 Hello, everyone and welcome to the fourth meeting of our monthly seminar series on central banking and digital currencies. On behalf of all the organizers, I want to thank you all for joining us again today. So our tradition in this series is to have each event hosted by a different institutional partner. Today's host is the Federal Reserve Bank of New York, and I'll now turn things over to our moderator for today, Antoine Martin. Thank you, Todd. Welcome everyone. So today, Michael Lee will present a paper joined with Rod Garrett on monetizing privacy with central banking and digital currencies. He has 25 minutes, then we're extremely happy to have Bruno be a who will discuss that paper he'll have 10 minutes and the rest will be for discussion. So Michael the floor is yours. All right, well, thank you so much for the organizers to for inviting us to present this paper. Here we're going to think about kind of the potential impact of CBDC, particularly with respect to privacy issues. I want to start by saying the disclaimer the views expressed are ours and not necessarily that of the New York Fed or the Federal Reserve system. So it's first kind of important to recognize the world that we live in. And right now, the majority of payments are electronic. And what that means is that virtually all electronic payments generate data. And this data is tracked, collected and activated. Now the reason is kind of simple, like amongst various forms of data payment data is particularly valuable. The data has detailed information on the identification of people involved with the transaction. And you can also have demographic and financial information about the users. And this makes it particularly useful in terms of informing decisions and overall enhancing the productivity of firms who are utilizing this data. So it's kind of no wonder we have a lot of non traditional players entering into the payment space. You know, the large tech companies have all kind of joined the race to control the flow of transaction data. So I'm going to come back to this idea of enhancing productivity. And my argument is actually that we already live in the reality where even very basic and generic goods are utilizing transaction data in order to inform decision making. So a prime example is Amazon. So one day I wanted to Amazon and I searched for a very generic good, which is a phone charging cable. So you think well phone charging cables very generic high competition. The first thing that popped up in when I did the search on Amazon is Amazon's very own in product, which is Amazon basics. And surprisingly enough, you can see that Amazon has to make a variety of different choices just to produce on the decision of how to produce this. So they need to decide what type of phones to cater to the choice on the color, the choice on the length, how to market it and the pricing and so on and so forth. How is Amazon making all of these product decisions. Well, it turns out that they may have actually been tapping into kind of their privileged access to granular and real time data about the listing and sales of on other merchants on their platform. And this was kind of the reason why you regulators in 2020 suit Amazon for anti trust practices. So this kind of goes back to kind of highlighting how customers private data that's generated from payments is actually really valuable for firms decisions and also productivity. On the one hand this data is really useful for firms when they're trying to analyze the data understand trends and pinpoint what type of products to produce a design. On the other hand this means that consumers are foregoing their privacy in the process. So, a natural question is our consumers adequately compensated for foregoing privacy. And we have kind of good reasons to think that they might not be. First of all markets with monopolies are going to lead to kind of the surplus not being redistributed to consumers. And secondly, in given the nature of data, the data can be valuable when it's aggregated but it might be difficult to quantify the value of data for individuals. Now adding on to that, even if data is valuable at the aggregate. It's really difficult to collectively bargain. So, as even as an individual I think it's very rare for us to try to renegotiate our user agreement on how our payment data and our derivatives are used by vendors. And this kind of leads to a difficulty in terms of how consumers might be adequately compensated. So in this paper we're going to ask kind of three key questions. First of all, we need to start by understanding how much surplus is generated from data and how it's divided. And second, how do policies that and the set of available payment instruments affect surplus and consumer welfare. Now our main kind of pivotal and final question is how does introducing a privacy preserving CBDC impact the real economy. And in order to tackle this question we really need to build an environment where three things are endogenously determined. These are constantly making choices in particular payment choices that can determine how data is collected. When they're using electronic payments data can be generated, and this data kind of leads to the distribution of data across firms who are able to collect this. Now firms can produce goods informed by this data, and this kind of information is going to be vital for how they're going to compete in the product market. So overall, the kind of when the dust settles the market structure is going to be formed based on how competitive firms are, and that's actually going to fit back into the types of options that consumers have available for them, and also the prices at which they can obtain these kinds of goods. So today I'm going to delineate kind of three key results of our paper. First of all, we show that payment data drives the formation of what we call a data monopoly. And here, we call this a data monopoly because data enables a firm to build and maintain its dominant position in the marketplace. And we show that actually under a monopoly the total surplus is maximized under data. And this is due to kind of the positive attributes of having a concentration of data in terms of firm decision making. But on the flip side, there's a limited benefit that accrues to consumers and this is kind of a concern from a consumer welfare perspective. So a natural consideration is to think about policies that can maybe level the competition and allow for consumers to better benefit from the gains from data. What we show is that certain policies can in fact restore competition, but actually it reduces total surplus and can even harm consumer welfare and that kind of leads to the trickiness in terms of designing a policy that can be beneficial to consumers. Finally, we introduce a privacy preserving CBDC or what we call digital cash. And we show that it actually can preserve total surplus and improve consumer welfare by enabling consumers to monetize their private information. So in order to kind of unhash kind of the intuition for these results, let me jump right into the model environment here. So we have an infinite horizon model with two types of consumer, two types of agents. The first stars is a measure of consumers indexed I, and we also have two firms indexed J. So on the get go, each consumer each day is seeking to purchase a unit of good, and they need to choose how to purchase this and they're going to have different types of payment options, physical cash, electronic and CBDC which we're going to introduce later on. Now the trade off. There's a trade off between these payment options and the trade off is going to be regarding privacy and convenience. So we assume that cash is going to be less convenient than electronic. And as a result, there's some disutility cost of using cash as a payment option. On the flip side cash also allows you to preserve privacy. And so for a consumer, you're going to have some gains from being able to conceal or protect your privacy. Now this is going to vary across consumers based on their preferences. And this is captured by alpha I, and it's going to, it's on the interval anywhere between zero and the maximum kind of value of privacy, which is alpha. Now on the, so we can think about the different types of payment options here, privacy preserving options are going to include cash and CBDC which we'll talk about talk about later. And on the other side we have digital payments such as electronic or CBDC, which is going to have this kind of convenience benefit in terms of payments. So on the other side firms on every period they're designing and producing goods, and these goods have various different types of characteristics. The idea is that every period there's some ideal design associated with each characteristic. And you want to match these characteristics because consumers enjoy products that are closest to this kind of theoretical ideal design. The issue is that this ideal design is not directly observable. What can firms do about it. Well they can forecast this ideal choice per characteristic by using historical data. So each firm has a stock of data that they have on this generated from electronic sales in the previous period. So specifically consumers are purchasing the firms good. They use electronic payments. This contributes to the stock of data that the firm is able to to collect and use in the next period, in order to inform their decision. Now, using this kind of stock of data what the firm can do is try to forecast or identify these particular design choices for a measure of characteristics. In order to capture this we have this mapping row, which maps the stock of data that the firm has to the number of characteristics that the firm is able to precisely pinpoint. So there's only kind of two characteristics that we use in order to classify what role looks like more data is generally thought to be good. In addition, having more data also allows for the firm to be more precise about how well they can pinpoint characteristics. We think of this as kind of a how deep how you can obtain deep preferences using more data. When you have a broader set of consumers who are using electronic payments. In the initial period there's some random stock of data that each firm starts with, and importantly each firm's data is exclusive, which means this data is proprietary data that the firm can use in order to develop their products and provide them with some edge. So each period the firms are competing on two dimensions they need to design a product, and they also need to decide the prices and specifically, they need to, they can set the price per payment vehicle which means that depending on whether you use physical cash, or electronic payments, they can be different prices that the consumers face. So we come back to the consumers decision now. So the consumers are making two choices they're choosing between which firms good to to consume, but also how to pay. What's the choice of payment vehicle, and this boils down to choosing the firm, the good and the payment that maximizes their utility. So the utility from purchasing firm J's good is given by three components, you can see here first is the consumption value of the firms good. The second is the price that you pay in order to consume, and there's some payment dependent utility that I've outlined earlier on. So looking at the consumption value of the good, you can see there's some reservation utility from just consumption. And on top of that, depending on how good the product is designed, there's some additional benefit, you can think of this as what a cruise from the data that the firm is able to use, and in order to produce a better good. Now on the payment dependent utility can see that if you're able to protect your privacy, using privacy preserving payments, you get this utility alpha I. In the case of physical cash, it's a physical payment and there's some this utility associated with using it. So if we put this together in equilibrium each period firms are developing these products, setting these prices per payment vehicle to maximize their expected profits, and consumers are choosing this product payment vehicle pair to maximize their utility. So we're going to focus on the long run outcome, or the steady state equilibrium, where it requires a stable market share per payment vehicle between the two firms. Now our baseline is going to be a world without CBDC. And in this world, consumers are going to face one of two payment options cash for electronic. As I mentioned before cash has this advantage that you can protect your privacy. It has the disadvantage that it's less convenient and so there's going to be this this utility minus capital. So, the main force to kind of keep in mind here is the fact that electronic purchases made by consumers enable firms to collect exclusive data. Now using this exclusive data firms can actually design their products. And this means that this data is vital for offering them with the competitive edge to producing attractive goods in the future. The conclusion that we can draw from this is the fact that if firms can what they want to do is actually discriminate between prices in order to influence consumers payment choices. They want more consumers to purchase their good, but not only that to use electronic payments in order for them to accumulate more data, so that they can use that to their advantage in the next period as they compete in the product market. This leads us to our first result, which is that payment data catalyzes the formation of monopolies. In particular we show that there's a unique steady state equilibrium, where a single firm dominates the market. And as I said before, we call this a data monopoly. Why will data here is acting as the key asset to maintain this monopoly status. Let's kind of quickly go through the intuition for why this happens. So suppose that we have one of the firms have a small informational advantage. Well, with this informational advantages, they can incrementally extend their market share relative to their competitor. Now if you have more market share, then that means that you can potentially acquire more payment data with more payment data. Now you can develop a better product and that allows you to widen your market share. And this feedback between the payment data that you're able to acquire and the market share that you're able to control leads to this world where one of these firms can now snowball this advantage and establish dominant control over both the flow of data and also the market as well. So naturally, we have this long run steady state with winners takes all market. Now the key question is what happens in terms of the aggregate outcome so as long as data is sufficiently valuable the monopolist wants to induce all consumers to use electronic payments and equilibrium. Now what that means is that the total surplus when we aggregate overall consumers is and the firm is given by the reservation value of utility. The cost of production and importantly here, the total surplus generated from data, you can see here because everyone is using electronic payments, the input here is what all of the data is being aggregated by the monopolist. Now, we really need to think about pricing though because pricing is what determines the division between the firm and the consumers in terms of who benefits from the use of this data. The dominant firm here is producing a great product which is V plus the maximum quality of data of a product that can be designed using the data available. And because the dominant firm controls all of the data, the competitor can only produce a really basic generic good that has the base utility. Now on the flip side the competitor can offer a really competitive price in particular just equal to the marginal cost of production. More this price can be provided for physical cash purchases as well. And this means that when the firm, the monopolist wants to capture the entire market. The firm needs to make sure that the prices are sufficiently low so that even the consumer with the highest privacy value would want to use electronic payments to purchase this good. Well, what does that mean? Well, the firm can charge extra for the quality of the good that it produces. So the gains from product quality is incorporated into the price. But in order to attract the most private types, it needs to provide a discount associated with the value of relinquishing the privacy. So if we think back to what the consumers are getting from the data, well they're actually reaping a limited benefit from the data surplus. And this benefit just equals the cost of acquiring this data. What's the cost of acquiring this data was equal to their outside option of kind of preserving privacy, which is alpha minus capital. Now from this, we can do kind of a quick experiment. What happens when cash becomes more inconvenient? In other words, what happens when capital rises? Well, in this case, the consumer share diminishes. And this is because the monopolist understands that in order to protect their privacy, consumers now need to take increasingly large inconveniences in order to do so. And this affects the way in which prices are determined and hence how consumer welfare is determined as well. So this kind of leads to a natural question of what policies might help redistribute the surplus to consumers, especially because it's a key policy concern in terms of how the nature of data can naturally lead to monopolies. And this intuition kind of is shown in our model as well. Well, regulator, one thing that the regulator can do, wants to do is actually to potentially level the playing field and promote competition. If you're able to do that, then you can lower prices and this would increase consumer surplus. The one policy that can implement this is to actually data sharing policy. So let's consider a world where the regulator jumps in and says, well, now we want to require firms to share any and all exclusive data derived from past activities. Well, the result is that actually it's pretty effective. So a data sharing policy can break the monopoly and in the economy transitions into a world where firms are applying equal share of the market. So by democratizing the data, now competitors can produce goods of comparative quality. We have better competition. This leads to a drop in prices. And as a result, consumers are able to extract the entire surplus generated from their data. So this might be the end of the paper, except we need to then think about the impact on aggregate outcomes. And the first observation is the fact that if you have a level playing field, this may also diminish firms abilities to actually price discriminate. So with the monopolist, the monopolist can price discriminate between different types of options and make sure that they're able to acquire more data as much data as they can. In contrast, when firms are competing, this can lead to firms competing both across electronic payments and cash, and this can lead to a drop in total data collection. So as a result, total surplus drops as we're seeing a significantly lower collection of data. But not only that, when this collection is too low, consumers consumers can also be worse off as well. In other words, even if consumers are reaping the entire benefits of the data, the total pie has now shrunk too much, and this leads to worse consumer welfare. So this kind of brings us to our main and final analysis, thinking about what we could do in terms of introducing a different payment option that's digital and provides consumers a way to preserve their privacy. So let's what we think of as a privacy preserving CBDC in the context of our paper are threefold. First, there's zero cost of using this. Second, it preserves privacy much like cash. And third, it's convenient like other forms of electronic means of payment. As I showed before, digital cash is in the intersection of these two properties that cash and electronic have in the context of our model. It allows consumers to benefit from privacy, and it also doesn't have this this utility associated with cash. Now, the observation from this is the fact that all things fixed digital cash is the dominant payment method. It's better than both cash and electronic in this world. So introducing this has the effect of with introducing digital cash in the economy, we show that the data monopoly now survives, but with lower equilibrium prices and higher consumer welfare. So specifically, the underlying market structure and also the data acquisition is preserved. So we have the same monopolist that continues to dominate, and the same monopolist continues to set the set lower prices in order to acquire the consumers data in the form of electronic payments. So, as before, we have that total surplus is maximized under the data monopoly. And this means that if prices are going down. Actually, consumer surplus is increasing. So what's happening here with digital cash consumers have improved bargaining position. They have a better ability to have an outside option that allows them to preserve their privacy. And in equilibrium this allows them to have an increased ability to monetize their privacy or monetize the private information that is generated from their payment activity. So relative to the data monopoly equilibrium previously, what's happening here is that all consumers are now reaping the maximum value of privacy equal to alpha. Now, this brings me to kind of a broader discussion on why CBDC. What is, how should we think about the central banks role in terms of the provision of this type of payment option. And to kind of think about this we need to kind of walk back and think about cash as well. So cash was not specifically created to provide privacy, but privacy kind of became a feature inherent in its use and its design as well. And kind of because of this, the feature of privacy is has been argued by some is just as important as its role for substituting credit relationships. The key that cash use continues to decline, particularly for transactions. This is relevant question of should central banks actually be providing this digital alternative to cash. So in order to kind of think about this let's kind of rehash kind of the payment landscape that we live in today. And before we have that one of the unique features is that we have physical cash here. And in terms of a user base actually using physical physical cash is low cost it's you can go anywhere they accept it for no cost. And also preserves privacy but it's not a digital form and and we've highlighted that this may lead to kind of exclusion from certain types of economic activity. So the kind of dominant payment source is credit and debit. And furthermore, a lot of innovation has taken place in terms of this quadrant here that's at the intersection between low cost and digital. So new new payment options have arose, and kind of the kind of the premise is that at the benefit of using a very affordable and easy to use convenient digital payment option. And these providers are trying to collect the data and control the flow of this pain data. Now the third option is certain types of cryptocurrencies that may provide privacy preservation. And for these though, they're not really low cost. There's a direct transaction costs associated with a clearing payments and settling payments. In other words, and in addition to that there's indirect costs such as environmental costs. Could the private sector provide digital cash. Well, we see that there are high costs associated with a privacy centric digital payments that we see here today. And for other types of options payment innovations we see that there's actually privacy erosion in a lot of these initiatives. Now in contrast the central bank, maybe better position to commit to safeguarding this data. First of all central banks don't have this profit motive to exploit payment data. Secondly, providing a ubiquitous low cost digital digital cash is costly, especially if utilization is low. In particular, if digital cash is value is not just from utilization but actually from existence. It makes it difficult for a private sector initiative to actually monetize on providing this type of service. Kind of what this kind of shows is that privacy should definitely be a part of the consideration for the design of CBDC, particularly as central banks are thinking about what are the social gains or social gains from introducing a CBDC into the payment landscape. So, just to conclude, we kind of present in this paper, a model that where market structure is endogenously determined by competition and consumer choices and data acquisition, all this is endogenously determined. Now we show that payment data leads to this formation of data monopolies, and this allows for large surplus to be obtained from the data, but consumers only marginally benefit. And furthermore kind of a data sharing policies can be used to restore competition, but it's not always the case that we want to alter the underlying competitive environment. On this backdrop we show that digital cash or privacy preserving CBDC can actually improve consumers bargaining power and allows them to monetize their privacy, while preserving kind of the inherent value the economic value of the data by allowing firms to utilize that to produce superior goods. All right. Well, thank you very much, Michael. So, next, Bruno is going to be discussing this paper, and I'll just note that people should feel free to continue asking questions in the Q&A I think Rod has been answering some of them, and then we, Michael and Rod can expand on some of the questions after Bruno's discussion. Bruno, the floor is yours. Thank you very much, Antoine. It's a real pleasure to see you all guys and to read this very interesting paper and to discuss with you these nice issues. It's very nice. This is a very good paper. I really like it a lot. It's, in a sense, it's providing a IO approach to central bank digital currencies. So it's an interesting paper. It bears on an important topic and it's an elegant model. You know, it's quite horrible for me because usually what I like is, you know, I get these horribly complicated models. And then I work and I come up with a very simple 2.1 period model that delivers the same economic message. But in this case I was totally unable to do that because the paper already was very elegant and streamlined. So I don't know what I'm going to do. So anyhow, so in this paper, so you have three types of payment, you know, traditional cash, which has a, it's good because it gives you privacy, which value of privacy is alpha. But it's inconvenient because you have to carry coins in your pocket. The cost of that is minus K. Then you have electronic payment. It's convenient. You don't have the cost minus K. It bridges privacy so you don't get the privacy benefit alpha. Plus, the company that is getting electronic payments is acquiring information and uses that to design better products that are more in line with what the customers want. And this has a value gamma. And then the final possible payment is CBDC. So like electronic payment, it's convenient. Like cash, it gives you privacy, but of course it doesn't give you the knowledge that is necessary to design better products. So this paper is looking at the interaction between these three means of payments. And so the interaction goes via firms. So you have firms that can produce a good at unit cost C. You have like traditional firms that they can't use data to just, you know, competitive firms and you have data driven firms and those are able to collect data when customers use electronic payments. And so basically, as Michael was telling us, if you have a large number of customers, then you get a lot of data. And if you get a lot of data, then you get a very good product. And if you have a very good product, you're going to have a lot of customers, etc, etc. In the end, that gives monopoly. In equilibrium, you will have only one firm that will be producing and that firm will attract all the customers and design the best possible product. And of course, because it's a monopoly, it's going to be able to extract all rents from the customers. So what's the, what's the utility of a cut. So now let's, I'm just following Michael, Michael's presentation, we just start with just the CBDC and then we go to the case, sorry, without the CBDC and then the case with a CBDC. So if you don't have a CBDC, you're a customer, you have the choice between buying from the monopolist or going to the traditional firms, the competitive firms. If you go to the monopolist V plus gamma is the value of the good that you get V is the, you know, initial value and gamma is the additional value created by the monopolist by using all the data it has, and then you have to pay the price to the monopolist. If instead you go to the traditional firm, you get V, you do not get gamma because the traditional firm does not have the information. But you get alpha, which is the value of privacy, which you didn't get with the monopolist. Plus, you also have to pay the cost of inconvenience C minus Kappa, which you didn't have with the monopolist. And then you pay the price, and the price is the competitive price C. Now the customer has the choice between going here and going there. And the monopolist is going to set the price to just make the customer indifferent between coming to the monopoly or going to the competitive firm. So the utility of the customer when going to the monopolist has to be equalized with the utility of the customer when going to the competitive firm. And that gives us the price that the monopolist is going to set the monopolist is going to set a price equal to its cost. Plus a market. And this is the market and the market has three components. First, it's a better quality product. Second, it's a convenient product. But there's a little bit of a discount, because it reaches privacy. So this is the price the monopolist can can can can post and still get all the market by doing this as Michael explained extract all the rents. Very nasty. So now what happens with the CBDC. Well, now. So basically what happened without the CBDC is you have the choice between the monopolist and the traditional firm. The CBDC is going to be better. It's going to be dominating what you get with the traditional firm. So now what you have the choice between is as a customer, you can either go to the monopolist or go to the CBDC. What I mean, what I mean by that is you're using CBDC to pay. Going to the traditional shop. When you go to the traditional shop, but you pay with a CBDC, you don't have the this utility of carrying the coins. So the cap is cap is no longer there. And you get the benefit from privacy. So that's that's your utility to go to the traditional shop and pay with CBDC it's the value of the good plus the value of privacy minus the competitive price that you pay. Now again, what the monopolist wants to keep all the market to be able to attract all the information and design the best products. So the monopolist is going to set the price so that the customers are indifferent between monopoly and competitive firm. And that gives us the price of the monopolist. So price is cost plus market. And marker has a gamma just like before, it has a minus alpha just like before, but it no longer has the K. We used to have a K here that is gone. The K was here because the monopolist could afford to charge higher prices, because frankly going to the traditional firm and having to pay with coins with such a pain. But now you can go to the traditional firm and pay with CBDC. So you don't have to pay cost K K. And so that's why it's become more attractive to do to go to the competition. And the monopolist has to be a little bit less demanding on the price. Now notice that the role of CBDC in this economy is off the equilibrium path. On the equilibrium path, you never use the CBDC. The CBDC is just there as a threat. And the monopolist knows that if he was to exaggerate and set two prices that will be too high, then people would go to the competing firms and pay with CBDC. So what you have to do is to do all the plumbing and the technology and the legal thing to to make the CBDC available while anticipating that no one is going to use it. So of course, no private private entity would ever do that. That's not a very profitable business. But central banks, as we know, specialize in providing public goods for free. And that's what they're going to do in this equilibrium. So I think this is very nice. It's a, it's, I never thought of that. And I think it's a very nice idea. So, so my, my, my two virtual sense on this is, when I look at the model, I think, Oh, the assumption that carrying coins around is, you know, that this is inconvenient. I find this assumption very plausible. I hate to have coins in my pockets. Now, the assumption that the data firm, you know, which is very important, the assumption that the data firm is, is using all this data to design better products. I couldn't think of any good example of that in my, in my personal life. You know, I ways, ways is good, you know, ways is using a lot of data from all the people and, and by using all these data is able to, to think of the best way to drive. That's nice, but they're not selling this. So it's not an example of what we have in the present model. And so Michael told us about Amazon, but personally, I never use Amazon. I was not convinced. And so I think it after reading this paper and really enjoying this, I came to the conclusion that data firms are not for dinosaurs. And so I'm sure the reason why I cannot think of any convincing example in my real life of in my personal life of what is described in this model is just because you know dinosaurs are not concerned here. And so that's, that's it. That's my discussion. Thank you Bruno. So, the floor is open for discussion so there's a number of questions, several have been, I've already had an answer. Others are being answered in the Q&A but Rod and Michael if you if you feel like you would like to elaborate on some of these answers please jump in on the Q&A answers and then, of course, panelists are welcome to unmute themselves and jump in. Well, I want to start by thanking Bruno for the, for the great discussion. I think it's true that in our in the context of our model we are taking certain simplifying assumptions in order to kind of delineate kind of what the impact of CBDC is. I think that's an interesting question that we're not able to explore in the context of this paper, but I definitely something that could kind of shed more light on thinking about the exact process, the economy through which this kind of payment data is actually informing firms decisions. Amazon is a particular case because they have the privilege of seeing everyone's transaction data on their payment on their platform. But the same thing is for we're going to see it increasingly so for other types of payment processors who also have synergies with the transaction data and the service they provide. You can think of advertisement as another example where now having that kind of customer data allows you to better attract trends to better pinpoint it and have better retention and throughput in terms of the tracking consumers were actually going to now purchase the good so so here we think of this as we call it good but it could be a service it could be a good it could be a platform it could be many different things. Definitely looking into kind of the more details of how the, how this is different across different types of markets would be an interesting evidence. Right, I haven't yet seen all the questions so please go ahead and, and take over. No, no, and thanks Michael know I've been I've been frantically typing in there's been a lot of really good questions and I apologize to the people whose questions I haven't gotten to yet so I would just invite them to speak up and ask them. Yeah so some folks who are into Q&A cannot speak up right so so if you think they're good questions, pick them up read them and provide an answer if you haven't so feel free to do that at least. Maybe I can I jump in Antoine please do. So, so thanks Michael and rod this, I think this paper is quite interesting as a very elegant framework I think as Bruno said, there are a couple of things that my comments are mainly more about how one could expand the framework on a couple of dimensions that either may or may not complicate the results so the first one is on the presumption of privacy and how you might model privacy for a digital asset. And so I would say this kind of ties to the practical direction that I think a lot of central banks around the world are thinking about central bank digital currency and that is that you probably cannot drive the privacy thing down to zero. In that you might be able to reduce the cost of privacy or sorry maybe you're not able to fully extract the benefit of privacy in the model, but you might be able to get a little bit of an extra degree of privacy but what I what I mean by that is that I think the, the arguments that you get when you introduce electronic payments from central bank digital currency perspective is that you are a first principle is that you're not going to design something that is completely private in the sense that it would deter or detract from the ability to prevent illicit finance so money laundering, terrorist financing, etc. So there's a practical dimension I'm wondering if, if you actually approximate but don't get to have an offer a product that is fully realizable that you get with physical currency does that fundamentally change the results or is there a way to sort of think about that from a from as a possible extension. The second one is sort of thinking another degree of complication around the sort of two side of markets literature that you get in card payments. So for that I, you know, thinking a little bit about, you know, sort of the benefits and the rent, the possible rents that could go to either the monopolist or the consumer in your model there may be, you know, this may be a general equilibrium effect but if you if you do think about different payments that might coexist. So this would be physical cash, maybe private digital payments and a central bank digital currency or potential threat of a central bank digital currency. You may have the sort of cross subsidization of different types of consumer users. I know it's not in your model, but I'm just sort of asking to think a little bit about that. Because often when we hear about in terms of some of the rent seeking behavior in the private sector when it comes to card payments, they are cheap and easy but you know consumers don't pay a lot of costs but merchants often do. And so this two side of markets literature obviously spends a lot of time and more of a partial equilibrium way to try to understand the question of how rents are allocated and any of inefficiencies that result. So in your model it's possible with different types of preferences of different types of users you might see a little bit of this subsidization from card users to cash users. And I'm kind of curious that when you introduce these other digital forms of payment whether they're provided by not the traditional banking sector but but these data firms like the big tech giants or a kind of central bank digital currency potentially internalize some of those frictions might be another way of the avenue or extension to explore. So hopefully I'm making myself clear but I think I think the model and framework is quite useful to start to think about some of the dimensions dimensionality and some of those complexities around around was often been focus focus of the card interchange fee debate, but also a bit on this this sort of constraint that you might have for a central bank to be issue a fully comparable private bearer instrument say that you might you might model in the first pass. So thank you. Yeah, Michael. You want me to. I, you can go ahead if you have. Look, you go first off. Okay, so I think that's an excellent question. I mean, there's a lot to unpack there. The question of implementation is something that we're not addressing here and it's actually really important. Kind of the goal of the paper in the context of the discussion on CBDC is to first propose. Okay, here's a clear channel through which CBDC can actually be beneficial. Kind of the next step I think is as you say thinking about, can we technically design something that balances the benefits of having privacy preservation, but also allows for us to have adequate control over the use of this in the sense that we still want to have similar standards of privacy. That's a challenging question. I think I would need to personally think more about that. But at least if we have kind of the means or the reasons to develop something like this, then it kind of give us gives us a reason also to think about implementation. Secondly, kind of a slightly different thing that I want to talk about in terms of cross subsidization, which is different from the two sided market context is the fact that in some sense, the amount that people are able to monetize in the context of our model is given by the, the maximum value of privacy. So in some sense, there's a cross subsidization across consumers, people that actually don't value privacy and all are benefiting the most because everyone's data is being used. They don't really care about privacy, but because there are staunch people who believe in privacy, it allows it pushes firms to provide greater concessions and allows them to actually unlock the most benefit from both the data and the privacy as well. So there's an interesting question I think there that we don't explore which is what happens when firms are able to better identify the privacy preferences I think that's another interesting question that we're thinking about as we're thinking about how the kind of the micro level aspects can affect also the macro, macro outcomes. Yeah, I see there's a lot of other questions so let me just add really quickly, David that that, I mean it is important to remember that the central bank can see all the data. And they can do AML and they can do KYC it's just about not giving the data to the, to the, to the firms. And, moreover, in an equilibrium that the central bank doesn't actually get the data anyway because the firms induce the consumers to continue to use the non privacy for preserving payment mechanism it's just that it's just that the existence of the central bank threat induces them to, to, to yield more of the surplus to the consumers. So it's, it's in a way that, you know, what's sort of neat about this is that the central bank action and this is by the way, another thing Michael didn't have time to get to one of the reasons why the central bank is so pivotal here is that, that a private entity wouldn't have the incentive to provide this private privacy preserving means of payment that the central bank would be doing this more as a policy that improves consumer welfare, but, but I see there's a lot of other questions. Yeah, so, Dirk had his hand up earlier so I'm going to check with him if he still wants to have a question then Jonathan and Darrell have to hand up and, and wait, so we would go to them in that order. Dirk, do you want to jump in that display. Yeah, thanks very much and thanks a lot to the presenters in the paper. First one, quickly following up on what Bruno said, it's a very expensive way here for the central bank to step in and to provide this outside threat they would probably prefer to do it by making cash more attractive right if you would have some some leeway maybe paying interest or not paying interest on cash but finding some other ways of making existing alternatives more attractive would probably be the preferable way to act for the government or the central bank. I had a question what I did not fully understand this, since you're focused on the steady state if you were to look at transition dynamics. Would you sort of have a particular time profile of how you would prefer to intervene in that context first to give incentives for firms to collect a lot of information. Early on or probably later on or how would you optimally time that, and could there be some time consistency issues related to that. Thanks. So conceptually I in practical terms, yes, it would be a big ordeal, presumably the creation of a CBDC that also has some kind of privacy preservation would not be the only reason why you would want to build a CBDC. But having said that, when you're thinking about how do you encourage people to use physical cash what are the benefits. You're starting to think about digital form of cash that has those types of features. The question is whether, can you escape that now it could be a different arrangement where the central back is collaborating with the private sector to do this. I think that's kind of open to debate. The second thing about the transition now we have a very frictionless environment in our model, which means that the transition is very quick. So you can imagine that in reality there are other frictions that allow for a very, very gradual kind of convergence to the long run. But in terms of the timing then what do you want to do well, kind of the premise here is that we don't really want to necessarily mess with the market structure, especially if that requires. We want to have a high concentration of data. So there's a certain level of agnostic kind of feature of when do we want to incorporate CBDC. You know, in some sense because you're not fundamentally altering the way in which the markets is being structured. This means that, you know, that's just really a question of, you know, the time discount of whether you want to have the privacy benefit to cooing early on, whether you want to have the monopolist form slightly faster and so on, which can lead to some some some trade offs. But I don't do to my something to have to think about more but there's not really a direct kind of timing consistency with respect to the provision of digital cash. Johnson. Yeah. So it seems to me an important feature of the model is that firms can price discriminate according to the payment choices of users, and this assumption seems to be important because when we compare with other paper right what has with Martin JP also about CBDC and privacy, where I think uses payment choice generate negative externalities others and CBDC is useful because now people can can can hide their information and because firms cannot price discriminate So combining the two results seems to me that that two aspects are very important right one is whether or not data generate positive or negative externality to the society and secondly, about also payment regulations such as such a charge rules, whether or not merchants can price discriminate according to the payment choices. Is that correct. Yeah I'll respond to that I mean I think you're right. Jonathan I mean it's, there's lots of ways that data is used, and you know it can be used for good or evil. And so, you know different models are going to pick up on different on different aspects. You know it's, it's the idea of whether or not you can price discriminate in the United States I believe, I believe this is largely regulated at the state level. And it differs across states, but but generally speaking, you know it doesn't have to be something is as explicit as charging different prices for different payment methods it can be reward programs or various other things so. You know your points well taken there's there's many uses, many uses of data and obviously we'd like to have a model that incorporated them all but yeah. Yeah I want to come back to Bruno's question about the dinosaur. So, the Amazon case fits very well with your model, because it mixes banking and commerce or payments and commerce. The Fed has been very reluctant so far at least to allow big tech firms like Amazon to enter the payment rails by getting accounts at the Fed. The more likely model with a hybrid CBDC would be that there would be a separation between the firms providing the goods and services who could benefit from in their designs from getting those data. And versus the payment service providers who are not designing the products for sale they are actually providing the payment services. My guess is the future state will be more of that that the payment service providers will get the data in a hybrid CBDC model. They may be able to use those data to bind payment customers to their businesses which is providing payment services. But the ultimate goods and services that are sold to the consumer. Those guys are probably not going to get a lot of those data, I don't see why they would benefit a lot more from selling a cup of coffee with cash. Or be worse off with selling the coffee with cash versus selling the coffee to someone paying with CBDC. So I just wanted to ask Rod and Michael if they can clarify who exactly do you envision getting the data to produce better goods and services in a world with and without CBDC. Do I need to allow mixing of banking and commerce, or are you going to follow the model in which there's a separation between the payment service providers and the producers of goods and services. Sorry for the long question. Well, I guess I'll answer first I mean, yeah I would I would I guess I would envision a world where that separation is no longer there. I think that, I mean, maybe drift into sort of more philosophical reasons but you know the central bank supports payments. And historically it supported the banks in this activity but it's not at all clear to me why in the future, they couldn't, they wouldn't consider supporting other types of entities. Again in this context, it's it's interesting that the CBDC doesn't get used. So again it just becomes a threat point that changes the bargaining position in some sense to collect the bargaining position of the consumers. But I think the, you know, the idea of of how we think about data I think is largely driven by the fact that most applications including my previous work focuses on the negative aspects of the data, and having to protect consumers from the way firms use data. But I think there's valid reasons why you know data can be beneficial. You know a great example that I was given once at a conference was was when people looked at how people paid for the tube in London, and they looked at people that were paying on a monthly basis, and they had enough data from their home to recognize that that these people had stable jobs and would be able to, were dependable or credit worthy. And so they offered these people loans to purchase yearly passes. So by having these people purchase yearly passes they're able to split the surplus of the difference between the price of the yearly pass and the monthly pass, and in some sense everybody's made better off and this happened due to data collection. So I think there's lots of examples and I think even the cell phone charger example is a good one where where firms like Amazon that are already in the market, and you know are leveraging data so I, your points well taken but I think, you know I think we're still in the early stages of thinking about how the central bank is going to support payments, who's going to be collecting the payments and how much separation there's going to be between payments and the goods providers. And let me just add to that, you know, the question is also whether these types of large tech companies can inject themselves and the kind of the chain of payment processing. And they've already kind of effectively done that even though they're may or may not they don't need to issue directly credit cards in order to inject themselves into the process. So that's a good example of how. If there are synergies between their product lines and being able to access this kind of detailed information on the consumers activity. It enables them to then kind of kind of exercise some level of exclusivity. Now, I think follow up just briefly just just to make sure that I get the point so with CBDC what prevents Amazon from saying oh you paid with CBDC and we're going to see the same payment information that we would have had had you paid with a Visa card. I don't understand that that distinction why why except with a very direct CBDC model where all the information is collected only by the central bank. So why would we see in a world of CBDC that the that the people in the London Tube or the Amazons don't actually get to see the same consumer payment data that they would in a world without CBDC. Let me that's actually a great question. So the way that I think about it. If you if you want to actually browse the Internet or get information digitally, there's ways to do that anonymously. So the reason why we implicitly share a lot of our personal information is because we know that when we're making a purchase, we either need to provide a personal information at the payment stage of a purchase, or we need to create and that naturally makes us encourages to also create an account and reap the different benefits. So you can imagine that if you have a CBDC in principle people should be able to not only browse things anonymously, but also purchase things anonymously to, or make purchases anonymously. So it kind of breaks the barrier where we've always assumed that when the digital platform and digital activity. There's some level of level of sharing of identity. Now, this still means that Amazon still going to see the purchase, but it's not that Amazon won't know who is actually making the purchase. And this to a certain degree then is similar to physical cash like if you buy things with physical cash, you know okay there were 10 widgets sold, but you don't know kind of the underlying demographic and information about their customers were making those purchases that can be really vital for supply and design choices in the future. Okay, so Antoine had a hard stop yet to leave but for those of you who can stick around for a few minutes. I'd like to continue the discussion. Wei, would you like to ask your question. Yes, I thought that this is a fascinating topic. So if you want to follow up with Darren's question about sort of privacy related sort of, you know, associated with CBDC. My understanding is that, you know, even for CBDC I'm sympathetic to this idea that the, you know, the data, you know, who can see data is going to be different on the CBDC but now there is sort of there's still sort of a privacy related issue here. In China's case right so this is the probably if there's going to be a CBDC soon this will be the first one. My understanding is that PBC can actually cannot really launch CBDC on its own right. It has to work with companies right. So in fact, my understanding there's a consortium of firms going to work with PBC to implement to launch the CBDC. I guess that means sort of a lot of parties still can potentially access the data that associated making payment through CBDC. Right. So, so in that sense, sort of this issue is still there but maybe in a different form. And also fundamentally I think sort of for PBC to launch the CBDC a very important motivation if I understand right is that the central bank also want to use the data, the detailed data actually to implement all sorts of policies right. Like, you know, do the stimulus actually discriminative discriminative, like say subsidies during the COVID and all that sort of in a sense sort of the PBC also wants to have to be able to use the data is not like if you use CBDC the data will be there and just like cash right that that sort of in a sense of the we're still going to face this trade off under CBDC privacy sharing data provide convenience, but the different sort of convenience, but at some cost of privacy right so you know for consumers you won't see your data to be fully seen by some platform like Alipay or you, you know, versus sort of being seen fully by central bank or you know, a potential company associated with the central bank. I think there's still sort of a trade off. Thank you. I think that's an interesting question and it kind of goes down to what the goal is in terms of actually creating a CBDC. I would argue that maybe the intent or the reason why China is developing a CBDC is not for privacy reasons. And in that case, it's not really a consideration where they're thinking about kind of other intermediaries, helping them produce this or kind of helping them develop this kind of system. Now, kind of caveat here is that when I was doing the presentation I talked about how central banks are better positioned to protect the data now. Kind of in between is kind of a huge array of kind of different levels of trust towards central bank as well and so kind of the underlying assumption is the fact that there is some credibility for the central bank to also commit to not use that data. Now, I think kind of more generally the question is whether the CBDC can be designed so that there's a certain time basic infrastructure where you can have it used in different ways. So you can opt into some sort of privacy usage, and you can also use other kind of options that can be used more useful from policy. And, you know, there I think I'm not sure if they're mutually exclusive, although there is a fundamental difference in terms of whether you want to have that private information of consumers in order to implement policies that are kind of enabled through CBDC, or you want to focus on the consumer welfare perspective and provide them with an additional way to kind of preserve their privacy. Okay. So, so given the time I guess we'll go ahead and wrap up so thank you again to Michael to Rod and to Bruno for a fantastic session. Thanks everyone for your questions and comments has been a really interesting discussion. I've done so before to encourage you to take a look at our website CB and DC dot net, you can find the videos of the past sessions papers and, and the presentation slides there. And you'll also see the schedule for our upcoming seminars. So our next session is July 30 Friday July 30. And it's going to be separate the Central Bank Research Association. Harold do league is presenting parallel digital currencies and sticky prices. Martin Uribe is discussing and Rafael power will moderate. So hope to see everyone again next month until then. Thanks, and goodbye. Okay. Oh, Rod's gone. I was going to ask. Michael, I just