This is a lightly edited version of a speech I gave at the Chillin’ Competition conference late last year where I gave the opening speech for the panel, “An Emerging Competition Law for a New Economy?”
I begin with the usual disclaimer that I’m speaking in a personal capacity and my views are not necessarily those of the European Commission.
In the early years of the “new economy”, lawyers had a strange tendency to use horse metaphors to describe what they thought was going on in the law.
Judge Frank Easterbrook – an excellent jurist and one never short of a strong opinion – speaking at an internet law conference with Lawrence Lessig lambasted the very idea of the conference. There was no “law of the internet” any more than there was a “law of the horse”. There were laws of contract, of animal husbandry, of gambling – there were laws that applied to horses, but there was no law of the horse. And there was no law of the internet. He had a good point.
There is no “new economy” or “internet” law just as there is no law of the horse. There are telecoms laws, intellectual property laws, data protection laws and competition laws. But no new economy law.
Maybe it’s a little bit more complicated than that…
In all of these areas there have been developments over the last twenty years that are directly caused by the new economy and the internet. Data protection laws have been stretched to deal with the problems of vast amounts of information being moved around the world at the speed of light and in the EU are now being fundamentally reviewed; telecoms laws have to cope with issues such as net neutrality and zero rating; intellectual property laws have had to find answers as to whether linking to copyrighted content or copyright-infringing content is in itself a copyright infringement; and competition laws have had to grapple with a range of problems which – though not unique to the new economy – are more prevalent, more important or simply more complex when applied to the new economy.
To assume that we need only look at competition law, data protection law, telecoms regulation, or intellectual property law as they have historically been cast, and further to assume that they need no reconsideration, no nuancing, no elaboration in the new economy is overly simplistic.
Judge Kollar-Kottelly, the judge charged with finalising the remedies after the Court of Appeal in the US Department of Justice’s action against Microsoft, also invoked a horse metaphor, describing the task of crafting an appropriate future-proof remedy as being like trying to “shoe a galloping horse”.
And yes, technology markets appear fast moving so crafting a remedy of more than merely temporary usefulness appears difficult. But although particular products may change from year to year – think of a new version of Windows, or a new model of iPhone – the underlying technology changes rather more slowly. If there is a competition problem deserving of a remedy, then it will almost certainly be linked to the underlying technology rather than a particular product release. So – for example – the Commission’s Microsoft interoperability remedies continued to apply beyond the version of Windows that was on the market at the time the remedies came into force. And the Federal Trade Commission’s consent decree in relation to Intel applied beyond the products that were on the market at the time.
Both Easterbrook and Kollar-Kottelly were superficially right, but were more fundamentally wrong.
The superficially right answer to the question posed by the title of this panel is that no there need be no new competition law, the existing principles suffice. How can the principles not suffice when they are couched so broadly?
That’s the simple answer, the “law of the horse” answer. Superficially right but more fundamentally wrong.
The principles can only suffice if they are properly understood. And properly applied to what may be novel factual circumstances.
New economy cases may have particular characteristics, alone but more often in combination. There are many-sided markets, markets with network effects, markets with high fixed and low marginal costs, markets where average costs keep decreasing, markets with zero or low cost interoperability both between complements and between competitors. None of these things taken individually are unique to the new economy. You’ll be able to find markets with each of these characteristics in the old economy – though the implications of these characteristics haven’t necessarily been tested in the old economy, so a new economy case may be the first time that particular problems have to be addressed. Perhaps more importantly though, technology markets often put several of these together. The combination quickly becomes complex.
Before we get to that I want to say a few words about the issue of timing of intervention.
On timing, it’s a more complicated version of the Goldilocks problem. If the porridge is too hot, then it may be too early to intervene. If the porridge is too cold, it may be too late.
But when it’s neither too hot nor too cold, neither too early nor too late, there’s another argument that gets used to oppose antitrust intervention: even when the porridge is just right, you still should not eat the porridge because something even better than porridge will come along soon. The transience of dominance. Market power is temporary and will be undermined in a wave of Schumpeterian destruction.
Let’s look at each of these concerns.
The too early argument I find hard to understand. When has the Commission ever intervened in a technology case at such an early stage that it was difficult to see how the market was going to develop? Or indeed where the market changed radically shortly after the decision such that the underlying competition concern disappeared?
Too late is perhaps more of a concern. As Philip Lowe – the former Director General of DG Competition – used to say, we should not wait until the patient is dead before we begin to operate. The extreme situation of too late to remedy the problem is rare, however. I suppose in those circumstances – if remedial action can’t restore competition to a market – you may instead want to look at behavioural remedies that limit exploitation of that lack of competition but I have not been able to think of a case where that would have been relevant.
The transience argument (you should not intervene even when the porridge is just right) also does not stand up under scrutiny. As I said at the start although products appear to be galloping by, technologies do not. Market power does not.
It is fashionable to think of Microsoft as yesterday’s news in terms of technology, and it is true that Microsoft seems to be more of a complainant than a defendant in technology cases these days. But is that because its market power proved transient?
Many people now use Macs, particularly laptops. So has Microsoft’s market power eroded? Well they still have a market share of over 85% on desktops and laptops and I suspect their share in the business market as distinct from the consumer one is even higher. That is still a pretty good business position.
Admittedly, portable computing devices like smartphones and tablets might change the size of the market – either by reducing it in absolute terms or by limiting its growth compared to what would have happened absent such devices. But that looks more like what happened to mainframes when desktop computers came along – the relative size of the market changed, but for those customers that needed mainframes, IBM remained a rather important company – hence the IBM interoperability case of just a few years ago. (And mainframes are still not dead – IBM announced a new mainframe product a few weeks ago.)
Another example. Should the Commission have intervened in the Intel case, given the variety of new processors now available, in particular for mobile devices? Intel is not doing so well there. Should the Commission not have been concerned at its practices for desktop and laptop CPUs?
To bring in my own horse metaphor, that’s like saying no-one should have been concerned about a price fixing agreement in the guild of blacksmiths, because at some point Dunlop would invent the car tyre. And again who is to say that for customers that continue to need blacksmiths services – people still use horses today – that a blacksmith’s cartel would not still be of concern? The market would be smaller, but the potential for exploitation as against any individual customer may still be as great.
So overall I do not see evidence of intervention that is too early; or evidence that new economy dominance is transient counselling an excess of caution even when market power exists.
So if the principles remain the same and timing is not too much of a concern, what is difficult about new economy cases?
Because you need new knowledge, and you need to challenge old assumptions.
What do I mean by new knowledge?
I work in cartels at the moment. One of the reasons cartels are outside of the sectoral organisation of most of the rest of DG Competition is that sectoral knowledge is often not that important. A price fixing agreement for a widget is much the same as a price fixing arrangement for a sprogget. (One exception may well be financial services cases where sectoral knowledge is probably more useful.)
For many new economy cases because the markets and products may be new and unfamiliar, you need a strong sectoral understanding. You need to know, for example, whether a request for interoperability information means making public APIs that were previously private, or whether it means writing APIs that don’t yet exist. (And you need to know what APIs are.)
Many years ago I remember trying to interest an MEP in a telecoms policy issue that would affect how consumers used the internet. It became apparent after a while that the MEP had no idea that consumer computers connected to the internet via telephone or cable networks. I’m not entirely sure what magical mechanism he thought allowed him to send an email, but he clearly didn’t have the requisite level of knowledge to have an opinion on telecoms policy. Things have improved since then.
The flipside of needing knowledge is perhaps understanding that not everything that is new is novel. One of the topics of great concern to some people at the moment is the competition law analysis of “big data”. I saw an article a little while ago that pointed out that data could be used as an input in a downstream product. So data about consumer preferences could be combined with data about a product, to create targeted advertising in relation to that product. So data can be an input. Explaining that takes about a paragraph and understanding it takes, I hope, about thirty seconds.
The rest of the multi-page article was devoted to how, assuming data could be an input, data issues could come up in a variety of competition law situations. But if you removed the word “data” and replaced it with “any input” you would have a fairly basic competition law introduction to vertical competition issues.
So you also need enough knowledge to know that even if something is new, it is not necessarily novel.
Aside from new knowledge, you also need to challenge old assumptions, because the characteristics of the new economy markets might mean that the assumptions no longer apply. This is a particular problem of competition law, because it suffers from a loss of nuance over time. Complex economic cases become reduced to black letter rules, rules which may be completely wrong applied to different complex cases. A sensible ruling loses its context and becomes a slogan.
Just as Cartes Bancaires recently told us that context is everything in judging whether a restriction is by object or by effect, so context should be everything in deciding whether and how to apply a legal ruling.
Taking the Magill criteria and applying them to every situation where A is asked to supply B with “something” makes little sense; it’s divorced from the nuances of why the Magill case was decided as it was. Or as Eleanor Fox has pointed out, it takes the fact pattern of the Magill case and calls it a legal rule.
One of my favourite misguided assumptions when it comes to the new economy is in relation to the application of competition law to interoperability problems. Justice Scalia in the US Supreme Court Trinko case set out three potential problems of mandated dealing.
- First it may reduce investment and innovation incentives for the company obliged to supply;
Second, it may require courts and competition authorities to engage in central planning and ongoing price regulation;
Third, it may facilitate the “ultimate evil” of antitrust – collusion.
He assumed, as do many, that these are characteristic of any mandated dealing case.
- But what if the “facility” to which access would be mandated is a by-product? The TV listing information in Magill is an obvious example? By definition there can’t be a reduced incentive to invest in a by-product.
- What if there is no price that needs to be regulated?
- What if there is no ongoing relationship between supplier and supplied so that collusion is not a concern? (And of course, they may anyway be in a vertical relationship so collusion seems … not obvious.)
In many interoperability cases, these three concerns may be wholly irrelevant. What then should be the legal test for determining whether a request for access to interoperability information is the right one? Maybe the concept of an essential facility – crafted to determine when it’s right to give access to rivalrous fixed infrastructure like a port – is not necessarily the right one when looking at non-rivalrous interoperability information.
There are plenty of other potentially problematic assumptions.
For example, it is trivially easy on the internet to find examples of pricing below average variable cost – there are plenty of free products out there – which if you simplistically apply Akzo looks predatory if you can find dominance. But if this is a multi-sided market you obviously have to look at all possible sides of the market and look at all revenues before you can draw any conclusions.
There is a flip side to this as well of course – I’ve seen arguments that because a product is given away for free there cannot be consumer harm (who can be harmed by a free product?). That’s equal nonsense for the same reason – you have to look at all sides of the market to see what is happening.
The risk of faulty assumptions do not stop there. Even taking into account the multi-sided nature of markets, you cannot necessarily assume that the predatory pricing test of Akzo is the right one. In high-fixed cost and low-variable cost industries then relying on a variable cost test rule may be as Herbert Hovencamp has argued, a license to predate.
And I have already criticised the assumption that because products change regularly, market power may be transient – flawed because it is often the technology that determines the market power and that changes much more slowly.
So, in summary, we do not need new rules or new principles; we do not need to challenge the fundamentals of competition enforcement. But we do need to understand the markets, understand the legal principles, and understand how to apply the latter to the former. So pretty much standard competition law practice.
Kevin, this is a tour de force and I agree with most of it. I wonder, though, about (1) the comments on essential facilities not being the right concept for non-rivalrous interoperability information, and (2) the argument that there can’t be a reduced incentive to invest in a by-product.
Why is the essential facilities concept not appropriate in tech markets? The motivations behind the essential facilities concept are not wholly different to those that Scalia identified (a desire to avoid free-riding except in cases where the establishment of competing infrastructure is near impossible). Innovative tech products that benefit of significant investment also merit protection from free-riding. Tech markets are often characterised by innovations being copied by rivals, which in turn creates incentives to further innovate. That seems like a positive cycle and one we should not interrupt. So the essential facilities test seems to pitch the threshold about right.
Also, why can investment incentives not be chilled by the threat of a requirement to share a by-product? Investors often make decisions based on the totality of a return being made available to them, including by-products. That potential upside may even be included in the investment analysis as a “known unknown.” That positive externality might also be something competition authorities should be slow to chill. Your thoughts?
Thanks for your comments.
On (1) my concern about essential facilities is that it’s used reflexively, without considering whether the – justifiable – reasons for it being an onerous test apply in all cases. I agree that free riding on significant investment is a concern, but I’m not sure that the fact patterns of all cases suggest that there has been significant investment. I’d be much happier focussing on the question of harm to consumer welfare, and asking a range of questions about that potential harm – ie not just sticking to the three questions of the Magill criteria. (Has there been significant investment? would be a good question.)
On (2) you’re quite right and I’ve made the mistake that I think others make with the essential facilities point – assuming something is generally applicable when it’s not. There are many circumstances where return on by-products affects investment decisions and I should have recognised that in the text. But there will also be circumstances where they won’t – if the by-product was an unknown unknown, if the value of the by-product is so small that it’s a rounding error compared to the value of the product itself, if there’s a legal obligation to produce the product itself and so on.
The general point is that the question “can I use part of your port facility?” may be a very different question to “can I interoperate with your software?” in terms of investment incentives, degree of intrusion, risk of collusion and so on. But that requires a much more detailed treatment than I’ve done here.
Thanks for the response Kevin. And as I mentioned, I thought this was an excellent analysis.
Kevin, I drop a line just to concur. This is a very good piece, and since I could not attend the Chillin’ conference, I am glad that I could read it here.