Amazon and the complexity of legislation regarding multisided platforms in the EU

(1)

Jaakko Sarhimaa

Amazon and the complexity of legisla- tion regarding multisided platforms in the EU

Subtitle

Metropolia University of Applied Sciences Batchelor of Business Administration International Business & Logistics Thesis

29.10.2021

(2)

Author(s) Title

Jaakko Sarhimaa

Amazon and the complexity of legislation regarding multisided platforms in the EU

Number of Pages Date

28 pages

29 October 2021

Degree Batchelor of business administration Degree Programme International Business & Logistics Specialisation option Finance

Instructor(s) Ross Kamarul-Baharin, Senior Lecturer

Amazon, one of the largest e-commerce companies in the world has gained massive success by working as a multisided platform where it works as the marketplace as well as competes against 3rd parties in that marketplace. Decision makers face dilemmas on how to regulate, when to regulate and perhaps most importantly should they regulate multisided platforms. Thesis provides deep dive into the methods that Amazon, one of the largest of the FAANG/Big Tech companies has used to gain its current positions as one of the largest retailer/e-commerce companies, as well as what means European Commission (EC) has in its disposal, and challenges the EC faces when dealing with multisided platforms. Looking at Amazon''s use of Network effects, Economies of scale, switching costs, Aggressive acquisitions, Platform domination and Predatory pricing, it can be argued that Amazon has acted against EU laws, more specifically Article 102 TFEU by distorting competition in e- commerce and should be punished for it. Historical antitrust cases forcing structural remedies (breakups) would likely be ineffective, which leaves only forcing behavioral remedies via regulation and control, which would likely raise same regulatory problems. To combat the regulatory issues this thesis recommends a method of leveling the playing field for new emerging companies in the way of Data sharing mandate.

Keywords Multi-sided platforms, European Commission, Antitrust

(3)

Contents

1 Introduction 1

2 Market study on Amazon 2

2.1 History of Amazon 2

2.2 Amazon Marketplace 4

2.3 Amazon Prime 4

3 Literature review 5

3.1 Multi-sided platforms 5

3.2 Network effect 7

3.3 Economies of scale 9

3.4 Switching costs 9

3.5 Breaking up monopolies 11

4 Analysis on European Legal structure 12

4.1 EU competition law articles 101 and 102 12

4.1.1 Forbidden cartels and article 101 12

4.1.2 Leniency policy 13

4.1.3 Article 102 and the abuse of dominant position 14 4.2 Concentrations (Council Regulation (EC) No 139/2004) 15

4.3 GDPR 16

5 Recommendations/conclusions 18

5.1 How to remedy antitrust 18

5.2 Data sharing mandate 19

6 References 21

Appendices

(4)

1 Introduction

During the pinnacle of the power of Standard Oil, the Rockefeller led company controlled 90 to 95 percent of all oil produced in America. The enormous advantage Standard oil held in the Oil market combined with anticompetitive actions by the company led to its eventual breakup and the beginning of the Progressive era. Comparing Standard Oil’s status to that of Google’s market share of 86 percent in the search engine market, Am- azon’s 49 percent in US ecommerce market and Facebook’s 71 percent in social media worldwide, generates a clear picture that the FAANG¹ companies are not far from where the Robber barons once stood. To achieve the positions that big tech companies currently enjoy, they have had, not only enormous technological achievements, but also their fair share of controversy and legal battles in both EU and the US. FAANG companies hold such an integral position in people’s lives that opting out from their services is simply not a feasible possibility anymore unless a person would like to spend the rest of their lives as a social outcast, a type of a lone hermit residing in the lone mountains.

Whereas customers might be limited on their realistic options on battling the tech titans, the regulators of US and EU are not. Questions that decision makers should be asking themselves are how to regulate, when to regulate and perhaps most importantly should they regulate?

This bachelor’s thesis provides deep dive into the methods that Amazon, one of the largest of the FAANG/Big Tech companies has used to gain its current positions as one of the largest retailer/e-commerce companies, as well as what means European Commis- sion (EC) has in its disposal, and challenges the EC faces when dealing with multisided platforms.

The concept of multisided platforms in itself is relatively new and as such does not bolster enormous amounts of top-quality literature as topics which have been more widely re- searched have. The leading voices, in general on the topic as well as in this thesis, come from the pioneers of the concept multisided platforms Nobel winner Jean Tirole and his coresearcher Jean-Charles Rochet who have together contributed on over 200 papers

1 Leading major Technology companies are often described with acronyms such as FAANG (Fa- cebook, Apple, Amazon, Netflix & Google) FAAMG (Facebook, Apple, Amazon, Microsoft &

Google) GAFAM (Google, Amazon, Facebook, Apple, Microsoft) or simply as “Big Tech” or even as “superstar companies” in some cases.

(5)

that have been cited thousands and thousands of times. Rochet and Tirole pioneered the concept of multisided platforms with their 2003 paper “Platform Competition in Two- Sided Markets”. (Schmalensee, 2014)

Another set of brilliant authors who have contributed to the research on multisided platforms are Howard W. Johnson Professor of Management Emeritus and Professor of Economics Emeritus at the Massachusetts Institute of Technology (MIT) Richard Schma- lensee and economist, writer and a cofounder of two consulting organizations: Global Economics Group and Market Platform Dynamics, David S. Evans, who released book called “Matchmakers” open and develop the concept of economics of multisided platforms by tying the theory into existing success stories of large corporations such as Fa- cebook and Uber.

2 Market study on Amazon

2.1 History of Amazon

“It’s all about the long term”

- Jeff Bezos 1998

Before using their digital platform for providing consumers and businesses all over the world with music, audiobooks, reading recommendations, fashion, consumer electronics, award winning TV-series and movies, movie databases, comics platform, smart home security systems, voice commanded personal assistants, delivery, and cloud computing services, Amazon.com simply sold books. Amazon.com was established in 1994 by Jeff Bezos as an online marketplace where of books were sold in great quantities from one platform where searching was easy. Amazon has portraited an image that it exists to and for its customers. This can be seen all the way back from the first letter to shareholders where Bezos stated that Amazon will “focus relentlessly on customers” and “obsess over customers”. (Bezos, 1997) Bezos stated that customers are loyal to businesses as long as there is no other better or cheaper service on the market. That statement will be reviewed later in this thesis, as research shows that competition might not be so simple in the online market platforms.

As a company Amazon has shown that it is heavily focused on future and that it is willing to sustain losses to establish dominance. Amazon has historically created relatively low

(6)

profits compared to its vastly growing revenue and it has only recently started to create consistent profitable quarters due to the vast success of its cloud computing services called Amazon Web Services (AWS) see figure 1.

Figure 1 Amazon Annual Net Revenue and Profits 2004-2019 (Coppola, 2020) (Sabanoglu, 2020)

From the late 90’s and the early 2000’s Amazon has grown into leading online marketplace which dominates the e-commerce in its home market of US as well as globally.

Amazons estimated market share in the US e-commerce in 2020 was about 44%, whereas its closest rivals Walmart, eBay and Target have market shares of 7%, 5% and 2% respectively. (Duggan, 2020).

In the EU Amazon has not reached same heights as in US, as it boasts market share of around 9,8% in the EU area. (Sabanoglu, 2020) Amazon operates in Europe under company Amazon EU S.à.r.l. which has its headquarters in Luxembourg, where it did not pay any taxes for the revenue of €44 billion it generated during the year 2020². (Satariano &

2 This was possible due the European division reporting loss of €1,2 billion to the Luxembourg authorities which made it exempt from corporate taxes.

-50, 0, 50, 100, 150, 200, 250, 300,

2 0 0 4 2 0 0 5 2 0 0 6 2 0 0 7 2 0 0 8 2 0 0 9 2 0 1 0 2 0 1 1 2 0 1 2 2 0 1 3 2 0 1 4 2 0 1 5 2 0 1 6 2 0 1 7 2 0 1 8 2 0 1 9 (in billion U.S.

dollars) Revenue

(7)

Gross, 2021) Still Amazon has been taken into close inspection by the European Com- mission and it has been a part of 5 competition related cases, out of which three have been on antitrust and one a piece in Mergers and State Aid. Due to the vast profitability of AWS Amazon has been able to take large losses caused by investments on other fields of business. Amazons stock value works in considerably peculiar manner where the stock will soar high when the company makes zero gains or even sustain losses. In 2015 analyst in New York Times’ interview stated that “Amazon’s stock price doesn’t seem to be correlated to its actual experience in any way”. (Streitfeld, 2015) Today it is easy to see that investors might have had a good reason to hold fate in the company as profits have been increasing by the year and Amazon’s stock value has risen to over 3200$ (17.2.2021) from measly 1.5$ in 1997 (Verizon Media, 2021)

2.2 Amazon Marketplace

Amazon Marketplace is an online commercial platform owned and maintained by Ama- zon.com INC. that shares the web address of the forementioned company’s name. Am- azon Marketplace started as an online book selling site in the summer of 1995. In 2021 customers can buy anything from fashion to videogames and food from Amazon’s own production or from third party sellers, which from the latest statistics contribute for over half of all units sold through Amazon Marketplace (Chevalier, 2021). In essence Amazon controls the marketplace, allows other parties to sell in there as well as competes with those third parties in the market that Amazon has created. This is one of the core issues this thesis tries to unravel.

2.3 Amazon Prime

Amazon Prime is Amazon’s a subscription-based loyalty program where customers can sign for monthly (12.99$) or annual (119$) fee to gain access to exclusive deals, fast and free delivery as well as endless amounts of reading, listening, watching, and consuming of everything from books and streaming services to clothes and medicine. By now Ama- zon Prime has reached more than 200 million Prime members worldwide according to CEO Bezos’ comments on the annual shareholders letter for 2020. Amazon historically has sustained heavy losses from Prime to buy customer loyalty and estimates made in 2019 concluded that Amazon lost between 1 and 2 Billion USD per year due from Prime.

(Reuters Staff, 2019) Amazon does not publish its detailed numbers regarding Prime, but it has been estimated that Prime brought in 7.9$ Billion on Q2 of 2021, Profitability

(8)

of Prime, or rather the lack of it, was an investment to the future by Amazon as they have definitely understood what analysis by Consumer Intelligence Research Partners found, which is that customers subscribed to Prime spend on average 27 times a year on Am- azon, which is nearly double the amount users without Prime membership shop. (Weise, 2021)

Amazon Prime works as a factor for lowering the barrier of entry for customers into Am- azons services, and at the same time it increases the switching costs of changing to Amazon’s competitors’ sites due to sunk costs of the membership fee. In other words it encourages customers to spend more in Amazon to get the “money’s worth” for the amount they spend for the Prime subscription.

3 Literature review

3.1 Multi-sided platforms

Multi-sided platforms, also known as Matchmakers, are a class of businesses that create value by bringing two or more different types of economic agents together and facilitating interactions between them that makes all agents better off. (Evans & Schmalensee, 2013) Some examples of multisided platforms are recruitment platforms where one group of agents are looking for new jobs and one group of agents is looking for employees to fill open positions in their companies. Other example of multi-sided platform could be flight booking services where the service connects agents that wish to browse flights to flight companies that wish to gain access to as wide group of the agents as possible looking for flights in routes that the flight company operates. Theory of multi-sided platforms is related to theories of network externalities and market or regulated multi-product pricing. From the theory of network externalities in multi-sided platform theory shows that there exists a set of noninternalized externalities amongst the agents and from the multi- product pricing theory it borrows “the notion that price structures are less likely to be distorted by market power than price levels.” (Rochet & Tirole, 2006) As the examples above indicate, there are some common features that most multi-sided platforms share.

Multi-sided platforms often enjoy two types of indirect network effect externalities in the form of usage externalities and membership externalities as described by Evans &

Schmalensee (pp. 5-6). Membership externalities occur most of the time when both or

(9)

all agents of the platform enjoy benefits when the platform is being used. In the recruitment platform example job seekers benefit from finding a job from the platform and the employer’s benefit from finding employees from the platform. In the case of video game industry where console manufacturers such as Sony with PlayStation and Microsoft with X-Box serve as the platform between game developers and players. Console companies try to get as many gamers as possible to use their consoles so that developers would create/port their games to that console platform. That has led to console companies to make deals of exclusivity with game developers so that they would have unique selling points in the form of specific game franchises on their consoles. See Microsoft with Halo franchise and Sony with award winning Last of Us franchise selling over 70 million and 20 million respectively. (Microsoft, 2016) (Reeves, 2019)

It is also possible that membership externalities are positive for one group of agents and negative for another, as long as the net benefit of the externalities for the group’s stay positive. See a newspaper that sells adds within its pages. It creates value on the adver- tiser’s side with wider reach of potential customers, but it does not necessarily crate extra value to the readers of the magazine, who might enjoy the magazine with more sub- stance and less adds. Consumers will join a platform should they find that “aggregated benefits of joining … increase with the number of participants in the other side”.

(Johnson, 2020)

Usage externalities on the other hand occur when the platforms value to its users increase, based on the growing amount of other type of agents using the same platform.

Usage externalities are often described by multitude of varying terms such as; network externalities, indirect network externalities, network effects and indirect network effects.

Usage externalities are often present when a platform charges fixed prices from its users.

Again, using the recruitment platform to demonstrate that with larger amount of job seekers using the platform is more valuable to employers as the pool of qualified potential employees from where to choose from increases and vice versa with higher number of potential employers to attract those skilled job seekers. (Rochet & Tirole, 2006)

One other key definition for multi-sided platforms is that the agents cannot go around the price structure of the platform to capture the gains of connecting and thus leaving the platform inutile. One way of testing whether a platform is single-sided or multi-sided is to apply Coase theorem to the platform that is being investigated. Coase theorem states that “If property rights are clearly established and tradeable, and if there is no transaction

(10)

costs nor asymmetric information, the outcome if the negotiation between two (or several) parties will be Pareto efficient, even in the presence of externalities.” (Rochet &

Tirole, p. 649) What this means is that should two or more agent groups in the platform be able to negotiate a contract amongst the participants that allocates resources effi- ciently without any single group being left worse off, Coase theorem applies and the market thus is single-sided. According to Rochet and Tirole transactions hosted by a multi-sided platform, user and seller interaction are depended on the price level but not its allocation which leaves Coase theorem unapplicable. Coase theorem in practice is a good benchmark, but it, like most theorems, does not apply to all cases such as cases where there is asymmetric information in bargaining or price setting.

Crucial aspect on competition between multi-sided platforms is the necessity of reaching critical mass. Achieving critical mass means that the platform has acquired enough of both (or multiple) types of agents to participate in their platform for indirect network effects to start boosting the platforms growth and creating value for all types of agents in it. Achieving critical mass is one of the greatest challenges on any new entries as it can require heavy investments on product development and marketing.

Literature on multi-sided platforms is relatively new and it doesn’t, as of today, appear in abundance, compared to literature on single-sided businesses. That leads to some of the questions related to multi-level platforms such as their definitions and comparison within the field of multi-level platforms (especially when discussing antitrust and competition laws) difficult to grasp on. One reason for that lies in the fact that rise of large global multi-level platforms has been recently propelled by the relatively recent developments in the fields of internet, mobile and information technology. (Evans & Schmalensee, 2013)

3.2 Network effect

Network effect occurs when the value of product or service increases conjointly with the increase in quantity of users of that product or service. (Katz & Shapiro, 1994) For example most of the modern digital services depend on their success in this effect. Face- book would create little to no value for average consumer should it not have the consumers acquaintances also in the site sharing their life for others to see, like and share.

Google would have never gotten to be so fast and precise in its search results should it not have had the revelation to use the “waste material” that the millions of searches made

(11)

by consumers produced to train and improve their search algorithms. At the early stages of the company search result quality was based on balance of power between people, who needed the search to learn and the search that needed people to learn. More searches meant more learning which translated to higher quality of relevant search results which again attracted more users and searches. (Zuboff, 2019, pp. 67-16) Currently Google’s network is so vast that one can search anything on the internet withing thou- sand of a second.

Yet the value gained from a network is not only dependable on the size of the network.

Afuah (2013) argues that networks structure and its conduct also have significant impact on its value to customers and network providers. Network’s structure consists of; multiple constructs such as number of possible connections, centrality, structural holes (or rather the filling of those), network ties, number of roles played by actors and distinctive capabilities. Network’s conduct consists of, opportunistic behavior, reputation effects and trust. These aforementioned factors are important to take into consideration when re- viewing the benefits of networking effect, instead of just focusing on the size of the network.

Currently the tech giants use the network effect not only to build up new services to provide the customers with, but also to improve the predictive capabilities of their algorithms that bring value to the services. Big tech companies collect and use enormous amounts of data to better their algorithms and other predictive services and to create value to the consumers. Amazon collects vast amounts of customers behaviour data when they are visiting the site. (Kelion, 2021) The data they collect varies from purchases to product wish lists and page clicks to more specific information such as time spent on product pages, searches, emails opened, and products reviewed and rated. (West, 2019) In the fields of advanced machine learning and training of artificial intelligence, the quantity of data almost directly translates to quality when training the models to, for example predict customer behaviour or preferences. (Mayer-Schönberger & Ramge, 2018) Am- azon uses predictive algorithms to enhance the quality of their products such as the digital assistant Alexa and search optimization on their website to, just like Google in searches, to attract more customers and so the loop continues.

(12)

3.3 Economies of scale

Economies of scale occur when companies’ concentrate production on fewer tasks and thus increase the cost efficiency of the production which gives the company competitive advantage over prices of the products. (Ethier, 2009) Exploiting economies of scale have historically been considered as important competitive measure for growing businesses and it has been tested to still carry a major role in advanced economies even with the rise of small-scale self-employment in many industrialized countries. (Congregado, et al., 2014) A research on the effects of economies of scale in the context of modern network economies reached slightly contradictory results as it concluded that modern companies gain more opportunities from cooperation withing locally based companies with local economic environment – support institutions, access to knowledge, finance and resources – comparing to opportunities gained from reaching higher economies of scale.

(Patacconi & Russo, 2017) Economies of scale work slightly more in benefit for modern internet based companies than mass production based companies such as manufactur- ing for the simple reason of ease of scalability. Software, unlike hardware, does not require almost any additional costs when scale is upgraded. For example, selling cars requires materials and worktime that is correlative to the number of cars sold – hundred cars take hundred times the materials and work to assemble - whereas software can be duplicated with little to no extra costs on materials or work.

3.4 Switching costs

Switching costs as a concept has many different definitions, but for this bachelor’s thesis we will apply Chen & Hitt’s (2005) definition that focuses more directly on Information Technology (IT) and switching costs. According to Chen & Hitt switching costs can be described as “perceived disutility a customer would experience from changing product or service providers”. Switching costs are closely related to network effects as customers who enjoy the benefits of enhanced services gained from the effects of networks, face larger resistance to switch services or products due costs such as terminating current relationship and starting a new one. (Chen & Hitt, 2005) Everyone who has enrolled on a customer loyalty program such as Finnish retail brand Kesko’s “K-Plussa”, which cu- mulates discounts and store credits called “Plussa-raha” when used would experience disutility if changing to different retailer such as S-Ryhmä or Lidl, as the gained benefits from the membership program are not applicable in other stores. Furthermore, companies may increase the switching costs, even at their own cost of implementation of the

(13)

barriers, as defensive measures to protect their customers from competition. Switching barriers as described by (Fornell, 1992) makes it costly for customer to switch for another service or product provider. Such measurements can be seen in modern companies in the forms of previously mentioned loyalty programs and instore credits that are gained by using the service for long periods of time and would be lost in the process of changing service/product provider. Switching costs can also be non-monetary costs such as lost conveniency that has been built by the tracking of user preferences and behavior. For example, switching from one video streaming platform to another customer would lose recommendations that the algorithm provides based on previous searches and reviews.

Switching costs have also been linked with possible monopolization of markets as Klem- perer shows in his (1987) paper. Klemperer presents three types of switching costs:

transaction costs, learning costs and artificial or contractual costs. Transaction costs are costs that occur for example when changing a bank and the opening of new account in the new bank costs the customer first on closing fees in the old bank and then opening fees in the new, even though the services that the two banks offer are next to identical.

Perhaps most common example of learning costs for anyone to understand in the 21^st century is when changing mobile phone from Android based operating system to iOS based or PC to MAC. Again, in the core Android and iOS or PC and MAC offer the same exact uses, yet the operating of the different systems is highly different and the learning curve on mastering new system is quite steep. Third costs, artificial or contractual, occur in loyalty programs that for example flight companies offer to their frequent flyers or instore credits that can be used for discount in future purchases in their platform as Japa- nese videogame company Nintendo offers in their online shop. Klemperer argues that switching cost affects mature markets in two ways. First switching costs reduce competition by creating submarkets inside the market, which companies may pounce to mo- nopolize, which then results in “(noncooperative) equilibrium” that “may be the same as the collusive solution in otherwise identical market with no switching costs”. (Klemperer, 1987, p. 377) The second way switching costs could affect the markets occurs when companies that have gained monopoly power in a segment, compete with others in a struggle of retaining the customers, which in the end might not be as beneficial due to high costs of acquiring new customers. Klemperer concludes that switching costs do not necessarily benefit companies but might end up hurting all relevant parties due to large competition in the early stages of markets development.

(14)

3.5 Breaking up monopolies

Current dominating view on antitrust is of an angle that was defined in 1970’s spear- headed by Robert Bork’s 1978 book “The Antitrust Paradox” where Bork weaves a framework where predatory pricing is essential, and for consumers, even beneficial at- tribute of competition. The soul theme of Borks argument for consumer welfare is to disregard everything else but the increase of competition which will enhance the market performance. This consumer welfare focused view of the antitrust laws can be seen jus- tifying actions that can be seen as uncompetitive, as long as the outcome would result in the gain of the consumers, often in the shape of lower prices. (Crane, 2014)

In 2017 Lina Khan “public enemy number one” for the big tech titans as described by Foroohar in Don’t be Evil (p.176). published a research paper “Amazon’s Antitrust Para- dox” (Khan, 2017) which has gained enormous amounts of interest and discussion in the field. In her paper Khan shows why the old ways of considering Monopoly law does no longer work for the new digital era. Khan argues that “it doesn’t matter if companies such as Amazon are making things cheaper in dollars if they are using predatory pricing strat- egies to dominate multiple industries and choke off competition and choice”. Khan argues that that the consumer welfare point of view to antitrust laws is no longer applicable and that companies (more specifically Amazon, as title of her article states) should not get away with anticompetitive behavior only because it keeps the customers happy with lower prices. In Khan’s eyes company that owns a platform where it directly competes against third party sellers has inherent unfair advantage and should be regulated. (Khan, 2017)

Tim Wu an author, policy advocate, and professor at Columbia Law School best known for his work on the development of Net Neutrality theory and writing on private power, free speech, copyright, and antitrust, is also arguing that the consumer welfare point of view is no longer applicable to the internet companies as their services often come with no direct monetary cost to the customers. Wu writes on this point in his October 2020 article on The New York times “Over the past few years, the experience of conducting a Google search has been getting worse — at least if your goal is to find information, as opposed to viewing ads. And in the absence of real competition, Google manages to get away with shamelessly tracking your shopping habits, video-watching preferences and the content of your email conversations.” (Wu, 2020)

(15)

Providing a different angle to Wu, Khan is Robert W. Crandall, who in his 2019 article

“The Dubious Antitrust Argument for Breaking Up the Internet Giants” argues that historical cases against monopolies in oil, tobacco, motion picture and telecommunications have not proven to be as beneficial to economic welfare as researchers often state. Us- ing historical cases Crandall points out the difficulty of applying antitrust laws against the internet giants as their businesses are largely divided and do not merit monopoly power in any one single field such as digital advertising. Crandall believes forcing structural remedies (breakups) are likely to be ineffective, which leaves only forcing behavioral remedies via regulation and control, which in his opinion would likely raise same regulatory problems that rose during the AT & T case. (Crandall, 2019)

4 Analysis on European Legal structure

4.1 EU competition law articles 101 and 102

European Union’s competition law is composed of three domains which try to ensure that companies behave in accordance with the idea of free-market economy. In other words, the laws try to establish and uphold a market of fair competition for all participants regardless of size or stature. The three domains include article 101 of Treaty of the functioning European Union which deals with prohibition of cartels, article 102 of the same treaty (TFEU) was created to make sure that companies do not abuse their dominant position in the market and lastly Council Regulation 139/2004 which makes sure that Concentrations of businesses are in line with the previously mentioned idea of fair and free market economy.

4.1.1 Forbidden cartels and article 101

According to European commission a cartel is a group of two or more independent companies which join together to distort competition. Forbidden cartels, instead of increasing competition diminish it with deals that are beneficial for the companies in the cartel group, while simultaneously reduces their incentives to innovate or provide better services at lower prices to customers. Article 101 of the TFEU defines distortion of competition as action that:

a) directly or indirectly fix purchase or selling prices or any other trading conditions;

b) limit or control production, markets, technical development, or investment;

c) share markets or sources of supply;

(16)

d) apply dissimilar conditions to equivalent transactions with other trading parties,

thereby placing them at a competitive disadvantage;

As cartels are strictly illegal it is quite rare that two or more companies have written agreement to establish a cartel. For that reason, European Commission can conclude from evidence from the market behavior of the concerned parties that companies are in fact engaged in concerted practice and thus are a functioning as a cartel.

Once it has been established that the companies are distorting the competition and are in a cartel, a third feature must be examined and that is: are they undertakings. According to settled EU competition law case (Ministero dell’Economia e delle Finanze v Cassa di Risparmio di Firenze SpA and Others, 2006) it was established that the concept of undertaking, “covers any entity engaged in an economic activity, regardless of its legal status and the way in which it is financed”. (European Commision, 2021, p. 2.5) Establishing whether there is an undertaking or not is important as an undertaking does not necessarily have to be a company. This means that if an entity which is not a company is engaging in economic activity, it still needs to take the previous parts of article 101 TFEU into consideration when doing business.

There are exemptions to every rule and so there are in article 101 of TFEU. If a cartel can be proved to “confer sufficient benefits to outweigh the anti-competitive effects” such as the necessity for improvement of innovation which couldn’t be achieved without the cartel and that in the end it will benefit the consumers. Other way of getting away with cartel is when it can be proved that when it leads to no substantial distortion in a given market. A rule of thumb in the threshold percentage of combined market share in vertical cartel is 15% and in horizontal cartel 10%. Vertical cartel is a form of cartels where for example a manufacturer has an agreement with retailers whereas in horizontal cartels a deal has been struck between group of manufacturers or retailers. The assessment whether a cartel is forbidden will be done by the European Commission.

4.1.2 Leniency policy

To combat forbidden cartels European Commission has created a method called “Leni- ency Policy”. The idea of leniency policy’s is to increase the incentives for companies to hand over inside evidence of forbidden cartels to the European commission for exchange of partial or full immunity from fines imposed to all members of the cartel. According to European Commission the policy also has a deterring effect on cartels as it increases the mistrust and suspicion inside the group.

(17)

4.1.3 Article 102 and the abuse of dominant position

Second feature in European competition law is the prohibition of abusing company’s dominant position in any given market or sector. Being a dominant player in any markets or market sectors in European free markets is not in itself illegal nor is it discouraged or by European Commission. But what is illegal is to use that dominant position in abusive manner to gain unfair competitive edge within that market or sector and to determine whether dominant position has been abused. Like Article 101 of TFEU, Article 102 uses a three-step procedure to assess whether a company is abusing its dominant position or not.

First step in article 102 is to assess whether concerned undertaking is dominant or not.

Before assessing dominance, the Commission has to define product market and geographic markets where the undertaking is partaking as dominant position can only exist on a particular market. Product and Geographic markets are described in European Commission published fact sheet as follows:

“Product market: the relevant product market is made of all products/services which the consumer considers to be a substitute for each other due to their characteristics, their prices and their intended use.

Geographic market: the relevant geographic market is an area in which the conditions of competition for a given product are homogenous.” (European Commision, 2013)

European Commission has tools in its disposal to determine whether different products exist in the same product market or not. A computer model called small and significant but non transitory increase in prices (SSNIP) test is one of those tools in Commissions arsenal. The SSNIP test works by artificially increasing the price of product A and see if consumers will change into subsidiary product B. If consumers change into product B, it can be concluded that the market is bigger than just product A, but if the consumers continue to purchase product An even with the price increase then it can be concluded that the market is just for product A. SSNIP tests results can be viewed using different analysis’s such as Critical Loss Analysis and Pricing Analysis, but in the end it is only used as a tool to support arguments on finding a relevant market, but it in itself does not give conclusive answers to the question. (Amelio & Donath, 2009)

(18)

4.2 Concentrations (Council Regulation (EC) No 139/2004)

The third domain of European competition law is implemented to make sure that concentrations between undertakings being in line with the idea of fair and free market economy. The Legal basis for the regulation of concentrations is provided with Council Reg- ulation 139/2004, also known as the EU Merger Regulation. Concentrations, like cartels and market dominance mentioned above, are inherently not harmful to the markets. On the contrary consumers may benefit from the increases in efficiency and improvement in quality that follows from conjoint activities between different companies. European Com- mission does not review all mergers within the EU area, but only the ones that reach certain turnover thresholds. There are two sets of requirements that decide which mergers will be reviewed:

The first alternative requires:

i. a combined worldwide turnover of all the merging firms over €5 000 million, and ii. an EU-wide turnover for each of at least two of the firms over €250 million.

The second alternative requires:

i. a worldwide turnover of all the merging firms over €2 500 million, and

ii. a combined turnover of all the merging firms over € 100 million in each of at least three Member States,

iii. a turnover of over €25 million for each of at least two of the firms in each of the three Member States included under ii, and

iv. EU-wide turnover of each of at least two firms of more than €100 million.

Should a proposed concentration or merger exceed the previous thresholds, European Commission must be notified, and the Commission must examine the proposition. After the examination has been finished by the Commission the mergers will either be prohib- ited or approved with or without conditions. The decision is based on the outcome of the examination on the merger’s effects in the competition in the EU. If European Commis-

(19)

sion finds in its examination that a merger is harmful for the competition it may still ap- prove it, should the merging parties commit to action that would correct the harmful effect on competitions such as selling part of the combined business or to license technology to another market player. (European Commission, 2020)

4.3 GDPR

The data collected from the customers is not only used to improve internal processes and services, but also to be packaged and sold to third parties such as marketers and even political decision makers who want to target directly their most optimal potential customers. In 2016 Cambridge Analytica managed to use data from Facebooks person- ality test and passed it to third parties such as Donald Trump’s presidential campaign, which then used the data to target potential voters with advertisements. (Grassegger &

Krogerus, 2017)

In the EU are privacy is protected by the General Data Protection Regulation (GDPR) 679 which was adopted in May 2016 with aims to take Europe safely to the digital age.

GDPR aims to protect the basic rights and freedoms of natural persons by stipulating rules on the usage of personal information and its free movement. Directive is enforced when citizens personal data that leads to direct identification such as name, age or address or indirect identification such as several sources about certain person such as address and age which can be used to deduce the identity of the subject of the data.

(Art. 2) The principles that GDPR stipulates for data usage are:

Table 1 Regulation (EU) 2016/679 of the European Parliament and of the Council of 27 April 2016 Art. 5.

1. Lawfulness, fairness and transparency

Data processing should never lead to violation of the law.

Data processing should not be done in misleading manner and when data subject wishes to know what kind of information about them is possessed, they should be able to find it out.

2. Purpose limitation Purpose for data processing must be defined and com- municated to the data subject.

(20)

3. Data minimization Data processor should not exceed the minimum amount

of data required for realizing the legitimate interest.

4. Accuracy Data processed should be correct and, where necessary, kept up to date.

5. Storage limit Data should not be stored for longer than it is necessary for realizing the legitimate interest.

6. Integrity and confidentiality Data should be processed in a manner that ensures the security of the data.

Should the data be used according to the principles listed above, the next step for the data processor is to be able to provide grounds for justification listed in Article 6 if the regulation 679. These grounds include; Consent, Necessary in contract, Legal obligation, Vital interest, Public interest and Etcetera (Art 6) Should the data handler not be able to adhere to these justifications, all handling of the data is deemed illegal. Furthermore, there are certain types of data that are typically not allowed to be processed. This data is often so private by nature that no company should have anything to do with it. Data such as sexual orientation, personal health information or political preferences in general fall into this category and such private data can only be processed, if deemed necessary, under heavier regimen described in Article 9.

Lastly GDPR declares certain rights for the data subject over their data. For example, subject has a right to ratification of their data when it proves to be inadequate. Perhaps the most known right within GDPR is Article 17 or the right to be forgotten. Right to be forgotten states that “The data subject shall have the right to obtain from the controller the erasure of personal data concerning him or her without undue delay and the controller shall have the obligation to erase personal data without undue delay” (Wolford, 2021) Yet the right to be forgotten has its critics stating that the right “allows Europeans to erase irrelevant information” and that “European policymakers failed to conform their privacy law to the Internet’s architecture.” which leads to failing to achieve meaningful privacy for those who “exercise” the right to be forgotten. (Cunningham, 2017)

(21)

5 Recommendations/conclusions

Concluding the above discussed factors and taking into consideration the European Commissions Statement of Objections to Amazon (2020)³, Amazon’s use of Network effects, Economies of scale, switching costs, Aggressive acquisitions, Platform domination and Predatory pricing, Amazon has acted against EU laws, more specifically Article 102 TFEU by distorting competition in e-commerce and should be punished for it. Fur- thermore, as discussed earlier regarding the historical antitrust cases forcing structural remedies (breakups) would likely be ineffective, which leaves only forcing behavioral remedies via regulation and control, which would likely raise same regulatory problems that rose during the AT & T case.

To combat the regulatory issues this thesis recommends a method of leveling the playing field for new emerging companies in the way of Data sharing mandate.

5.1 How to remedy antitrust

Antitrust remedies can be roughly divided between structural and behavioral remedies, usage of a combination of the two is possible. Structural remedies are defined in Council Regulation (EC) No 1/2003 as “changes to the structure of an undertaking”, but no generally accepted definition within the discussion and literature of structural remedies exists. (Maier-Rigaud, 2013) Most commonly the change to the structure leads to

“breakups” which generally means divesture of a part of a company. The purpose of the remedy (also applies to other remedies) is to dispense or reduce the ability of the undertakings conduction that impedes or eliminates effective competition within the market.

Once sanctioned, structural remedies require little to none ongoing monitoring from me- dium to long term (Maier-Rigaud & Loertscher, 2020)

“Structural remedies should only be imposed either where there is no equally effective behavioural remedy or where any equally effective behavioural remedy would be more burdensome for the undertaking concerned than the structural remedy.”

-Council Regulation (EC) No 1/2003 (12)

3 Statement of Objections is not a sentence, but rather start of the investigations.

(22)

Behavioral remedies are targeted on the anticompetitive conduction that the undertaking is taking a part in. Whereas in structural remedies attempt to turn the undertaking away from uncompetitive actions with a single blow to the incentives of those actions, behavioral remedies endeavor to affect the actions that lead to the abuse by either requiring or prohibiting certain conduct. Behavioral remedies imposed on Microsoft in

Robert W. Crandall, who in his 2019 article “The Dubious Antitrust Argument for Breaking Up the Internet Giants” argues that historical cases against monopolies in oil, tobacco, motion picture and telecommunications have not proven to be as beneficial to economic welfare as researchers often state. Crandall’s article looks at antitrust from an US view- point where many antitrust cases with high public interest have occurred in the IT field.

US antitrust legislation mostly revolves around Sherman act sections 1 and 2, which target restraints on trade such price fixing, market division etc. and monopolization, respectively. Using historical cases such as U.S. V. Standard Oil, U.S. V. American To- bacco U.S. V. AT & T, prime examples used in almost all antitrust cases afterwards. U.S.

V. IBM, a case that was dropped after 13 years of hard work and considerable expendi- ture due speed of technological innovation that led to products that were in the centre of the case not even produced anymore. U.S. V. Microsoft corporation, the latest big tech related case where divesture of Microsoft corporation nearly took place. With these cases Crandall points out the difficulty of applying antitrust laws against the internet giants as their businesses are largely divided and do not merit monopoly power in any one single field such as digital advertising or search. Crandall believes forcing structural remedies (breakups) are likely to be ineffective, which leaves only forcing behavioral remedies via regulation and control, which in his opinion would likely raise same regulatory problems that rose during the AT & T case. (Crandall, 2019)

5.2 Data sharing mandate

One form of behavioural remedy that has been suggested to help level the playing field for competition and new entries is mandated sharing of data. This practise has been championed by Oxford professor Viktor Mayer-Schönberger and The Economist writer Thomas Ramge in their book “Reinventing Capitalism in the Age of Big Data” (2018) and in their cowritten article “A Big choice for Big Tech: Share data or suffer the consequences” (2018). Which offers a different, perhaps more modern, option to the historical trustbusting methods of breaking big companies.

(23)

Mayer-Shönberger and Ramge believe that big tech companies collect and use enormous amounts of data that to better their algorithms and other predictive services are what in the end create value to the consumers. In the fields of machine learning and training artificial intelligence, quantity of data almost directly translates to quality when training the models to, for example predict customer behaviour or preferences. Thus, when limiting the amount of data available to the companies, via separating their different fields of businesses into different companies, their predictive services would conse- quently suffer and that would in term extract from consumer welfare. Data sharing mandate would lower the barriers of entry for new companies in the fields that are heavily AI/Data driven as gathering the relevant data for training predictive models of algorithms would be more accessible for smaller companies. Basis of Mayer-Shönberger’s and Ramge’s theory lies in the statement that the modern “superstar firms” do not so much compete in the traditional markets but are in themselves the markets. Mayer-Shönberger and Ramge also believe that existing regulatory tools are not adept in tackling the FAANG companies ever rising power in the markets. For that reason, they bring forward progressive data sharing mandate.

“Under this system, every company above a certain size, say, those with more than a ten percent share of the market, that systematically collects, and analyses data would have to let other companies in the same market access a subset of its data. … Data would be stripped of personal identifiers, augmented with metadata to make clear what sort of information the data provided and where it came from, and selected randomly”.

This would limit the role of regulators to assessing market shares and enforce the access to data. Mayer-Shönberger and Ramge believe that progressive data sharing mandate would offer several advantages such as; it would not “cost” anything to the companies, unlike taxes, it would allow broader innovation from the same data due different views of using it and the competitive edge that companies might acquire would not be dependable on the sheer amounts of data that they have managed to extract, but rather the insights they are able to gain from the data. Mayer-Shönberger and Ramge address the worries that data sharing would increase the power that firms have over customers in a rather cynical way of stating that; so far regulators have anyway failed to counter power imbal- ances between users and companies by strengthening individual privacy rights and that even in Europe where privacy protection is relatively strong, consumers just blindly click

“OK” without considering the options. (Mayer-Schönberger & Ramge, 2018)

(24)

6 References

About Amazon Staff, 2021. Why Amazon warehouses are called fulfilment centres.

[Online]

Available at: https://www.aboutamazon.co.uk/amazon-fulfilment/our-fulfilment- centres/why-amazon-warehouses-are-called-fulfilment-centers

[Haettu 26 September 2021].

Afuah, A., 2013. ARE NETWORK EFFECTS REALLY ALL ABOU. Strategic Management Journal, Issue 34, pp. 257-273.

Amelio, A. & Donath, D., 2009. Market definition in recent EC merger investigations: The role of empirical analysis. Law & Economics Concurrences, Osa/vuosikerta 3.

Bezos, J. P., 1997. Jeff Bezos' original letter to shareholders is reprinted from Amazon's

1997 Annual Report.. [Online]

Available at: https://www.aboutamazon.com/news/company-news/amazons-original- 1997-letter-to-shareholders

[Haettu 13 January 2021].

Brodkin, J., 2019. WSJ: Amazon changed search results to boost profits despite internal

dissent. [Online]

Available at: https://arstechnica.com/tech-policy/2019/09/amazon-changed-search- algorithm-to-favor-its-own-products-wsj-reports/

[Haettu 30 March 2020].

Chen, P.-y. & Hitt, L. M., 2005. Information Technology and Switching Costs. [Online]

Available at:

https://static.aminer.org/pdf/PDF/000/326/425/information_technology_innovation_and _competition_in_the_presence_of_switching.pdf

Chevalier, S., 2021. Share of paid units sold by third-party sellers on Amazon platform as of 2nd quarter 2021 , s.l.: Statista.

Congregado, E., Golpe, A. A. & van Stel, A., 2014. The role of scale economies in determining firm size in modern economies. The Annals of Regional Science, 52(2), pp.

431-455.

Coppola, D., 2020. Annual net income of Amazon.com from 2004 to 2019 (in million U.S.

dollars). [Online]

Available at: https://www-statista-com.ezproxy.metropolia.fi/statistics/266288/annual-et- income-of-amazoncom/

(25)

Crandall, R. W., 2019. The Dubious Antitrust Argument for Breaking Up the Internet Giants. Review of Industrial Organization, 54(4), pp. 627-649.

Crane, D., 2014. The Tempting of Antitrust: Robert Bork and theGoals of Antitrust Policy.

University of Michigan Law School Scholarship Repository, Osa/vuosikerta 79, pp. 835- 853.

Cunningham, M., 2017. Privacy Law That Does Not Protect Privacy, Forgetting the Right to Be Forgotten. Buffalo Law Review, 65(495).

Dougherty, C., 2017. Inside Yelp’s Six-Year Grudge Against Google. The New York Times, 7 January.

Duggan, W., 2020. Latest E-Commerce Market Share Numbers Highlight Amazon's

Dominance. [Online]

Available at: https://finance.yahoo.com/news/latest-e-commerce-market-share- 185120510.html?guccounter=1&guce_referrer=aHR0cHM6Ly93d3cuZ29vZ2xlLmNvbS 8&guce_referrer_sig=AQAAAIcZ1MwNL5NK42xb9AjbegBT6Dw0cSvFXaSgHUdHgX9 Plq9DAnYGP_Mic-q7S99HhtOpFBpjI1HK-n3hFgi4ukDJkWRY8N9HC

Ethier, W. J., 2009. Economies of scale. Teoksessa: The Princeton Encyclopedia of the World Economy. Princeton: Princeton University Press, pp. 332-337.

European Commision, 2013. Competition: Antitrust procedures in abuse of dominance.

[Online]

Available at:

https://ec.europa.eu/competition/publications/factsheets/antitrust_procedures_102_en.

pdf

[Haettu 12 August 2021].

European Commision, 2021. COMMISSION NOTICE. [Online]

Available at: https://eur-lex.europa.eu/legal-

content/EN/TXT/PDF/?uri=CELEX:52010SC0411&from=EN [Haettu 12 August 2021].

European Commission, 2020. Mergers. [Online]

Available at: https://ec.europa.eu/competition-policy/mergers_en [Haettu 12 August 2020].

Evans, D. S. & Schmalensee, R., 2013. The Antitrust Analysis of Multi-Sided Platform Businesses, Cambridge: National Bureau of Economic Research.

Fornell, C., 1992. Journal of Marketing, 56(1), p. 6.

Foroohar, R., 2019. The Network Effect. Teoksessa: Don’t be evil. New York: Currency, pp. 129-147.

(26)

Foroohar, R., 2019. The Network Effect. Teoksessa: Dont be evil. New York: Currency, pp. 134-136.

Grassegger, H. & Krogerus, M., 2017. The Data That Turned the World Upside Down, s.l.: Vice.

Johnson, M., 2020. Home advantage? Who wins in multi-sided platform competition?.

[Online]

Available at: https://www.oxera.com/insights/agenda/articles/home-advantage-who- wins-in-multi-sided-platform-competition/#nav-site

Katz, M. L. & Shapiro, C., 1994. Systems Competition and Network Effects. The Journal of Economic Perspectives (1986-1998), pp. 93-115.

Kelion, L., 2021. Why Amazon knows so much about you. [Online]

Available at: https://www.bbc.co.uk/news/extra/CLQYZENMBI/amazon-data [Haettu 13 January 2021].

Klemperer, P., 1987. Markets with consumer switching costs.. Quarterly Journal of Economics, 1(102), pp. 375-395.

Maier-Rigaud, F. P., 2013. Behavioural versus Structural Remedies in EU Competition Law. European Competition Law Annual 2013, pp. 207-224.

Maier-Rigaud, F. P. & Loertscher, B., 2020. STRUCTURAL vs. BEHAVIORAL REMEDIES. CPI Antitrust Chronicle .

Mayer-Schönberger, V. & Ramge, T., 2018. A Big Choice for Big Tech: Share Data or

Suffer the Consequences. [Online]

Available at: https://www.foreignaffairs.com/articles/world/2018-08-13/big-choice-big- tech

[Haettu 26 October 2020].

Microsoft, 2016. news.xbox.com. [Online]

Available at:

https://www.google.com/url?sa=t&rct=j&q=&esrc=s&source=web&cd=&ved=2ahUKEwj f2_ThoojzAhUlxosKHebECV0QFnoECCMQAQ&url=https%3A%2F%2Fnews.xbox.com

%2Fen-us%2Fwp-content%2Fuploads%2Fsites%2F2%2FHalo-Franchise-Fact- Sheet_May-2016-FINAL.docx&usg=AOvVaw0XzO9Qt7RnB

Ministero dell’Economia e delle Finanze v Cassa di Risparmio di Firenze SpA and Others (2006) European Court Reports I-00289.

Oremus, W., 2013. The Time Jeff Bezos Went Thermonuclear on Diapers.com, s.l.:

Slate.com.

(27)

Patacconi, G. & Russo, F., 2017. Rethinking Economies of Scale Towards Network Economies. Development, 58(4), pp. 521-527.

Reeves, B., 2019. The Last of Us and Uncharted 4 Reach Incredible Sales Milestones.

[Online]

Available at: https://www.playstationlifestyle.net/2019/10/14/the-last-of-us-sales- uncharted-4-sales/

Reuters Staff, 2019. U.S. retail sales post biggest gain in one-and-a-half years in March.

Reuters, 18 April.

Rochet, J.-C. & Tirole, J., 2006. Two-sided markets: a progress report. The Rand Journal of Economics, 37(3), pp. 645-667.

Sabanoglu, T., 2020. Amazon’s share of online retail sales in selected regions as of

September 2020. [Online]

Available at: https://www.statista.com/statistics/1183515/amazon-market-share-region- worldwide/

Sabanoglu, T., 2020. Annual net revenue of Amazon from 2004 to 2019 (in billion U.S.

dollars). [Online]

Available at: https://www-statista-com.ezproxy.metropolia.fi/statistics/266282/annual- net-revenue-of-amazoncom/

Sainato, M., 2020. ‘I'm not a robot’: Amazon workers condemn unsafe, grueling conditions at warehouse. The Guardian, 5 February.

Satariano, A. & Gross, J., 2021. Amazon had a big year, but paid no tax to Luxembourg, its European headquarters. The New York Times, 4 May.

Satariano, A. & Isaac, M., 2018. Facebook Used People's Data to Favor Certain Partners and Punish Rivals, Documents Show. The New York Times, 5 December.

Schmalensee, R., 2014. An Instant Classic:. CompetitionPolicyInternational, 10(2).

Streitfeld, D., 2015. Amazon Reports Unexpected Profit, and Stock Soars. The New York Times, 23 July.

Verizon Media, 2021. Amazon.com, Inc. (AMZN) NasdaqGS Real Time Price. Currency

in USD. [Online]

Available at:

https://finance.yahoo.com/chart/AMZN?p=AMZN#eyJpbnRlcnZhbCI6IndlZWsiLCJwZX Jpb2RpY2l0eSI6MSwiY2FuZGxlV2lkdGgiOjEsImZsaXBwZWQiOmZhbHNlLCJ2b2x1b WVVbmRlcmxheSI6dHJ1ZSwiYWRqIjp0cnVlLCJjcm9zc2hhaXIiOnRydWUsImNoYXJ0

(28)

VHlwZSI6ImxpbmUiLCJleHRlbmRlZCI6ZmFsc2UsIm1hcmt

[Haettu 17 February 2021].

Weise, K., 2021. Amazon’s quarterly profit jumps to $7.8 billion.. The New York Times, 29 July.

West, E., 2019. Amazon: Surveillance as a Service. Surveillance & Society, pp. 27-33.

Wikipedia, 2020. List of mergers and acquisitions by Amazon. [Online]

Available at:

https://en.wikipedia.org/wiki/List_of_mergers_and_acquisitions_by_Amazon [Haettu 20 March 2020].

Wolford, B., 2021. Everything you need to know about the “Right to be forgotten”. [Online]

Available at: https://gdpr.eu/right-to-be-forgotten/

Zuboff, S., 2019. The Age of Surveillance Capitalism. New York: Public Affairs.