Abstract
Although across-platform parity agreements (APPAs) are increasingly scrutinized, their exclusionary effects remain ambiguous. This article posits that the practice can either prevent entry or lead to the exit of rival platforms. Assessing this foreclosure effect presupposes, however, using one of the available tests, namely, the no economic sense test, the as efficient competitor test, and the consumer welfare test. Whether APPAs lead to the foreclosure of rivals depends ultimately on the applied test and, sometimes, on the way the test is applied. This reflects the paradox of APPAs. In an effort to identify the most appropriate test to deal with APPAs’ foreclosure effect, this article also suggests that the consumer welfare test is far superior to the others.
Keywords
I. Introduction
The rise of the digital economy challenges in many aspects the common ways of dealing with competition concerns. A particular manifestation of this metamorphosis is the emergence or sometimes the renewed interest in new forms of vertical restraints, one of which is across-platform parity agreements (hereinafter, APPA).
APPAs have been labelled as a “special category of price relationship agreements.” 1 Within this special category, several versions have been identified. 2 The present article focuses on the most widespread version of APPAs, which is an agreement between a producer and a platform 3 whereby the producer “undertakes to charge on that platform a price that is not higher than the price charged on other platforms, including the new entrants.” 4
The consequences of such a practice remain ambiguous, as it has received little attention from both economic and legal literature. Recently, the clause gained more visibility due to high-profile cases such as Booking investigations, 5 the E-Book case, 6 and Amazon’s investigations. 7 Despite the fact that those investigations gave rise to major divergences between National Competition Authorities (NCAs), 8 it also appears that the practice has been mostly approached through the lens of Article 101 of the Treaty on the Functioning of the European Union (TFEU) by assessing its collusive effect. 9
This article deviates from such an approach. It focuses on the potential anticompetitive foreclosure of the practice 10 by asking whether the use of an APPA by a dominant platform inevitably exclude rivals. In practice, an APPA may obstruct entry or expansion of rival platforms by reducing their ability to attract users through lower commissions or by implementing a differentiated business model. 11 However, assessing this foreclosure effect is a difficult task due to the absence of precedents 12 and the lack of any reference to APPAs, in both the European Commission’s Guidelines on Vertical Restraints 13 and guidance on article 102.
That is precisely where the interest of this article lies. I assess the potential anticompetitive foreclosure of the APPAs using three different tests, namely, (1) the consumer welfare test, (2) the no-economic-sense test, and (3) the equally efficient competitor test. By doing so, the article highlights the merits of each test and identifies the most appropriate one for purposes of defining APPAs’ anticompetitive foreclosure. This is in substance the focus of the fourth, fifth, and sixth parts.
Part II analyzes the functioning of competition in the absence of an APPA, highlighting the importance of an agency business model for a foreclosure effect to occur and the key factors for competition between platforms.
Part III explores the functioning of competition between platforms when an APPA is introduced by highlighting the impact of the practice on market players and the consequences flowing therefrom with regard to market structure.
II. Platform Competition in the Absence of an APPA
This part intends to briefly review the competitive environment of platforms. First, it describes the role of platforms; and second, it reviews the key determinants to competition between them and which matters for the analysis of foreclosure effects.
A. The Role of Platforms: Agents or Resellers?
From an economic point of view, the distinction between platforms acting as agents or as resellers is not devoid of interest.
In the wholesale model, platforms are seen as retailers. As a result, the supplier loses the capacity to set retail (final) price. In practice, suppliers sell the product to the retailer, who in turn sells it to consumers at the price it chooses. In such a setting, suppliers cannot be asked to match retail price offered by other platforms as their influence is limited to the wholesale price. Even if a wholesale price parity agreement is introduced, the retail price can still diverge from one platform to another due to diverging cost structures. 14
By contrast, in the agency model, the supplier preserves its freedom to set retail prices. Knowing the commission charged by each agent-platform, a supplier is in fact the one to determine the final retail price and to bear the economic and financial risk. 15 The platform is simply endowed with a “matching” 16 function that consists of bringing together users on both sides of the platform. This does not, however, mean that competition between platforms is excluded. Quite the opposite, platforms can still do so by offering a lower commission to suppliers. This is exactly the context under which APPAs raise competition concerns by neutralizing the impact of commissions on platforms’ rivalry.
In the remainder of this article, the focus will be exclusively on the impact of an APPA in an agency model.
B. Key Determinants of Platform Competition
1. Indirect network effects
Network effects are a critical factor in competition between platforms. 17 In this regard, the concept of indirect network effects, which refers to the fact “that the value that a customer on one side realizes from the platform increases with the number of customers on the other side,” 18 is particularly relevant.
Yet the concept’s competitive implications are not always clear. In some cases, indirect network effects may have procompetitive effects. This happens when a growing number of users interact with each other leading to an improvement in both the quality and the value of the platform. 19
In other cases, however, indirect network effects may give rise to barriers to entry and expansion at the platform level. 20 New entrants or existing rivals must first reach a critical mass in order to efficiently operate in the market. 21 Ultimately, this may drive the market to tip in favor of a particular platform.
Remedying this situation presupposes convincing consumers to switch to the new platform. Two elements play a critical role in this process, namely, the ability to multi-home and differentiation.
2. Multi-homing versus single-homing
The degree to which competition operates between platforms depends, inter alia, on consumers’ reliance on an exclusive platform (single-homing) 22 or their use of several competing platforms (multi-homing). 23 More precisely, if for a given transaction a consumer compares prices on one platform and concludes the transaction on that same platform, it is said to single-home. On the contrary, when a consumer uses several platforms to compare prices and then concludes the transaction on one of those platforms, the consumer is said to multi-home. 24
This consumer behavior affects the competition dynamics between platforms. If, in a given market, consumers are known to single-home, competition is likely to focus on capturing their attention. 25 This is due to the fact that once a consumer is attracted to the platform, the transaction is likely to be concluded on that platform. However, when consumers tend to use various platforms in parallel, it is likely that competition will occur with regard to the price dimension. 26 Indeed, it is not enough to draw consumers’ attention; a platform needs to offer consumers the best deal in order to drive them to conclude the transaction on the said platform and not turn to another one. Consequently, price becomes decisive in platforms’ competition.
In accordance with the above, suppliers react in the following terms: When a large fraction of consumers single-home, suppliers are more likely to multi-home. They need to be present on different platforms in order to reach the largest possible share of consumers. By contrast, when consumers tend to multi-home, suppliers do not need to be listed on every platform. 27 A supplier can focus on some major platforms and remains competitive given that consumers compare prices and offers of different suppliers.
As will be discussed below, suppliers’ tendency to single-home or multi-home impacts greatly the extent to which an APPA can lead to foreclosure effects. The reason for this lies in the fact that the practice interferes with competitors’ possibility to reach a minimum efficient scale. 28
3. Platform differentiation
Differentiation is another determinant in the competition between platforms. It can be vertical or horizontal. 29
In the absence of an APPA, horizontal differentiation, which focuses on nonquality attributes, is mainly what allows for the coexistence of several competing platforms. 30 The reason for this lies in the fact that, even when products are offered at the same price across different platforms, the use of which does not entail any cost differences, consumers would not agree about the platform on which it is preferred to buy the product. In other words, platforms’ horizontal traits are good for some consumers but not for all of them. This is what explains the emergence of several platforms, each one corresponding to the preferences of a given category of users. Such a possible horizontal differentiation generates possibilities for entry in markets with strong network effects.
Vertical differentiation can be said to play an important role if consumers were to agree to purchase on a given platform a product that is offered at the equal price across various platforms, the use of which is at equal cost. Quality will play a decisive role in orientating consumers toward one platform or another. Yet in double-sided markets, quality does not only depend on the experience the platform offers but is also intrinsically linked to indirect network effects. 31 The larger the number of users on both sides of the platform is, the higher its utility for consumers and thus its perceived quality. 32
However, for successful entry in the platform market, differentiation needs to be carefully assessed by acknowledging the double-sided nature of the market. 33 As explained by Evans, “for multi-sided platforms, by determining the customers on one side, horizontal and vertical differentiation affect demand on the other sides. Because of these interdependencies, a platform must usually make differentiation decisions jointly for all of the sides it serves.” 34
However, as will be shown below, the presence of an APPA can alter this assessment by influencing “the type of entrant measured by how vertically differentiated it chooses to be,” 35 which can ultimately lead to a loss in terms of differentiation.
III. Platform Competition in the Presence of an APPA
Now that the dynamics of platforms’ competition have been briefly reviewed, it is important to analyze how the introduction of an APPA changes this setting. To this end, the next part will examine the impact of an APPA on market players’ behavior and the potential consequence on market structure.
A. The Alteration of Market Players’ Incentives
1. For consumers
The introduction of an APPA can distort consumers’ behavior on a platform in two ways.
First, an APPA can increase the levels of single-homing. Horizontal differentiation is what results in consumers multi-homing. Price and platform features are decisive in this regard. However, price plays a limited role in the presence of an APPA. Knowing that the price quoted on one platform will never be lower than the one on other platforms, consumers are unlikely to check several platforms in their search for a given product or service. When this happens, single-homing rates among consumers may increase, even though it can be counteracted by platforms’ overinvestment in features that appeal to groups of customers. 36
Second, if an APPA is agreed upon, consumer choice is likely to be limited, which may result in some consumers being forced to switch or select a platform they do not prefer. To understand this, one has to keep in mind that there are always consumers with a low willingness to pay and who prefer using low-cost platforms. 37 However, a parity agreement reduces the impact of price as one of the most important differentiating factors, especially from the perspective of consumers wishing to use low-cost platforms. As low-cost competition is no longer possible, platforms maybe driven to ignore the preferences of those consumers and focus on high-value ones. Consequently, as put forward by Corts, APPAs “may be limiting options targeting those consumers, driving them in effect to purchase bundled platform amenities they do not value.” 38
2. For platforms
From the perspective of platforms, the existence of an APPA shifts the focus from price differentiation to retailers’ differentiation. As price differentiation is no longer possible, it becomes a lot more economically viable to attract consumers by focusing on non-pricing parameters. Two main risks are associated with this option.
First, turning to “any other mean of differentiating itself” 39 could shift platforms’ focus to overinvestment in features that appeal to different groups on the buyer side. Empirical results show that, in those circumstances, an APPA is bad as it “causes the intermediary to over-invest in buyer-side benefits.” 40 One can easily imagine that this excessive investment can make market entry more difficult.
The second risk, closely linked to the first one, refers to excessive advertisement spending. With an APPA in place, the possibility to achieve a differentiated product becomes more centered on the ability to mobilize substantial marketing investments. 41 This can lead platforms to spend even more on advertisement. 42 In this way, advertisement can become an increasingly important barrier to entry in the platform market. This analysis is consistent with the finding that consumers are likely to single-home in the presence of an APPA, leading platforms to fiercely compete for their attention by investing heavily in advertisement. 43
This is also in line with Boik and Corts’ finding according to which a new entrant platform would, in the presence of an APPA, favor a quality level similar to that of the dominant platform. 44 In other words, vertical differentiation is somehow neutralized in the presence of an APPA. As explained by Hviid, “This result arises because the APPA reduces the intensity of competition and hence the desire to differentiate as much for the incumbent.” 45 This argument does not contradict the argument discussed above regarding overinvestment in platforms’ features. The overinvestment consequence is not necessarily in meaningful quality traits but in features that appeal to different categories of users.
3. For suppliers
The question here is mainly approached from the perspective of how an APPA affects suppliers’ incentives with regard to entry at the platform level. Given that APPAs favor the emergence of homogenous platforms and reduce the scope of price differentiation, suppliers are likely to be encouraged to single-home.
This is, however, unlikely to be the case in practice. Most of the time, suppliers’ behavior needs to be assessed in the light of APPA’s impact on users on the other side of the platform. The most important effect in this regard flows from the fact that an APPA favors consumers’ single-homing. This results in suppliers being more inclined to list on several platforms so as to reach the largest share of consumers.
An additional impact for suppliers relates to the restricted possibility to differentiate between platforms in term of prices. It is true that an APPA maintains suppliers’ freedom to set prices on a given platform. However, this freedom is immediately restricted when suppliers decide to multi-home. This restricted possibility to differentiate limits suppliers’ ability to sponsor or promote 46 the entry of a new platform or the expansion of an existing one. This is because suppliers cannot price more competitively in those platforms as a reaction to a dominant platform imposing a higher commission fee.
From this perspective, an APPA can be said to lock in suppliers by limiting their ability to stimulate competition between platforms. In case of a high commission fee increase, a supplier’s reaction would mostly be limited to assessing the cost and the benefits from being on the dominant platform or delisting. 47
B. The Consequences for Market Structure
The intervention of an APPA impacts the market structure by favoring a competition to dominance outcome and at the same time raising barriers to entry.
First, two-sided markets are predisposed to high market concentration given that indirect network effects favor a winner-takes-all outcome. 48 In absence of an APPA, other characteristics such as differentiation and multi-homing oppose market concentration allowing for several platforms to coexist. Nevertheless, with an APPA in place, the impact of differentiation and multi-homing as competition drivers is mitigated. 49 In clear, the practice manipulates indirect network effects, sometimes, to the profit of dominant platforms. This ultimately enhances the risk of a competition to dominance outcome.
In theory, differentiation can help mitigating network effects. It is “a key reason why many industries with multi-sided platforms have multiple competitors even though indirect network effects and sometimes economies of scale would seem to propel them to monopolies.” 50 Any restriction on the degree of differentiation between platforms may prevent the emergence of a new platform by restricting possibilities to achieve a critical mass. 51 However, an APPA leads to a loss in terms of differentiation particularly the vertical dimension. Support may be drawn from Boit and Corts’ findings that the practice may “distort the entrant’s choice away from a lower-end business model (and toward a model more similar to that of the incumbent).” 52 Similarly, the practice reduces also possibilities for differentiation with regard to prices and the emergence of new business model. All of this strengthens the impact of indirect network effects. 53
The same analysis applies to multi-homing. The ability to use several platforms increases substitutability between them and thus reduces the likelihood of a winner-takes-all outcome. However, the analysis of market players’ behavior has revealed that APPA favors a specific users’ adhesion pattern consisting in competitive bottlenecks whereby suppliers multi-home while buyers single-home. 54 This configuration cannot be said to favor overall market concentration, given that one side still multi-homes. 55 However, the issue has to be carefully assessed as the multi-homing side can be charged very high commission. Consequently, this may limit platforms’ attractiveness to suppliers, which ultimately can decide to focus on their direct sale channels.
Second, an APPA can, in parallel, contribute to reduce the degree of market contestability. This effect is assessed in part IV.A.1.b, below. Suffice it to say here that the limited possibility to differentiate favors a concentration of similar or nondifferentiated platforms. Differentiated platforms either in terms of price or business model may be prevented from efficiently entering the market.
In general, platforms benefiting from indirect network effects should not strive to strength their market position by manipulating those network effects. 56 The key element here is the prevention of any strategy that may strength the anticompetitive impact of network effects by locking in users and preventing rival platforms from reaching a critical mass. The previous analysis has revealed that an APPA can reduce the impact of competition’s driving factors, namely, differentiation and multi-homing and, thus, render it more difficult for actual and potential rival platforms to reach a critical mass in order to efficiently operate in the market.
Consequently, such a process can immunize dominant platforms from competition. It then becomes important to scrutinize practices like APPAs, which may lead to such an outcome.
IV. Evaluating APPAs Under the Consumer Welfare Test: A Straightforward Test?
The implementation of the consumer welfare test presupposes that APPAs could foreclose a competitor. 57 Consumer harm resulting from such a foreclosure needs to be balanced against efficiencies benefiting those consumers. 58 If the overall effect is positive, the conduct can be declared lawful.
The test echoes the court’s view as expressed in Post Denmark I, where it stated that a practice leading to the exclusion of rivals can still be justified on efficiency grounds. 59
In line with the above definition, the following analysis will assess APPAs’ net effect on consumers.
A. Capturing the Anticompetitive Foreclosure Impact of APPAs
To begin, it is important to understand that APPAs are vertical agreements with a horizontal impact. 60 In the vertical dimension, a dominant platform defines conditions under which suppliers will be dealing with the platform’s competitors. Low-cost platforms will be forced to display a similar price to less efficient rival platforms. In doing so, the dominant platform prevents rivals from obtaining cheap inputs. In this way, the dominant firm worsens the conditions of supply for rival platforms. In other words, the incumbent platform reshapes the competitive constraints it has to face and that would otherwise have been determined by market forces.
At this stage, it is however important not to draw a hasty conclusion about the existence of an anticompetitive foreclosure. It is not enough to identify that, in the absence of an APPA, rivals could obtain cheap input leading the dominant platform to face a serious competitive constraint. The identification of the anticompetitive foreclosure requires “the likelihood that, as a result of the exclusion or marginalization of competitors, the dominant undertaking will be in a position profitably to increase or maintain prices above the competitive level or negatively to affect the non-price parameters of competition.” 61 In short, it is important to establish that the dominant platform has both the capacity and the incentives to foreclose rivals.
1. Ability to foreclose under APPAs
Proof of the dominant platform’s capacity to totally or partially foreclose rivals is indispensable. 62 This effect can affect either actual or potential competitors.
a. APPAs’ impact on actual competition
The starting point is the existing competition in the market prior to the practice. 63 If the market was already dominated by the implicated platform prior to the implementation of the APPA, any additional restriction on competition, however small, may suffice to characterize an elimination of effective competition. 64 The evaluation has, however, to be comprehensive. It must take into account the different dynamics of competition in the relevant market.
In this regard, it is important to remember that APPAs have a direct impact on some parameters of competition in particular prices. They eliminate both actual and potential price-based competition. Sometimes, the impact extends to major quality and quantity aspects. 65 Accordingly, the assessment should establish whether demand in the relevant market is a price-sensitive one. If price is an essential parameter of competition, the impact of APPAs on effective competition will be substantial. On the contrary, when quality is a determinant factor, a restriction on prices will have limited impact on the intensity of competition unless the scope of the APPA extends to major qualitative aspects.
In weighing the dominant platform’s capacity to foreclose competitors, it is important to investigate if residual competition focuses on parameters other than the ones covered by the APPA. The purpose is to establish whether the incumbent platform enjoys a net advantage in this respect. It is probable that competition will be eliminated if it appears that the dominant platform enjoys a real competitive advantage in respect of key non-price parameters (namely, reputation effect, market anteriority, established customer base, etc.) while, at the same time, rivals cannot compete and differentiate themselves in term of prices.
The evolution of the market position of different platforms is an additional crucial parameter. 66 This condition helps understanding the competitive implication of APPAs on the ability and the willingness of rivals to react by adopting counter strategies. 67 This is the case when, for instance, a dominant platform loses its leading position in favor of a competitor despite the implementation of an APPA. 68 In the same way, whether a platform has managed to achieve a dominant position despite the fact that rivals were implementing an APPA can also be telling. 69
The extent to which effective competition is eliminated is furthermore dependent on how widely the relevant sector relies on APPAs. The impact on competition is enhanced if several platforms implement such agreements. 70 This is more likely to materialize when the supplier side is highly fragmented 71 or when suppliers need to multi-home. In such a case, every platform, even with limited market power can impose such an agreement. 72
b. APPAs’ impact on potential competition
Whether effective competition is eliminated varies according to APPAs’ impact on potential competition.
73
The focus is on how APPAs alter the conditions of entry by making the possibility of new entry on a significant scale more or less likely.
74
(i) Stimulating entry or artificial advantage for potential rivals?
The stimulating impact of an APPA on platform entry is mainly linked to the price effect of the practice. The intuition behind this argument is that an APPA combined with an agency model leads to high profits, 75 which, in turn, can attract new entries. From a foreclosure perspective, this triggers the question of whether an APPA can, at an initial stage, help new platforms developing in the market and subsequently try to become more efficient than the dominant operator.
This sort of logical premise has been partially raised in the German HRS case. 76 The case shows, through the example of Hotel.de, that an APPA can, at an early stage, foster market entry by enabling a new platform to benefit from prices similar to those of the dominant firm. 77 However, later on, the APPA becomes more of a burden on the growth strategy of the new hotel platform. 78 In fact, given the impossibility to present a differentiated product, the new entrant finds it hard to attract consumers and suppliers and thus build a critical mass on both sides.
The German experience demonstrates that APPAs give new entrants an artificial advantage. A potential platform can truly access the market but cannot efficiently enter that market. The impossibility to initiate a growth phase, meaning to efficiently operate in the market by enhancing demand-side advantages, can dissuade future entry attempts. (ii) An adverse effect on potential entrants
In considering the likelihood of any entry on a significant scale, it is crucial to examine both the conditions under which the APPA is entered into force and the prevailing market characteristics.
As regards conditions under which the APPA is agreed upon, it emerges that the larger its scope is, the more difficult successful entry in the market will be. This is because an APPA covering a wide range of parameter such as prices, products’ availability, inducement and discount possibilities and other conditions leaves no room for other platforms to sufficiently differentiate their offers. In other words, the potential entrant cannot offer consumers a more attractive alternative to what is already available in the market.
This is relevant in two main scenarios, namely, when an APPA is able to neutralize new entrants’ capacity to fiercely compete on the basis of prices and when it prevents the emergence of differentiated business models. 79
In the first scenario, new competitors cannot inject price competition in order to gain market shares. A potential entrant is deprived of a powerful tool to convince users to switch to a smaller platform. 80 Entry or expansion is unlikely if the new platform possesses no competitive advantage other than its cost structure. 81 The APPA can be said to have a deterring effect as it signals to potential entrants that the incumbent platform will be always meeting their prices. This deterring effect may be strengthened when the APPA is coupled with other clauses such as price-beating guarantee. The connection to such a clause could signal incumbent platform’s readiness to aggressively compete in case of new entry.
In the second scenario, an APPA can prevent a new entrant from developing an innovative and competitive offer and thus insulates the dominant platform from competition. For instance, in sectors offering perishable products, new entrants may try to position themselves in the “last minute offer” segment. Given the perishability argument, suppliers have incentives to grant these platforms discounts on last minute offers. However, such an entry is unlikely to occur in the presence of an APPA. In the digital economy context, such a foreclosure effect can be extremely damaging. This is because the emergence of new business models is the conventional way to challenge the dominant position of incumbents. 82
In assessing the likelihood of those two scenarios, it is essential to take into account existing market characteristics 83 and the way they interact with APPAs. The most relevant market feature relates to barriers to entry. 84 Besides a traditional assessment of these barriers, a particularly crucial question relates to how APPAs exacerbate the problem of existing barriers to entry.
Two particular barriers to entry are common in the platform market. The first one relates to the need to reach a minimum efficient scale. An APPA may exacerbate such a problem. It can for instance exclude cost advantages as a mean to induce consumers to switch to new platforms or at least multi-home. In the absence of such a possibility, platforms will be prevented from reaching the minimum critical mass needed to operate efficiently in the market. This question can also be approached from suppliers’ perspective, asking how APPAs affect their willingness to deal with potential new platforms. In fact, a supplier may lack economic incentives to multi-home knowing that the new entrant cannot offer better deals than the dominant platform. In such a case, the new competitor may find it hard to access the market and reach the minimum viable scale. Therefore, an efficient entry is unlikely.
The second barrier relates to the existence of high advertisement costs. These costs can be characterized as sunk costs, which can prevent entry when they are high. In practice, this may happen if the potential new platform has to incur high sunk investments by, for instance, spending large amounts of money in advertisement and brand investment 85 in order to challenge the strongly built reputation of the dominant platform. In such a case, the presence of an APPA can further discourage the possibility of an entry as it increases the commercial risk 86 inherent to market entry. An APPA reduces the chances and possibilities for an efficient market entry. When this adds to the existence of high sunk costs, entry may be prevented.
Whilst barriers to entry are crucial for assessing APPA’s impact on potential entry, other market conditions can be equally important. Indeed, the timing of the adoption of an APPA can indicate a potential anticompetitive effect. 87 This is particularly the case if the practice is introduced as a reaction to an imminent entry or a recent one.
2. Incentives to foreclose under APPAs: A dynamic assessment?
The existence or inexistence of incentives to foreclose rivals is a matter of profitability. If the incumbent firm foresees a possibility to raise its prices as a consequence of the practice, it is said to have incentives to foreclose rivals. 88
To investigate these incentives, it is first important to recall that, when an APPA is entered into by a dominant firm, rival platforms could be prevented from getting access to cheap input and consequently see their costs increasing. At the same time, it is equally important to understand that the rise in profits in the upstream markets is likely to benefit suppliers and not the dominant platform. The latter is generally not vertically integrated. As a consequence, the classic upstream foreclosure model where a firm sacrifice profits upstream and recoup it in the downstream market, or vice versa, does not hold in the case of APPAs.
This does not however mean that the dominant platform has no incentives to foreclose rivals. Instead, a more suitable approach would be a dynamic assessment of the incentives to foreclose. According to such a view, incentives to foreclose rivals arise from the benefit the dominant platform derives when protecting its competitive advantage in the platform market and marginalizing competitors. The feasibility of this protection depends on the existence of mechanisms able to render the competitive advantage persistent. 89 Such a view imposes “to identify not only a clear source of downstream advantage, but also a clear mechanism that translates this into a bidding advantage and a mechanism that translates this bidding advantage into further downstream dominance.” 90
This virtuous circle can be easily duplicated in the case of APPAs and two-sided platforms. It is important to recall that two-sided markets are characterized by important network effects. These network effects are a substantial competitive that advantage dominant platforms enjoy. An APPA can render this advantage binding given that it manipulates indirect network effects to the favor of the dominant platform by reducing the scope for differentiation 91 and reinforcing competitive bottlenecks. 92
As feedback effects have already materialized at this stage, indirect network effects become self-reinforcing and allow the dominant platform to further enhance its dominance by increasing its critical mass. Ultimately, the dominant platform can more easily impose a commission fee increase and then become stronger.
B. Consumer Harm in the Framework of APPAs
The identification of the consumer harm asks whether foreclosure of rivals is likely to result in more consumer harm than in a market situation where foreclosure did not take place. 93 This harm needs to be investigated with respect to prices, innovation and consumers’ choice.
Firstly, APPAs can impact prices either directly 94 or indirectly as a consequence of foreclosing a competitor. The second hypothesis raises the question of whether foreclosure leads to any additional effect on prices, meaning that the dominant platform would be able to charge prices higher than what would have otherwise prevailed absent the exclusionary effect. 95
To answer the question, it is important to look at the constraints the dominant platform would face post-foreclosure. In this regard, three factors are crucial, namely, the sources of competitive constraints in the relevant market, the constraint exerted by the excluded operator, and finally the ability of remaining market players to constrain the dominant platform.
As regard the sources of competitive constraints, a crucial aspect relates to whether a dominant platform is more constrained by rival platforms than by suppliers’ own websites. 96 In this respect, the elimination of a rival platform is likely to relax the constraint the dominant platform faces. Three arguments confirm such a view as put forward by the Competition and Markets Authority (CMA) in the private motor insurance investigation. 97 First, platforms benefit from advertisement scale economies and can thus reach consumers cheaply compared to suppliers’ direct sale channels. 98 This argument echoes what I discuss in Part IV.C.1.b about the cost reduction function of platforms. Second, due to the importance of interbrand competition at the platform level, suppliers tend to price more aggressively on platforms than on their direct sale channels. 99 Finally, intrabrand competition coming from other platforms can also constraint the dominant platform. This intrabrand competition is likely to be “the main driver of competition between PCWs via commission.” 100 The elimination of rival platform is likely to affect the most important source of platform competition, which is already weakened by the existence of an APPA. The combination of those three factors leads to the view that rival platforms are the most important competitive constraint a dominant platform would face in the market.
In a subsequent stage, the analysis moves to whether the foreclosed rival is an important one and if the remaining platforms could constraint the dominant operator. If it appears that foreclosure affects an important rival while the remaining operators cannot sufficiently constrain the dominant platform, it is likely that the incentives of the incumbent platform to rise its commission price will grow, which in turn will impact the final price.
Secondly, APPAs can lead to consumer harm by preventing the emergence of new and innovative products. In other words, nonequivalent business models cannot enter and efficiently develop in the market. However, by excluding the emergence of differentiated business models, APPAs are likely to relax innovation efforts of both incumbent platforms and potential entrant to the disadvantage of consumers. This deprives consumers from enjoying the benefits of innovation and ultimately affects the available variety of choice. Innovation is the conventional and most successful way of challenging the position of digital dominant firms. In general, reproducing the incumbent business model is unlikely to contest the position of the dominant platform if the market has already tipped. 101 Competitive changes are more likely to depend on disruptive innovation. 102
A final aspect that requires further scrutiny is consumers’ choice. Whilst APPAs favor the emergence of business models similar to those already established in the market, platforms with “differentiated, non-equivalent business models” 103 are likely to be foreclosed. In this case, consumer harm can be very tangible given that foreclosed platforms are differently perceived by consumers. The more differentiated platforms’ business models are, the greater the harm to consumer choice is. This is because platforms are not simply resellers, but often undertake several investments and use various techniques that improve consumers experience in order to generate a higher demand.
C. Possible Efficiency Gains of APPAs
Recent court decisions have clarified the approach to assessing procompetitive effects of exclusionary practices. 104 This approach is in line with the European Commission’s one as detailed in the Guidance on Article 102 TFEU. 105 In this guidance paper, the Commission states four cumulative conditions any efficiency argument needs to fulfil in order to be justified. 106
The efficiency argument may succeed if (1) the efficiencies have been, or are likely to be, realized as a result of the conduct; (2) the conduct is indispensable to the realization of the claimed efficiencies; (3) the conduct is proportionate meaning it does not eliminate all or most existing sources of actual or potential competition; and (4) the negative effects of the conducts are fully counterbalanced by the efficiency benefits.
Here, it is important to note, that any decision on the merits of APPAs’ social benefits should be made in the light of the first three cumulative conditions. The fourth condition reflects the idea of balancing, meaning weighing procompetitive and anticompetitive effects that are reached once these effects are separately assessed.
1. Efficiency gains likely to result from APPAs
a. APPAs’ ability to contain free-riding
Most of the cases dealing with APPAs have acknowledged the risk of free-riding. 107 That risk materializes when the benefits of the investments realized by an investor-platform are appropriated by other platforms or producers offering competing products. 108 The fact that platforms are remunerated in terms of commission fees, meaning only when a transaction is concluded, renders that risk credible. 109
To take the case of hotel platforms, customers can use a platform to benefit from features such as ranking, reviews, products’ presentation in different languages, photos, and videos and then book the room on other channels offering a better deal. Rivals can thus capture demand by offering lower prices without investing in any one of the above-mentioned features. 110 Empirical evidence of free-riding in Italian hotels shows that “an increase in the price differential between hotels resulted in a substantial decline in the conversion rate.” 111 However, this is not sufficient to establish that there is a real free-rider problem. To exist, several conditions need to be met as stated in the Commission’s guidelines on vertical restraint. 112
Firstly, the risk of free-riding must be real. The likelihood of this risk materializing depends on whether rivals can divert costumers in the absence of an APPA. 113 This is likely to occur in two different hypotheses.
In the first hypothesis, the supplier itself can free-ride on the efforts made by the platform. Sometimes, a supplier acts both as a manufacturer and a competitor to the platform. This setting may encourage the supplier to divert consumers from the platform-investor to its own online sale channel simply because this may discharge it from paying commission fees. 114 Suppliers can do so easily by charging a higher price on the platform or even in extreme cases by asking to be listed on the platform without offering any product or limiting the available quantity. 115
In the second hypothesis, free-riding occurs between platforms. This possibility materializes either when a supplier tries to channel demand toward platforms charging a lower commission or when consumers themselves do so. 116 The Internet has increased market transparency and reduced transaction costs for end consumers. Consequently, it becomes possible for those consumers to use the platform offering the most advanced features in order to identify a given product and then to conclude the transaction on another platform granting a lower price.
Secondly, free-riding has to occur on presale services. In most of the cases, consumers and suppliers benefit from the value a platform provides before a transaction is concluded. 117
Consumers can enjoy, among other things, extensive and improved product presentations in different languages and using different digital supports. They can also benefit from a reduced information asymmetry thanks to the possibility of exchanging reviews on platforms. In short platforms, offer several presale services that help consumers in their decision. In addition to that, by investing heavily in the auction of keywords and buying links on search engines, 118 platforms make it possible for suppliers to benefit from targeted advertising campaigns and enhanced visibility on the Internet.
Clearly, platforms invest mainly in presale activities. 119 However, their profitability could be negatively affected by the free-riding problem. The reason is that the platform’s conversion rate, which is the number of visitors the platform converts to buyers, can remain very low while at the same time the number of visitors increases. 120 Obviously, this can lead to the deterioration of platform’s ad efficiency. A platform has to pay more to advertising media (pay-per-click) and perceive lower commissions given that, gradually, fewer visitors are becoming paying customers. That is a clear evidence that services provided by the platform are capable of being diverted.
Thirdly, the offered services must be necessary for consumers in order to buy the product. The question is highly factual and depends on the characteristics of the products each platform offers. Undoubtedly, when it comes to booking a hotel, presale services are highly necessary as clients need a detailed presentation of the hotel in a language they understand (location, room, services, etc.), photos, and sometimes even video illustrations in order to book a room. The necessity condition may not be satisfied in the case of the other platforms.
Fourthly, the Commission insists that “the product must be of a reasonably high value as it is otherwise not attractive for a customer to go to one shop for information and to another to buy.” 121 The relevance of such a condition when free-riding exclusively occurs in an internet economy context is questionable. The reason is that identifying the best offer and benefiting from the necessary information is less time- and cost-intensive for consumers on the Internet than in a physical point of sale. 122 With few clicks, consumers can switch from one platform to another without the necessity of moving from home or incurring any additional costs. 123
Lastly, the Commission excludes the existence of free-riding when the victim of such a behavior can prevent it by imposing promotion or service requirements on the other parties. This condition joins the indispensability condition as discussed in Part IV.C.2. In practice, a given platform cannot contractually impose on other platforms or suppliers to offer presale services. Suppliers may not have the necessary know-how to offer such services, while other platforms have no contractual relation with the investing platform.
b. APPAs’ ability to generate cost efficiencies
(i) Reduced transaction costs
In the presence of APPAs, end consumers can benefit from a constantly improved information. 124 An APPA makes it possible for platforms to invest in demand-enhancing services (detailed product presentations, reviews, ratings, and others services), which enable consumers to dispose of better information on the quality of an offer. In this way, users benefit from a reduced asymmetry of information and make better decisions. This is of particular importance in the Internet economy market for two reasons. First, consumers are not always able to physically investigate the product before concluding the transaction. 125 Second, consumers may strive to identify a reliable and reputed seller given the large amount of information they need to process on the Internet and the possibility for sellers to be exclusively present online. 126 This enhanced information translates in fact an improvement in consumer welfare. Better yet, knowing that consumers are better informed, suppliers will put every effort to keep on improving the quality of their products. 127
Furthermore, end consumers can benefit from reduced search costs. Comparing different offers on a platform is less time- and cost-consuming. In the absence of an APPA, consumers will have to go through different websites, compare prices, offer conditions, delivery costs, and other variables without being sure to pick up the best offer.
However, this potential efficiency gain has to be carefully analyzed. In fact, the reduction in transaction costs is not necessarily specific to APPAs given that nonintegrated price comparison website and search engines can offer such services.
128
Accordingly, this transaction costs argument is likely to fail the first condition according to which the gains have been, or are likely to be, realized as a result of the conduct. The assessment of APPAs’ contribution to reduce transaction costs will consequently stop here. (ii) Reduced distribution costs
Thanks to APPAs, consumers may indirectly benefit from supplier’s reduced costs of distribution. In this regard, platforms play what Ezrachi calls a “risk and cost mitigation” function. 129
Whilst, in theory, suppliers’ products can be seen by consumers around the world, 130 in practice it is far from being true given the large number of available online businesses. 131 In this context, suppliers need to engage in online advertisement in order for their business to be found. Platforms offer such an advertisement service. They constitute a “technological innovation in the advertising” 132 systems of suppliers with the advantage of lowering their advertisement costs thanks to marketing related-scale economies. 133 This has led platforms to bear substantial marketing costs instead of suppliers.
In the absence of APPAs, investments in different services including advertisement can be compromised with the consequence that suppliers will have to engage in direct advertising. However, the capacity to engage advertising expenditure varies from one supplier to another. In practice, large suppliers are more likely to benefit from such a system. They may have their own online shop or booking system and the necessary resources to advertise it on search engines and web-comparators in order to bring in new costumers. 134 In contrast, small suppliers with no private online distribution channel will be forced to rely on platforms. 135 Such a change could lead to an increase in distribution costs for two reasons.
First, if large suppliers rely more on direct advertising, small suppliers may see their distribution costs increasing. This can be explained by the fact that platform’s advertising costs will be borne by a limited number of suppliers. 136 Therefore, the advantages of mutualizing costs, resulting from being part of a group, are wiped out.
Second, suppliers, whether larger or small, may see their costs increasing as they lack economies of scale needed for efficient online advertising. When a customer visits a platform, there is higher probability that he converts into a buyer given the possibility to choose from a large variety of products or services. Individual suppliers lack the diversity of supply that increases the conversion ratio. By pooling different offers, platforms advertise for all of them as if they were advertising for only one and can thus achieve high-efficiency gains with regard to paid advertisement.
Nevertheless, the German Bundeskartellamt has challenged such an efficiency gain in the Booking case by claiming that “hotel portals which carry out online advertising such as Booking are conducting self-promotion, but not advertising that benefits all of the listed hotel companies equally.” 137 Hotels that are well ranked on the platform are more likely to benefit from this advertisement effort than the others. 138 This shows again the need for a careful assessment of any efficiency gain put forward by platforms.
2. Are APPAs’ efficiency gains indispensable?
The question to be asked here is whether less restrictive mechanisms appear conceivable to reduce the risk of free-riding and thus maintain both costs and quality efficiencies. 139
As regards the free-rider problem, several less restrictive options can be envisaged in order to ensure that a platform is remunerated for its presale services. Their effectiveness appears, however, to be questionable.
First, it is possible to think of a mechanism whereby platforms are remunerated in terms of generated visits instead of the number of sales or bookings. The remuneration can take the form of a flat rate amount or a sum proportionate to the number of clicks. 140 In both cases, the risk is that the focus will be less on converting visitors into payers than generating a high number of visits. Consequently, suppliers and platforms would be targeting divergent aims as the latter would have less incentive to promote the conclusion of a transaction. 141 The end result will be less demand volumes for suppliers. In addition, the risk of an increase in supplier’s distribution costs due to platform’s low incentives to convert visitor into payers will be exclusively born by suppliers and not the platform able to better absorb that risk due to its ability to pool costs. 142
Second, a service fee can be envisaged as an alternative to the commission system. A service fee is an income that platforms can directly charge to suppliers for the offered services and which is not dependent on the sale of supplier’s product. 143 This system has the advantage that platforms’ costs are passed on all suppliers benefiting from the platform and not only those that effectively sell a product. 144
The proposition does not, however, solve the problem. 145 The service fee is a cost that suppliers have to reflect in the price offered on each platform. Naturally, platforms that invest more in different services and features will face higher costs and subsequently charge a higher service fee. Consumers may prefer using these advanced platforms and then conclude the transaction on another one. Such a behavior may harm investments made by a platform in two ways. In the worst-case scenario, it can prompt suppliers to turn away from that platform when they realize that it does not enhance their sales. Even in an optimistic scenario, the mechanism can drive the concerned platform to underinvest and thus lower its service fee so as to preserve its competitiveness vis-à-vis other platforms.
Under the precedent alternatives, the compensation is not linked to demand with the risk that platforms will have less incentive to invest in demand enhancing features. Ultimately, the potential existence of alternatives to APPAs depends greatly on the nature of goods or services the platform offers, the quality of the presale service and the willingness of consumers and suppliers to move to a new business model. 146 For instance, consumers may be reluctant to switch to a fee-based business model that can cause the number of customers to decrease and ultimately lead to “a significant reduction in the value of the platform.” 147 Similarly, suppliers may be unwilling to implement a pay-per-click model given the reversal of the burden of risk and the potential loss in term of business volume. 148
With regard to the second efficiency gain, namely, the reduction in distribution costs, it is possible to doubt that the APPA will pass the less restrictive means condition. The existence of an APPA may be seen as a mean to protect platforms’ integrated business model 149 at the expense of price comparison websites and direct sales channel. Such a protection may ultimately lead to the disappearance of direct sales channels. Regulators would consequently question whether the reduction in distribution costs could justify such a restriction to choices available to customers. In particular, the question arises as to whether reduction in distribution costs can be achieved by a narrower parity agreement. Similarly, it is possible to wonder if a sophisticated system “for cost effective bidding on keywords which are central for search marketing” 150 can increase the conversion rate and thus lower distribution costs. These can be seen as alternatives which may justify rejecting the argument that no mechanism less restrictive than an APPA allows the dominant platform to realize distribution cost savings.
3. Do APPAs lead to the elimination of all effective competition?
This condition is confusing. It not only lead to a duplication of analysis 151 but may also rule out any chance a dominant company has to justify its conduct. 152 The idea behind the condition is to verify whether the dominant firm is still subject to some degree of competition pressure that will incite it to “to create and pass on efficiency gains.” 153 Any assessment of these incentives needs to be done with regard to both residual competition and the foreseeable threat of entry. 154
As regards residual competition, market shares are usually relevant. The more dominant a company is, the greater the impact on effective competition will be. This is, however, not systematic. 155 Sometimes, the actual market share offers a superficial prediction on the elimination of all effective competition, hence the need to couple it with other indicators.
On key type of evidence relates to rival platforms’ ability to react and restore competition. From this point of view, APPAs undoubtedly narrow rivals’ possibilities to react to a decrease of competition resulting from the foreclosure. 156 Price competition and disruptive innovation that can be rewarded in terms of market shares are hindered. 157 A counter argument to this is that rivals may not be able to react and grow by offering lower prices but they could do so by competing on quality, which the high price allows them to do.
The prevailing conditions in the relevant sectors, in particular whether the market is a mature one or still growing, may also indicate if rivals can react and compete with the aim of expanding their market share or not.
With respect to the threat of entry, conditions relating to market circumstances and characteristics; the growth prospects on the market; and barriers to entry, 158 in particular the ease of getting users on both sides on board and the costs associated with this, are relevant. 159 When the threat of entry is real, it is likely that efficiency gains translate into concrete benefits for consumers. Most importantly, the dynamics of entry have been extensively analyzed above. This analysis has shown that APPAs have an adverse effect on platform entry.
V. The No Economic Sense Test: Does It Make Sense for APPAs?
In application of the no economic sense test, a conduct is not exclusionary “unless it would make no economic sense for the defendant but for its tendency to eliminate or lessen competition.” 160 Put in other words, the test focuses on identifying whether the defendant would derive any profits from the practice beyond those that are amenable to the exclusion of a competitor. As we can imagine, a conduct would make economic sense if the nonexclusionary profits are higher than exclusionary ones. The test has, however, been questioned in some instances. This imposes to verify whether it is an appropriate one for the purpose of assessing APPAs before turning to its concrete application.
A. Adapting the No Economic Sense Test to APPAs
The application of the no economic sense test to APPAs raises the question of its suitability. This is due to the fact that APPAs have two characteristics that may call into question the appropriateness of the no economic sense test. 161
First, in application of APPAs, exclusion may prove costless or simply cost little. Unlike predatory practices, where the defendant may incur significant losses before a rival is driven out of the market, an APPA entails no negative sacrifice. 162 In practice, the existence or not of a sacrifice influences greatly the profitability assessment. 163 Consequently, APPAs may allow the argument that the practice makes economic sense for the defendant as they entail no negative sacrifice. Given that the focus is no longer on conducts that sacrifice revenue but practices that raise rivals’ costs (hereinafter, RRC), it would therefore seem legitimate to ask whether a profit sacrifice test may work. Supporters of the test believe it would, provided some adjustments 164 to the but for scenario as will be suggested in the following part.
Second, an APPA can be characterized as an efficiency-enhancing mechanism. Besides its potential to generate profits from excluding rivals, an APPA can simultaneously lead to other various sorts of benefits that may provide an economic rationale for the practice. For instance, in the French Booking case, 165 the defendant has submitted several arguments that show the efficiency rationale of an APPA. 166 This demonstrates that APPAs can be a key element in the business model of some platforms and consequently make at least “some rather than economic sense.” 167 In such circumstances, there is a risk that the test may again prove deficient.
Overall, keeping the above caveats in mind, it is possible to proceed with the test, beginning with the identification of the but for scenario in line with the adjustments suggested by supporters of the no economic sense test.
B. Implementing the No Economic Sense Test to APPAs
Melamed has suggested that the “but for” determination for conducts that raise rivals’ costs but have efficiency properties requires to ask “what the price would have been if rivals’ costs had not been increased in order to determine whether the conduct would have made sense even if it had not raised rivals’ costs.” 168 This definition appears to remedy criticisms highlighted above and which relates to the test’s potential failure to capture RRC practices and the bias toward the benefits of the practice.
The question that immediately emerges relates to how to apply this determination to APPAs. In our view, this entails two successive questions. First, one has to ask what the prices charged by other platforms would have been absent the APPA. The answer to this question will, in a second stage, help determine whether a platform would have implemented the practice even if it had not raised other platforms’ costs.
1. Establishing the but for scenario
Obviously, in the absence of an APPA, prices would differ from one platform to another, even though, in some instances, it could be difficult to predict prices accurately.
In theory, platforms that invest more in order to attract users will not be able to offer the same final price than a low-cost platform due to their different cost structures. Indeed, all other things being equal, a platform
Nonetheless, an APPA changes this situation by linking the price of the same products on different platforms. It thus neutralizes any cost advantage a rival may enjoy. Evidently, the question of prices that would have prevailed with no APPA in place requires more investigation than a simple comparison of prices. The reason behind this is to identify how the characteristics of the parity agreement contribute to raising rival platforms costs. Two particularly relevant aspects should be analyzed. 169
The first one relates to whether suppliers are willing to offer platforms a given product at the same price (net of commission) in order for the final price to reflect intermediaries’ divergent cost structures. The problem here is that an APPA forces suppliers to engage in a sort of price discrimination by absorbing any difference in commission fees. Therefore, suppliers end up offering lower prices to some platforms and higher ones to others. In the previous example, it would mean that the supplier gets
The second aspect depends on whether rival platforms find themselves unable to compete with the dominant firm as a consequence of the cost increase. A given platform may or may not find it materially impossible to compete with the defendant depending on its business model and the extent to which the market is price-sensitive. To understand this, it is important to recall that when an APPA is introduced, price is no longer an important element of competition. In fact, competition becomes mainly centered on platforms’ quality, notoriousness, and visibility. 171 In this context, a low-cost platform may not be able to compete with the dominant platform if the latter enjoys a net advantage on nonprice competition parameters. By contrast, a platform that targets higher-paying consumers (say business travelers in the case of hotel platforms) may still be able to do business in the same market than the dominant platform despite the cost increase. Raising rivals’ costs does not systematically place the affected rival in an incapacity to compete. Thus, APPAs can be said to have a divergent impact on the ability of rival platforms to do business depending on the variety of the competing business models.
Besides that, mostly, there is however no need to ask whether competing platforms can avoid the cost increase by switching to other equivalent alternative solutions. 172 The reason for this lies in the network effect explanation. Platforms need to attract a large fraction of suppliers in order to attract customers and conversely. A platform may not be able to reach the critical mass to operate if it turns away from all the suppliers that have engaged into a parity agreement with a given competing platform. This is even truer given that APPAs have an intrinsic cumulative effect in the sense that they are more likely to apply to all manufacturers that use the dominant platform and multi-home at the same time.
To sum up, this first question shows how APPAs are effectively able to weaken the competitiveness of a rival platform by rising its costs. Retail prices would have certainly diverged in the absence of an APPA regardless of the resulting market power for the defendant.
2. Establishing APPAs’ profitability
Based on the above-identified market scenario, the second question focuses on whether the defendant would have engaged in the practice had it known that rival’s costs would have not increased. In other word, would the APPA be implemented in the absence of the cost increase for rivals? This presupposes a comparison of the cost and the benefit of APPA in the hypothetical world previously identified. Whether the practice makes sense or not depends ultimately on the outcome of this balancing. If the revenue resulting from entering into a parity agreement is higher than the avoidable cost, then the practice makes economic sense and is lawful. 173
a. The sacrifice associated with APPAs
Besides transaction costs needed to renegotiate contract in order to implement an APPA, very few costs can be associated with the practice. As raised before, the implementation of the practice entails no negative sacrifice which makes it less likely for the APPA to be labeled as making no economic sense.
The sacrifice assessment is, however, not limited to those costs strictly necessary to the implementation of the practice. In fact, account should be taken of all the avoidable costs, 174 meaning costs the defendant would not incur had it decided not to embark upon an APPA. The problem here lies in the fact that the test labels as a cost or a sacrifice investment associated with an APPA but that can have a procompetitive rationale. 175 With an APPA in place, the focus shifts to investment in platform features. 176 From the perspective of the no economic sense test, those platform features investment can be seen as a short-term sacrifice. The expectation that this investment will help increasing the platform demand and accordingly its revenue is irrelevant for the qualification of the investment as a sacrifice. 177 Similarly, the possibility for the incumbent platform to offer cashbacks, consumers’ gifts, and postsale discounts when they are not covered by the APPA can be seen as a short-term sacrifice even if the long-term goal is to induce consumers to join the dominant platform or continue to use the said platform. The same reasoning applies to potential bonuses or cashback offered to suppliers in order to compensate them for any decrease in demand as a consequence of the introduction of an APPA.
b. The profits derived from APPAs
When a platform, in a given market, contemplates the benefits of the parity agreement, it asks if the practice is going to be profitable by allowing more sales and thus earn more commissions, to reduce its costs 178 or even perceive a higher commission. 179 The question is not limited to benefits that serve the platform’s self-interest but extends to all the foreseen efficiencies to the exclusion of those resulting from the foreclosure effect. 180
To avoid duplicating analysis, efficiency gains are discussed above as a part of the consumer welfare test. For the sake of completeness, however, it should nevertheless be pointed out that the dominant firm can bring forward different arguments to demonstrate that it actually derives several benefits from the practice.
As exposed above, an APPA enables the concerned platform to focus on investments and improving the platform’s features. This may be seen as a way of enhancing the quality and the value of both the platform and the offered products, which in turn might lead to a shift in the platform’s demand curve. The free-rider argument intending to preserve those investments could also be read from the same perspective. 181 Added to these advantages are other benefits an APPA can bring about notably the reduction in distribution costs. 182
Another, rather curious, argument is that APPAs may allow a platform to increase its commission. The literature on APPAs price effect acknowledges the practice’s potential to lead to higher commissions even if no rival is excluded. 183 The no economic sense allows for such an argument as it represents a nonexclusionary benefit that the incumbent platform may derive from the practice even if not shared with consumers. As explained by Salop, the most important point is that the practice is “economically rational (i.e., maximally profitable) absent a reduction in competition.” 184
Above all, the fact that other nondominant platforms use the APPA can be a sign of justifications other than the exclusionary potential of the practice. Indeed, since those competitors do not have the ability to exclude, the only reason for them to use such a parity agreement would be because it has other merits.
c. The allocation of the profits and sacrifices associated with an APPA
Turning to the allocation of the above-mentioned benefits, several challenges may arise. Exclusionary and nonexclusionary benefits can be seen as two faces of the same coin. 185 That is to say, if an incumbent platform is able to invest in features benefiting consumers and thus increasing the quantity it sells, it is because an APPA is implemented and producing a foreclosure effect vis-à-vis to low-cost potential entrant.
The result of the balancing and thus the no economic sense test depends on the way nonexclusionary profits and sacrifices are allocated. If investments related to platform features and inducements offered to consumers are labelled as a sacrifice, there is a risk that the test results in false positives. This risk is all the stronger if, on the other side, the benefits derived from the practice by the dominant platform are not properly taken into account because, as shown above, they are intertwined with the potential exclusionary effects. In an instance such as this, it is likely that the practice will be found unprofitable and makes no economic sense only but for the resulting increase in market power.
Alternatively, if the above-detailed benefits are fully acknowledged, chances of the APPA being approved are high. Those benefits are likely to outweigh the sacrifice given their broader scope. With such a permissive approach there is, however, an omnipresent risk that the test results in false negatives. The APPA will make economic sense or at least some economic sense 186 but ignores the impact on competition. However, acknowledging that the APPA makes economic sense does not mean it also makes sense from the perspective of competition law. This is simply because a practice that makes economic sense from the perspective of the dominant platform can still damage the process of competition. 187
VI. The As Efficient Competitors Test: A Deficient Assessment of APPAs?
The rationale behind the as efficient competitor test (hereinafter, AEC) is that no firm should be condemned just because of its superior efficiency. Therefore, a practice is, at first sight, anticompetitive when it leads to the exclusion of a rival that is equally or more efficient than the dominant operator. 188
The AEC test is ideal for pricing abuses. 189 Its use for nonpricing practices, such as APPA, is however questionable. 190 In spite of this, one can proceed with its application to APPAs by asking whether an equally efficient rival can, absent the parity agreement, profitably compete on the market.
The immediate challenge relates to the very definition of an equally efficient competitor and its application to a network industry. The problem stems from the risk that platform markets tip in favor of a given platform due to the importance of indirect network effects and scale economies. In such circumstances, it is not uncommon to find a rival that, in overall term, is as efficient as the dominant firm meaning that it has an “identical cost curve to the dominant platform.” 191 Nevertheless, given the absence of a minimum efficient scale, that rival maybe seen as less efficient when its efficiency is assessed at the unit level. 192 This translates the exact issue one has to face when applying the AEC test to APPAs.
More precisely, the particularity lies in the fact that reaching a point of critical mass depends on the existence or not of an APPA. In the presence of such a practice, an overall equally efficient competitor may be prevented from reaching the dominant platform efficiency level per unit. Consequently, the emergence of an equally efficient competitor becomes unlikely. Such a position echoes the European Court of Justice’s view in Post Denmark II, where it held that, in some instances, “applying the as-efficient-competitor test is of no relevance inasmuch as the structure of the market makes the emergence of an as-efficient competitor practically impossible.” 193
Therefore, the identification of an equally efficient competitor depends on the importance attached to demand-related advantages the dominant platform enjoys but rivals may find complex to replicate in the presence of an APPA. Against this background, two options can be contemplated: A first option ignores this caveat and applies the test in a static manner, while the second one acknowledges this problem by applying the test in a dynamic way.
A. A Static Application of the AEC Test to APPAs
In this first option, the test is strictly applied. To be considered as an equally efficient competitor, the rival should have access to the same advantages than the dominant platform. This means that a competitor who is not yet as efficient as the dominant platform, but could become equally or more efficient if the parity agreement is waived, cannot be treated as an equally efficient competitor. 194
In application of this first hypothesis, a dominant platform engaging in exclusionary behavior vis-à-vis an overall-equally efficient competitor may still be relaxed. If adopted, such a reading would ignore the potential underinclusive nature of the AEC test. 195
The Commission itself recognizes this risk in the case of a price-based exclusionary conduct. It asserts that “in certain circumstances a less efficient competitor may also exert a constraint which should be taken into account.” 196 The Commission goes on to confirm that in such circumstances it “will take a dynamic view of that constraint, given that in the absence of an abusive practice such a competitor may benefit from demand-related advantages, such as network and learning effects, which will tend to enhance its efficiency.” 197 The same analysis can be extended outside the pricing context. This leads to a second possible hypothesis.
B. A Dynamic Application of the AEC: What About Potential Equally Efficient Competitors?
The second possibility is less lenient with the dominant platform. It accepts that a competitor can be as efficient as the dominant platform but only in theory. In practice, however, that rival may prove less efficient as it does not enjoy the same demand-related advantages as the dominant platform.
Those advantages relate to indirect network effects and scale economies. The assessment should build upon the importance of those characteristics to online platforms. Indeed, an APPA, which acts as a barrier to entry, prevents rivals from differentiating and thus attracting users on both sides of the platform with the view to achieve the same advantages as the dominant platform.
Therefore, the question to be asked is whether the rival could, absent the APPA, easily access both consumers and suppliers and thus enhance its operative efficiency. The objective is to determine if the APPA is what prevents a rival from exploiting its efficiency or enhancing it within a reasonable time frame. In asking this question, the investigation turns, to some extent, to an analysis of the actual effect of the practice on the market. 198
To begin with, it is important to identify the characterizing tendency of the relevant market, either a single-homing or a multi-homing one. The distinction is crucial. It interacts with the question of how APPAs affect platforms’ ability to benefit from demand-related advantages.
Even when the single-homing rate is high, the APPA can still have a limited impact on the possibility to reach the minimum efficient scale for two reasons. On the one hand, competition will, in this case, focus on consumers’ attention rather than price. Price competition will be limited given that consumers tend not to compare prices. Accordingly, rival platforms can still attract consumers by focusing on advertisement even if they cannot differentiate in term of prices. On the other hand, knowing consumers’ tendency to single-home, suppliers would have strong incentives to be present on different platforms. In this context, rival platforms will find it relatively straightforward to attract suppliers. In general, even in the presence of an APPA, a platform can still attract both sellers and consumers and thus reach a minimum efficient scale.
By contrast, if large fraction of consumers tends to multi-home, price competition assumes a more significant role. In this case, an APPA is more likely to interfere with rivals’ ability to reach a minimum efficient scale. A new entrant may not be able to offer lower prices in order to attract consumers. At the same time, suppliers will be more inclined to limit their presence to the dominant platform and probably very few other large platforms that have the largest consumer base and offer the best prices thanks to the APPA. Accordingly, attracting both consumers and suppliers may be obstructed in the absence of price differentiation.
In this context, the critical factor becomes the size of the foreclosed share of supply. 199 This requirement means that if a substantial share of the supply is covered by an APPA, it is more likely to see an equally efficient platform or a platform that can become equally efficient prevented from reaching a minimum efficient scale. To sum up, the larger the covered proportion of supply is, the harder to expand and reach a critical mass on both sides of the platform it becomes.
Such an analysis must, in addition, be supplemented by a specific examination of rival’s ability to reach a minimum efficient scale. This will depend on whether there will be a demand for the rival’s product or not. In other words, the investigation needs to establish if users would be inclined to switch platforms and use the rival one.
There is, however, no need to take into account factors relating to rival’s cost structure and the likelihood to reach a minimum efficient scale given it was assumed that, in overall terms, the rival is as efficient as the dominant firm. In fact, demand-related advantages is what prevent the said rival from being equally efficient when operating in the market.
In some cases, this demonstration entails a serious limitation. This is the case when the effect of the practice may be simply potential. For instance, a rival may be denied access to the market in order to show it is equally efficient or potentially equally efficient. 200 A platform may consider entering the market but refrain from doing so, knowing the existence of an APPA that limits the possibility to differentiate in order to reach the minimum efficient scale. In such an instance, it is impossible to prove the actual effect of the APPA, meaning that the parity agreement is what prevents the potential entrant from reaching an efficient minimum scale, simply because the entry has never occurred in the first place.
VII. Conclusion
APPA is a novel business practice that can be assessed under the no economic sense test, the AEC test, and the consumer welfare test. The article discussed the practice’s foreclosure effect from the perspective of these different tests and came to the conclusion that the consumer welfare test is more appropriate than the other ones. Indeed, whether APPAs lead to the foreclosure of rivals depends on the applied test and sometimes on the way the test is applied.
As regards the assessment of APPAs under the no economic sense, the test appears too complicated and, therefore, unmanageable. Many modifications have to be made to adequate it to APPAs, which could lead to error costs. In practice, a proper evaluation of APPAs requires an examination of the competitive impact of the cost increase for rivals. However, this appears to be something that the no economic sense test is incapable of properly assessing. The question raised by the test resembles more an inquiry into the objective justification of the practice. In short, the test focuses on the dominant platform’s nonexclusionary profits and sacrifice, which are of little relevance to assessing its market power and the consequences of the practice on consumers. Consequently, there is an omnipresent risk that the test results in both false negative and false positive when applied to APPAs.
Despite its great potential to result in consumer harm, the practice can escape liability. The reason is that the implementation of the APPAs is costless while the dominant platform can still derive some nonexclusionary profits from its introduction. Similarly, the test may result in false positives if the dominant firm engages in substantial investments that can be labeled as sacrifice. Accordingly, it is possible to conclude that the test is of limited utility for the stake of assessing APPAs.
When the AEC is applied to APPAs, serious concerns arise. In two-sided markets, where demand-related advantages play a critical role, it is irrelevant to compare competitors’ efficiency to that of the dominant platform. In such a context, APPAs’ foreclosure effect would, in most of the cases, affect a potential and not an actual equally efficient competitor. Therefore, an appropriate assessment of the exclusionary effect presupposes a dynamic application of the AEC test. There are, however, intrinsic risks with such an approach as it may lead to the protection of competitors instead of competition. For that very reason, it was shown that there is a need to look at the actual effects of the practice by assessing if the APPA is what prevents a competitor from becoming as efficient as the dominant platform. To the difference of the static application of the test, it is probable that, when looking for a proof of the actual effect of the practice, the APPA is found to be leading to the foreclosure of potential equally efficient.
Lastly, the consumer welfare test appears to address the above-mentioned enforcement risks. In particular, it safeguards against the risk of underdeterrence associated with the two previous tests. The reason is that the test imposes a comprehensive assessment of the practice. The exclusion of a rival is necessary but not a sufficient condition. Additionally, it is necessary to prove the existence of consumer harm and investigate the potential efficiency justifications to assess the net effect of APPAs on consumers. It is true that such open-end balancing entails significant administrative costs. However, dominant platforms would not be allowed too much freedom to engage in harmful exclusionary APPA simply because the practice does not entail a sacrifice or because the market characteristics render the emergence of an equally efficient competitor unlikely.
Footnotes
Acknowledgments
The author would like to thank Francisco Beneke, Driss Mansour, and Sanna Widen for helpful comments on this article.
Declaration of Conflicting Interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The author(s) received no financial support for the research, authorship, and/or publication of this article.