 Hello, everyone. Welcome to this new edition of the online platform seminar. So today, the speaker is Shijun Chen, who will talk about data-driven measures. So Shijun will have 40 minutes, and then there will be a discussion by Greg Taylor. So the seminar is recorded, and it will be so far an hour. After an hour, we'll turn off the recording, and we can have a more informal discussion. For those of you who don't want their words to be immortalized. But in the meantime, during the seminar, you can ask questions through the chat or raise your hand. And Shijun, you should try and maybe stop sometimes to take questions. But we'll have a Q&A session at the end for maybe a discussion. So please focus more on clarifying question during the talk. OK, I think I've spoken too much already. So Shijun, if you want to share your screen. OK, thank you very much. OK, thank you, everyone. So as I mentioned to Jacques, I have to say good morning, good afternoon, and good evening to everyone. So it's a great pleasure to present this paper in this seminar. This is joint work with my colleague, Chong Chou, and my co-host, Jia Jia Chong, from Fudan University and Noriyaki Matsushima from Osaka University. So as you can see, this paper was motivated by the data-driven mergers, in particular, as a Google Fitbit merger case. The recent years, we have seen a large amount of data-driven mergers. Mainly, they are so paid or dominated by the major players like Google, Amazon, and Facebook. Just according to the recent survey, we find more than 300 acquisitions between 2008 and 2018. And the target companies are particularly young and startups. One common feature is these business generates a large volume of consumer data. So it's quite clear the primary purpose of such mergers is about consumer data. So about the Google Fitbit merger, I think you all know about this, but just to introduce a bit of the background. So Google offered a 2.1 billion bidder for the Fitbit. And this merger case is under regulatory investigation in many jurisdictions, including HLC and the European Commission. So Fitbit actually is a small player. Its market share is less than 5% in 2019. So a question raised is, why should we concern about this merger of a small firm? It seems a merger would not generate a huge impact on the market for wearables. Also, Google said the deal was about devices, not data. And that they promised Fitbit data would not be used for Google online placement. So however, the fact is Google can harvest 30 million of Fitbit users detailed health data in 24 hours, seven days, by selling the wearable devices. So the main concern is Google will try to monetize health data and harm consumers. So why we have this concern? So in order to interpret this, let's first try to reveal Google's strategic plan or business plan behind this merger. And you will find that this merger is actually an important step of Google's ambitious plan in health care market. So Google started this plan many years ago. So first they started a so-called Google Health project, started 10 years ago, and restarted again in 2018, which aims to develop a specific search engine for medical records. Another project called Project Nineteen Gale, it was secretly initiated in 2018, a joint project by Google Cloud and Ascension, which is the second largest health care system in the United States. According to this joint agreement, Google can access to the complete set of health records of millions of Americans. Another important project involving a company, which is not well known called Verily, the company was founded in 2015 by Alphabet, focusing on research in health care first. Starting from 2019, it entered into the insurance market by collaborating with John Hancock, which is a big company in the United States. And in 2020, just a month ago, it created a company called Coefficient Insurance Company with Swiss Re-Insurance Group, which is one of the largest re-insurance group. As mentioned by various presidents, so we are hoping to be more personalized in the way we offer health solutions. So personalization clearly is one of the important purpose. So now, when we have this big picture, it seems quite clear that so Google's, as a Google Fitbit mergers of primary purpose is for health data. So why is this case become quite hot? Because the problem is traditional Google, sorry, merger guidelines may not be suitable for such kind of merger. Indeed, so we are involved in the, in say, MSCAP brief offered to the European Commission, and we have some interaction with the European Commission. And they also face some kind of dilemma. So these mergers are not horizontal, not vertical, and accomplishing authorities may want to show home these mergers into conglomerate mergers. It's this kind of merger between different firms in different markets. But conglomerate mergers are mostly not harmful. In the United States, for example, no conglomerate mergers have been ever challenged since 1980s, accomplishing authorities always ask this question without the evidence of harm. This is the language from the European Commission. So I think we need a new framework and a guideline for such kind of mergers. This is a normal or traditional type of merger. So as a first step, I think it really important we try to develop a new theory of harm for data-driven mergers. So in order to develop such kind of model, we started with market definition. Of course, there are lots of rated markets. So here we would like to focus on two types of markets, particularly relevant. First type of market we call the markets for data collection, we call it the market B. They are mostly focused on digital devices such as wearables, Google Nest or applications related to internet of things. One main feature is devices in the app are offered with very low prices or mainly free. The second type of the market is called market for data applications, we call it market A. These are the main market that the platforms will use data, big data for personalization in their products and services. Personalization seems to be the future of these markets. In particular, we are concerned about the healthcare market become a new battlefield for personalization. Questions? So these two markets are highly curated and the products are complementary. Meanwhile, dominant platform may not sell the product directory into market A. What we found in the Google case is they try to launch joint ventures and with partner firms in their market like Hancock and Swiss Regions Re-Insurance Company. So any questions on this introduction? Okay, no questions. So let's start with modeling mass storage. As you will see that actually the model is very simple. We consider two hotel lines because the two markets, each line is for one market. So each market is served by two firms that we normalize the marginal cost to zero. For trackability purpose, we assume the perfect correlation of consumer's taste. That is a consumer's taste X in market A is the same of his taste in market B. Of course, this assumption is a bit extreme but we believe that even in the imperfect correlation case, some kind of main results could still preserve. So consumers taste X is uniformly distributed in hotel line. Consider market A first. We here assume that market A is symmetric hoteling competition because the mainly we think current is the insurance market are quite competitive. So a consumer with taste X obtains the utility VA minus X from insurance company A1, whereas VA minus one minus X from product A2. So A1 is located at zero end and A2 located as end of one. Without merger, each firm offers a standard version of product. We call product A1 or A2 and the charges hotel and uniform price here. We use alpha to denote as the uniform price that equal to one. So very simple. Now market B is a bit different. We introduce asymmetric hoteling competition in market B before the merger. Why asymmetric hoteling? Because market B is the kind of wearable markets in which we have asymmetric players. Think about Fitbit is much smaller than Apple. So it's a bigger company. We think Apple watch offers better values than Fitbit and to characterize this dominance or competitive advantage we introduce a parameter gamma. So that is a consumer with taste X obtains a utility VA minus X from Fitbit, whereas VA plus gamma minus one minus X from Apple watch, let's say. So before the merger there's asymmetric hoteling competition and you can see the equilibrium prices. Beta one is margin of Fitbit and beta two is a margin of Apple watch and the cutoff threshold is X tilde here. Okay, any questions? So you may wonder why we can model competition in health insurance use a hoteling model. Actually, we check the literature and find the hoteling model is commonly adopted in the analysis of healthcare market. This is because health insurance or healthcare products are highly differentiated for two reasons. First, the health plans are kind of complicated covers a broader range of different treatments and offer different co-payment rates. For instance, take for example, two treatment, one is knee, I have knee problem and the dental treatment, I don't have teeth problem. So suppose insurance plan A one offer 40% of cost for dental treatment and 60% of cost for the knee treatment while A two is in opposite. So you can see these two plans are differentiated. Also second and the more important consumer health conditions vary across persons. In my case, of course, I like the insurance plan A one because I have knee plan, a knee problem while some other people may like insurance plan A two. So roughly we try to use this permit X and hoteling competition to model competition in health insurance. Okay, the second remark. Yes. Can I ask you a question? But that means your assumption of that the costs are independent of types, doesn't Oh, yeah. It would be very strange with this interpretation. Certainly, yes, Jack. Your comments are very important. Here we assumed away lots of ingredients. So we normalize the cost and also we assume that there's no cream skin effect. So just in order to focus on that, very basic accomplishing. Yeah. Thank you. Any other questions? Yes, it's around the same line. I had the same comment and been working on health and adverse selection is a huge issue in health. But also there are a lot of regulations that prevent you from personalizing offers to... So this would affect the use of data in a sense that would be specific to health. It's not like selling shoes and... Yes, Bruno, your comments are very important. Actually, we were a bit worried about this, but after we found some very interesting evidence, we are a bit competent about personalization healthcare. So look at this. So Mackenzie actually estimated that annual value of applying big data is $100 billion. And actually according to what we found that starting from years ago, several insurance companies begins using personalization. So one important health care plan is actually initiated by Ernst and Jung to call the pay as you live product. So basically an insurance company offers you, Bruno, a wearable device is tracking your health information. If you agree on this, you will receive some kind of discount and on top of that, you will also receive different kind of accommodations. In United States, there are several companies doing such kind of practices. In European Union, at least 50% of firms expect their willingness to do all they are doing experimental or experimenting of such kind of personalization. Yeah. Thank you very much for your comments. Any other questions? Okay. So these are the two separate markets. Now, one important features we want to model the merger. What's the impact on merger? What's the main feature of the merger? So the most important feature of the merger is this merger will hinge two separate markets together. The way of such kind of linking is Google can incorporate wearable devices into its ecosystem and package it with other Google product. So in order to capture such kind of benefits, we assume that combining health data with non-health data allows Google to deliver extra value omega within one access for a substitute for consumers. That is consumer, when they use personalized healthcare products, combining with other Google's product will generate a so-called consumption signature. Here consumption signature omega measures from C's competitive advantage. So in presentation here for simplicity, we assume B A equal to one and C equal to zero. Sorry, I skipped one slide. Modeling personalization. Now suppose firm B1 is acquired by Google who is partnering with A1, an insurance company. So Google can consolidate consumer data and improve its capacity in personalization. Here we introduce some kind of efficiency gain for personalization. We assume that the cost of personalization in variable market is very high because of these hardware. Instead, Google can use consumer data for personalization in market A. As we mentioned, this is health insurance market and deliver a personalized version of health insurance. For instance, when Google knows exactly my health conditions and they know I have a new problem and all the insurance company, the partner insurance company can offer me a personalized insurance plan. Let's say, hi, Zijun. We are going to cover 100% of your new treatment. Are you happy with that? Of course, I'm happy with that. But then the problem is they can raise the price. So we here assume that personalized product and deliver an improved matching value. So instead of having VA minus X here, after I receive this personalized product, my matching value improved by VAX. So become VA minus one minus VAX. Here, VA measures platform or Google's capacity of data analytics. That is personalization can reduce welfare loss due to mismatching by this amount, VAX. Of course, there's cost of offering personalized product. So in the presentation here, we will assume VA equal to one and C equal to zero for simplicity. Okay, so this is the modern methodology for personalization. Now, timing of the game, because there are two markets. So we would like to introduce a kind of a bit dynamic feature of the market. So we assume in period one, firm set prices in market B that is the market for wearables. And then consumers by either B1 or B2. And then in period two, firm set prices in market A and consumers by A1 and A2. We also assume that consumers are not for the looking, they are myopic. And we serve for the equipment used in backwarding actually. Okay. So let's consider the sub again in period two. Suppose... I'm sorry, Jijun, Jacques speaking. Can you go back to the previous slide? Yes. I'm somewhat worried about your thing that consumers are not forward looking because it means that the consumer would not realize that if he buys a fit bit, he's going to get better insurance from Google. That's basically what you are assuming that the consumers are unaware of the benefits of matching. And that's clearly going to buy. It seems to me that I'm not sure it's clear, but it would seem to me that this is going to bias the analysis against the merger. Okay, Jacques, thank you very much. Yes, this is just for simplicity of analysis. So we definitely should consider extension to relax this assumption. Yeah, I doubt the result will change a bit. Okay, thank you very much, Jacques. Yes? Can I follow up on this? So maybe one way of capturing this none more less forward-looking consumers could be saying that I wanted to ask a question about the data externalities. So because here you only allow the firm to personalize using the data of the consumer from one market and the personalized product on the other market. How could we interpret your model if you want to capture also this Google knowing something about some group of consumers and correlating this information with what it knows on other consumers and basically doing personalization also on those others who did not buy from Google in the first part market. So could that be a way of capturing this in a reduced form approach? Maybe then that might be a way of capturing this also not forward-looking aspect which could be generated due to, I mean, these are not the same things but just in terms of the reduced form how much it would play a role in the model. Okay, thank you very much. Yeah, so we did not consider this inter-consumer externalities, but that's really important. Yes, we should consider this in some extension. Yeah, thank you all then. Any other questions? Okay, so now let's solve the sub-game in period two first. That's a very simple game. Suppose consumers have already made a decision and the firm sees market share is given by X star. So this is determined by beta one, beta two. So then because now firms see as consumer data from zero to X star, these consumers have purchased the Fitbit wearables and then Google can deliver personalized products to these consumers with improved value, PA. So one important feature of personalized products we would like to capture is personalized products that is offered privately to a particular consumer. That gives the firm some kind of flexibility. Naturally, a personalized product comes along with personalized price because it's offered in personalized way. So we also assume that firms can prevent target consumers from accessing its standard product that is Google can block your search to the standard product and offer you the personalized product only. In this case, we assume that firms set uniform prices first because they can be observed and then firms see offers personalized prices. First, let's think about the competition between personalized price and the uniform price. When firm A2 set a uniform price to R5-2 firm C can best respond to this price. So a target of consumers now has two options. He will receive personalized offer. Meanwhile, he can compare the offer from A2 standard product. So using or consuming the personalized product gives extra value as we mentioned. Now it's matching values VA, not VA minus X and extra value concept signature Omega. So a consumer is willing to pick this only for VA plus Omega minus PAX. Here PAX is personalized price is greater than outside option from A2. So we assume that firm A2 is willing to accept this offer as well as in different. So this gives PAX equal to Omega plus X plus this one. So here clearly Omega plus X is the extra benefit from personalization and from the merger. But look at this formula. It's very clear that Google can capture all these benefit from consumers. That is if the target consumer accept the offer from personalized product he is left with the same surplus from outside option as if consuming the product A2 and efficiencies again actually Omega plus X from personalization are all extracted by Google through personalized pricing. So this generates a very strong effect of exploitation we call the exploitation effect. Actually you can imagine that targeted consumers can be worse off because before the merger the symmetric hotel competition only the marginal consumer located at X equal to one half will be indifferent between A1 and A2. All consumers on the left hand side of one half are better off because they prefer A1. Now all consumers, all targeted consumers receive the same surplus as if they are forced to or they have to buy A2. In this sense all the consumer surplus exceeding the outside options are extracted through personalization. Okay, the second effect we call the proper squeeze effect of personalization. This terminology is not perfect. That is personalization actually makes firm see very aggressive in price competition and could squeeze the rival's profit. Because of personalized offers firm see is ready to offer each targeted customer the lowest price here X equal to zero without reducing the price to others. For instance, I'm going to make offer to Bruno and suppose he Alex is competing with me I can offer Bruno if he's a marginal consumer price equal to zero without affecting my offers to other consumers but Alex cannot do this because Alex has to offer a uniform price to all consumers. So this gives firm see some kind of flexibility and makes this firm very aggressive. On top of that, that is firm see can secure consumers from zero to X star on top of that, it can also compete for the other consumers from X star to one on these consumers from see does not have data has to offer a uniform price but this uniform price can be lower. So what you can find actually is the pricing Caribbean in this game is from see can offer personalized price to consumers located from zero to X star on top of that, it can offer a uniform price alpha one which is lower than alpha two from to consumers between X star and X hat and then from to smart to share actually squeezed from one half to this part as a result we can see actually from to support that can be reduced dramatically. So these are the equilibrium price in the sub game given X star, okay. So now let's go back to stage one or period one we can solve for the equilibrium pricing market B here we assume that omega is so large so we can write down firm sees total profit because it's profit from market A is very clear from zero to X star will charge a personalized price P a X and then uniform price to serve these consumers on top of that it can make a profit to market B by charging the margin beta one so so X star so after reorganize we can find this formula and the firm B to the profit is very simple there are two times one minus X star so solving this whole ten type of competition actually is quite easy and we can find the equipment prices that are related to gamma and omega so beta one star equal to this clearly you can see that beta one star is from sees equilibrium margin in the wearable market assuming omega greater than gamma greater than one immediately you can see that beta one star is below zero that is we can we call this equilibrium is a Caribbean with accommodation because from a to B to still still in the market and this figure gives you clear connection of the hinge between two markets so beta one is below cost the beta two of course above cost and X star is a market share for firm C in both markets while in market A you can see that surprising given like this way so solving for this simple game we find several important effects first we find a cross-up relation between markets that is firms C will price its wearable product below cost this bill cost pricing is driven by two forces first sees incentives of exploitation because here when omega increases its profit from market A it's increasing with X star it has extra incentives to expand its market share and it's also driven by the rival's competitive advantage gamma because when the gamma is higher it imposes a barrier for firm C to expand it has to reduce meanwhile we find the two types of externalities of car markets the first externality is a negative externality from market A to B that is consumption signature imposes next externality of the pricing market B you can see that both beta one beta two decreases with omega so interesting state there's a positive externality from market B to market A imposed by apple's competitive advantage say gamma you can find that today's uniform prices in market A increases with gamma so that is because from B to that's the apple actually brings contributing power which can restrict firm C's expansion in market B and this of course benefits from A to so when interesting finding is a dominant player in market B such as apple could protect a small player in market A so while this is one type of equilibrium when this equilibrium rises when omega is reasonably big not very not sufficiently large when Google's efficiency gain or consumption signature is sufficiently large you can imagine that Google has incentives to monopolize both market because the benefit from expansion of exploitation is huge so Google can set a predator pricing and which can actually exclude the competitors in both market so monopolization can arise of course when important policy recommendation is competition or solution ban predator pricing so questions you are five minutes left okay so quite good now I've finished a model let's now go to this policy implications first we would like to sum up what we find a series of harm from this merger we find that Google can leverage its dominance into wearable markets and hurt small competitors Google can sell Fitbit watch for cost or enter agreement with other insurers it can recoup this loss while other small firms cannot and Google will monetize health data and harm consumers consolidating health data with health data can make it yes I have a question your second point here I didn't see the small competitors in your model that's a good question yes we in particular in market B we do not incorporate this so that's sort of the conclusion of your model that's your opinion okay thank you very much okay so what's the policy implications we would like to offer is because assessment of this merger is really complicated of course block the merger will preserve the current paying field for and benefit competitors they also can protect consumers but there's a short run welfare loss for consumers because of this concept and Google is good at offering personalized product which can further benefit consumers okay so if we do not consider blocking the merger so some kind of remedies may be important but once again designing a suitable package of remedies becomes very difficult here we just we would like to try to offer some kind of workable remedies first we find that restriction on bill cost pricing could be helpful because it can prevent Google from excluding the rival in both markets but in the short run this will raise a price may harm consumers but could benefit them in the long run second provision of tie-in bundling and other exclusionary strategies can benefit competitors we consider requirement of data sharing that is if completion authority can required Google to share the insurance the health data with other competitors there's of course pros and cons of sharing the side effect of Google will have less incentive to track the health data and there also the data privacy concern issues okay so finally I would like to just have a brief discussion of rated literature so this paper benefits a lot from two very important reports on digital market paymax with others and Scott Morton's also we would like to say this paper is highly complemented to Alex and Greg's paper while maybe they would disagree and also it was a massive amount of money price and they use a form where we try to provide a micro foundation for such a merger and this paper also builds on a large literature of personalized pricing and personalization okay conclusions so this paper is motivated by the Google fit merger we develop a very simple model to examine the impact of such merger so model captures two main features of such mergers first a merger firm merger firm can use consumer data for personalization second the merger hinges market together and it can create cross market facts so we found that personalization leads to exploitation which can harm consumers and hurts competitors in particular when concept signature is sufficiently high so merger can result in monopolization this is a risk we also discuss policy implications consider various merger remedies okay that's the end of my presentation thank you very much I should stop sharing thank you thanks for the opportunity to discuss this interesting paper I think everyone on the call knows that this is really an important topic especially at the moment and I think this is an area where we need to get a better handle on how to do policy so I was particularly glad to see in the latest version of the paper what I think are some relevant policy issues really addressed head on one of the things that I thought was quite neat in this paper that I haven't seen in a lot of others is that it has this conceptual dichotomy between on the one hand using data for analytics so how good is my analysis technology and on the other the idea of the scale of my data set and this is something that I've seen a lot in the debate is this about the amount of data I have or is it about the quality of my algorithm or both and so I wondered if there are ways to push that idea a bit further in the model one thing I noticed for example is that in the model analytics plays an important role in determining how well I can personalise my product for consumers but it doesn't really affect how good I am at personalising my price for those consumers I can always do that perfectly accurately and so I was curious if there's a meaningful way to get at the idea that a better algorithm might also help me to personalise my price as well as the product I guess that you won't be too surprised to hear that one of the things that was most interesting to me was the idea about the spillovers between the two markets given that that's something that we've been working on as well and particularly I like the results you have about the countervailing power that a big firm in market be can exert I guess the results you have on those spillovers they're sort of culled in a way by the assumption that the locations of the consumer in the two markets is perfectly correlated and you say you do that for tractability say perhaps it's not possible to relax that assumption but it also seemed to me that there could be interesting questions to ask if you were able to relax that for example it seems very natural to me that if Google were to acquire Fitbit and Fitbit's mode of operation stops being about competing in the wearables market and it becomes much more about gathering data to use in some other market then Google might want to reposition future versions of Fitbit's product to help better serve that objective by changing the types of consumer it attracts and so it seems like there's a lot of interesting questions there that you might be able to get at I think the Fitbit example is obviously nice because insurance is a market where personalized pricing and also product personalization are obviously important I shared a little bit I think the reaction of Jacques and Bruno in that it's very hard for me to think about that market and not worry about issues like adverse selection so in the paper you talk about this idea of a risk group but if I understood correctly you really would need whatever data is being gathered is somehow orthogonal to the risks implied with a consumer and it was really hard for me to understand how it can be that I can get more data about my customers and not cause bigger adverse selection problems from my rival and more to the point not make my rival want to redesign its base insurance packages in response to that and so I think when motivating the example it's important to bear in mind that this is a model where you have products that are sort of immutably differentiated and the data is about is this consumer a better fit for me or my rival and be very clear about how you tie that to the example that you use a couple of quick points to finish up that I guess are sort of not just specific to this paper but more comments on how economists are tackling these issues in general one is that you know we know in this TSA Viva's style model that we get the effect of intensified competition when firms know more about their consumer consumers I think that's a little bit special to the TSA Viva's framework and so it's important to bear in mind you know that those two things go hand in hand when thinking about this context that we might apply the model to and the other is that you know some of the other examples besides the Google Fitbit merger are cases where prices don't seem to be very important if I think about Facebook WhatsApp for example and so then it seems to be more about product personalization than pricing personalization in a way and I guess as economists something we know how to do really well is model price discrimination so there's always a tendency to want to do that it's not always clear to me that that's the first order issue and so I wondered in particular in your model whether you can say anything about the case with product personalization but not price personalization and whether that could also yield something of interest so I guess that's all I have to say for now and I will hand back to Alex Thanks Greg so let me start by a question and actually which is related to a point that Greg made so you make the assumption that you have perfect correlation between the type on market A and on market B but then so first I think that there's a bit of maybe there's a bit of a problem in terms of interpretation because if you interpret the position on market A as being say the relative state of your knee versus your teeth then it seems hard to me to imagine that this would be correlated with your preference between Apple Watch and Fitbit right so I think that's the first issue that I have and so in a sense I would like to I would be curious to see what happens if you don't make this assumption and I think it's not so innocuous it seems important to me because it seems like so the firm is going to learn the willingness to pay of consumers that are in its tough on market A only whereas if you had and so it can extract all the value from those guys and those guys are not going to be poached by the rival whereas if you have a different story in which what you learn in which you may you will learn something about a consumer who might be maybe he's going to buy a Fitbit but he would be very close to 8 sorry I'm getting confused with the A and the B so he's going to buy Fitbit but he would be very close to A2 and so in that case the fact that you have information would allow you to offer him a very low price and this could really intensify competition so in the end I mean you went a bit quickly so I wasn't completely sure what happens on the overall on consumer surplus on market A I was a bit lost there but it seems to me that this assumption of perfect correlation sort of makes it harder for you to get pro-competitive effects and now if someone has any question feel free to ask I have a question or comment but I'm a bit puzzled by story about the merger I mean there is a market for data so why do you need to buy a company to get the data and how the model would work if you have a market where a firm B just selling data to the firm on market A and I guess that if firm B has full bargaining power the model will give the same outcome or something so it may depend on the balance of bargaining power some old up problem something but I don't see the story that distinguish for the moment merger and sell of data can I answer this question? yeah but you know your question is very important so here the main difference is if you buy data you will not affect competition in market B but if you acquire this company in market B you will have incentives to crack more data in order to crack more data you will be aggressive to expand your consumer base so I guess this is the main difference yeah but the firm in market B know that he can sell the data so they will compete to buy the data to acquire data to be able to sell so there would be also an effect that is similar yes that's a really good question in principle I guess there will be a similar effect but also here with another effect we model this combining this data will have consumption synergy but according to your idea there could be a similar if I may add something when I read I think it was a report from the Australian competition authority regarding this precise merger I think what they found was that Fitbit was about to enter the market for data anyway so they were going to start the respective of the merger they will start licensing data to other players maybe they will stop yes after it's acquired maybe start doing so which report I think the ACCC had a report inquiry data inquiry it's in June I think it was specific to I will send you the link I can add something on that because this is where Google gets into the picture if you're going to sell data on a competitive market but if the firm on A let's say Google is a very powerful company who is able to exert market power and the market for data will not give the full value to the seller on market so if you use a Nash bargaining solution there will be no problem in the market that would justify the merger in this case Google can fully acquire the benefit of data becomes like a monopoly for these consumer data otherwise if Fitbit can sell data it cannot commit only sell to Google or maybe this other opportunity yes okay thank you very much questions sorry I have a question or comment I was wondering what the role of Apple is in your model like fitting it to the actual Google Fitbit merger because Apple obviously not only allowed to play in the wearables market but it's also very active in the medtech market and they also have tie ups with insurance companies developing apps and using data so actually if you model that aspect as well it's really sort of competition becomes competition between two C type firms thank you very much so here we model Apple as a contributing power in market B2 as you mentioned actually Apple is preparing to enter an insurance market so maybe we would expect a second merger if this merger is approved maybe Apple is going to acquire some insurance company not to the wearable devices I understand they actually in 2019 they already partnered with Aetna which is I think one of the big insurance players which company? Aetna A-E-T-N-A A-E-T-N-A can you find on this side? you just search Apple Aetna partnership specifically for this the same kind of idea okay so that would be very interesting I think the other thing that's missing in terms of what's actually going on in the merger is the operating system which I think maybe an important motivation for the merger which is Google has its wearable operating system which is not Android it's specific to wearables and it's trying to get more and more wearable manufacturers to adopt their operating system but Fitbit was not adopting it so by acquiring Fitbit it gets a major player and through the network effects could then get other manufacturers on there to dominate the operating system in competition with Apple's own operating system for wearables and that might be another motivation for the merger which potentially could be more important oh that's important because I did not see any kind of discussion on this I was thinking they all used the Android system so Google has developed it so Google actually wearables but was not successful it hasn't been so successful but I think there are a few wearable manufacturers adopting it the operating system quite a few but obviously if it gets Fitbit on there with so many users then it will get more developers developing the operating system and the network effects could lead to more and more manufacturers adopting it oh sure yes it's called Wear OS Wear OS okay good thank you very much are there any other questions comments I have a question after we've stopped recording dangerous one okay so in that case maybe now is a good time to thank Shijun so I will stop the recording now and so thank you very much