 Welcome everyone to the fourth online platform economic seminar. I'm so glad we moved online. This actually is a completely new experience. It's different but in many many ways it's better we can have a wider audience and it's great to have so many good people in terms of speakers and attendees. Very interesting new format. So, my name is Hannah Haloborda. I will be moderator today and we have Oslem Bedre Defoyle. Did I pronounce it correctly? That's fine. Defoyle. That's the French part of the surname. That's fine. Yes, it's really difficult to guess which one is which language family. So, Oslem is going to present a paper joint with Gary Biglaser on platform competition for exclusivity with Marquis Sellar. And the way this new format works is we are going to have 40 minutes of Oslem presenting the paper. It is going to be broken up every now and then. Oslem is going to stop and take questions. If you have questions, mostly clarifying questions at this point, please type them in the chat, either to myself or to everyone if you want. When Oslem will stop, then I will ask the questions. At the end we are going to have another 20 minutes for general Q&A and at that point there will be more free flowing questions, not just clarifying questions. We will stop recording after an hour, but we will hang out and you are welcome to ask more questions and have more discussion after that one hour. So this is the format. The seminar once again is recorded. If you do not want to be recorded, then act appropriately. Don't ask questions. Or if you want to ask questions anonymously, I can do it for you. So with that, Oslem, the screen is yours. Thank you. Thank you very much, Hannah, for being the moderator of this talk. And thank you very much to the organizers for giving me the opportunity to present my joint work with Gary Becklaser, who is also here with us today, attending the webinar. So let me start by sharing my screen. Hopefully that works as expected. So go to the full screen mode. Is it fine? Yeah? Yes. Good. Excellent. So this paper is about platform competition for exclusivity with the marquee product seller. As I said, this is joint work with Gary, and I would like to first start by giving some motivation for this work. We observe platforms imposing restrictions on multi-homing or switching behavior of their customers. In particular, many industries platform signs exclusive deal with some premium content provider or a product seller. For instance, in pay TV markets, we see pay TV providers signing exclusive contracts with some popular content providers like show series and sports events. And in the app stores market like Apple and Google Android, signing some exclusivity closes with some popular apps with the app developers. And music streaming platform signing exclusivity contracts with the famous artists. And game platforms have been using this type of exclusivity closes with popular games. And recently ebook platforms have also started using exclusivity with some popular authors or podcasters. And indeed this type of exclusivity was already existing in traditional two-sided markets like shopping malls, signing exclusivity closes within the radius, basically restricting those anchor stores not being available in a competing shopping center. And we also see some non-exclusive content product being available on those competing platforms. So what is the main motivation for this work? There have been significant antitrust concerns arising from the exclusive dealing provisions. And in particular, they might dampen competition in the market and also for the market. And they have been dealt under the abuse of dominant position closes. Basically they were only concerned as anti-competitive if they are used by a dominant player or a platform. And what were those concerns? They might lead to tipping in the market. This is particularly a concern for markets with network effects where we expect tipping or monopolization of the market being more likely. And that might be indeed triggered by exclusivity closes or in any market they can reduce variety. They can weaken competition in the market and also they might foreclose a potential rival in particular if it's more efficient or valuable that would be detrimental for the welfare. And there have been several cases both in the US and Europe basically banning exclusivity closes used by dominant players. For instance, Microsoft cases in both Europe and US and also MasterCard and Visa were banned from using exclusivity closes with their merchants basically preventing them from accepting AmEx cards. But very recently, European Commission fined Google for the usage of exclusivity closes with the Google Android phone manufacturers, basically Google imposing them to use the Google search as the exclusive search engine on those phones. And of course, exclusivity means competition for exclusivity and that might also increase the prices for those very valuable content, which is the case for instance, Champions League games and that also has raised some significant public attention basically those products contents being only available on very expensive channels. So then what we do in this paper is try to understand profitability and implications of exclusive dealing provisions in those markets where we have network effects and we would like to make this distinction between the exclusivity with a marquee product or a popular product between the product seller and the platforms. And what are the key elements of the model? We have initial asymmetry between the platforms. So one platform has a larger locked in base than the other one. And we also want to capture this cross group network effects between buyers and sellers as well as we would like to allow for within group negative or congestion effects arising from competition between different qualities of products. And in our model, platforms will be choosing the quality endogenously via two channels by determining number of unpopular what we call the fringe products at the same time deciding whether to have the marquee product and if so exclusively. So these will be the two ways of achieving the quality for the platform so the quality will be endogenous in our model. And the key questions that we would like to answer in this paper are when do we expect to see exclusive dealing in equilibrium? If so, who wins the exclusive dealing with the marquee product seller? What are the welfare implications? And then we also raise the question of if we allow platforms to price discriminate basically set different prices to their locked in consumers and the new consumer segment. What will the implications for the profitability and the welfare implications of exclusive dealing? So these are our research questions very briefly what do we find as main results? So the big platform we find that wins the exclusive dealing with the marquee product seller in the unique equilibrium if the marquee product matters for the quality of the small platform. And I will be more precise about what we mean by this. And we also find that there exists nonexcusive equilibrium otherwise in other words if the marquee product does not change the quality level of the small platform. And exclusive dealing whenever it arises in equilibrium lowers welfare by reducing variety basically number of fringe products on the small platform and also the average quality of the small platform and price discrimination makes exclusive dealing more likely to arise in equilibrium when the initial asymmetry between small and the big platforms in their locked in basis is large. However, it increases the total welfare when this initial asymmetry isn't small when the platforms are similar. So we have different basically unclear welfare implications for the price discrimination. We are not the first analyzing exclusive dealing there is a very well developed literature and the vertical contracting. Basically, most of the literature try to answer the key question that the Chicago school school economists were raising. How could an inefficient or less valuable incumbent compensate buyer not to buy from the more valuable entrant or a rival? So that was the Chicago school question and they gave different lines of answers basically saying under which conditions we might see exclusive dealing occurring in equilibrium in an inefficient way. What we are doing in this paper, of course, we want to look at the exclusive dealing profitability but we try to shut down those channels that the previous literature were emphasizing as the exclusive dealing mechanisms that could arise in equilibrium. So we try to shut down those channels because we would like to mainly focus on those markets where we have network effects so we try to understand how the network effects matters for the profitability and the implications of exclusive dealing. So to be a little bit more precise so there is also literature analyzing exclusivity contracts with network effects. Our key differences from this literature is going to be we will look at initially asymmetric platforms so one is big in the locked in base and the other one and also we will introduce heterogeneous content in other words heterogeneous qualities of those products and we will have both cross group network effects as well as meeting group congestion or competition effects between the products. So this will be the key differences from this existing literature and lastly just let me emphasize that the key finding of this literature was that when there are strong network effects they mainly consider the cross group network effects. When the network effects are strong or the platforms are close competitors then the exclusive dealing is more likely to be profitable in equilibrium but these are the cases also the exclusive dealing is less likely to hurt welfare. This literature talking about network effects basically give us this common finding saying that look when there are network effects it's less likely to harm exclusive dealing if it's profitable it's less likely to harm the total welfare and we will add to this literature by looking at also this congestion effects on top of the cross group network externalities. So let me pose here and see if there are any clarification questions. So I guess that's not so much clarification but G. Joy is asking whether exclusive dealing can be used by entrant to break into the market and how does it fit with your results. So indeed that's a very good question. So we have like for instance Bruno's very early paper made this point of divide and concure type of strategies might help an entrant to come to the market by breaking this chicken and egg problem to be viable that that you know there are not so many papers finding entrant winning the exclusivity in equilibrium but there is a paper by Robin Nie showing that indeed in the video game console market entrant might benefit from exclusive dealing with the popular games we have predictions that that can happen in equilibrium under some circumstances of our model so indeed we will obtain this in equilibrium under some conditions so we allow for this to be happening also. Thank you. Then let me continue to describe the model that we are looking at. We are modeling two platforms platform B and platform S which platform B has a locked in base measure of beta B platform S has a locked in base measure of beta S and platforms are competing for a mass of new consumers on the line so basically they have this locked in basis backyards but they are competing for the new segment new consumers are single homing in other words they choose either platform B or S to visit and then platforms we assume that they are effectively competing so the market is covered so the transportation parameters is slow enough even without the market seller and we also would like to analyze different qualities of products that the platforms could have so that's why we have this fringe product which is our on average low quality product each platform can get access to there is no competition for fringe products and each product costs F and there is some product called marquee product which is our high quality product for which platforms are competing effectively in particular if the marquee product seller signs an exclusive deal that I'm going to be a little bit more precise about what it is and marquee product has a fixed cost FM so this is the broad structure of our model let me be a little bit more precise about our assumptions so we assume that the platform B is our big platform which has a larger locked in base than platform S so beta B is larger than beta S platform S might have a locked in base of zero so that would correspond to an entrant platform but we don't have any fixed cost of entry in order to again shut down this channel of economies of scale from the entry that previous literature emphasized the potential for having exclusivity in equilibrium so we give the highest chances for the platform S to be viable in this market so there are no fixed costs of entering for a platform which even doesn't have a locked in base and then we assume this marquee product seller has this quality the M which is certain the French products have uncertain values for consumers so we assume that the French product could have a value l with probability p and value h with probability 1 minus p h is greater than l binomial distribution is not crucial for us I will try to comment on it at the end but we assume that the marquee seller has this quality which is above the average French firm so this is the average quality of a French product so marquee seller is on average better than the French here we could also have some uncertain value for the marquee again this is not crucial for our model we try to capture the main effects that we are interested in in the simplest way so that's why we have this specification here okay so just after the assumption so these red were our assumptions let me be precise about the timing of the interactions so stage one try to read this slide from the bottom to the worst up so in stage one the firms platforms choose their contracts basically they sign a contract they specify a lump sum transfer to marquee seller conditional on the market structure so lump sum transfer TBE of the platform B basically offered if marquee seller is exclusively available on platform B and lump sum transfer of TBE and E is offered if marquee seller is non exclusive meaning that available on both platforms and the same for platform S and platforms offer those contracts simultaneously I didn't say previously but there's also important assumption in our model we assume that the platforms have rich enough contracts with the fringe product sellers as well as the marquee seller that enables each platform to control the number of fringe products in equilibrium which is MB here that notation in equilibrium why are using this contracts with the fringe product sellers and at the same time we assume that platforms can control the products prices in equilibrium why do we assume that because we want to get rid of any contractual inefficiency arising because of any contractual you know distortions here we really want to focus on that if there are no distortions in the contracts between the platform and the sellers even in that situation do we expect to see exclusive dealing arising because then again previous literature emphasize contractual inefficiencies in the vertical chains could lead to exclusive dealing in equilibrium and we want to shut down those channels this is on purpose that we assume those contracts are perfect but this is really to focus on the key mechanism here that we would like to analyze and in those contracts firms the platforms also choose their number of fringe products so since we assume that the platforms can control them perfectly and be is the number of fringe of platform B if the market structure is exclusive with the marquee seller and MB and E is the number of fringe products that the platform B has if market seller is available on both platforms and the same for the platform S so we have this to you know the menus of contracts that the platform set in the initial stage and in the second stage the marquee seller decides which contracts to accept and acceptance decisions are observable however the terms of the contracts are not observed by the rival for instance if exclusivity of the platform B is accepted by marquee seller then platform S does not observe the number of fringe products set by the platform B so in a sense they are setting their number of fringe products as if it was set simultaneously because they don't observe it and that's why we are looking for Bayesian Nash equilibrium of this game and in stage three platforms set their subscription fees SJ for platform J and the product prices again we assume that the platforms can control those fringe product and the marquee product price if they have the marquee product in equilibrium and this SJ is a subscription fee to the buyers and they set the same SJ to their locked-in base or to new consumers but in the case when we allow price discrimination we allow platform to set different subscription fee to their loyal base versus new consumers that's what we name as price discrimination in this model and in stage four consumers observe which platforms have which platform has the marquee if both or one and they also observe the subscription fees set by each platform and the product prices and then those new consumers choose one of the B or S and the locked-in base consumers decide whether to visit their preferred platform or not and the final stage once their consumers are in the platform they draw and J draws of products from the pool of the fringe products on the platform and each draw has probability P low value and probability H high value so these draws are IID so this is an important assumption for our model and consumers then after getting those draws decide whether to buy the fringe product or the marquee product if the marquee is available and they decide whether to buy the fringe or not if there is no marquee available so there's multiple stages so we look for a bias in Nash equilibrium again of this model again and I would like to pause here to see if there are any questions on the model and the assumptions Yes, Oslo so Jacques Kramer is asking whether this is a hoteling line I guess you discussed a little bit more about the model but I think it would be useful to explain what is the role of the hoteling line and I guess it only applies to one side of the market, right? Yes, so basically our consumers in the new consumer segment are like hoteling line but what is crucial for us is not just the uniform distribution of these new consumers so this is not crucial but what is crucial is horizontal differentiation between platforms so this is important assumption but the uniform distribution of those consumers is not an important assumption in the model and the consumers in the backyard they have homogeneous types so there is no elastic demand in the descents they either buy from their preferred platform all of them or not but again this is not crucial for the model we could have had also heterogeneous types on the backyards as well so this is just to simplify the model And there is no horizontal differentiation for the product There is kind of a horizontal differentiation so what we have in the model is that fringe products have uncertain types for consumers, right? So it could be high type or low type but consumers they draw those types in the platform before coming to platform they don't know it so they only know the number of fringe so they have an expected value of those draws before coming to the platform and in the model in the benchmark case you don't have uncertainty on the market product type but we could also have had this in the model again that wouldn't change qualitative results but that would make just our lives a little bit more complicated in the solution So a couple of people ask whether or how many products a consumer can buy whether it's one or more Very good question This is a unique demand so they draw those products on the platform and they look at the value and they decide whether to buy the fringe product or the marquee so one or other basically this is the unique demand assumption and I will comment at the end in the discussion how this is important for us and if there is no marquee they just decide whether to buy the fringe product or not so this is a unique demand model and this is important One more how sensitive is the model to the assumption that the platform cannot change the number of fringe products after the marquee seller has rejected its offer by Roberto Circusia It can change Sorry, maybe it was not clear from my presentation so if for instance platform B has an exclusive contract accepted by the marquee seller in this stage platform S observes that it lost the exclusivity and it lost marquee seller then platform S could choose its own number of fringe products at that point so they don't allow basically we want to have them the optimal numbers of fringe products in any configuration of the market structure so the contracts are not just set at stage one and take it or leave it, they can be adjusted so basically if M accepts exclusive contract of B say, then B does not have an incentive to adjust MBE so this is renegotiation proof in the sense however if M accepts the contract of B then S might be interested in changing its number of fringe products and that we allow it to be the case in the model. Okay, so this would be in stage two. Exactly Exactly. So one more from Bruno Julien who observes NB and NS and when in the timing? So NB and NS can only be observed right before consumers so the NB and NS, so it's not observed when the platforms are choosing again it's not observed before stage two, okay, but after stage two it's gonna be observed. Very cool. By all parties basically when the platforms are choosing their subscription prices and the product prices. Okay. So I have one more question so as I understand the quality, the expected quality of fringe products are the same, is exactly the same on both platforms. So I know you will be talking about comparing the qualities of the platform, so the quality of the platform is what? Is it just whether the market? I'm gonna define it now. I'm going to define it explicitly. I think it's very important question. Yeah. Okay, so I guess we are ready to move on. Okay, excellent. Thank you. Again, please ask questions if continuation doesn't answer your question. Okay, so in the equilibrium the first point that we are finding is that the platform said their product prices at their marginal cost which is assumed to be zero in the model because they could capture expected surplus of consumers from the consumption via subscription fees. It's similar to Armstrong and Vickers model so that's why we are model predicts positive subscription fees and zero variable fees. For instance, if you want to think about the real world implication that would be like Netflix charging subscription fees positive but not charging per view views of its content. Okay, the expected quality to answer your question Hanna is VJ in the notation if the platform does not have the more key that's going to be P to the power NJ times L because this is a probability that the consumer will accept expect to draw low quality fringe product on the platform and with the probability one minus P to the power NJ consumer expects to draw high quality fringe product on the platform. Okay, again this P to the power NJ is because of the IID assumption so each draw is independent so that's why I expect the low draw of all these draws with this probability. Okay, and this is the expected quality of the platform if platform does not have the marquee but has number NJ of fringe products and if the platform has the marquee then whenever the draw of the fringe product is low consumer expects to consume the high quality marquee VM because VM is higher than L and whenever the draw of the fringe product is high consumer again expects to consume the high quality fringe product. So then this is the expected quality of platform J if platform J has marquee product that's why the notation VJM. And a couple of observations adding more fringe products increases the expected quality without marquee or with the marquee so in other words partial derivative of this VJ and VJM are positive with respect to NJ. But there are decreasing returns from the fringe products in other words adding more and more fringe products gives you less quality improvements if you have already a lot of number of fringe products. So there are decreasing returns from adding fringe and more valuable marquee so when the value of the marquee product goes up that also increases the expected quality of this platform which has the marquee product as you can see that's lower straw basically and more interestingly when the marquee product becomes more valuable returns from adding fringe products goes down in other words if you want to increase your quality by adding fringe products that pays off less if you have higher quality marquee why because what marquee does is basically cutting the left tail of the distribution and instead of consuming the low quality fringe you consume high quality marquee product if the draw happens to be the low quality fringe draw. So adding marquee basically adding the fringe products are less valuable if your marquee product is more valuable. That gives us this interaction between two products and that's how we generate the congestion effects indeed for the quality provision of the platform. What we do is basically we consider this model instead of saying platforms are choosing their number of fringe products and deciding whether to have the marquee or not at the end this is a hotel in competition as Jack pointed out in the benchmark model so platforms are just choosing their quality levels in this competition and then they decide how to achieve this quality at the minimum cost so basically which number of fringe firms and whether to have the marquee or not will affect their cost given the optimal quality they would like to achieve. That's why we want to invert these quality functions and think about what is the number of fringe a platform J needs in order to achieve quality BJ without the marquee and what is the number of fringe that the platform needs to achieve the quality level BJ with the marquee. That's how we write down these equations. Why do we do that because we are going to write the cost function of the platform for quality level BJ when it doesn't have the marquee this is going to be just the fixed cost of the fringe firms times this function that I draw in the previous slide not explicitly but just define it and J BJ so basically that will be the cost of achieving quality level BJ when platform does not have the marquee okay and then we write down the profit of the platform J at the equilibrium subscription piece. This is a standard hotel in competition with different qualities but the difference is that the firms have different backyards or the locked in bases okay so basically this is the profit expression of the firm J platform J when it has quality J rival has quality V minus J and the firm J base locked in bases beta J and the rival locked in bases beta minus J and we are evaluating these variable profits at the optimal subscription fees they chose in equilibrium so I'm not showing this part of the solution because it's pretty standard what is important is that this cost of achieving quality level BJ will determine the optimal quality of each platform because platform is going to maximize this profit with respect to BJ to determine its optimal quality okay and then we solve this question of optimal quality question when the platforms don't have the marquee product and then we show that in the first lemma the big platform chooses higher quality level than the small platform without the marquee and the number of fringe products on the big platform is higher than on the small platform again when there is no marquee in the market so without the marquee so this is the benchmark to compare us with the marquee case why is it the case because the big platform because it has a larger locked in customer base it can amortize fixed cost of the fringe over a larger base so it can have a larger profit from quality improvements starting from equal quality levels okay so that's why big platform having larger subscription base is going to offer higher quality before the marquee product seller is available okay so then the next question what's going to happen if we have a marquee on the platform cost function will be different because now I will need a different number of fringe firms to achieve the same quality though that's why I denoted NJM because now I have the marquee which basically cuts the left tail of the fringe distribution so that I need a different number of fringe firms to achieve a given quality level than before and of course I might have some other fixed costs this is exogenous cost of the marquee seller product and this is the endogenous part of the cost which is depending on how much lump sum transfer I'm paying to the marquee seller in equilibrium but that would not matter for my optimal quality choice because the variable part is the crucial part that will determine that and to be a little bit more careful here when I have the marquee the platform when I say the platform has the marquee its minimum quality level will be BN when the platform does not have the marquee its minimum quality level will be L okay because if it says zero number of fringe firm if it has only one fringe firm and if the draw happens to be low for instance that's going to be the minimum quality for a platform but if platform has the marquee product even the low quality fringe arises consumer will choose by the marquee product so this will be the lowest quality of the platform okay so after that what we show is that if the platform has the marquee product then the marquee and the fringe products are substitutes for the platform so when the value of the marquee goes up platform chooses less number of fringe firms why this is interesting because we know that there are this cross group network effects so marquee product attracts more buyers to the platform so this is the traffic effect due to the cross group network effects normally this should increase the demand for the fringe products as well but because of the fact that fringe products and the marquee products are competing against each other and here this unit demand assumption gives us the highest substitution between those products basically then what we are showing is that this congestion effect dominates and the products are substitutes from the viewpoint of the platform and the intuition is very simple I have already shown that when I have a high more valuable marquee product adding one more fringe is less valuable for improvement of the quality of the platform that's why having the higher value for the marquee product lowers the number of fringe products that the platform chooses in equilibrium and the level three shows when and how does the marquee product affect optimal quality of the platforms the first bullet point is basically state in condition when it doesn't affect so if without the marquee small platform was already choosing a quality above VM then having the marquee on one platform or both wouldn't affect the equilibrium quality of the platforms and if the platform J's quality before the marquee was below VM then having the marquee product will increase the quality of that platform so just to give very brief intuition for this result if small platform had a quality level above the VM then it must be the case also the big platform has a quality level above VM without the marquee product but what we did in the paper is that we showed that the marginal cost of quality is the same with or without the marquee and we believe that this is driven from the IID assumption of this binomial distribution but not the distribution of the type of the distribution and because of the fact that marginal cost is the same for the quality if the quality levels were already above VM having the marquee does not affect their optimal qualities and more importantly we showed that the cost of quality is indeed a complex function what it means is that the production function for quality is concave so there are this degreasing returns from adding more the fringe products in the market to achieve a given quality level okay so let me close here and see if there are any questions so Jacques Kramer says I do not understand the role of the explicit modeling of the quality in the absence of the marquee product why speak about the fringe products rather than just a vertical quality component so in the absence that's a relevant point so in the absence of the basically marquee product it doesn't matter so much okay so but what we want to capture is that when I see more number of fringe products available on the platform my expected value of getting a better match is higher okay so this is what we are capturing with this fringe product distribution and again binomial is not crucial any IID distribution will give us this type of expectations this is important and again when I see the marquee at the same platform then we need this cross derivative of the quality increase from the fringe is getting more less important when the marquee is available so we have this cross effects between the marquee and the fringe so that's why we need it but of course I mean he's right pointing out that if we did not have the marquee we don't need to model it in this way but we want to capture this competition between the marquee and the fringe product so that's why we have this type of modeling thank you there are no other questions at this point okay and Aslan we are at the 40 minute mark so if you still have a lot of slides to show us maybe we can postpone all the other questions until the Q&A afterwards okay maybe that's a good idea but we started like five minutes yes so five minutes yes I hope I will be done by then excellent then proposition one basically shows that there exists an unexcused equilibrium if the marquee does not matter for the quality of the small platform and this is the case if the small was already choosing a quality level above the marquee product value without having the marquee and what we show is that there exists no non-exclusive equilibrium otherwise okay why is it the case if the small was platform was already choosing a quality above the VM then we cannot affect the rival quality by excluding the rival from the marquee again this is because of the fact that in that case having the marquee will not change the quality level of the S if we only change it is total cost of achieving this quality but not the quality level okay and in that case we have a non-exclusive equilibrium however if S was choosing a quality level below the VM before having the marquee then having the marquee will increase would increase the quality of the S and then in that case excluding S from the marquee gives competitive advantage to the B okay so how does it work basically consider a candidate non-exclusive equilibrium it must be the case that the marquee is indifferent between the sum of the non-exclusive tariffs and one of the exclusive tariff of the each platform starting from this equilibrium B has a profitable deviation by offering slightly more for exclusivity to the marquee because that will enable B to make a discrete jump in its profit by gaining this competitive advantage against the rival okay so this is how we kill non-exclusive equilibrium if marquee matters for the quality of the small platform and they show that there exists an exclusive equilibrium where the big platform wins the marquee and there exists no exclusive equilibrium where S wins the marquee in the benchmark model what is the intuition remember without the marquee B would like to achieve a higher quality level in order to do that B was having more number of higher number of fringe products but then if B gets the marquee it can save more fixed costs from those high number of fringe products than S so that's why starting from the same number of fringe products B can generate more profit than S having the marquee exclusively because it has a larger base okay and so indeed in equilibrium B can do even better so that's why we can always overbid S in the exclusive dealing competition okay and allowing very quickly allowing price discrimination at the beginning I promise that if we allow platforms to offer different prices to loyal and the locked in base customers what we find is that price discrimination will make the competitive segment more competitive so it will lower returns from improving quality in that segment at the same time price discrimination will increase prices for the locked in base customers and that will increase the quality improvement returns from the locked in base but B will benefit more from that than S so what we find is that price discrimination will induce B to choose a higher quality however it might induce S to choose a lower quality if initial asymmetry is significantly large just to consider for instance beta S to be 0 for instance in that case what price discrimination does is basically lowers the old returns from quality improvements for S so that the S will choose a lower quality okay you can find then the critical threshold for beta S sufficiently low below which this is going to happen but then what it means for our model is that it makes exclusive dealing equilibrium more likely to be the unique equilibrium why because it lowers the S quality so then this VM will be more binding for this threshold of uniqueness and in the exclusive dealing equilibrium S will choose lower quality and B will choose higher quality when we allow price discrimination and then platform initial asymmetries are very close to each other price discrimination will increase the quality of both platforms so shrinks the region where exclusive dealing is unique equilibrium okay so I think I'm running out of time this is my last slide for welfare implications and then I will conclude so what are the critical welfare implications of this model first space in our paper is having the marquee product on both platforms and exclusive dealing is the unique equilibrium where marquee product matters for S quality and exclusive dealing harms welfare by lowering quality and the variety on the small platform and B prefers exclusive dealing in our model to capture more rent from the marquee seller so when the initial asymmetry between platforms is large enough price discrimination might lower welfare by lowering quality provision by the small platform however when this initial asymmetry is not big then price discrimination increases the welfare okay so just let me then move on I don't have time so just go back and conclude because we also look at this case where the number of fringe firms is fixed so not endogenous and I will give you those results just in this conclusion slide so this paper aims to understand platform competition profitability for competition for exclusivity with a marquee product seller and the platforms initially asymmetric in their locked basis we capture both cross group network effects and between group congestion or negative competition effects between these different qualities and allow platforms to choose quality endogenously by choosing their number of low quality products and deciding whether to have the marquee product or exclusively or not and in the unique equilibrium the big platform wins exclusive dealing with the marquee if marquee matters for the quality of the small platform and exclusive dealing always lowers welfare in our model and this is also true when the fringe number is fixed not endogenous in the model okay and price discrimination might harm welfare when initially asymmetry is large price discrimination indeed might help us to win exclusivity this is the part I wanted to come at the end I hope there is a question on that when the fringe number is fixed not endogenous in the model why fringe number fix is important because indeed in some context maybe platforms cannot condition fringe condition on the market structure so then we also look at this case and analysis okay let me stop here I have some discussion slides but hopefully there will be some questions that I can use these points to answer those questions okay thank you so I think that now it's time for Q&A and the way the best way to do it is if you have a question just raise your hand and then and then either I will call on you and you will either I will unmute you or you can unmute yourself but this avoids a kind of collision where a lot of people start talking at the same time so are there any questions and comments I see no raised hands so how about that may be too quick but what we can do is as long as you can I would be very interested to hear a little bit more of your discussion slides and especially if you go back one slide before your last bullet point that you said that you are going to discuss why exclusive dealing as can win with exclusive dealing when the French number is fixed can you elaborate on that it goes back to one of the first questions we had yes happy to do that so let me then just go to this slide as I said maybe the platforms cannot condition their number of French products on the market structure with the Marquis seller this could be because of antitrust scrutiny or it could be some content available already under long-term contracts or some reputation concerns for instance pay TV and on some content so they can't just withdraw the content from the platform okay if this happens basically exclusive dealing of the B with the Marquis the unique equilibrium always when there is no prices combination so that means there is no non-exclusive equilibrium why is this a case because the Marquis will always increase the quality of the rival in that case when the NBs and NS is exogenous given in the model such that the big platform has a higher number of fringe than the S platform okay so that's an interesting point to make because basically when we fix the French number we get rid of non-exclusive equilibrium and there is non-exclusive equilibrium which is harmful however starting from that if we allow platforms to do prices discrimination set different prices to their locked in base and the consumers need consumers that will enable the S to win exclusivity with the Marquis seller and we did some parameters specifications how this could happen basically this can happen if big platform does not care so much having the Marquis exclusivity for instance if the number of fringe products is sufficiently large in this graph the horizontal axis shows number of fringe products on the big platform compared to the small one if it is sufficiently large or if the small platform base the initial base is sufficiently large we are in this blue area blue area is the area where the highest bit of the small platform is higher than the highest bit of the big platform for exclusivity okay and again this is a parameter specification this is in line with the prediction of the lead paper basically in that area small platform can win exclusivity and if you want to understand how the competition will affect that again if the platforms have sufficiently low degree of competition differentiated market but again still covered market assumption holds or the Marquis product has sufficiently high value we are in this blue region again where the S platform can win the exclusive deal okay so that's it just to say that if we can't have platforms choosing their fringe numbers at the same time they contract with the Marquis it might happen that S platform wins exclusivity with the Marquis seller when platforms are allowed to price discriminate between sales and new consumers can I encourage now? yes please, yes for now I was a bit lost, it was very fast so to get and then yet on this point I didn't get what matters it was that the L is on dodging so it's a quality dimension as Jacques said and whether the Marquis seller cares about it why do the Marquis seller care about the BONS it does not care about the MBNS it just cares about its compensation on their different market structures so could you write a more general model in which you have endogenous quality for getting platforms and then you have a seller of some input which is kind of important input and if you buy input then you have a different cost for quality let's say lower cost okay and then you could have just the same analysis in the sense almost without the platforms, the platform is one case where this cost function is a particular shape but there may be other case where the cost function is a different shape so that becomes a problem of exclusivity of input and I don't know the literature on that but you see what it's a rebound on Jack's comment actually Jack might to some point okay so basically you are thinking of a model of competition and qualities where firms can prevent each other to get access to particular technology that might enable them to reach that quality level at a more efficient way that's a good point and definitely that's very similar to what our model does I need to think about a little bit more about so basically modeling of the fringe products makes sense if you want to capture this platform effects right we want to have this other product sellers number being endogenous in the model but in the model that you suggested basically what is my technology to achieve this quality level I don't care how I get there it's just some kind of a technology I think if we shut down this black box of sellers the model might indeed apply in this other broader context I think we are essentially we are raising the cost of achieving a certain quality level by having an exclusive deal yeah that's true second question is about the game so you assume to take it a little bit also by the platform but this is a marquee seller so you could regard the timing and the marquee seller could form a side that was my discussion point if the marquee seller the entire bargaining power in our model then there won't be exclusive dealing in equilibrium because then marquee seller would capture all the profits generated from its presence on both firms so then there is no reason to go for exclusivity to get larger rent but then there is not an issue with competition on the consumer side that participates around competition for example competition may be quite intense so basically it's always efficient to have Bruno it's always efficient to have not exclusive dealing and then the marquee seller has all the bargaining power he can internalize any externalities if he can have a full contract let's come back to the point you said before he has a complete contract setting and can control the price of the product and the feature for consumers but he can do that through his contract right? we need a very complex contract it depends on the number of consumers on the platform that can control for the price of the other product and a complete contract setting would be right so we're allowing it for a rich contract space to try to focus on a particular mechanism but I think Bruno's question was hinting on if the marquee is available on both platforms it's kind of like competing away it is a higher quality and if it was available only on one it creates this differentiation effect previously Tisha was also mentioning so content might prefer to go to one platform to capture larger rent if the competition is very stiff and it could be interesting to take a look at if the contracting is not perfect yes we haven't done this analysis we haven't given the bargaining power to marquee but we just had the conjecture that in that case there won't be rent shifting incentives so that's why we were expecting exclusive dealing and the same for the vertical integration between one of the platforms and the marquee seller but it could be interesting to see if the contracting is not perfect how does this would change so so let us go to some other questions Julien Wright has a question Julien can you unmute yourself yeah I had a very simple question just wanted to understand what happens if you go back to the baseline model and take the case where there's no asymmetry in the installed basis yeah to remove any asymmetry just wanted to watch the equilibrium then so basically then to existing literature yeah so then basically if they are initial so if I remove the symmetry coming back to this first lemma then without the marquee platforms will set the same number of fringe products or the same quality to start with okay and starting from there of course having the marquee would matter if this quality level was below the marquee quality level okay so the same arguments would apply so if you are in this world that the initially platform were choosing a quality below the marquee quality value level then excluding my rivals still would give me competitive advantage so there exists always an exclusive equilibrium but which platform wins is not determined because they are symmetric but they always want to beat yeah so this is multiple equilibrium yeah exactly thank you you're welcome