1 00:00:00,000 --> 00:00:14,920 Welcome to the Oxford Undergraduate Law Podcast where we discuss the law and its implications 2 00:00:14,920 --> 00:00:18,440 on our relationships, our markets, and our futures. 3 00:00:18,440 --> 00:00:19,800 I'm Siobhan. 4 00:00:19,800 --> 00:00:26,080 And I'm Bianca, and we're your hosts for this series. 5 00:00:26,080 --> 00:00:31,360 Today we are incredibly honoured to be speaking with Professor Ariel Ezrachi on the role of 6 00:00:31,360 --> 00:00:35,280 competition law in regulating big tech companies. 7 00:00:35,280 --> 00:00:39,240 Professor Ezrachi is the Slaughter and May Professor of Competition Law at the University 8 00:00:39,240 --> 00:00:45,160 of Oxford and the Director of the University of Oxford Centre for Competition Law and Policy. 9 00:00:45,160 --> 00:00:49,800 He's the co-editor-in-chief of the Journal of Antitrust Enforcement and the author and 10 00:00:49,800 --> 00:00:55,480 co-author of numerous books, including Competition Overdose and Virtual Competition. 11 00:00:55,480 --> 00:00:59,600 Professor Ezrachi's research and commentary have been featured in The Economist, The New 12 00:00:59,600 --> 00:01:07,760 Yorker, The Wall Street Journal, and many other influential international outlets. 13 00:01:07,760 --> 00:01:11,380 So to start off, what is competition law? 14 00:01:11,380 --> 00:01:18,120 So there is a basic assumption that we have as a society that competitive markets deliver 15 00:01:18,120 --> 00:01:23,000 welfare to us, to the community, to society, to customers. 16 00:01:23,000 --> 00:01:29,600 And the reason is that when you have competition, each one of us in a market tries their best 17 00:01:29,600 --> 00:01:36,080 to make money, and to do so we lower the prices so we can sell more, we improve the quality, 18 00:01:36,080 --> 00:01:37,960 and we improve the service. 19 00:01:37,960 --> 00:01:42,640 So you probably heard about Adam Smith, the invisible hand of competition. 20 00:01:42,640 --> 00:01:49,840 The general idea that it is at the heart of modern markets is that more competition benefits 21 00:01:49,840 --> 00:01:50,840 us. 22 00:01:50,840 --> 00:01:56,520 Now, competition makes life better for us as consumers, but it can make the life of 23 00:01:56,520 --> 00:01:59,080 sellers rather difficult. 24 00:01:59,080 --> 00:02:02,560 Because of course if you're a seller and you operate in a competitive market, you have 25 00:02:02,560 --> 00:02:04,000 to work very hard. 26 00:02:04,000 --> 00:02:08,560 You need to improve your product, you need to lower the price which requires you to become 27 00:02:08,560 --> 00:02:10,360 more efficient. 28 00:02:10,360 --> 00:02:16,440 And because of that, sometimes sellers may try to dumpen the competition. 29 00:02:16,440 --> 00:02:21,560 They might engage in cartels where they fix the price or share the market. 30 00:02:21,560 --> 00:02:26,920 Or you might have a very powerful company, we call them dominant companies, monopolies, 31 00:02:26,920 --> 00:02:28,880 that will abuse its market power. 32 00:02:28,880 --> 00:02:32,000 And this is where competition law steps in. 33 00:02:32,000 --> 00:02:38,560 So competition law is a set of regulations and laws that are aimed at protecting the 34 00:02:38,560 --> 00:02:44,120 dynamic of competition, making sure that we have undistorted competition, and making sure 35 00:02:44,120 --> 00:02:47,480 that we maximize consumer welfare. 36 00:02:47,480 --> 00:02:55,840 So it's a set of laws that prevent companies from engaging in anti-competitive agreements, 37 00:02:55,840 --> 00:03:02,160 prevent monopolies from abusing their powers, and also is relevant for mergers and acquisitions, 38 00:03:02,160 --> 00:03:07,680 because what we try to make sure is that as companies merge, this does not result in market 39 00:03:07,680 --> 00:03:11,180 power that can later distort competition on the market. 40 00:03:11,180 --> 00:03:17,360 So that's the basic idea behind competition law, making sure that markets work. 41 00:03:17,360 --> 00:03:21,880 Is there a difference between digital markets and regular markets? 42 00:03:21,880 --> 00:03:27,280 So in the regular market, we rely on the invisible hand that I mentioned earlier. 43 00:03:27,280 --> 00:03:33,720 And that means that you have demand and supply, and these powers, they determine what is the 44 00:03:33,720 --> 00:03:38,960 market price, and they determine how much choice you might have on the market. 45 00:03:38,960 --> 00:03:43,720 If you shift all of this into the digital arena, what you have is something that looks 46 00:03:43,720 --> 00:03:51,120 very much like a real market, but in reality, it is a market that is controlled by algorithms. 47 00:03:51,120 --> 00:03:54,160 And this is where the difference lies. 48 00:03:54,160 --> 00:04:00,000 You have, for instance, platforms or websites that control the interface. 49 00:04:00,000 --> 00:04:05,720 So if you go on a website to shop, someone can decide what it is that you see on that 50 00:04:05,720 --> 00:04:08,560 website and what it is that you don't. 51 00:04:08,560 --> 00:04:11,640 They can decide how much information is available to you. 52 00:04:11,640 --> 00:04:17,540 They can decide whether they want to target you specifically and engage in what we refer 53 00:04:17,540 --> 00:04:22,080 to as discriminatory pricing, whether to give you a discount. 54 00:04:22,080 --> 00:04:28,760 So the invisible hand in some ways is pushed aside and is replaced by a digitalized hand. 55 00:04:28,760 --> 00:04:33,560 It's a controlled environment that looks very much like real markets, but in fact, it's 56 00:04:33,560 --> 00:04:34,560 not. 57 00:04:34,560 --> 00:04:39,080 Let me give you just a few examples on what it means. 58 00:04:39,080 --> 00:04:46,520 Because I can, as an online operator, harvest quite a lot of your data, it means that I 59 00:04:46,520 --> 00:04:49,560 will know about your search history. 60 00:04:49,560 --> 00:04:52,840 I might know about your address, your postcode. 61 00:04:52,840 --> 00:04:56,480 I might have information about what you like, what you don't like. 62 00:04:56,480 --> 00:05:02,120 And I'm able, based on the patterns that I identified, to estimate what is your willingness 63 00:05:02,120 --> 00:05:04,820 to pay for a service. 64 00:05:04,820 --> 00:05:10,240 And what that does is enable me to charge you a specific price. 65 00:05:10,240 --> 00:05:15,560 I can charge you a higher price, making it look as if this is the market price. 66 00:05:15,560 --> 00:05:20,520 And indeed, we have a lot of examples of that and there are a lot of systems in place that 67 00:05:20,520 --> 00:05:24,000 engage in such discriminatory pricing. 68 00:05:24,000 --> 00:05:28,960 It's just something that you can do in terms of power, added power that you have because 69 00:05:28,960 --> 00:05:32,100 of big data and big algorithms. 70 00:05:32,100 --> 00:05:35,000 But it's not limited just to products and prices. 71 00:05:35,000 --> 00:05:39,760 If you think about the market for ideas, it also is relevant there. 72 00:05:39,760 --> 00:05:45,960 On the digital market, I will know much more about what you do, who do you interact with. 73 00:05:45,960 --> 00:05:50,580 So I can build a profile and I can micro target you accordingly. 74 00:05:50,580 --> 00:05:54,160 So of course, you heard the stories about Cambridge Analytica and you know the stories 75 00:05:54,160 --> 00:06:03,640 about how I can influence people's beliefs or political affiliations. 76 00:06:03,640 --> 00:06:08,920 All of that can be viewed also from a competition law perspective because it is all about market 77 00:06:08,920 --> 00:06:09,920 dynamics. 78 00:06:09,920 --> 00:06:12,640 It is all about the exchanges that we have. 79 00:06:12,640 --> 00:06:17,840 And in this digital arena, what competition law is trying to do is to ensure that markets 80 00:06:17,840 --> 00:06:19,680 are not distorted. 81 00:06:19,680 --> 00:06:25,560 So you have cases that deal with either using algorithms to fix the price. 82 00:06:25,560 --> 00:06:29,080 We have cases that deal with discrimination. 83 00:06:29,080 --> 00:06:37,000 We have a lot of cases and this is the main area of focus these days that look at platforms. 84 00:06:37,000 --> 00:06:42,960 The power of the platforms, either the ability to manipulate, but very often the ability 85 00:06:42,960 --> 00:06:44,220 to exploit us. 86 00:06:44,220 --> 00:06:48,480 This is really the focal point of a lot of what is discussed at the moment. 87 00:06:48,480 --> 00:06:52,800 So can you tell us a little bit more about some of the issues we face today with digital 88 00:06:52,800 --> 00:06:54,140 markets? 89 00:06:54,140 --> 00:06:59,340 So I guess the thing I will start with is the asymmetry of information. 90 00:06:59,340 --> 00:07:08,520 When you go shopping online, when you join the social network online, what you see is 91 00:07:08,520 --> 00:07:12,400 determined by engineers that sit on the other side. 92 00:07:12,400 --> 00:07:19,000 They have much more information about you than you think and they control much more 93 00:07:19,000 --> 00:07:21,000 than you are aware of. 94 00:07:21,000 --> 00:07:27,840 In fact, I tend to say that when you shop online or when you interact online, often 95 00:07:27,840 --> 00:07:34,520 you will walk on a path with a real sense of autonomy without appreciating that someone 96 00:07:34,520 --> 00:07:41,680 created that path especially for you and is leading you in a very specific direction. 97 00:07:41,680 --> 00:07:49,020 So most of us are completely unaware of the ability to use dark patterns, to use nudges 98 00:07:49,020 --> 00:07:51,840 or manipulation to affect what we do. 99 00:07:51,840 --> 00:07:57,060 And that's true of course in marketing and everywhere, but online there is just this 100 00:07:57,060 --> 00:08:02,920 enhanced capacity and this just changes the dynamics of competition. 101 00:08:02,920 --> 00:08:08,160 Linked to that, and that's another challenge, is the stealth mode of these elements. 102 00:08:08,160 --> 00:08:12,440 The fact that it's very hard to detect them. 103 00:08:12,440 --> 00:08:17,840 Even if you're aware that something is done from a strategic viewpoint, it's quite difficult 104 00:08:17,840 --> 00:08:24,320 to put your finger on the exact way in which dark patterns are utilized, in which someone 105 00:08:24,320 --> 00:08:28,320 is manipulated or the way it affects the price. 106 00:08:28,320 --> 00:08:34,980 Moving more toward the way markets operate, one of the key challenges are what we refer 107 00:08:34,980 --> 00:08:38,080 to as network effects. 108 00:08:38,080 --> 00:08:44,320 And network effects make it difficult for new companies to challenge the incumbent, 109 00:08:44,320 --> 00:08:47,320 challenge the companies that are already established on the market. 110 00:08:47,320 --> 00:08:52,360 And this is something which is a characteristic of many digital markets. 111 00:08:52,360 --> 00:09:02,580 So for instance, if I now invent a new social platform and I invite you to join and I tell 112 00:09:02,580 --> 00:09:08,920 you of all the benefits and you ask me, so how many other people are on your platform? 113 00:09:08,920 --> 00:09:11,520 And I tell you, well, you're the first one. 114 00:09:11,520 --> 00:09:16,380 That's not a very attractive proposition if you compare it to the billions of people that 115 00:09:16,380 --> 00:09:23,800 you have on other platforms, which means that it is quite difficult for newcomers to actually 116 00:09:23,800 --> 00:09:29,440 be able to surpass the barrier to entry to those markets. 117 00:09:29,440 --> 00:09:35,560 And network effects not only affect us as users, they also affect the sellers, the application 118 00:09:35,560 --> 00:09:39,280 developers, the providers, the advertisers. 119 00:09:39,280 --> 00:09:46,240 If I'm a new platform and I need to attract revenue, I need to provide a sufficiently 120 00:09:46,240 --> 00:09:50,560 large scale of users. 121 00:09:50,560 --> 00:09:55,200 And if I don't have that, then advertisers and all of those that actually pay for the 122 00:09:55,200 --> 00:10:01,680 services, because we call these multi-side markets, they will prefer to go to the leader, 123 00:10:01,680 --> 00:10:03,680 whoever is leading the market. 124 00:10:03,680 --> 00:10:12,680 And because of that, some of the markets that we have tipped in favor of the dominant company. 125 00:10:12,680 --> 00:10:17,960 And when you have a dominant company that feels very comfortable because it is not subjected 126 00:10:17,960 --> 00:10:24,520 to real potential competition, then they can start and change the way they behave. 127 00:10:24,520 --> 00:10:30,280 If you think of Facebook in the early days, privacy was a big thing. 128 00:10:30,280 --> 00:10:39,160 So when at some point Facebook changed some of the default definitions in terms of privacy, 129 00:10:39,160 --> 00:10:41,400 there was a real criticism. 130 00:10:41,400 --> 00:10:47,520 And back then, they apologized publicly and they said, we're so sorry, this is important 131 00:10:47,520 --> 00:10:51,000 for us that you feel comfortable and so forth. 132 00:10:51,000 --> 00:10:57,520 At some point when you gain power, you have to think slightly less about, we look at it 133 00:10:57,520 --> 00:10:59,680 as the quality of the service. 134 00:10:59,680 --> 00:11:03,400 Privacy can be regarded from a competition perspective as a reduction of quality. 135 00:11:03,400 --> 00:11:08,520 I can reduce privacy, reduce the quality that I offer you, because I know that you don't 136 00:11:08,520 --> 00:11:10,880 have outside options. 137 00:11:10,880 --> 00:11:13,800 It is harder for you to leave Facebook. 138 00:11:13,800 --> 00:11:15,160 Why? 139 00:11:15,160 --> 00:11:17,520 Because of the network effects. 140 00:11:17,520 --> 00:11:22,400 Of course you can, in theory, but you will have to take quite a lot of circles of friends 141 00:11:22,400 --> 00:11:23,680 with you. 142 00:11:23,680 --> 00:11:25,840 You might have already preferences. 143 00:11:25,840 --> 00:11:31,040 You might have affiliated products that are linked to that network. 144 00:11:31,040 --> 00:11:35,760 So the longer we stay, the harder it is for us to say goodbye. 145 00:11:35,760 --> 00:11:42,720 And the more I know about what makes you stay with me, the more I can make sure that I deliver 146 00:11:42,720 --> 00:11:48,040 that and I can start giving you less of other things that are very important for me. 147 00:11:48,040 --> 00:11:53,960 Because the less privacy I give you, it means that I have more data about you. 148 00:11:53,960 --> 00:11:58,200 Data is power and money because then I can profile you. 149 00:11:58,200 --> 00:12:01,960 I can sell that data to advertisers who can target you. 150 00:12:01,960 --> 00:12:09,440 I can sell that data for possibly political organizations who want to target you to convince 151 00:12:09,440 --> 00:12:11,800 you to join their cause. 152 00:12:11,800 --> 00:12:16,880 So micro targeting works really well when I know a lot about you and I give you very 153 00:12:16,880 --> 00:12:19,280 little protection. 154 00:12:19,280 --> 00:12:23,780 It's an interesting thing because when we started and you asked me about competition 155 00:12:23,780 --> 00:12:30,520 and competition law, I explained that markets work for us, the consumers. 156 00:12:30,520 --> 00:12:36,220 And in fact what I describe to you now is a situation where the market doesn't necessarily 157 00:12:36,220 --> 00:12:40,360 work for you as a user. 158 00:12:40,360 --> 00:12:46,920 And this is why some people argue that there is room for intervention because the market 159 00:12:46,920 --> 00:12:49,000 does not deliver. 160 00:12:49,000 --> 00:12:55,240 So are there any solutions that we can implement to combat these issues? 161 00:12:55,240 --> 00:13:01,080 It's a very good question and it's a simple question with a complicated answer. 162 00:13:01,080 --> 00:13:03,720 So the solution can come from various directions. 163 00:13:03,720 --> 00:13:09,540 For example, you can focus on the data and you can say who owns the data? 164 00:13:09,540 --> 00:13:12,840 So this can be a debate. 165 00:13:12,840 --> 00:13:17,880 Is the data yours or is it, let's say, the platform's? 166 00:13:17,880 --> 00:13:26,040 And once the platform worked on the data and engaged in analytics, is the product that 167 00:13:26,040 --> 00:13:29,840 they created yours or is it the platform's? 168 00:13:29,840 --> 00:13:32,880 And let's say that it's not yours. 169 00:13:32,880 --> 00:13:37,000 Can they transfer it to third parties? 170 00:13:37,000 --> 00:13:40,360 Do they have to allow access? 171 00:13:40,360 --> 00:13:47,840 And even if you look at that very simple interface, you can see that there is friction here between 172 00:13:47,840 --> 00:13:51,220 privacy and competition. 173 00:13:51,220 --> 00:13:58,560 Because your privacy might suggest that if I am your provider and I got your data, you 174 00:13:58,560 --> 00:14:02,400 say, well, it's one thing that you have my data, I don't want anyone else to have that 175 00:14:02,400 --> 00:14:04,140 data. 176 00:14:04,140 --> 00:14:06,200 That's my privacy. 177 00:14:06,200 --> 00:14:13,000 Whereas from a competition law perspective, we might say if we force the dominant player 178 00:14:13,000 --> 00:14:19,160 to give that data to third parties, there will be more competition and you will benefit. 179 00:14:19,160 --> 00:14:25,120 So it's not that all the policies are aligned because some policies protect our autonomy, 180 00:14:25,120 --> 00:14:28,520 some policies try to protect the market. 181 00:14:28,520 --> 00:14:31,520 So it can turn a little bit complicated. 182 00:14:31,520 --> 00:14:36,360 So I just mentioned this is an example of data, then you have privacy. 183 00:14:36,360 --> 00:14:43,500 And of course, you can have regulation such as GDPR that will protect your privacy. 184 00:14:43,500 --> 00:14:45,360 You can have the e-privacy regulation. 185 00:14:45,360 --> 00:14:50,880 You can have various rules that are there to protect your autonomy. 186 00:14:50,880 --> 00:14:57,980 The area that I focus on is the possible role of competition law in this story and the possible 187 00:14:57,980 --> 00:14:59,840 role of regulation. 188 00:14:59,840 --> 00:15:07,320 So competition law can impact in two or maybe even three significant ways. 189 00:15:07,320 --> 00:15:11,760 It can deal with what we refer to as abuse of dominant position. 190 00:15:11,760 --> 00:15:17,960 And if you heard about, for example, the Google case, the Google shopping case, so the European 191 00:15:17,960 --> 00:15:22,840 Commission accused Google of abusing its market position. 192 00:15:22,840 --> 00:15:28,240 Or if you heard about the Facebook case in Germany, the Bundeskartelln accused Facebook 193 00:15:28,240 --> 00:15:36,160 of exploiting its position and using it to harvest excess information from users. 194 00:15:36,160 --> 00:15:38,000 So this is one way to deal with it. 195 00:15:38,000 --> 00:15:42,440 Or through, if there is an agreement, you can also apply the law to anti-competitive 196 00:15:42,440 --> 00:15:44,480 agreements. 197 00:15:44,480 --> 00:15:50,240 Another way is to try and approach this in the area of mergers and acquisitions. 198 00:15:50,240 --> 00:15:56,700 A lot of those leading platforms engage in routine acquisitions of smaller companies 199 00:15:56,700 --> 00:16:02,920 that could become potential competitors and could maybe disrupt the market. 200 00:16:02,920 --> 00:16:09,040 So if we have better tools to monitor these transactions, we could have markets that are 201 00:16:09,040 --> 00:16:11,160 more competitive. 202 00:16:11,160 --> 00:16:15,080 And this could also help us resolve some of the problems that we discussed. 203 00:16:15,080 --> 00:16:20,720 Because if there is competition, then I need to start giving you a better service because 204 00:16:20,720 --> 00:16:22,840 I want to get your business. 205 00:16:22,840 --> 00:16:27,760 And if this is what I do, then if we spoke about privacy, I will have to offer you better 206 00:16:27,760 --> 00:16:30,600 protection when it comes to privacy. 207 00:16:30,600 --> 00:16:35,480 And you can also engage in something that is sometimes called market investigations. 208 00:16:35,480 --> 00:16:40,460 It's a preemptive tool where you just look at a market and you try to see how can we 209 00:16:40,460 --> 00:16:44,080 increase the competitiveness of the market. 210 00:16:44,080 --> 00:16:46,400 So antitrust law can work in two ways. 211 00:16:46,400 --> 00:16:47,400 One is reactive. 212 00:16:47,400 --> 00:16:54,640 This is sometimes referred to as ex post intervention, and one is proactive, preemptive. 213 00:16:54,640 --> 00:16:58,360 This is ex ante intervention. 214 00:16:58,360 --> 00:17:08,360 And the main challenge that we have is that often by the time you have a decision under 215 00:17:08,360 --> 00:17:12,480 antitrust laws, it might be slightly too late. 216 00:17:12,480 --> 00:17:14,200 The market has been distorted. 217 00:17:14,200 --> 00:17:16,560 There is very little competition. 218 00:17:16,560 --> 00:17:22,880 And many people would say that even in cases that were successful, the remedy did not restore 219 00:17:22,880 --> 00:17:25,260 competition on the market. 220 00:17:25,260 --> 00:17:29,600 And that affects the quality, the service, and the price. 221 00:17:29,600 --> 00:17:33,560 And in online, often the price is not money that we pay. 222 00:17:33,560 --> 00:17:38,520 We kind of pay with data usually, with information about ourselves. 223 00:17:38,520 --> 00:17:45,140 And because of that, because of the limitations of antitrust, this is why you probably came 224 00:17:45,140 --> 00:17:50,000 across a lot of reports on proposals for regulation. 225 00:17:50,000 --> 00:17:58,000 So the idea is to supplement and enhance the power of competition by having specific regulations 226 00:17:58,000 --> 00:18:04,640 that target platforms, target gatekeepers, target companies that have strategic power 227 00:18:04,640 --> 00:18:06,200 on the market. 228 00:18:06,200 --> 00:18:10,480 So the idea is to have this dual mechanism. 229 00:18:10,480 --> 00:18:15,760 Often law will deal with abuse of dominant position, anti-competitive agreements, of 230 00:18:15,760 --> 00:18:17,600 course mergers. 231 00:18:17,600 --> 00:18:24,040 But in some areas where you can preempt and you can identify a problem, we will have a 232 00:18:24,040 --> 00:18:25,880 specific regime. 233 00:18:25,880 --> 00:18:31,760 And that regime will apply to begin with on companies that are very powerful on the market. 234 00:18:31,760 --> 00:18:38,200 Basically, there will be a law that they will be subjected to and will restrict their power 235 00:18:38,200 --> 00:18:44,720 either when it comes to their behavior with relations to consumers or with relation to 236 00:18:44,720 --> 00:18:46,600 service providers. 237 00:18:46,600 --> 00:18:52,400 So it can have a lot of different implications depending on the market. 238 00:18:52,400 --> 00:18:58,880 So specifically relating to mergers and acquisitions, sometimes some startups enter the market with 239 00:18:58,880 --> 00:19:03,440 the sole purpose of being acquired by a bigger company. 240 00:19:03,440 --> 00:19:08,960 So why is it that mergers and acquisitions is the subject of increased scrutiny when 241 00:19:08,960 --> 00:19:18,200 sometimes it is both sides who, it is the goal of both sides to either buy or be bought? 242 00:19:18,200 --> 00:19:25,120 When we think about mergers and acquisitions in the digital environment, it's really interesting. 243 00:19:25,120 --> 00:19:30,360 On one hand, mergers can be extremely efficient because what you do is you create scale, you 244 00:19:30,360 --> 00:19:35,200 create access to new markets, you create greater network effects. 245 00:19:35,200 --> 00:19:39,800 All these things are super efficient and we can all benefit from them. 246 00:19:39,800 --> 00:19:48,240 And indeed, a lot of the merger transactions that we have in digital markets are approved. 247 00:19:48,240 --> 00:19:53,920 A lot of them involve very large platforms buying smaller companies. 248 00:19:53,920 --> 00:20:02,160 What is the problem or why is it that merger analysis is the subject to increased scrutiny? 249 00:20:02,160 --> 00:20:09,920 Because one of the fears is that large platforms are able to identify the future disruptors 250 00:20:09,920 --> 00:20:14,040 that are likely to inject competition into the market. 251 00:20:14,040 --> 00:20:19,400 And what they do is they buy them in order to kill competition. 252 00:20:19,400 --> 00:20:22,520 It has a term, it's called killer acquisitions. 253 00:20:22,520 --> 00:20:25,480 Now we don't think that all the transactions are killer acquisitions. 254 00:20:25,480 --> 00:20:30,800 In fact, probably very few of them, but that's one of the problems that we have. 255 00:20:30,800 --> 00:20:37,400 Another problem is that when you enable them to engage in all these acquisitions, they 256 00:20:37,400 --> 00:20:41,360 can also control the innovation paths. 257 00:20:41,360 --> 00:20:51,220 So if I am a dominant platform and you started to develop technology that will disrupt my 258 00:20:51,220 --> 00:20:56,720 business model, it might be that the best thing for me, because I have very deep pockets, 259 00:20:56,720 --> 00:20:58,880 is just to buy you. 260 00:20:58,880 --> 00:21:04,600 And what I'll do is I'll align your innovation with my value chain. 261 00:21:04,600 --> 00:21:10,320 Basically, I will kill the product, the future product that you were developing, but I will 262 00:21:10,320 --> 00:21:16,400 still use whatever it is that you do to develop a product that suits my value chain. 263 00:21:16,400 --> 00:21:21,560 And because of that, there are calls for better scrutiny of these transactions. 264 00:21:21,560 --> 00:21:23,680 What is the problem? 265 00:21:23,680 --> 00:21:28,440 That it's very hard in those markets that are very dynamic at the point that the merger 266 00:21:28,440 --> 00:21:34,880 takes place to say, well, in the future, that tiny company will become a relevant, viable 267 00:21:34,880 --> 00:21:36,840 competitor on the market. 268 00:21:36,840 --> 00:21:44,100 In fact, it is so speculative that it's almost impossible for a competition agency to establish 269 00:21:44,100 --> 00:21:46,240 that this will be the case. 270 00:21:46,240 --> 00:21:50,920 So some calls are, for example, to switch the burden of proof, to require companies 271 00:21:50,920 --> 00:21:55,840 to show that the merger is efficient rather than the competition agency to show that the 272 00:21:55,840 --> 00:21:57,620 merger is problematic. 273 00:21:57,620 --> 00:22:00,800 That is one approach that has been proposed. 274 00:22:00,800 --> 00:22:07,820 But there is also a warning, because if you will have a very strict regime when it comes 275 00:22:07,820 --> 00:22:14,120 to those transactions, you might also affect innovation by ruining the exit strategy of 276 00:22:14,120 --> 00:22:18,800 many of those small companies, because many of those small companies actually operate 277 00:22:18,800 --> 00:22:25,720 on the market expecting to be purchased because this is their way to achieve growth. 278 00:22:25,720 --> 00:22:29,920 Seeing that we have all these different types of solutions that we could implement, looking 279 00:22:29,920 --> 00:22:35,880 at how they're implemented in practice, do different jurisdictions have different regulation 280 00:22:35,880 --> 00:22:38,080 schemes? 281 00:22:38,080 --> 00:22:44,000 Different countries sometimes have different ideas as to the role of the government and 282 00:22:44,000 --> 00:22:46,160 the role of the state. 283 00:22:46,160 --> 00:22:54,320 So some countries will have this sense that the state is responsible to designing the 284 00:22:54,320 --> 00:23:01,880 market and because of that, they may engage in more competition law enforcement and also 285 00:23:01,880 --> 00:23:05,240 use regulation much more. 286 00:23:05,240 --> 00:23:09,160 And this would be, the EU is a good example. 287 00:23:09,160 --> 00:23:17,640 We have a proactive regime and of course we have proposals, the Digital Market Act, the 288 00:23:17,640 --> 00:23:26,140 Digital Services Act, proposals for significant regulation that will affect the role of platforms. 289 00:23:26,140 --> 00:23:32,460 If you shift to the US, traditionally they had a somewhat different approach. 290 00:23:32,460 --> 00:23:42,360 They believe much more in the power of markets to self-correct and they also tend to believe 291 00:23:42,360 --> 00:23:45,720 that the government should have a very limited role. 292 00:23:45,720 --> 00:23:50,200 Now there is an ideology behind it because of course governments, when they engage in 293 00:23:50,200 --> 00:23:53,120 regulation, they might get it wrong. 294 00:23:53,120 --> 00:23:56,020 We call that government error. 295 00:23:56,020 --> 00:24:01,360 Sometimes they might be subjected to biases, to capture, and all of that might result in 296 00:24:01,360 --> 00:24:05,800 sub-optimal markets or sub-optimal regulation. 297 00:24:05,800 --> 00:24:10,960 Sometimes also when you have dynamic markets that change very often, if you have a regulatory 298 00:24:10,960 --> 00:24:17,440 regime, something heavy, that moves very slowly, it might be that the market changed but the 299 00:24:17,440 --> 00:24:22,880 regulation is still the same regulation of five years ago. 300 00:24:22,880 --> 00:24:25,580 So it is the wrong tool for the task. 301 00:24:25,580 --> 00:24:32,600 So it is not very effective and it might even accelerate some disruptions or distortions 302 00:24:32,600 --> 00:24:34,960 on the market. 303 00:24:34,960 --> 00:24:40,000 And because of that, if we focus first on competition, antitrust enforcement, you tend 304 00:24:40,000 --> 00:24:48,400 to see in the US a much more limited enforcement of antitrust and in the EU greater enforcement 305 00:24:48,400 --> 00:24:53,400 and this is because of different ideas about the ability of the market to self-correct 306 00:24:53,400 --> 00:24:58,160 and the costs of over-enforcement. 307 00:24:58,160 --> 00:25:04,960 And on top of that, when it comes to regulation, in the EU regulation is seen as the way forward. 308 00:25:04,960 --> 00:25:11,460 In the US that has yet to be endorsed as the solution. 309 00:25:11,460 --> 00:25:18,560 So we had at the end of last year, we had the Congress report, very interesting report 310 00:25:18,560 --> 00:25:26,280 from the House Judiciary Committee on antitrust and when they looked at Google, Apple, Facebook 311 00:25:26,280 --> 00:25:34,960 and Amazon, they tried to identify the problems, the difficulties and also tried to give a 312 00:25:34,960 --> 00:25:37,120 sense of what is the solution. 313 00:25:37,120 --> 00:25:45,800 And their solution is not introduce more regulation but apply greater antitrust, break up some 314 00:25:45,800 --> 00:25:49,560 of the giant companies, have better merger control. 315 00:25:49,560 --> 00:25:56,120 So it might be, it's a different way of delivering the medicine, let's say, because we all differ 316 00:25:56,120 --> 00:26:01,640 in our understanding of how markets operate, in our philosophy, do we trust markets, do 317 00:26:01,640 --> 00:26:03,320 we trust companies? 318 00:26:03,320 --> 00:26:09,040 It's quite interesting, different people from different jurisdictions will have very different 319 00:26:09,040 --> 00:26:13,280 ideologies even if not consciously. 320 00:26:13,280 --> 00:26:21,880 I find it quite interesting that in Europe people tend to distrust companies and they 321 00:26:21,880 --> 00:26:25,260 feel comfortable with their governments. 322 00:26:25,260 --> 00:26:32,840 In the US, generally speaking, people tend to distrust the government and will trust 323 00:26:32,840 --> 00:26:33,840 companies. 324 00:26:33,840 --> 00:26:36,440 So this is just something, it's not directly linked to our discussion but of course it 325 00:26:36,440 --> 00:26:41,160 will affect the toolbox that you will create. 326 00:26:41,160 --> 00:26:48,040 So there's no easy answer to those questions but this is also what makes it a fascinating 327 00:26:48,040 --> 00:26:50,240 area. 328 00:26:50,240 --> 00:26:56,920 And we focused a lot on privacy just to highlight what about algorithms, what about auditing 329 00:26:56,920 --> 00:27:03,120 algorithms, what about deciding whether an algorithm can harvest your information, whether 330 00:27:03,120 --> 00:27:06,560 an algorithm can analyze your information. 331 00:27:06,560 --> 00:27:10,700 Is it okay that someone is micro-targeting me? 332 00:27:10,700 --> 00:27:18,240 Is it okay that there is an algorithm that compares prices and is able then to assess 333 00:27:18,240 --> 00:27:20,760 the possibility of increasing the price? 334 00:27:20,760 --> 00:27:22,920 So we have a lot of automation. 335 00:27:22,920 --> 00:27:29,040 There is a wonderful experiment by an Italian group that created artificial intelligence 336 00:27:29,040 --> 00:27:30,040 algorithm. 337 00:27:30,040 --> 00:27:36,600 So they're using Q-learning which is a form of artificial intelligence, relatively simple. 338 00:27:36,600 --> 00:27:43,640 And those algorithms are able to increase the price in the market without infringing 339 00:27:43,640 --> 00:27:45,720 competition. 340 00:27:45,720 --> 00:27:47,440 That's quite remarkable. 341 00:27:47,440 --> 00:27:52,440 So they engage in something we call tacit collusion and they are able to align their 342 00:27:52,440 --> 00:27:56,620 policies without entering into an illegal agreement. 343 00:27:56,620 --> 00:28:02,080 So there is so much that is changing and if there is one thing that I think we should 344 00:28:02,080 --> 00:28:11,200 just appreciate is that this is not a simple evolution of the way markets operated. 345 00:28:11,200 --> 00:28:19,160 These are completely new tools that require us to think outside the box in many ways. 346 00:28:19,160 --> 00:28:26,040 Just to appreciate that the dynamics and forces that shape modern markets are very different 347 00:28:26,040 --> 00:28:31,160 and the hope is that enforcement regulation will not lag behind. 348 00:28:31,160 --> 00:28:38,240 They will be able to remain effective because what you hope is that markets eventually will 349 00:28:38,240 --> 00:28:43,560 continue to work for consumers rather than the other way around. 350 00:28:43,560 --> 00:28:49,120 Consumers working for the market which you can argue is sometimes the case when you think 351 00:28:49,120 --> 00:28:52,240 about some of the leading platforms. 352 00:28:52,240 --> 00:28:55,120 Is there a better regulation scheme to implement? 353 00:28:55,120 --> 00:28:59,460 Is one jurisdiction doing a better job? 354 00:28:59,460 --> 00:29:04,520 So if you think of regulation or competition enforcement as a medicine, think about you 355 00:29:04,520 --> 00:29:06,280 going to the doctor. 356 00:29:06,280 --> 00:29:11,200 The doctor prescribes you with a medicine and you know that you have to take exactly 357 00:29:11,200 --> 00:29:13,920 what was prescribed. 358 00:29:13,920 --> 00:29:18,540 If you undertake it you will continue to be ill. 359 00:29:18,540 --> 00:29:24,520 If you overtake it you might become even worse or you might even die. 360 00:29:24,520 --> 00:29:31,280 So this is why you really rely on the doctor to give you the right amount of medicine. 361 00:29:31,280 --> 00:29:35,840 And we're lucky that we have science so it's very clear in most cases what exactly is the 362 00:29:35,840 --> 00:29:36,840 right treatment. 363 00:29:36,840 --> 00:29:42,240 Now if you think of competition and regulation as a medicine the problem is you don't have 364 00:29:42,240 --> 00:29:43,720 a doctor. 365 00:29:43,720 --> 00:29:48,880 You don't have science, accurate science that tells you how much of the medicine you need. 366 00:29:48,880 --> 00:29:57,080 So the debate that you see is basically the debate on how much of it should we prescribe. 367 00:29:57,080 --> 00:30:02,520 In the EU if we speak about competition enforcement the attitude has been for many years that 368 00:30:02,520 --> 00:30:09,880 we need more of the medication because the US is under-medicating. 369 00:30:09,880 --> 00:30:12,080 And US criticized the EU. 370 00:30:12,080 --> 00:30:18,360 You had a lot of statements from the US in the Google cases, in other cases saying the 371 00:30:18,360 --> 00:30:21,840 EU is over-intervening. 372 00:30:21,840 --> 00:30:26,860 And if you over-intervene you will distort the market and you might chill competition. 373 00:30:26,860 --> 00:30:32,280 You might create a disincentive and companies will not compete as fiercely as they would 374 00:30:32,280 --> 00:30:34,480 otherwise. 375 00:30:34,480 --> 00:30:40,860 The interesting thing is that in the past two years the US has gone through a major 376 00:30:40,860 --> 00:30:47,480 change in attitude starting also already with the Trump administration because of several 377 00:30:47,480 --> 00:30:54,680 issues that he had with the leading platforms, him believing that they are restricting his 378 00:30:54,680 --> 00:31:03,080 access when it comes to customer base and scale when it comes to Twitter, Facebook, 379 00:31:03,080 --> 00:31:04,080 Google. 380 00:31:04,080 --> 00:31:10,480 So what happened in the US that they completely shifted from under-enforcement into something 381 00:31:10,480 --> 00:31:18,560 that much more resembles now the attitude in the European Union and indeed if you look 382 00:31:18,560 --> 00:31:26,520 at the very recent Google and Facebook claims that were launched in the US they are even 383 00:31:26,520 --> 00:31:31,440 more expansive than what was done here in the EU. 384 00:31:31,440 --> 00:31:38,280 So it tells you something about this debate on how much of the medicine we should use 385 00:31:38,280 --> 00:31:44,080 because you have a real-time experiment where you have the doctor from the US who changed 386 00:31:44,080 --> 00:31:49,720 his or her opinion on how much medicine is supposed to be used and now is prescribing 387 00:31:49,720 --> 00:31:52,260 much more medicine. 388 00:31:52,260 --> 00:31:57,640 The logic behind it is that if you understand that markets are not easily self-correcting 389 00:31:57,640 --> 00:32:04,200 then under-dosing on competition law will result in under-enforcement. 390 00:32:04,200 --> 00:32:09,040 And this is why there is a preference to overdose to try and correct it. 391 00:32:09,040 --> 00:32:13,880 So this is why you have differences and it's very hard to say who's right or who's wrong 392 00:32:13,880 --> 00:32:17,120 because it's not accurate science. 393 00:32:17,120 --> 00:32:19,600 And the same would apply to regulation. 394 00:32:19,600 --> 00:32:26,360 The big proposals that we have at the moment in Europe, the DMA and DSA, are still in the 395 00:32:26,360 --> 00:32:28,040 pipeline. 396 00:32:28,040 --> 00:32:34,160 We have more than a year before we will start to see how the final product might look like. 397 00:32:34,160 --> 00:32:36,200 So there are costs. 398 00:32:36,200 --> 00:32:44,640 If you limit the regulation only to dominant companies, costs on other companies are lower 399 00:32:44,640 --> 00:32:50,320 or non-existent, which means that you minimize the distortion that the regulation creates. 400 00:32:50,320 --> 00:32:57,000 If you over-require those companies to give access to other companies, let's say for data, 401 00:32:57,000 --> 00:33:02,420 when it comes to data, then maybe you will create free riding. 402 00:33:02,420 --> 00:33:09,000 If there is a dominant company that is obliged to give me access to the data that it collected, 403 00:33:09,000 --> 00:33:14,980 as a third party I will become lazy because I'll say I can use the law and force you 404 00:33:14,980 --> 00:33:17,120 to give me the data you have. 405 00:33:17,120 --> 00:33:23,840 So almost any action that you take in this game can have a benefit and can have a cost. 406 00:33:23,840 --> 00:33:28,720 And the great challenge when you engage in public policy, either in competition law enforcement 407 00:33:28,720 --> 00:33:32,800 or regulation, is to engage in fine-tuning. 408 00:33:32,800 --> 00:33:40,000 And every measure that you put in place, you have to think what is the benefit, what is 409 00:33:40,000 --> 00:33:41,900 the potential cost. 410 00:33:41,900 --> 00:33:50,640 Even the GDPR, which governs data protection in Europe and we all benefit from, is a great 411 00:33:50,640 --> 00:33:53,400 regulation but it also comes with a cost. 412 00:33:53,400 --> 00:34:00,480 For example, it created an opportunity for those who control the data to limit access 413 00:34:00,480 --> 00:34:05,560 to the data when it comes to third parties. 414 00:34:05,560 --> 00:34:13,080 So every action that we take as the state, as policymakers, as enforcers, has some sort 415 00:34:13,080 --> 00:34:16,040 of implications. 416 00:34:16,040 --> 00:34:21,800 And the aim is to try and predict those implications, try to think how this regulation might be 417 00:34:21,800 --> 00:34:29,280 misused and what is the best way to communicate or to reach the desired outcome. 418 00:34:29,280 --> 00:34:34,840 So again, like if we go back to the idea of the medicine, how much do you prescribe the 419 00:34:34,840 --> 00:34:37,560 medicine is very challenging. 420 00:34:37,560 --> 00:34:42,560 I think generally speaking there is a consensus that at the moment we under enforce when it 421 00:34:42,560 --> 00:34:44,640 comes to merger control. 422 00:34:44,640 --> 00:34:48,240 But again, the details are what is important here. 423 00:34:48,240 --> 00:34:55,520 There have been so many reports now that try to look at it in detail and really identify 424 00:34:55,520 --> 00:35:00,400 better or more accurate benchmarks for intervention. 425 00:35:00,400 --> 00:35:06,000 Since these technology platforms have such a global reach, do differences in regulations 426 00:35:06,000 --> 00:35:10,120 across jurisdictions pose any difficulties? 427 00:35:10,120 --> 00:35:14,420 You're a company and there is one state that tells you do this and the other state tells 428 00:35:14,420 --> 00:35:16,060 you do that. 429 00:35:16,060 --> 00:35:18,880 And if those two instructions conflict, it's a real problem. 430 00:35:18,880 --> 00:35:21,160 There is a real cost there for you. 431 00:35:21,160 --> 00:35:26,760 So we try to minimize system friction and this is why there are constant calls for cooperation 432 00:35:26,760 --> 00:35:28,280 and collaboration. 433 00:35:28,280 --> 00:35:32,480 And it's true for taxation, it's true for everything, of course also for competition, 434 00:35:32,480 --> 00:35:38,120 law, data protection, but there are limits to cooperation because of these ideological 435 00:35:38,120 --> 00:35:39,120 differences. 436 00:35:39,120 --> 00:35:45,560 Because if I believe in free market and minimal state, minimal government, and you believe 437 00:35:45,560 --> 00:35:51,120 in large government and more intervention, then of course we will find it more difficult 438 00:35:51,120 --> 00:35:54,300 to agree on certain things. 439 00:35:54,300 --> 00:36:04,480 But beyond that, when it comes to us as consumers, what it could suggest is that, let's say, 440 00:36:04,480 --> 00:36:13,240 customers in the EU will be subjected to different regime or strategies than customers in the 441 00:36:13,240 --> 00:36:17,920 US or Canada or South Africa or New Zealand. 442 00:36:17,920 --> 00:36:19,600 So I'll give you an example. 443 00:36:19,600 --> 00:36:30,120 In Germany, Facebook users get at the moment greater privacy than Facebook users in the 444 00:36:30,120 --> 00:36:32,640 US. 445 00:36:32,640 --> 00:36:38,680 Specifically looking at the UK and in light of its recent exit from the European Union, 446 00:36:38,680 --> 00:36:43,960 do you think that there will be any significant changes in regulation of big tech companies 447 00:36:43,960 --> 00:36:45,880 in the UK? 448 00:36:45,880 --> 00:36:54,500 So in broad lines, the UK has legislation that is very similar to the EU. 449 00:36:54,500 --> 00:37:02,160 But of course, Brexit means that the UK can now move beyond and develop its own approach 450 00:37:02,160 --> 00:37:04,020 that might be very different. 451 00:37:04,020 --> 00:37:13,200 So if you think on the topics that we discussed, so in the UK we have a digital market unit 452 00:37:13,200 --> 00:37:20,480 that the idea is that it will govern this whole area. 453 00:37:20,480 --> 00:37:30,160 And then there is this idea of creating a pro-competitive ex-ante framework, so a pre-emptive 454 00:37:30,160 --> 00:37:34,600 framework that will be quite different from the digital market act or the digital service 455 00:37:34,600 --> 00:37:35,600 act. 456 00:37:35,600 --> 00:37:37,880 It will deal with strategic market status. 457 00:37:37,880 --> 00:37:41,520 And so I guess what I'm saying is the mechanism starts to differ. 458 00:37:41,520 --> 00:37:46,440 You can already see that here in the UK we're taking a slightly different approach. 459 00:37:46,440 --> 00:37:53,320 The aim is very much the same, but again, now the UK is using some, in the EU in general, 460 00:37:53,320 --> 00:37:59,240 if you look at the proposed regulation, it looks like a shopping list of various activities 461 00:37:59,240 --> 00:38:04,160 that the company might engage in and saying if you do that, that will be illegal. 462 00:38:04,160 --> 00:38:10,920 The UK has taken an approach that tries to create a rule and basically says if you have 463 00:38:10,920 --> 00:38:19,280 strategic power, we will tailor for you a regime that will address the circumstances 464 00:38:19,280 --> 00:38:21,080 in which you operate. 465 00:38:21,080 --> 00:38:25,000 So something that looks a little bit more nuanced and it will continue to evolve in 466 00:38:25,000 --> 00:38:33,680 a way where we all try, I believe, to achieve the same goals, but we will start to see differences. 467 00:38:33,680 --> 00:38:36,040 We will see differences in merger control. 468 00:38:36,040 --> 00:38:42,120 We will certainly see differences in the regulatory regimes and we'll also see differences in 469 00:38:42,120 --> 00:38:48,080 the way competition law is applied here in the UK and in the European Union. 470 00:38:48,080 --> 00:38:52,960 Are we likely to see more regulation aimed at data sharing? 471 00:38:52,960 --> 00:39:00,880 So data access, data portability, data mobility, these are big things and here you have an 472 00:39:00,880 --> 00:39:06,880 intersection between intellectual property rights, between privacy, between competition 473 00:39:06,880 --> 00:39:13,320 law and you really see the complexity in that junction when all of them meet. 474 00:39:13,320 --> 00:39:20,320 There is certainly, if your focus is on markets exhibiting barriers to entry, there is certainly 475 00:39:20,320 --> 00:39:29,000 an attempt to increase access to data and I think it is likely that we will see if we're 476 00:39:29,000 --> 00:39:34,440 basing it on the reports that we had in the UK, also in Europe by the way, if we base 477 00:39:34,440 --> 00:39:41,520 it on general statements from stakeholders and policymakers, it is likely that we will 478 00:39:41,520 --> 00:39:44,600 see something along those lines. 479 00:39:44,600 --> 00:39:49,600 But again, the details are the important thing here because it's not all data. 480 00:39:49,600 --> 00:39:52,400 Because of course, A, we have the issue of privacy. 481 00:39:52,400 --> 00:39:55,400 So what kind of data? 482 00:39:55,400 --> 00:39:56,860 Is it aggregated data? 483 00:39:56,860 --> 00:39:58,000 Is it personal data? 484 00:39:58,000 --> 00:40:01,580 And all these things will require much more fine tuning. 485 00:40:01,580 --> 00:40:07,240 Another thing is data mobility to enable you to just say, I'm just taking all my data with 486 00:40:07,240 --> 00:40:14,160 me and going to another provider and here as well we have interoperability, a question 487 00:40:14,160 --> 00:40:21,160 of can that data actually be used easily somewhere else? 488 00:40:21,160 --> 00:40:26,040 And if you take your data, do you take your data or do you take the data that was subjected 489 00:40:26,040 --> 00:40:27,040 to analytics? 490 00:40:27,040 --> 00:40:29,560 Because obviously the analytics were not yours. 491 00:40:29,560 --> 00:40:34,280 But certainly, generally speaking, certainly this is a direction that we are likely to 492 00:40:34,280 --> 00:40:41,400 see either through the attempt to increase data mobility, portability, or to increase 493 00:40:41,400 --> 00:40:42,400 data access. 494 00:40:42,400 --> 00:40:45,120 That's all the questions that we have for today. 495 00:40:45,120 --> 00:40:49,880 Thank you so much, Professor Ezrachi for joining us and discussing the role of competition 496 00:40:49,880 --> 00:40:53,200 law in regulating these big tech companies. 497 00:40:53,200 --> 00:40:55,360 My pleasure. 498 00:40:55,360 --> 00:41:00,760 That was Professor Ezrachi speaking with us on the regulation of big tech companies. 499 00:41:00,760 --> 00:41:05,920 For more interesting legal discussions and writings, visit the OUULJ's blog and read 500 00:41:05,920 --> 00:41:08,120 our annual publications. 501 00:41:08,120 --> 00:41:10,760 This is the last podcast featured for Trinity. 502 00:41:10,760 --> 00:41:13,200 Join in to make a mess for more exciting interviews.