines as a direct benefit of the authority, for the reason set out above). The economic benefits of competition and (where this is the appropriate counterfactual) competition law enforcement can be measured in terms of productive and allocative efficiency, enhanced dynamic competition, enhanced market functioning, and wider effects on other sectors in the economy. As to the category of economic costs (negative market impacts), we note that while the objective of competition policy is to improve market functioning, actions by competition authorities can have (unintended) adverse consequences for the market as well. One indirect benefit that competition law may generate is making interventions consistent and providing clear guidance, such that regulatory certainty among businesses is enhanced.

Table 8.2  Main categories of (p. 404) costs and benefits to be assessed in the CBA



Direct (administrative) costs of the authority

Direct costs of companies

– Competition law compliance costs

– Costs of specific competition proceedings

Economic costs to the market in question (negative market impacts)

Economic benefits to the market in question (positive market impacts)

– Allocative inefficiency

– Allocative efficiency

– Productive inefficiency

– Productive efficiency

– Distortion of incentives (reduced dynamic competition/innovation)

  • – Enhanced dynamic competition/innovation

  • – Increased product/service quality

– Reduced product/service quality

– Enhanced market functioning

– Restriction on market functioning

Indirect regulatory costs

Indirect regulatory benefits

– Legal uncertainty

– Legal certainty

– Likelihood of regulatory capture

– Deterrent effects

– Improved quality of competition regime

Social costs (if relevant)

Social benefits (if relevant)

– Distributive costs

– Distributive benefits

– Reduced security/quality of supply

– Enhanced security/quality of supply

– Negative effect on vulnerable customers

– Positive effect on vulnerable customers

Source: Based on Oxera (2004).

8.8.5  When to stop measuring: Precision and priorities

8.119  The economics literature has developed a range of quantitative techniques that can be applied in CBA.101 Quantitative analysis should be undertaken where feasible in order to obtain robust results, but this analysis should establish rough orders of magnitude of the various costs and benefits rather than seek (often spurious) precision in the calculation.102 The optimal degree of quantification of the costs and benefits will depend on the circumstances. Not all costs and benefits can be readily quantified, either because of a lack of data or because the effects depend on various indirect economic interactions which are difficult to measure. Nevertheless, in practice, the assessment of rough orders of magnitude may be sufficient to gain insight into the costs and benefits of the policy decision or remedy in question.

8.120  In a CBA, you can often conclude that the consumer welfare benefits of intervening in price-fixing cartel cases—focusing on the total cartel overcharge paid by consumers—will be so great that they exceed the direct costs incurred by the competition authority by various orders of magnitude. Baker (2003) shows that rough estimates of the direct benefits of cartel actions in the United States far outweigh the estimates of the total costs of US antitrust enforcement (see Chapter 9 for a discussion of the overcharge levels that have been observed in past cartels). The additional indirect benefits of the interventions—in particular the enhancement of dynamic competition and the deterrent effects on other cartels—can be described in qualitative terms, because they work in the same direction and thus would reinforce the conclusion. One tentative policy conclusion that follows is that there is merit in giving priority to cartel enforcement in larger markets, as there the welfare benefits will be greatest. However, a qualification to this conclusion is that intervention against cartels in smaller markets can still fulfil an important signalling function. A handful of such actions in smaller markets may achieve a deterrent effect. Another reasonable conclusion is that if the objective of measuring costs and benefits is to show that competition policy benefits the economy as a whole, the case can be made based just on rough approximations of the benefits of the actions against cartels.

(p. 405) 8.121  The above approach to assessing the benefits of cartel actions cannot be readily applied to mergers, agreements other than cartels, or unilateral conduct. As noted throughout this book, remedial actions against these other practices are not as unambiguously beneficial as actions against cartels. The practices produce some efficiency benefits as well as anti-competitive effects, so intervention by the authorities can lead to negative market impacts. The extent to which a competition authority is able to strike the right balance between these two effects will depend on the quality of the underlying analysis in its investigations, and will also inherently involve a degree of judgement. The extent to which market forces can be trusted depends on the views of policy-makers and competition authorities, and on whether they consider it more desirable to avoid false positives or false negatives (false positives result in over-enforcement; false negatives in under-enforcement).

8.122  A specific action against a merger or business practice can therefore not automatically be presumed beneficial to social welfare. Indeed, CBA of a merger or abuse of dominance case runs the risk of becoming circular if it starts from the premise that the intervention is beneficial. An example of such circularity can be found in an analysis by the OFT of the effects of intervention against predatory pricing (Office of Fair Trading, 2005). In this CBA, the welfare loss of predation (and hence welfare benefit of intervention) was taken as the NPV of the low prices to consumers during the predation period (treated as a consumer welfare gain) and the high prices during the subsequent ‘recoupment’ period (treated as a consumer welfare loss). In calculating this, the OFT effectively assumed that, first, recoupment was indeed going to be feasible, and second, predation was about to become successful at the point of OFT intervention (i.e. the predator would switch from low to high prices at that point). However, these are precisely the factors that make predatory pricing cases so complex.

8.123  Should a formal CBA be required for every competition law remedy? This policy question would perhaps merit a CBA of its own. Too formal a requirement on competition authorities could make competition law enforcement more like certain other policy areas where the possible effects of specific interventions are sometimes argued over for a considerable length of time. The general lesson for competition law is that CBA can be a useful policy tool that helps competition authorities develop their thinking about a competition problem and how to solve it, even if not all costs and benefits are quantified with precision in every case.(p. 406)


1  Comment made by Tad Lipsky at the 2007 FTC hearings on s 2 of the Sherman Act (transcript of 28 March, p. 47, at: <http://www.ftc.gov/os/sectiontwohearings/docs/transcripts/070328.pdf>). The quote is attributed to William Baxter, the Department of Justice Assistant Attorney General, who in 1982 broke up AT&T, a case discussed in this chapter.

2  Department of Justice (2008), ‘Competition and Monopoly: Single-firm Conduct Under Section 2 of the Sherman Act’, September, Ch 9. The DOJ withdrew this report in 2009, but the chapter on remedies remains a useful contribution to the debate.

3  For example, European Commission (2005), Deloitte (2009), Organisation for Economic Co-operation and Development (2012b and 2012c), and Kwoka (2015).

4  See, for example, Commission notice on remedies acceptable under Council Regulation (EC) No 139/2004 and under Commission Regulation (EC) No 802/2004 (2008/C 267/01); and Federal Trade Commission (2012), ‘Negotiating Merger Remedies’, January. The distinction between structural and behavioural remedies is not always clear-cut. For example, some authorities regard the licensing of IP as a structural remedy while others treat it as a behavioural remedy.

5  Unilever/Sara Lee Body Care (Case COMP/M.5658), Decision of 17 November 2010; and Colgate Palmolive/Sanex Business (Case COMP/M.6221), Decision of 6 June 2011.

6  Department of Justice (2008), ‘Competition and Monopoly: Single-firm Conduct Under Section 2 of the Sherman Act’, September, p. 123.

7  Verizon Communications Inc v Law Offices of Curtis V. Trinko, LLP 540 US 398 (2004).

8  European Commission (2006), ‘Guidelines on the Method of Setting Fines Pursuant to Article 23(2)(a) of Regulation No 1/2003’ [2006] OJ C210/02.

9  Pernod Ricard/Allied Domecq (Case COMP/M.3779), Decision of 24 June 2005; and Competition and Markets Authority (2014), ‘CMA accepts Scotch whisky business sale’, press release, 31 October.

10  Novartis/GlaxoSmithKline Oncology Business (Case COMP/M.7275), Decision of 28 January 2015; and Mylan/Abbott EPD-DM (Case COMP/M.7379), Decision of 28 January 2015. We advised the acquiring party in the latter case.

11  Competition Commission (2013), ‘The Royal Bournemouth and Christchurch Hospitals NHS Foundation Trust/Poole Hospital NHS Foundation Trust’, 17 October, at [9.206].

12  See Organisation for Economic Co-operation and Development (2012b, pp. 131–2 and 293; and 2012c, pp. 94–5).

13  State aid No. N 428/2009—United Kingdom Restructuring of Lloyds Banking Group, C(2009)9087 final, Decision of 18 November 2009; and State aid No N 422/2009 and N 621/2009—United Kingdom Restructuring of Royal Bank of Scotland following its recapitalisation by the State and its participation in the Asset Protection Scheme, C(2009)10112 final, Decision 14 December 2009.

14  Fortis/ABN AMRO Assets (Case COMP/M.4844), Decision of 3 October 2007. We advised the merging parties during this inquiry.

15  Ryanair/Aer Lingus (Case COMP/M.4439), Decision of 27 June 2007; Ryanair/Aer Lingus II (Case COMP/M.5434), Withdrawn on 23 January 2009; and Ryanair/Aer Lingus III (Case COMP/M.6663), Decision of 27 February 2013. We advised Ryanair on the most recent case.

16  Fix-it-first remedies are defined in the Commission’s Remedies Notice. Commission notice on remedies acceptable under Council Regulation (EC) No 139/2004 and under Commission Regulation (EC) No 802/2004 (2008/C 267/01), at [56–7].

17  Competition Commission (2013), ‘Ryanair Holdings plc and Aer Lingus Group plc’, 28 August; and IAG/Aer Lingus (Case COMP/M.7541), Decision of 14 July 2015.

18  United States and State of New York v Twin America LLC et al., Civil Action No. 12-cv-8989 (ALC)(GWG).

19  In Re NYC Bus Tour Antitrust Litigation, No. 13-CV-0711 (ALC)(GWG)(S.D.N.Y. 21 October 2014). Those having been on a hop-on, hop-off tour through New York could apply for compensation on <tourbussettlement.com>.

20  Standard Oil Co of New Jersey v United States 221 US 1 (1911).

21  United States v AT&T Co 552 F Supp 131 (DDC 1982).

22  Verizon Communications Inc. v Law Offices of Curtis V Trinko, LLP 540 US 398 (2004).

23  United States v Microsoft Corp., 97 F.Supp. 2d59 (D.D.C. 2000).

24  United States v Microsoft Corp., Nos 98-1232 and 98-1233 (D.D.C.), Revised Proposed Final Judgment, 6 November 2001.

25  European Parliament (2014), ‘MEPs zero in on internet search companies and clouds’, press release, 27 November.

26  CEZ (Case AT/39727), Decision of 10 April 2013, at [79].

27  German Electricity Wholesale Market and German Electricity Balancing Market (Cases COMP/39.388 and COMP/39.389), Decision of 26 November 2008; RWE Gas Foreclosure (Case COMP/39.402), Decision of 18 March 2009; and ENI (Case COMP/39.315), Decision of 29 September 2010.

28  Monopolies and Mergers Commission (1989), ‘The Supply of Beer’, March.

29  Competition Commission (2009), ‘BAA Airports Market Investigation’, March. BAA in fact ended up selling more airports and is now called Heathrow Airport Holdings Limited.

30  Competition and Markets Authority (2014), ‘Private Healthcare Market Investigation’, final report, 2 April.

31  For an international overview, see Organisation for Economic Co-operation and Development (2012a).

32  See Niels et al. (2003); and Financial Conduct Authority Payment Systems Regulator (2015), ‘A new regulatory framework for payment systems in the UK’, policy statement, 27 March.

33  For an overview, see Armstrong et al. (1994) and Decker (2014).

34  For an overview, see Department for Transport and Office of Rail Regulation (2010).

35  Liberty Global/Corelio/W&W/De Vijver Media (Case COMP/M.7194), Decision of 24 February 2015; and Liberty Global/Ziggo (Case COMP/M.7000), Decision of 10 October 2014. We advised the acquirer on both cases.

36  There was also a divestment remedy further upstream, as the deal combined the two main premium film channels in the Netherlands, and one of these had to be sold off.

37  Arriva The Shires Ltd v London Luton Airport Operations Ltd [2014] EWHC 64 (Ch). [2014] UKCLR 313. We acted as experts for the claimant in this matter.

38  Standard & Poor’s (Case COMP/39.592), Decision of 15 November 2011.

39  A classic, and very accessible, work on the economics of regulation is Kahn (1988) (first published in two separate volumes in 1970 and 1971). A leading textbook that economists often refer to is Laffont and Tirole (1993). More recent textbooks are Viscusi et al. (2005), Crew and Parker (2006), and Decker (2014).

40  Attheraces v British Horseracing Board [2007] EWCA Civ 38, 2 February 2007.

41  Albion Water Limited and ors v Water Services Regulation Authority and ors [2008] CAT 31, Judgment on Unfair Pricing, 7 November 2008.

42  European Commission (1998), ‘XXVIIth Report on Competition Policy (1997)’, p. 26.

43  Albion Water Limited and ors v Water Services Regulation Authority and ors [2006] CAT 36, Judgment, 18 December 2006 (this was an earlier judgment in the same case cited above in the discussion of LRIC pricing); and Commerce Commission v Telecom Corporation of New Zealand Limited and Telecom New Zealand Limited CIV 2004-404-1333, Judgment of 9 October 2009, High Court of New Zealand.

44  Skyscanner Limited v Competition and Markets Authority, [2014] CAT 16, 26 September 2014. We advised Skyscanner on this matter.

45  Ofcom (2005), ‘Provision of technical platform services: A consultation on proposed guidance as to how Ofcom may interpret the meaning of “Fair, Reasonable and Non-discriminatory” and other regulatory conditions when assessing charges and terms offered by regulated providers of technical platform services’, November; and Microsoft (Case COMP/C-3/37.792), Decision of 24 March 2004.

46  Rambus Inc v FTC 522 F 3d 456 (US Ct of Apps (District of Colombia Circuit), 2008).

47  Ibid., at [5].

48  Rambus (Case COMP/38.636), Decision of 9 December 2009.

49  European Commission (2009), ‘Antitrust: Commission closes formal proceedings against Qualcomm’, press release, MEMO/09/516, 24 November.

50  Motorola—Enforcement of GPRS standard essential patents (Case AT.39985), Decision of 29 April 2014; and Samsung—Enforcement of UMTS standard essential patents (Case AT.39939), Decision of 29 April 2014. See also European Commission (2014).

51  This option is described as the ‘Shapley solution’ in Layne-Farrar et al. (2007). The underlying thinking is derived from Shapley (1953).

52  In re Innovatio IP Ventures, LLC, Patent Litigation, No 11-cv-09308 (N.D. Ill. 3 October 2013), Memorandum Opinion, Findings, Conclusions, and Order, at [37].

53  Georgia-Pacific Corp. v United States Plywood Corp., 318 F. Supp. 1116, 166 U.S.P.Q. (BNA) 235 (S.D.N.Y. 1970).

54  Oracle America, Inc. v Google Inc, 798 F. Supp. 2d 1111 (N.D. Cal. 2011).

55  Ibid.

56  Microsoft Corp. v Motorola, Inc., No. 10-cv-1823 (W.D. Wash.), Findings of Fact and Conclusions of Law, 25 April 2013.

58  See Financial Conduct Authority (2015), ‘General insurance add-ons market study—proposed remedies: banning opt-out selling across financial services and supporting informed decision-making for add-on buyers’, consultation paper, March; and London Economics and YouGov (2014).

59  Ofcom (2011), ‘Automatically Renewable Contracts: Decision on a General Condition to prohibit ARCs’, 13 September.

60  Competition and Markets Authority (2014), ‘Private Motor Insurance Market Investigation’, final report, 24 September. We advised one of the insurers in this investigation.

61  Competition Commission (2006), ‘Market Investigation into Personal Current Account Banking Services in Northern Ireland—Provisional Findings’, 12 October, at [8] and Appendix 4.3. We advised one of the banks in this investigation.

62  Competition Commission (2009), ‘Market Investigation into Payment Protection Insurance’, 29 January. We advised one of the PPI providers during this investigation.

63  Barclays Bank PLC v Competition Commission, [2009] CAT 27, 16 October 2009.

64  Competition Commission (2010), ‘Payment Protection Insurance Market Investigation: Remittal of the Point-of-sale Prohibition Remedy by the Competition Appeal Tribunal’, 14 October.

65  Microsoft (Case COMP/C-3/37.792), Decision of 24 March 2004.

66  Ibid, at [845].

67  Ibid, at [870].

68  Microsoft (Tying) (Case COMP/39.530), Decision of 16 December 2009; and United States v Microsoft Corp., 97 F.Supp. 2d59 (D.D.C. 2000).

69  Microsoft (Tying) (Case COMP/39.530), Decision of 16 December 2009, at [81].

70  Microsoft (Tying) (Case COMP/39.530), Decision of 6 March 2013.

71  European Commission (2010), ‘Antitrust: Commission probes allegations of antitrust violations by Google’, press release, 30 November; and European Commission (2015), ‘Antitrust: Commission sends statement of objections to Google on comparison shopping service; opens separate formal investigation on Android’, press release, 15 April 2015.

72  European Commission (2014), ‘Statement on the Google Investigation’, press conference, 5 February.

73  European Commission (2014), ‘Statement on the Google investigation’, press conference, 5 February.

74  See the European Commission document ‘Cartel Statistics’, which is regularly updated, at: <http://ec.europa.eu/competition/cartels/statistics/statistics.pdf>.

75  Microsoft (Case COMP/C-3/37.792), Decision of 27 February 2008; Case T-167/08 Microsoft Corp. v Commission, Judgment of 27 June 2012; and Microsoft (Tying) (Case COMP/39.530), Decision of 6 March 2013.

76  Intel (Case COMP/C-3/37.990), Decision of 13 May 2009; and Case T-286/09 Intel Corp v Commission, Judgment of 12 June 2014.

77  See under ‘Kartellverbot’ on <www.bundeskartellamt.de>.

78  Autorité de la concurrence (2014), ‘Home and personal care products sold in supermarkets—The Autorité de la concurrence fines concerted practices between manufacturers a total of 345,2 € millions and €605,9 € millions on each market concerned’, press release, 18 December. Note that not all competition authorities have jumped on the high-fines bandwagon. The ACM in the Netherlands prefers to resolve matters by holding a dialogue with the companies under investigation first, and persuading or nudging them to change their ways. See, for example, Vane (2015).

79  European Commission (2006), ‘Guidelines on the Method of Setting Fines Pursuant to Article 23(2)(a) of Regulation No 1/2003’ [2006] OJ C210/02.

80  Leading works include Becker (1968), Landes (1983), and Polinsky and Shavell (2000).

81  Levitt and Dubner (2005), Ch 1. The original study was Gneezy and Rustichini (2000).

82  Case T-53/03 BPB v Commission, Judgment of 8 July 2008, at [336].

83  For example, a review of regulatory penalties for the UK government recommended that regulators should be transparent in the methodology for determining or calculating administrative financial penalties. See Macrory (2006).

84  European Commission (2006), ‘Guidelines on the Method of Setting Fines Pursuant to Article 23(2)(a) of Regulation No 1/2003’ [2003] OJ C210/02, at [6].

85  Chloroprene Rubber (Case COMP 38.629), Decision of 5 December 2007.

86  See, for example, the ECJ ruling on the copper tubes cartel: Case C-272/09 P, KME v Commission, Judgment of 8 December 2011.

87  Joined Cases C-93/13 P and C-123/13 P, Commission and others v Versalis and others, Judgment of 5 March 2015.

88  Case T-23/99 LR af 1998 A/S v Commission, Judgment of 20 March 2002.

89  European Commission (2006), ‘Guidelines on the Method of Setting Fines Pursuant to Article 23(2)(a) of Regulation No 1/2003’ [2006] OJ C210/02, at [35].

90  There is some high-level guidance in European Commission (2010), ‘Inability to pay under paragraph 35 of the Fining Guidelines and payment conditions pre- and post-decision finding an infringement and imposing fines; Information note by Mr Joaquín Almunia, Vice-President of the Commission, and by Mr Janusz Lewandowski, Member of the Commission’, 12 June.

91  Bathroom fittings and fixtures (Case COMP/39.092), Decision of 23 June 2010, at [18–20] of summary of decision.

92  European Commission (2009), ‘Impact Assessment Guidelines’, SEC(2009) 92, 15 January. See also European Commission (2014), ‘Guide to Cost-benefit Analysis of Investment Projects’, December.

93  Competition Commission (2008), ‘The Supply of Groceries in the UK—Market Investigation’, 30 April. We advised one of the retailers in this investigation.

94  [2009] CAT 6, Case 1104/6/8/08 Tesco PLC v Competition Commission, 4 March 2009, at [111].

95  Ibid., at [162].

96  Competition Commission (2009), ‘Groceries Market Investigation: Remittal of the Competition Test by the Competition Appeal Tribunal: Decision’, 2 October.

97  The reverse situation—a competition authority without a competition law—seems less plausible. Yet this was the situation in Jersey for some time. The Jersey Competition Regulatory Authority (JCRA) was set up in 2001, but the Competition (Jersey) Law did not come into force until 2005 (in the intervening period the JCRA did have certain regulatory responsibilities in the telecoms sector).

98  European Commission (2013), ‘Antitrust: Commission fines banks €1.71 billion for participating in cartels in the interest rate derivatives industry’, press release, 4 December 2013; and Microsoft (Tying) (Case COMP/39.530), Decision of 6 March 2013.

99  Figure from ‘DG Competition Annual Activity Report 2013’, Annex 3, available at <ec.europa.eu/atwork/synthesis/aar/index_en.htm>.

100  Assuming that the proceeds from the fines do indeed ultimately flow to taxpayers. We don’t know how realistic this assumption is.

101  See Sugden and Williams (1978) and Boardman et al. (2010).

102  Manski (2013) provides a sound health warning on the limitations of CBA in an uncertain world.