Edited Transcript of a CNE Market Insights Event
19 September 2002
"Tying and Bundling: From Economics to Competition Policy"

Professor Xavier Vives Professor of Economics and Finance, INSEAD
Professor Paul Seabright Professor of Economics, Université de Toulouse I
Moderator: David Lawsky of Reuters

My name is Tim Evans and I'm the new President and Director General of the Centre for the New Europe. I'm delighted you could all join us for this event today. As I'm sure most of you know, the CNE is the leading market-oriented think tank in Brussels. I'm delighted that you can make this event today, which is part of our Markets Insights programme. We hold a range of these events looking at various areas of competition policy every year. Without further ado, can I introduce our moderator for today, David Lawsky from Reuters. David.

David Lawsky: I'd like to introduce today's speakers. Xavier Vives, as a Professor of economics and finance at INSEAD de Paris, researches and teaches economic policy. For a decade he was the director for the institute of economic analysis in Barcelona, he got his Ph.D. in economics from the University of California at Berkley, and has taught at Harvard and other U.S. institutions. He edits a number of journals including the Journal of the European Economic Association and is the author of "Oligopoly Pricing", published in 1999 by MIT.

Paul Seabright on my left, at least physically, received his Doctorate in economics from Oxford, taught at Cambridge and is now at the University of Toulouse. He is the author and co-author of a number of books, including "Merger in Daylight", which is the first book linking the study of the European Union and merger regulation. He's written also about competition policy with John Fingleton, known to a number of you in this room, the head of the Irish Competition Agency. Xavier…

Professor Vives: Thank you very much. I'm grateful for this opportunity to share some thoughts with you on tying and bundling from economics to competition policy. Tying and competition policy, as you know, and that's probably the reason you are here, is a contentious area-an area in which there is a lot of debate, in which there has been some transatlantic tensions. It's a complex issue, and so I will try to concentrate my remarks on the economics of tying, the simple economics of tying, and also on potential competition policy implications that may be of use.

As you know tying is the situation in which the supplier sells a product, the tying good, to a customer. The customer agrees to purchase all its requirements for another product, the tied good, from the supplier. Okay, so the tying good may be a stapler, a stapling machine, and the tied good may be staples. Maybe the tying good is something more sophisticated, a computer, or some software programme, and the tied good may be the (punch card), the IBM (punch card). There is a famous case-this looks like ancient history; I was going to say from the last century. In fact, it is from the last century.

What's the benefit of tying? Why do firms tie? That's the first thing we have to ask: Why would firms want to tie and bundle products? Well, first of all, because typically it is more efficient. Indeed, we do not want to buy a car, the separate parts of the car, and then we bundle it together at home. This would be extremely costly. So it saves labour, there are economies of scale, so the first reason is efficiency. The second reason is more complicated. It depends on completeness of information and market imperfections. For example, in terms of quality, how to ascertain the quality of a product? By bundling and tying and setting the brand, this can be overcome. So these are good reasons if you want efficiency reasons. There may also be other reasons, which we may call strategic reasons, and then here firms may use tying in two different ways. One ias as an accommodating strategy, when they want to compete with rivals in the market. And another one, and this is stronger in the sense it is an exclusionary strategy-and here comes the strong word-foreclosure. That we'll see later.

When may this be a worry? Well, this is worrisome when there is market power at some level. Otherwise, we better forget about it. So there has to be market power somewhere-capacity to raise prices substantially above costs. What is an accommodating strategy? Well, an accommodating strategy would be tying or bundling to try and soften competition, so competition in the market is not so head-to-head, to price discriminate maybe. And here, in fact, I will not deal with this very much. Typically the consequences are ambiguous in terms of welfare for the competitors and consumers and so, overall, it depends really very much on the circumstances, and I would like to concentrate more on a more contentious issue many anti-trust cases have tried to attack, which is the possibility of foreclosure.

What is foreclosure? Well, let me just say one definition: actions that decrease the profitability of rivals in the short-run, with the hope that you will expel them from the market; so that you force them to exit in the long-run. Why may firms want to foreclose? Why? Well, first firms need to have the ability to do so, to be able to do so and, second, they need to have the incentive to do it. Let me just present a very conceptual frame of analysis, very simple. You have this firm, and two market segments A and B. This firm has a strong position in market A, that is, a substantial market power in market A, but faces competition in B, and A and B are complementary markets. So that's a software platform and other types of software for example. Now, in this context lets try to think about different economics points of view, and let me start with a classical doctrine by the Chicago school.

What does the Chicago school doctrine say? Well, it said foreclosure is not a rational strategy for a firm. This is why: There is no incentive. Why? Because there is only one monopoly profit to get and the firm can get it by pricing appropriately in the market in which it has a monopoly. This is particularly true for example thinking about it. So, if you think about complementary products, if you produce in A, if you produce nuts, well definitely you would like bolts to be very cheap. This way you will sell a lot of nuts. This doctrine applies to certain circumstances, and was a little bit a part of classical thinking. But more modern thinking based on strategic analysis and game theory, for example, has concluded that there are some circumstances in which it does pay for firms to foreclose. If it does pay, they have the incentive to foreclose. And let me just concentrate very briefly on two or three such instances.

One instance is what's called the leverage theory: a firm may try to leverage its market power from A (that's where the firm is strong) to B (where it's not strong), so, where it faces competition. So this is possible and, in fact, the basic idea is that by tying the goods A and B, the firm commits to being very aggressive, a very tough competitor, and then this may make rivals exit. But this is true only if, first, this firm really commits to bundle by integrating the product for example. Also, it must be successful because if the rival does not exit then, in fact, it is the worst possible result, an extremely competitive outcome: very, very harsh competition with the rival. And also the goods cannot be very complementary because otherwise we go back to the Chicago view.

So that's one possibility. Another possibility is that a firm may want to foreclose to reduce the empty threat in its base market (in A), so A is the base market where the firm is stronger. So suppose that then the firm ties product A and B. Then why may it deter entry in A, in the base market? Well, by forcing the exit of the competitive supplier in B, an entrant may not enter because it would need to enter also in B to compete effectively with the firm. That's to say, it raises the barrier to entry in the sense of saying, well if you enter, you have to enter in both A and B. And this may be very difficult. So that is another possible strategy. But, again, the circumstances under which this strategy makes sense are quite restrictive. The entry in A may not be sufficiently uncertain; it may be very easy to innovate in B. The profitability of entry in A, if the product A has a life of its own, may not be that much affected by B. In a sense, barriers to entry may be relatively low. So you need high barriers of entry for this to make sense. Final example, and this is very important because it's very much at the heart of the debates of whether tying and bundling are good in our information economy, is tying and innovation. So how does it affect innovation incentives? Because these are the most important dynamic incentives that we have. In fact, at the end of the day, this is what matters.

So here economic theory is relatively clear in that tying and bundling strategies tend to decrease the innovation by rivals of the tying firm but tend to increase innovation of the tying firm. So that's relatively well established, and there are two mechanisms by which this happens. One mechanism is the following: tying makes successful entry prospects in complimentary component markets, A and B, more uncertain. So basically it's the same argument as before. The entrant has to succeed in both markets. So if you bundle, then the entrant has to succeed in A and B, and to innovate in segments A and B at the same time may be much more difficult. So that's one reason why the entrant may have less incentive to bundle. First reason. Second reason: the incumbent, when innovating in market B-let's say we're thinking of software platform and applications-internalises the profit generated for segment A, the platform where the incumbent, but not the entrant, has market power. So he may have more incentives to innovate.

So what's the summary of this from the welfare point of view? Well the summary is that the welfare analysis is ambiguous: we do not know. So that's the truth. We do not know. Because if we think that what matters is the aggregate incentive, the aggregate incentive to maximise social welfare through competitive pricing-let's think this is what matters-then it depends on whether the reduction in the rivals are indeed more than compensated for by the increase in the incumbent's contribution to overall welfare through lower combined prices on the bundled goods. So that's the bottom line. Which means, basically, you need to know a very, you need to have a very detailed knowledge of the industry and of the parameters of the industry to see in what case you are.

So let me summarise the welfare analysis of tying, which is at the base of competition policy, of any competition policy recommendation. The welfare analysis has judged the consequences for rivals, the old firm, and consumers. In the short run, tying and bundling tends to decrease prices; it has a competitive effect. This is basically because it internalises complimentarily some of the products of the firm and then it creates incentives to lower prices to get market share, so there is a short run competitive increase. Number one.

Number two: it may decrease variety because, and this may be bad because the first was good obviously to decrease prices, it may decrease variety because obviously it may make mix and match. What is called mix and match? To buy one product from one firm, another product from another firm, is much more difficult and it may induce price privilege. And again this is ambiguous. This can be positive or negative. So these are the short-run.

What about the long run, the dynamics? Well, first, there is a very important dynamic efficiency that is sometimes forgotten; that's important to realise the efficiency of product integrations for consumers. These may be tremendous so let's think of a car again. We do not want an unbundled car. I don't even want unbundled furniture; some people do, but I do not. Number one. Number two, and this is a potential negative: exclusion of rivals if a foreclosure strategy is successful. Number three: impact of innovation. We know, we said it's ambiguous.

Okay, so in general, is there any rule of thumb? Well the only rule of thumb that one can think of is that efficiencies may be presumed if there is no exclusion of rivals, and that's a very rough, extremely rough, rule of thumb, but it's a rule of thumb. What's the challenge for competition policy then, to conclude, given that the welfare analysis of bundling is ambiguous? So tying can be anti-competitive but the conditions under which it is are very restrictive and hard to check and depend on parameter conditions, i.e., the degree of market power in segment A. Competitors must be unable to replicate the strategy or have an incentive and this, for example with bundling, is clear. It may depend on whether demand is elastic or not. So, for example, a best response to a bundling strategy by an incumbent may be to bundle or not, depending on the characteristics of demand. So therefore you must have very good demand estimates to ascertain whether this may be anti-competitive or not. And finally you have to draw a fine balance between the short-term benefit of the tying and bundling strategy and the potential discounted long-term cost if there is foreclosure, in the case there is foreclosure. This means that the burden of proof on the competition agency is high and it needs a very detailed empirical analysis. And to do this with short deadlines and few resources is very difficult.

So what do we do with this? Let me just end with two thoughts or issues, related in fact to the transatlantic, if you want, tension, which, for example, the GE-Honeywell case has provoked. Which I think is a case in which it has not been really the debate. Sometimes people have said it was over whether in one part of the Atlantic or the other, one protects competitors and not consumers. This is not the debate. I think, the debate is about different things. Thinking about the strategic behaviour, and in terms of merger policy now, is whether merger policy should deal with conditions that, although increase competition in the short run, make exclusionary and predatory behaviour, like foreclosure for example, more likely. So thinking in parallel to mergers, sometimes when things are mergers that improve or give facilitating conditions for collusion, that something should be done about it.

Is there a parallel to this foreclosure or exclusionary or predatory behaviour? And this is very difficult. The balance, because of what I said before, because the welfare analysis is ambiguous and because there is the danger of overshooting if this is taken into account, or undershooting if this is not taken into account. And this depends on the type of errors you value more-type 1 or type 2 errors? In any case, if a competition authority wants to follow this route it should, I think, I'm thinking more concretely for the European Commission, for example, to think along two lines. I think, first, this line of reasoning requires very important resources devoted to ground research and in fact to be prepared (exactly) with industry knowledge in-house, not only probably, not only externally, because the deadlines are very short and one needs very, very detailed knowledge. Otherwise it's very difficult to do a proper analysis. That's number one.

And number two in the context of the review of the merger procedure and thinking specifically about this issue, I also think it would be the case-and we can discuss this later if you want-to apply somewhat more flexible criteria in terms of substantial lessening of competition instead of the dominance criteria in Europe. I think this would probably fit better with economic analysis, in particular in this area. Thank you.

David Lawsky: Thanks. Now Professor Paul Seabright.

Professor Seabright: Thank you, I would like to begin by addressing a remark to the person who wrote on their reply to the invitation for today: "Thank you very much for this opportunity to listen to Professor Xavier Vives, whom I have long wanted to hear. Can you explain to me why it is necessary to listen to Professor Seabright as well? Could it be arranged that me might have one without the other?" I'd like to say to this person, whose name I shall not reveal, but you know who you are, I feel your pain, but I'm sorry it's a package deal. (Laughter)

Tying is everywhere. My day began, as they say, in a hotel in Paris which had bundled the breakfast with the hotel room. Unfortunately, they also bundled with a very pleasant croissant one of the most undrinkable cups of coffee that I have tasted in a long while, thereby incidentally denying access to my custom for the café across the street. I then took the Thalys from Paris to Brussels, which bundled the journey with newspapers, thereby denying access to my custom to the newspaper vendor at the Garde du Norde. I arrived in Brussels and took a cab, in which the ride was bundled with sundry observations on life and politics on the part of the cab driver, thereby denying access by my brain to various other thoughts I might have chosen to occupy myself with. And I arrived here where, of course as you see, the dessert is bundled with the main course and with the starter.

Bundling is absolutely everywhere. It is one of these phenomena that is so ubiquitous that we have ceased to notice it and I think that has important applications, implications rather, for the way we think about its role in competition policy. Now it's interesting to observe, over time, the way in which bundling practices change. For example, some of you will remember the time, in Europe at least, buyers of electrical equipment always had to buy a plug which they then put onto the electrical equipment. Those days, I'm glad to say, have gone. Mostly you can more or less guarantee that when you buy your wife goods from an electrical appliance store, that the plug will come bundled with it. It's a bit tough for those people who used to sell separate plugs, but I have to say that, you know, in a world where there are still one billion people going hungry, I don't waste a lot of worry over them. On the other hand, unbundling has become easier in some respects too. We know that, for example, you can go to the website, and I've just bought a Dell Computer, and you can actually specify a lot of the components. You can say "no, I don't want that I want this please", and by and large, Dell has made a marketing tool out of responding to your needs.

We know that the criteria that drive this are complex. They are driven partly by changing preferences for how willing people are to put things together. You know, there used to be, do you remember, radio enthusiasts? There were a lot of people around, like when I was at school, people who liked to put their own radios together. There are not too many of them now so it's not as easy to buy electronic equipment, electronic components unbundled, as it use to be. Similarly, I talked about plugs and appliances. You know, not so many people like doing that stuff anymore and so practices have changed. Transactions costs have gone down with the Internet but we know that it's the job of manufacturers and service providers to bundle. They do it all the time: a chef does it, and one of the reasons why they do it is precisely to make use of the information and skill and the judgement that they have about the things that go together. Any chef will tell you that good cooking… half the secret to good cooking is good shopping. Because, what does a chef do? A chef thinks carefully about who the suppliers are, how good their quality is. And when a chef's supplier's practices change, there are a lot of tears. You know, the person who says, well I used to supply La Gavroche with my very best organic vegetables and they don't want my stuff anymore. Well, you know, things change and it's part of, in a sense, what a provider of a service or goods offers you is their skill and judgement of putting the components together. Sometimes it's the physical effort, the transformation; sometimes it's the selection.

Now why am I telling you all this? You may say its very obvious. Well, the reason I am telling you is that I think tying and bundling, as I said before, have this character of being so ubiquitous that we forget they are there. So when we suddenly notice them for some reason, you know, competitors made a complaint for example, we suddenly think, gosh what does this mean. And I think it's important to emphasise that when we look at the anti-competitive impact of tying and bundling practices, we must avoid thinking that their mere presence is some kind of smoking gun. I agree with absolutely every word that Xavier said before. There's nothing that Xavier has told you that is not absolutely the conventional, and I'm proud to subscribe to the conventional analysis on this topic.

The more difficult problem is to know how we turn this kind of analysis, which says tying and bundling is really difficult, how do we turn it into a practical policy? And I want to say that the first thing we need to bear in mind is that its very ubiquity means that we have not got to treat it like a sudden surprise intruder to the banquet when we look at a case. When we see competition involving issues of tying and bundling, we have to treat it as something which is normally very common and which may, in certain circumstances which Xavier has described extremely well and carefully, which may indeed have anti-competitive effects but which should not be presumed to be dangerous by the fact of their mere presence.

Now you may say, aren't I flogging a dead horse, don't we all know that I'm not so sure. I've been asked to be controversial so I will talk about some real cases and I think I should preface what I'm about to say by the fact that nothing that I say about real cases is going to imply that I take any view in one way or the other about what way the case either should have gone or should go in any subsequent hearings, but I merely want to describe some of the reasoning as laid out in parts of a decision. I'm struck by that fact that in GE-Honeywell, in a discussion of bundling practices, a certain amount is made at one point of the fact that bundling has been a common practice in the industries in which the parties are involved. I'd like to know of an industry, I'd be interested if somebody in the questioning afterwards could suggest some industries, in which bundling of one kind or another isn't a common practice. I really would be surprised to find any. And, consequently, the discovery that it's a common practice in industry X doesn't tell us anything about either the intentions of the parties or the likely consequences of their undertaking of bundling.

Now the second point I think that we need to bear in mind in thinking about how to use competition analysis carefully to find the bad cases, the cases where bundling leads to anti-competitive practices, is that we need to think about why it happens in order to understand its possible efficiency rationale. As I said to you, any chef will tell you that half the secret of good cooking is good shopping. What that means is that why bundling happened in a particular industry will be the result of any number of decisions including ones that exploited information about production processes, about preferences of consumers which were known to only some of the parties involved. So for example when Dell tells you that they can offer you these choices for your computer but on the other hand can't offer you others. Nowadays you can chose to have an Ethernet card or not in your computer but it comes bundled with an ordinary V56 modem and if you say to them, why do we have to have the V56 modem when you give us a choice on the Ethernet card, the answer is, well its easier to do it that way-believe us we run the logistics-we know, okay.

Now, so the difficultly which I would add to what Xavier said at the beginning, is that not just is it that the welfare analysis is complex and ambiguous, it's that the very reason for doing the bundling in the first place involves knowledge, detailed technical knowledge, which was only accessible to the parties. So if you can somehow get at the knowledge, or if you can be absolutely sure that the normal efficiency rationales for bundling are not present, then by all means you may want to treat this as a serious problem, but you have to correspondingly raise the barrier for your evidence to make sure that you can be confident in your findings.

Now, the other thing that we need to bear in mind is the importance, the absolute importance, I think, of restricting attention to cases where there is really substantial market power in the tying market. Xavier has mentioned this and I would underline this conclusion. In other words, given that bundling occurs almost everywhere, if you try to make it a potential offence in too many markets then the ones, the cases that you actually end up cracking down on, will be determined by the accident of who has the inclination or more likely the financial power and the legal resources to mount a challenge. You want to avoid that arbitrary application of justice by restricting attention to cases where there's really significant and evident market power in the tying market. Once again I want to make an observation about GE-Honeywell, where as many of you will know the possibility that GCAS might tie engines with aeroplanes, or indeed does tie engines with aeroplanes, is given a certain amount of weight. Now without expressing a view as to whether that argument can be made good in the long run, I do want to say that it is only a good argument if you can establish absolutely that GCAS has market power. Now there are remarks in the decision and so on to suggest that the Commission believed that it did, but that should not be a side part of the decision. That should be a central part of the decision. If you want to make tying an offence, the market power, the existence of market power in the tying market, has to be a very clear and evident fact.

So how can we think a little bit about how we should proceed? I think the first point I want to suggest is that if you restrict attention to cases where very substantial, significant, market power exists, then you need to bear in mind that cases in which you can credibly establish an offence of anti-competitive tying are going to be relatively rare. That's not to say that they won't happen; they will happen but we should not expect a large number of them. Secondly, we need to be clear in analysing them, what the particular rationale has been. As Xavier has pointed out, there is a whole set of potentially anti-competitive uses of tying which rely upon tying of essentially unrelated products, which are not complimentary, in order to leverage market power from one market where it exists to another market where it doesn't. And all of those cases can be anti-competitive on the part of the tying firm only if they essentially succeed in driving a competitor out of the market, or so weakening it through network effects and so on that it ceased to be an effective competitor. Otherwise, the effect of the tying is just to make competition more intense, which competitors won't like, so they'll complain like hell, but which will be good for competition even if particular competitors are not pleased.

Now, there's a quite different set of cases where we talk about complimentary products, and as Xavier has emphasised and I want to give you as the take home message from the complementary products side as it were, is that, in general, it doesn't make sense for a firm to weaken the producers of complementary products. Why? Because complementary products increase the demand for your own. And so one of the things we need to bear in mind is that a firm which ties complements to its own product in which it has market power, makes it own product less attractive if it thereby weakens the ability of producers of complements to supply good and effective complementary goods. And that is one of the reasons why, as he suggested and I would emphasise, a firm can use tying to exclude rivals. That's to say it would be possible if it did tie in a way that brought complements together and denied rivals access to its market, it's possible that it could weaken those competitors, but since they're producers of complements, this would be bad for it. In other words, it would have the ability without the incentive.

So I think the take-home message from here is when complements are involved, ensure that the firm, which is allegedly committing an offence in tying, not only has the ability to tie but also the incentive. And again, I have been through a number of decisions recently where a lot of effort was concentrated on showing that the firm allegedly committing a tying offence, or allegedly was about to commit a tying offence, had the ability to tie without establishing either, at all or to an equivalent degree of rigour, that it had the incentive to tie. And I think both of those have to be a central parts of any case.

I want to conclude by one more slightly controversial observation, which is that I think it is fair to say that the court in the Airtours decision observed and indeed criticised a feature of the Airtours decision, which in my view does exist to some extent in a number of other decisions. I would cite Tetra Laval/Sidel as one that had a tendency to run through a list of possible anti-competitive consequences of a particular transaction with an air of saying, well if you don't believe that X will happen you can look at Y, and if you don't believe Y will happen you can look at Z. Now, I think that to some extent that's an inevitable consequence of, first, the fact that decisions are made under tight deadlines and, second, that you work with dominance rather than the substantial lessening of competition as Xavier has already pointed out. I think that a possible move to a substantial lessening of competition might make it less of a temptation to do that, to put in a large number of additional possible consequences to try and make the dominance case. But I do think that with transparency increasing, with the scrutiny of commission decisions increasing, it's going to be important to ensure that allegations of tying or possible tying are either made very rigorously-by strong emphasis on establishing market power in the tying market plus a careful analysis of the different ways tying works in substitute and complement cases-or else it should be dropped. I think putting it in as a possible consequence in a list is going to be, both rhetorically and practically, rather harder to convince the process of legal review of it's soundness. And I think that the final conclusion is really to come back to where I began, which is, as any of us looking at the meal around us can confirm, tying is ubiquitous. And when firms acquire market power, they don't suddenly stop tying.

Tying and bundling is, roughly speaking, what the modern firm does. It's the rationale. It puts things together and offers them in packages to consumers. So cases when tying can be convincingly established to be anti-competitive, and such cases exist and will continue to exist, cases like that are bound to be exceptions to a phenomenon that is everywhere around us. Thank you.

Professor Paul Seabright is Professor of Economics at the Université de Toulouse I. Professor Seabright has been Professor of Economics, Fellow, All Souls College, Oxford and Assistant Director of Research, University of Cambridge. He is currently Managing Editor of Economic Policy and the Maitre de Conférences, Ecole Polytechnique (Paris). He is the co-author of Merger in Daylight: the Economics and Politics of European Merger Control and Trawling for Minnows: European Competition Policy and Agreements between Firms. Professor Seabright received his B.A. in Philosophy, Politics & Economics; his M.Phil. in Economics; and his D.Phil. in Economics from the University of Oxford.

Professor Xavier Vives is Professor of Economics and Finance, INSEAD, and the Portuguese Council Chaired Professor of European Studies. Professor Vives is also a Member of the Board of Governors of Fundació Empresa i Ciència. He is the co-author, with K-U. Kühn, of "Information Exchange Among Firms and their Impact on Competition,” authored in 1995 for the European Commission. Professor Vives is also a columnist for La Vanguardia and Expansión; the Editor of the European Economic Review; Co-editor of the Journal of Economics and Management Strategy; Associate editor of the Rand Journal of Economics and Recherches Economiques de Louvain; and a Member of the editorial board of Economía Industrial.