Antoine Chapsal, "Should We Regulate the Data Economy?"

Presented at the Legal Challenges of the Data Economy conference, March 22, 2019.

Transcript

ANTOINE CHAPSAL: Well, so the question I ask is should we regulate the dead economy, OK? So it's actually a very broad question, and I will only focus on antitrust. So I'm afraid we won't have tons of issue to discuss with Etienne.

So let's start with having a kind of basic description of what data is from an economic perspective. And, actually, I have to say that that has a kind of weird and heterogeneous economic object. It's an input that is probably important but not unique in order to develop various kinds of applications, OK? Data is a non-rival, so which is also another interesting thing. We can have values that a scientist or algorithm, actually, that can use the same data at the same time.

And, finally, data's value is quite hard to estimate. You have economies of scale you have economies of scope. And the evolution of the value of the data across time is very difficult to assess, OK? And, actually, the relationship between the age of the data and what you can do with it is extremely hard to assess.

Another thing that we should mention is that, actually, the data economy is a kind of special world. And if we look at the players that operate within this world, you will see that, first, consumers are also producers, OK? So from an economics perspective, it's a bit weird, and it's interesting, and I think it's relates to what Randall explained in details earlier. So consumers use data. Consumers produce data. And what is interesting is that consumers provides firm with data, so they are both consumers and producers.

And if we look at all the players, like government, for instance, they are both players and referees. OK? So they use data, they generate data, but they can impose rules on the data world that affect, actually, data users. OK? And I saw that the steps. So the French telecom regulator is basically trying to grab some data in order to improve its regulation.

And if we look at the very last player in this economy, we will see that firms are also a bit special. And you know this, of course, quite well. And there are, most of the time, multi-sided platforms. OK? And so we have been talking about multi-sided platforms for, like, 20 years. When it came up, I was at the beginning of my career.

I was starting as a young economist. And I had the feeling that it was a kind of old wine in a new bottle. Like only packaging. I really believe that now it's a bit more, because we use it as a way to deal with anti-trust issues in a way that is a bit special and that makes probably a multi-sided platform more than an old concept. OK.

So well, here is our world. So our question is, well, it's Randall's question. Does this market clear? OK? And this is the only question we should ask. Is there something wrong here? And so are these weird specificities?

Are these weird features of both data and the data economy in general likely to harm consumers in the short run or in the long run? And more globally, could we believe and could we prove that these markets failed? And they can fail because of information of information asymmetry. This is what Randall explained very precisely. And this is a big piece of the puzzle. Market power.

This is really interesting because I think this is related to the very definition of regulation and the distinction between regulation and competition policy. Sometimes we say-- and it is true-- that regulation is ex ante and competition policy is exposed. OK.

This is probably true. But I'm afraid that when we are saying this, we are missing the main idea, which is that, well, you have to regulate the market if, basically, competition is not working well and yields inefficient resource allocations.

So basically, if you face perfect competition, well, you don't need to have competition policy or regulation. If you need natural monopoly, which is the definition of the market that works the worse, then you will have the easiest way of regulating the market, which is actual regulation.

So this can be considered as a kind of global framework in order to try to understand whether we should regulate this data economy or whether we should try to adapt our competition policy framework.

So let's start with information asymmetry. Well, we have two situations here. The first one is that, well, data, in some cases, actually reduces information asymmetry. And this is something that this is a fact, OK? And we can take a very basic example to understand the main consequences of that.

The taxi are regulated for a very basic reason. Taxis are regulated to protect consumer from holdup situation. And what is an holdup situation? It's the fact that the rider that hailed a taxi does not know the driver's duration of the ride, the price of the ride, and so on and so forth. And if you don't impose price caps, for instance, or other kinds of regulation, the driver could charge an unfair price at the end of the ride.

And this is a pure market failure. This is a pure information asymmetry. And this could lead to an under-provision of the services. If I know that I may face this holdup situation, I will probably decide not to grab a taxi. Yeah.

What Uber did? Well, this solved this issue. And basically, when you take Uber, you have all the information you need to make sure that the holdup situation won't occur. You have all the information about the price, you know the duration of the ride, and so on and so forth.

So at the end of the day, there is no reason why Uber should be regulated if you apply this framework. And we can go one step further and say, well, probably data may be used to deregulate markets. OK? So for instance, if taxis were able to provide me with all the information I would need to avoid this holdup problem, to reduce the information asymmetry, well, a price cap won't be needed.

But sometimes-- and this is the other case-- data reduce the information asymmetry. And this is what Randall explained. In my world, this is the way I can understand privacy. And why do we-- so we can go quite quick here.

We can, of course, explain that disclosure of personal data can be a source of efficiency, of course. But probably targeted advertising are great. I mean, I probably prefer to have a relevant ad than a random one. But at the end of the day, they are quite intrusive. And this is the same for personal prices.

So here we have a significant information asymmetry. And we have a significant information asymmetry between the individual who offers that data and the platform that requests it. And this is your silence thing. And this is exactly what you explained. And there is a huge discrepancy, actually, between the individual, short-term, tangible benefit from the data it provides the firm with and potential long-term harm.

And this is actually linked to another market failure, which is abundant rationality, which is that we are kind of smart people here, but we won't enter into a kind of discussion of what Facebook, or Google, or Uber can do with my data. I have no idea, and I don't have time to try to deal with this issue.

So probably in this kind of-- and I am perfectly convinced, because I think that here information asymmetry is a huge failure, that regulation is needed. Then you need to understand what kinds of regulation you need, OK? So basically, you have objectives. And the objective would be to prevent firm from using data in a way that harms consumers. Very clear. It could also facilitate entry in the data economy, like data portability.

But we need also to, in this kind of cost-benefit approach of regulation, basically, you also need to measure the risk of such a regulation. And some rules can make collection and use of data more difficult. So increase static entry barriers. So it's connected to the second aspect of my talk about market power and creating market power.

And actually, these rules may also affect the internet ecosystem, OK? It will decrease ad revenues and it may also decrease the quality of free services. So now you are playing in the world described by Randall, where, basically, you don't know what's happening if you change a bit one important parameter.

OK, so let's switch to market power. Here I just discovered something very, very strange to me. I tried to make a review of the existing literature on the link-- the economic literature on the link between data and market power.

And it's clear that this is an empirical question. What I found quite surprising is that this is highly debated. So you have economists saying that, well, you can have tons of data, you won't have market power. And frankly, I was thinking otherwise. But this is something that is well explained in many economics papers.

So then what we should probably try to understand is, why, actually, Amazon and all these multi-sided platforms have market powers? And probably it's a mix of two things. It's a mix of existence of data, for sure.

And it's also the way data is generated. And I think that both things is something-- I mean, these both-- market feature data and the way data is generating, so indirect networked effect is clearly what creates market power. Probably not only data.

And actually, to go a bit quicker, if there was only a question of data, it would be quite simple, I have to say, because as I told you, data is non-rivaled, so we can say, OK, if we really consider that data is absolutely necessary to let a rival enter the market and, basically, data is giving you a structural monopoly position, then the solution would be quite simple. I'm afraid it's much trickier.

So actually, this raised two main questions. Data plus indirect network defect creating a natural monopoly. And is our competition policy framework adapted to multi-sided platforms? And I think this is linked to one of the last part of the Etienne's discussion.

So this is an interesting question. And the first thing that you need to understand is that our competition policy framework is more or less based on the idea that some practices are likely to be anti-competitive. I'm not saying that they are illegal per se, but we really believe that they are likely to be anti-competitive. This is the case, for instance, of exclusivity increases.

And the problem with multi-sided markets is that some practices that are often considered as anti-competitive in no more market in one-sided markets are pro-competitive in multi-sided markets. And I can give you an example of that about exclusivity and inclusivity. So I assume that I have two platforms and I have exclusivity encloses, so Brand 1 can be purchased only through Platform 2 and Brand C can be only purchased through Platform 1.

Well, you may have a strong competition between these two platforms, but then if there is no longer any exclusivity encloses and I can't find exactly the same kind of brands in both e-commerce platforms, then at the end of the day, what's going on is that everybody is going to switch to the bigger platforms. OK?

So basically, what I am telling you here, the intuition behind this, is that exclusivity encloses increased platform differentiation, and basically, differentiation is something in a two-sided market. This is the opposite in standard market. But in a two-sided market, it's something that actually increases the intensity of competition. OK?

And why? Simply because this is going to give you strong incentive to multi-home. And I think we could make one remark on a solution [INAUDIBLE]. Probably the fact that if you can have four or five real estate website, it could be the case that you really believe that you will find your flat only using one of them. If you are sure, for sure, that all flats, all houses are listed in the four website at the end of the day, you won't multi-home.

So this does not mean, actually, that such kind of practices are always pro-competitive. And I really believe that, again, we can understand how you can have a very serious anti-competitive infringement if you try to mix data and this kind of two-sided market effect, which is the indirect network effect.

And market power comes sometimes from the fact that the platforms enters one side of the market. And this is the case for Google. And this is, in one sentence, the Google Shopping case. And this is the same for Amazon. Amazon has its own private ladles.

And if we are in this situation here, well, you are in a situation where excluding arrival is extremely easy. And for instance, a simple way of doing that is just to favor your own services, just to favor your affiliate in one side of the market. And this is the Google Shopping story.

We can think about other kind of example of that. For instance, indirect network effect and data, both of them will help you to-- and this is probably the case for Amazon-- to produce more attractive goods and services. And so you will be able to actually-- whoops. You will be able to increase the attractiveness of your products. OK.

So in just four words of conclusion, so the way we can, from an economic perspective, think about whether we should regulate the data economy, is try to understand to what extent these markets clear. We have a very serious information asymmetry, and this is what Randall explained. But we also need to understand that sometimes data reduces information asymmetry.

One thing I'm convinced of is basically the idea that data in itself is really important. But the way data is generated is-- and actually, both data and data generation are two things that, combined together, yield the most important antitrust challenges that we'll have to face. Thank you.

[APPLAUSE]

Thank you.