Randal C. Picker, "The Crossroads of Regulation and Antitrust"
Presented at the Legal Challenges of the Data Economy conference, March 22, 2019.
RANDAL C. PICKER: Thank you so much for having me here. I was happy to see but Dominique made sweaters safe on the stage, so I felt that I could make an adjustment. What a dazzling setting in this beautiful city, so thanks for having us here. It's so exciting to be here.
So when Omri asked me to talk at this conference, he said, give me a title that'll be interesting, maybe even sexy. And that's what I came up with. So I suspect that says more about me than you really need to know. What it really was was a place holder for me. I didn't know what I wanted to talk about, and I was waiting for something that seemed really interesting. And then in February, the German Cartel Office did exactly that. So that's what I want to talk about.
I've never really thought about Facebook very carefully, and this seemed like a chance to do that. The German Cartel Office decision seemed interesting, so that's where we're going to get to. But it's going to take me a little bit of work to get there.
So I want to start by talking about the economics of targeted advertising. I think that's an interesting, important topic. I think what we know about economics online and offline are pretty different, and I want to try to highlight some of those differences-- why that matters for how we think about Facebook. So this is the most basic construct we have in economics, I think-- supply and demand, price quantities. Life occurs at the intersection of those two lines.
And when I took price theory from Gary Becker in econ grad school, who would go on to win a Nobel Prize, Gary was pretty clear on two things. People optimize subject to constraints and markets cleared. And if I thought I needed more than that to solve a problem, I wasn't working hard enough.
So that's the construct we have. But part of that is the assumption that we know what the prices are and we know what the quantities are. And I think that works in the offline space, the physical space that we've spent so much time in. I think it works pretty well there. I'm a little more skeptical at how well it works online.
So if you go into a Starbucks in my neighborhood in Hyde Park, this is what you will see. And the board there will have something like this. The day I walked in, you could order a cinnamon shortbread frappuccino. You probably did that this morning in one of your Parisian cafes, I assume. I have no idea what that drink actually is.
But you have a sense of it. You see the picture of it. You see the prices, right? Price, quantity-- we know what that is. I was just in a store a few minutes ago, and I knew exactly what I was getting and how much I had to pay to get it.
And the companies-- we haven't really-- and [INAUDIBLE] said we haven't talked about the gaffe very much yet. No, that's right. We haven't. And maybe that's what we ought to do.
So apple has a different approach to these markets, right? Apple wants your cash. Apple wants all of your cash, right? Very much a premium model-- that's always been the model from Apple. When they started with the Macintosh, and obviously even before that, the Apple II, though the Apple II wasn't quite as pricey. Certainly the Macintosh was pricey, and certainly the entire iOS line has been pricey. It's not based upon data. It's based upon cash.
It is easy to forget that apple almost died in 1996. I'm in the middle of a book on sort of the history of law and computers. And Apple faced low ball tender offer in 1996 when Apple was dying. And you can see the resurrection of Apple, first with the iPod, and then eventually the iPhone. But Apple is a cash-based business.
But if we turn to business is based on data and media, then I think we're working in different markets. How much data gets taken from you is not particularly salient. So if you are paying in data, how do you know how much you are paying, right? Again, I just handed 2 euros to somebody for this Diet Coke. I knew exactly how much I was paying, right? How much did I pay this morning-- Twitter-- in data when I was posting a bunch of tweets about the conference? I just don't know.
And so the structure we need to understand is how do we think competition works when prices are not visible, what I call it the bottom here priceless competition. How does that competition work? And what's the nature of that competition? I think that competition is very different than the competition we're used to seeing in physical space.
So and as Omri mentioned this morning, the nature of that is then is that firms basically grab as much data as they can. They do not feel constrained by a consumer response because it's not salient. It's not visible to you. And so the flashlight app that Omri mentioned that was grabbing all this information-- I think you should have assumed that that's exactly what they were doing, right?
They were giving you something for free. There are no free lunches. That's another Chicago Economics idea-- Milton Friedman, I guess. So you should assume they were getting something. In a world in which you can't see what they're getting, you should assume they're getting all of it now.
How do you deal with that? And the answer is through regulation, I think. So if the prices aren't salient, it's hard to see how consumers are going to respond. And so our Federal Trade Commission, the United States Federal Trade Commission-- it's deceptive practices powers-- has been acting fairly aggressively to try to control some of these data flows in a context where they can say, here are reasons why the market's not going to work very well. And so we've seen a series of actions like that.
When you stare at this 150 times a day-- is that what I was told? I am not a millennial, so I don't do that. I'm probably no more than 50 or 60 times a day. Isn't that a frightening figure? But when you stare at that screen, I think you want to recognize that the competition to be on that screen is really a different kind of competition.
And I want to think because you're not competing over price, as it were, and you can't see the data that's being taken, I think it's a kind of pure quality competition. Your experience of the app, how you see this app as providing you services, is the only thing that matters. The price as such doesn't matter because there's no price, right? That's what's sort of interesting about this market.
And most, again, physical markets are situations where people make price-quality trade-offs, right? You walk into a McDonald's. The food you get there is not the same as you get in a medium French restaurant. It's not the same you get at a high-end French restaurant. There are price-quality trade-offs there. That's how the world works in the physical space. I don't think it works that well in the online space.
And I think that almost hypothetical exercise you want to undertake is to say, if I could only have a handful of apps on my phone, if what really constrains my use of the apps is how much attention I can bring to bear on them, not price, but how much attention I can bring to bear, which apps would I have, and why? If I can only have one app, what would I have?
And we sort of know the answers to that, right? Those are the two apps you might have, right? You would have Facebook and Google. How do we know those are the answers? Well, we mentioned at dinner last night this UK study came out last week on the digital economy. They had a very helpful chart in that on time spent on different websites and apps by UK consumers. And it's really interesting to look at that and see the dominance of Google and Facebook there.
These other competitors, the gaffe or gaffum, as I think of them-- throw Microsoft into there-- very different models from an intentional standpoint. You can see where Microsoft is, where Amazon is, where Apple is. Apple's not based on upon attention. They're based upon cash. Delighted to see how well the BBC was doing.
Now, some of this is going to change, right? Apple, I think, is reaching the limits of how much they can extract on us for devices. If they move into services, that's going to become data. Amazon is a really tricky situation, as I think Laurent was suggesting. And there's an EU investigation there. But our big data companies so far, in a big way, at least, have probably been Google and Facebook-- both media-based companies and both data-based companies.
All right, I went onto Facebook. You know what Facebook looks like. I was careful about privacy. I didn't want the GDPR to descend on me as I was speaking. Would that happen? I don't know. So I fuzzed the page up, as you can see. But you know what this page looks like. They've got ads here. This particular ad is about some sort of grammar program. I couldn't tell if that was commentary on my grammar or they realized, as a lawyer, I care about grammar. I hope the latter-- the marketplace.
But here is really the genius of Facebook right there. Right, so we talk about the information we put into Facebook. Facebook's ability to generate tools that produce information-- that's really the genius of what they've doing. So like button, which sounds like this really simple thing, in some sense, I think that's the work of genius because of the kind of information that it creates. And Facebook, of course, is very much an information-based business.
All right, but we are used to markets in which we get free things. And, again, Laurent said, is anything new here? I think that's a terrifically important question. And I want to say I think the data piece of it is maybe new, though, again, there are some past analogs that we can talk about and I'll do a little love. But certainly the fact that these are media companies, that's not new.
So in the United States, at least, over-the-air radio and over-the-air television was all finance-based through ads, right? You didn't pay with dollars to listen to radio in the United States. There was no licensing fee, no taxation. As it were, you paid, as it were, by consuming ads.
And if you asked, well, if you were a media firm with market power against consumers, how did you use that market power? The answer is you made them listen to more ads. That's the way you raised the price for the product. You effectively degrade the quality of the product by adding more ads and charging more. So the fact that it's free as it were doesn't mean that market power can't be exercised.
And we should always ask the question what would have happened in competition, right? And that should be our baseline. That's not to say but they're necessarily violating the law if there are more ads than there would be in competition, but at least it gives us a sense of whether we think we're seeing the exercise of market power. I think that's the right conceptual exercise. And, again, that's not new. We've been working with these markets for a very long time.
What I think is probably new is the data and the role of data. And we heard some dark stories about data. My colleague Omri Ben-Shahar talking about data pollution and negative externalities, and we talk about positive externalities, too, so we have a longer conversation about that.
But I want to say, at least with regard to targeted ads, the thing to focus on, I think, as a first cut is the idea that advertisers clearly want to see the use of data, right? The famous line in the United States from John Wanamaker, an early retailer-- I know 50% of my advertising dollars are wasted. I just don't know which 50%. Right?
That's always been the way these markets have worked. And so the promise of data is to get these ads to customers who care about them. We do not have a dog in my house Do not show me dog food ads. That's just a waste.
And at the same time, for consumers, if you think of experiencing ads as a kind of disutility, ads that are more targeted may be more valuable. So I used to get many years ago, like, three computer magazines at my house during the height of the PC computer era. I loved looking at those ads. Those were as valuable to me as the stories because those were super targeted ads.
Now, can we cross a line here in the idea of creepiness? Absolutely, though retailers don't want to cross that line. That's not the best way to sell your product. So I think data can be incredibly valuable here, both for the advertisers and the consumers in this two-sided market. And that, I think, is new compared to the old media that we've had before. All right, that was, if all goes well, the economics of targeted ads.
I want to talk just briefly about legal data restrictions, though I'm going to fly through that-- GDPR. OK, we now know what that is. No, you know what the GDPR is, so I don't want to linger there. It is clearly, as was suggested, I think, on the last panel, the most comprehensive effort at trying to regulate data protection and obviously is enormously important.
But I do want to say but there have been other efforts to deal with data in particular situations. Again, I know the United States best. So Laurent mentioned the telecommunications industry as a comparison here. Yeah, I think that's right. I think that's a good comparison.
I teach a course in this-- the next quarter-- in eight days, I'm teaching a course at the law school called platforms and network industries. It's basically about the regulation and natural monopoly. We'll start with the railroad and the telegraph. Then we'll end with like Amazon and Facebook and the like. And so I think those are all within the same zone, and the zone we want to think about.
And when we passed the 1996 Telecommunications Act in the United States-- and we had not passed major legislation in that area in 60 years. We said, oh, but telecommunications companies, they see a lot of data that you might care about-- who you're calling and when you're calling them. We're going to control their use of that data.
And so those ideas have been a long time. The question is when will we bring those to bear on these new kinds of industries? We haven't done that in the United States yet. The FTC has done it in sort of an episodic way. So it has a settlement with Facebook that's based upon deceptive acts or practices. The truth of the matter is, given that you have these privacy policies and you're required to have those by California, the chances that you're going to screw that up seems to be enormous.
So the FTC could have one of these with everybody. They have one with Google they have one with Facebook. They have one with Myspace. The Myspace one will run 20 years. Who thinks Myspace is going to be here in 20 years? Not possible.
But we've regulated this way, and we are starting to see a little bit more efforts at the state level. So California has passed, in June of 2018, a new interesting privacy regulation. So that's happening. And, again, we should think of that, I think, as a sensible response to the problems associated with data salience and the way in which that interferes with the normal market mechanisms that we expect to work in the offline world.
All right, Facebook-- how did we get to Facebook? I want to start here because I think it's important to appreciate this point. Facebook starts in 2003, not in 2007. In 2007, Facebook said, we are going to build a platform. We are going to create an API and give people access to that. And if you are a competition person, you go, that's great. Right? Don't close this, right?
Building interconnections is an important part of the path we've seen in network industries. The electricity business used to be local natural monopolies. Eventually, we decided that power generation could be subject to competition, but no one wants to build a power plant if you can't access the grid. So we build a connectivity regime for that. The 1996 Telecommunications Act did exactly the same kind of thing.
So creating these kinds of connections is something that we see in these public utility stiles industries and query. And, again, that was suggested in the morning. Query whether we need to think about the Facebooks and the Googles of the world in those terms. But Facebook took a step down that path in 2007 opening the network up, and that got us Cambridge Analytica. The path from there to there is almost immediate, right?
The reality is that engineering these kinds of networks, both from a physical Institute from a physical in the case of the electricity grid engineering and the case of data engineering, I think, is hard. I think the institutional engineering is hard. And so I think we should be aware of the possible problems that will arise when we go down that path.
Now, I am a Cambridge Analytica victim. Facebook told me that one day. I've never known what that really meant happened to me. I'm not aware of any meaningful harm that I've suffered from that. Maybe I'm not paying enough attention.
But in any event, in the past, we have regulated this kind of interconnection and these network industries. And the question, I think, for us is whether we're going to do that for data. Indeed, that's what the UK report called for last week, a kind of data public utility, as someone said on the last panel. I think doing that is actually going to be tricky and complicated. All right, I'm watching the clock here.
Facebook-- they do the Instagram deal. No one complains. They do the WhatsApp deal. No one complains. The EU thinks about it a little bit, but says, you know, they're really in different businesses-- not a problem. And, oh, by the way, there's no data concentration here. There's data everywhere. Facebook doesn't have a particularly unique set of data. I don't know how to think about that, but the EU thought about it hard, and that's what they the conclusion they came to.
So Facebook builds in this way. I think there's some-- and, critically, for this point, they say privacy is completely different from that. That's not a job for the competition authority to sort. OK, I think there's some regret about that. I'm going to skip the fine.
And I think the regret maybe about these deals is to recognize that the hard thing in these businesses is not particular features or services. The hard thing is getting hundreds of millions of users. And maybe that's what we should have been focusing on in these mergers, but we didn't seem to.
All right, to Germany-- So the German Cartel Office, they say, we've looked at this. We're going to do something about it. That Facebook is getting information off of third-party websites and that Facebook wants to combine the information between Facebook, Instagram, and WhatsApp. We're not going to let them do that, at least without consent.
So this is what the German Cartel Office says. They're going to create a kind of data silo, again, absent consent that requires these chunks of data within Facebook to stay separate. Germany calls this an internal divestiture of the Facebook data. And that is what the order has done, though. It's on appeal in the German court system.
Now, what's the reasoning and how should we think about that? Well, I want to say we've put on the table two types of remedies with regard to Facebook. One of them I'll call a division remedy, where you separate the data. That's what the German Cartel Office wants to do. The other one I'll call a multiplication remedy. That's, I think, what the UK expert report wants to do-- make copies of the data. Multiply the data. Give it to other people. Hard to engineer, but that's the conception.
I want to say, the EU has gone down this path before. So when the regulation of Microsoft through antitrust, we had a subtraction remedy, where Microsoft had to issue versions of Windows with and without the Media Player. We had an addition remedy, where the browser ballot or the browser choice screen was presented, where the first time you would turn on a Windows 7 computer in the EU, you would be presented with-- I think it was 14 different browsers. Require them to must carry-- distribute those in that way. And most recently, we've done the Google Shopping exercise.
And I'll say, speaking just for me, but I don't see any of those as being successful. The Windows Media Player case-- Microsoft eventually put out figures saying we've distributed 35 million copies of XP with the Media Player, 1,787 without the Media Player. There seemed to be no market demand for that. The browser ballot was broken 17 months. No one really noticed. We can talk about Google Shopping. The clock ticks. I won't.
But I want to say, I don't think this is a record of success. And I think engineering this for live data, which is continuously being generated, is not an easy problem. OK, the German reasoning, though, was sort of completely different. The German reasoning was really focusing on the idea of an exploitative abuse. That's what Facebook was doing here. And I think it's a distinctive German situation, though I'm careful about how much I know about this.
And what I mean by that is that if you compare the idea of exploitation that you see in this particular situation with the more standard EU version that you see in one 102 A there, which talks about excess prices and the like. I don't think that's this situation. So the understanding of exploitation, again, uses some sort of competitive benchmark on the cost side, some sort of approach with regard to the economic value. I think neither of those applies very well to Facebook.
So when you think about is Facebook taking more data than we would see in a competitive framework? The reality is because of the salience problem on data, the flashlight app is what the competitive framework looks like. We see companies which have zero market power grabbing huge amounts of data, right? That's sort of shocking, but explicable in a world in which data salience doesn't matter. So do I think Facebook is getting a disproportionate amount of data? I don't think so, though I don't know.
And then if you turn to the pricing side of this, there's some recent experimental literature which says, oh, I'll take your Facebook away for a month. Well, let's do some tests. How much is it worth to you? $100 a month. $100 a month. So are people getting this free product that they would value at $1,200 according to the experiments-- is there a value gap there? That doesn't seem right to me.
Instead, what the German ruling focuses on are the constitutional rights of German citizens. I can't speak to those, right? I can't tell you how to make those trade-offs. But I do think as soon as we're saying that, we're no longer properly in the land of competition policy. And it's fine to embrace other values, right? We do not have to have competition policy at top of the queue. But let's just be honest about what we're doing if that's the path we're going to go down. Do we think there's some failure in the data protection regime that somehow the competition needs to jump in?
Facebook is reorganizing. I don't know what that means. I don't think anyone else does. We heard a little bit about China this morning. Is Facebook going to move down that path and a monetization path? Maybe that's what they're going to do. I don't think it's a response to the German Cartel Office yet.
Finally, what do we take away from this? I do think the salience problems on data are substantial, and I think it means that the economics of online markets and offline markets are meaningfully different. We may change data merger policy going forward, but I think there's going to be a call to move to this world of maybe public utility-style regulation. That'll be a super interesting conversation, but a difficult one, and one that I think we should be nervous about in the sense of I think it's really hard to do. Thank you very much.
SPEAKER: Thank you very much. Thank you very much. Before-- [INAUDIBLE] just one short remark. You reminded us about Myspace. And when we think backwards in 2004, 2005, it strikes me that, at the time, Facebook presented itself as a privacy-centered alternative to Myspace in a way and won the market in a way insisting on its privacy settings. I find it odd that today they're on a privacy breach and they need what you called a privacy reset.
RANDAL C. PICKER: Well, and I'll say, if you haven't read the original-- switching a little bit, but completely the same point-- if you haven't read the original Brin and Page paper. But they wrote for the Australia conference in 1998, where they described this new thing called Google. They've got this appendix, where they talk about advertising, and they talk about the inherent conflict between advertising and having an honest search engine. And their conclusion is it's important to have a search engine that remains in the academic realm. Somehow they got around that.
SPEAKER: Thank you.
RANDAL C. PICKER: Yeah.Big data