 And starting now. All right, so let me open this series of webinars. So I'm Marky Valdi, the actual president of the International Transportation and Economics Association. I would say good morning, good afternoon, good evening, given that there are people from the West Coast of the US and people from also on the West Coast of China. So well, given the outbreak, we could not organize the annual conference. It was the conference should have been in Benjing. And unfortunately, we could not do it. And so finally, we have decided to organize these four webinar sessions. So maybe Leo will have time to present them. We have made a selection of paper out of the one that have been submitted to the Benjing event. I think we did that because I think it's very important. We thought that it is very important to show that our community of transport economics is alive and is very productive. And as you will see, I think that the papers that are going to be presented are very interesting papers. So if the pandemic situation continues, IT will take other initiatives so that you could share your most recent recent recent. So I will not speak too much. Given the timing and just one thing, as president, it's normal to recall you that you have to please renew your membership to IT or join IT. It is very, very important. I hope the only thing I can say is I hope to see you in person next year. It would be in Rome. And in two years, you will be in Benjing because we have moved this opportunity, this case. I mean, our Chinese colleague and Yakan has offered to organize it again in 2022, which will be in the same time as the Winter Olympic game in China. So just wish you that good health. And I wish you that you will enjoy this webinar. We are already almost 100 connected, which is already quite good. And I will give the floor to Ricardo who is going to chair the session. Thank you very much. OK, thank you, Marc. Thank you to all the attendants to these webinars. It's my pleasure to be here to chair this session. We have four speakers and the idea is the following. Every speaker is going to share his screen with the rest of us. So the speaker is going to manage the pace of the presentation. And then they are going to have 20 minutes to present their papers. And then there will be 10 minutes for questions or general comments or even a short discussion if possible. So the way it works is the following. You have to, the people in the audience who have eventually questions, you have to write down the questions in the chat of the Zoom application that is going to be on the right hand side of your screens. OK, so you have to write down your questions. You cannot interrupt the presenters while they are presenting. OK, and at the end we will see we have time to do some interaction. But in principle, you have to write down your questions in the chat. Then I will collect them and I will propose them to the speakers at the end of the presentation. OK, so that's the way we're going to do it. And I mean, we chose this way of proceeding just to be sure that we don't waste any time so that we can really stick to the schedule that we have set up. OK, so that's all from my side. Thank you again for being there. Good health to everyone. And then I will give the word to our first speaker, which is Guillaume Pomme from Paris School of Economics. And the title of the paper he's presenting is How to Regulate Modern Airports. OK, so Guillaume, the floor is yours. From now, you have 20 minutes. So try to stick to your time and that's it. Go ahead. OK, thanks. So let me just share my screen one second. Is that good? Can you see it? Yes. Yes, OK, so I'll start. Thanks. So first, thanks for having me in this particular condition. Thanks to Marc, Leonardo, Ricardo, the IT team for the organization. I guess it's not easy to organize this, so I'm very grateful that we can still present things. So I'm going to present you a theoretical paper about regulating airports, which is joint work with David Martinot and Jérôme Pouillet. So let me directly introduce what we have in mind, what we call modern airports. So nowadays, airports are more and more privately owned or managed. So as economists, we generally feel that they may abuse market power. So that's why we think that we need to regulate them. But on top of that, there are some specificities of airports that make it difficult to regulate them. So first, airports are multiproject entities. So on the one hand, they provide aeronautical activities. So this is just their car business. They say flight tickets to passengers. But on the other hand, they also provide more and more nowadays commercial services inside the airports, such as food, car parking, and other non-related activities, non-related to aeronautical activities. So this means that airports have multiple sources of revenues and also that there is some kind of demand complementarity between those different activities that we need to take into account. And third, airports also need to incur large infrastructure investments to build terminals, runways, and so on. And this affects consumer demand. This affects congestion. And also this affects profits. So this is a large concern for the regulator as well. And finally, airports generally deal with airlines. And we think that this is in a vertical relationship way. And this may also affect the way you have to regulate airports. You have to take into account the airline sign as well. So this is how I'm going to present to you this paper. I'm going to introduce the model and how do we model this different feature. Then I will talk about optimal regulation and the main lessons that we can learn from them. Then I will present to you how to implement this optimal regulation. And I will talk about traditional regulations, such as price cap regulation, rate of return, or talk a little bit about the debate between single-till versus dual-till regime. And if I have time, I will also go deeper in the airport-aligned relationship. How does it affect regulation and the case in which investments are non-observable. So let's directly dive into the model. So here we stick to a very simple theoretical model. We are going just to assume that there is one airport and one airline. And for now, let's say that they are vertically integrated. So they are the same entity. On the consumer side, consumers can buy erotical services at price P. So they can just buy a ticket and fly. And then they become passengers. And only if they become passengers, then they can also buy commercial services at price P. So they can consume food. They can rent some car parking, be in a particular space in the airport to be comfortable, and so on. And we are going also to assume that the airport is responsible for investments decision. And this investments are going to be called E. And those are going to positively affect the demand for erotical services. I would be more precise on how model this impacts of investments in a minute. So to be more precise on the consumer side, you have a continuum of consumers who have valuation for different activities. So V is going to be the valuation for erotical activities. And V0 is going to be the valuation for commercial services. Notice that the commutative decision function for V also depends on E. So in the next slides, I'll explain exactly how we model the relationship between erotical services and investments. So this gives rise to the interactivity of consumers. And we model this in a very particular way. That is, you can see from the slides that if I consume as a consumer erotical services, then I can get V minus P. So just the valuation minus the price of erotical services if I decide to fly. And if I fly, and only if I fly, I can enjoy the V0 minus P0, the surplus from consumption of commercial services. That is, when I decide whether or not to go to the airport, I only take into account the difference between the valuation for erotical services and the price. I do not take into account commercial services at all. So it means that if initially I don't want to fly, it's not because the airport is offering some cheap car parking or very good food that I will ultimately decide to fly. So the commercial activists cannot compensate. The surplus and the commercial activists cannot compensate for the surplus that I derive on the erotical services. And if we model demand that way, then the demand for erotical services is going to be one minus F of P and E. And the demand for commercial services is going to be one minus F times one minus G. That is, the demand for commercial services is directly conditional open consumption of erotical services, but the reverse is not true. So now what about investments? So investments, again, cost E to the airport. And we have in mind that investments favor the demand for erotical services. This can be because of an increase in capacity. You build an additional terminal. You build more runways. You offer more choice to the consumer, a better service quality. So how does the investment affect the consumer? It's simply, we're simply going to assume that if you have a larger amount of investments, then valuations for erotical services are going to be more likely to be higher. So in the first order, Stochastli dominates the set. So if you invest more, you are more likely to have consumers with higher valuation. So this is the way we model the impact of investments on the airport services. So obviously investments directly affects the demand for erotical services, but it also ultimately indirectly affects the demand for commercial services because if there are more consumer that would like to be passengers, then you have more passengers and then you have more people also consuming what's inside, what the other offers inside the airport. Okay, so what's the, given those demand and investment setting, what is the optimal regulation? So we are just simply going to assume that the regulator can jointly control both prices and the level of investments at the airport. And what the regulator does, it simply maximizes social welfare, which is the sum of the consumer surplus, CS, the profits of the airline and airport as a vertical utility entity, which is PR. And this minus one plus lambda t is simply going to be the cost of subsidizing the airport. So t is going to be the amount of public subsidy that the airport is giving to the airport. And lambda is just the cost of public fund. That is how much it costs for the regulator to spend one euros of subsidy, which is a pretty standard way of modeling those problems. And so the regulator maximizes the social service subject to the condition that the airport at least breaks even. So just to be a little bit more precise, if you look at the profits of the airport and the airline as an entity, you see that you have both profits from commercial activities and aeronautical services. The minus e is how much it costs to the airport to invest and the plus t is how much the airport receives in terms of public subsidy. And C and C zero are marginal costs of production for aeronautical services and commercial services that we assume to be a constant here. So we, and also, sorry, we assume implicitly that this is a single tier regime that is hold the profits of the airport are taken into account to cover the costs of investments. So once you have that and then to solve the problem, you obtain, and I'm not going to enter very much into details here, but you obtain, not surprisingly, hamster water pricing. That is the, you know, the prices, they depend both on the cost of public funds and they are also inversely related to the price elasticity of aeronautical services and commercial services. So what's important here to notice is that the point for the commercial services is to look at the costs that we take into account is the social marginal costs. And the social marginal costs, not only take into account the marginal cost of production, but it also takes into account the fact that when you let more consumer become passenger, you also increase the revenue in commercial activity side and also this increase the consumer surplus. So that's why the social marginal cost is divided into three parts. So what are the main lessons of this optimal regulation? So on the aeronautical service side, you can see from the previous side, sorry, you can see that in that slide, the optimal price for aeronautical services is strictly above the social marginal costs of those services. And so this stems directly from the fact that giving a subsidy to the airport is costly for the regulator. That is we may let the airport makes positive profits or maybe not too negative profits because it's costly to subsidy the airport so that it breaks even. So we price strictly above social marginal costs. What we can also see is that sometimes when for instance the demand for aeronautical services is very elastic, the price is very elastic or if the revenues derived from commercial services are quite important, it is even possible that the optimal regulation entails that the airport makes strictly negative profits without a subsidy. So even if the cost of public funds is positive, even if it's costly to subsidize the airport, it's might be optimal to set the price of the airport even below the private cost, let it make negative profits and offer a subsidy. So those are the main ideas here. You have to balance between the cost of public funds and the fact that you would like to diminish the price of aeronautical services to increase consumer surplus, but also because you attract more consumers of commercial services that way. What we can learn on commercial services side is that if this side is not regulated, then the airport is going to set the monopoly price on that side. So even if you regulate aeronautical services with our model, you find that commercial services are ultimately going to be chosen to be the monopoly price by the airport because consumers are captive once they enter the airport. So these goals are highly in favor of not only regulated the price of aeronautical services, but also regulated the price of commercial services. This is even more important when you look at investments. I do not provide the details here, but the optimal level of investments also depends on both prices, directly for aeronautical services and then directly for commercial services. So ultimately, if you want to regulate investments well, you cannot do that just by regulating the price of aeronautical services. You have to regulate both prices. Okay, so that's about the main feature of the optimal regulation. Now, how can we actually implement this optimal regulation? And for that, we'll just look into one of the most common way of doing that. It's the price cap regulation. And for that, what we propose is a global price cap formula where the regulator sets a global price cap, which is the speed of the line, and also some weights, alpha and alpha are zero. And then the airport freely chooses the price of those two services freely as long as they satisfy the constraints. So this is usually what's been done. It's a little bit more complicated in practice, but what we find here with this model is that if you don't regulate investments along with the regulation, along with the price cap regulation, then not only you have an inefficient level of investments, but also you have inefficient price for aeronautical services and commercial services. So if you just implement the price cap regulation, not only you fail in providing the optimal level of investment, but you also fail in providing the correct prices. For aeronautical and commercial services. And this is because the airport does not fully internalize the consumer surplus, and the price cap regulation is not enough to make the airport fully internalize those additional surplus. And then this is even worse if only the price of aeronautical services is regulated, which is generally the case in practice, this is even worse. That is you don't achieve an efficient level of investments and price out. So what can we do to actually solve this problem? I'm going to go very quickly on this. There is a less common now regulation, which is the rate of return regulation, but it is well known that it's generally prone to other investments. And in all cases, it's also the case. So we just get rid of this rate of return regulation. And instead, what we do is that we propose. Return the right again, just a nice reminder that you should control your time. Okay. So you have for about four minutes or so. Okay. Yes, I have a timer, thanks. Okay, thank you. Thank you. Okay, so what we have is a subsidy penalty regulation. And so we have that the global price cap regulation must be supplemented by this subsidy penalty regulation. And this is actually a very simple regulation. It's just that, look at this formula with the S, it's simply that when the airport invests less than the optimal level of the choose less than the optimal level of investment, then we offer the airport a subsidy. When the airport invests more, then this becomes a penalty. And this marginal subsidy is chosen to be exactly equal to the surplus that the airport doesn't internalize. So when you do that, when you supplement the price cap regulation with this particular regulation on investment levels, then you can actually achieve the optimal regulation. If you just remove one of them, then you fail to implement the optimal regulation at all on both sides, prices and level of investments. And also interestingly from this optimal regulation and implementation problem, we are able to address the single till with this dual till regulation problem. That is whether you should use all the profits of the airports to cover the cost of investments or if you have to allocate shares of this investments to different revenues. And the result here is that in our model, actually the optimal regulation is independent of whether you choose a single till dual till regime as long as the regulator can himself choose how to allocate the burden of investments. That is here, there is no, you can gain nothing by choosing one regime or the other. And so this is because you control initially all prices and investments, then it doesn't matter whether you choose one regime, I mean, you can perfectly choose the regime that fits the best, but none is providing you with an advantage in the regulation. So yes, I have less than two minutes. So I'm just going to very briefly tell you about the specificity of the airport and the airline relationship. So if you assume that the airport and the airline are non vertically integrated, and instead that the airport is offering some price, double with the airline, and then in turn the airline freely sets the electrical price, in that case, the regulator does not have direct control on the price of electrical services and now only has control on the price that the airport offers the airline. So the optimal regulation, can we implement the optimal regulation as before? And very briefly, what we obtain here is that this highly depends on whether or not the airport can price discriminate the airline. So in case the airport has access to two-part tariff, that is the airport can also have, also fix an access charge to the airline, that is some constant amounts of money that you have to pay regardless of the amount that you consume, then the optimal regulation is unchanged and the regulator can implement it exactly as before. But if you forbid price discrimination, then the optimal relation has to be changed and you have to increase the price of electrical services. So the take away from this is that apparently you should not, it's not a good idea to forbid price discrimination here because forbidding it generally creates distortion in how much you can control what the airlines does and then it ultimately affects the design of the optimal regulation and it's become worth. So let me just give you a little summary if that's okay with the time. Yes, you should control the game. Yeah, perfect, thanks. So the big things here is that if you want to implement correctly the optimal regulation, not only you need to use a global price cap regulation on both prices and not only on electrical prices, but you also have to supplement this with some specific policy on investment. In that scenario, single-till and dual-till regime are actually equivalent and you cannot gain anything by choosing one or the other. And finally, for what I've been telling you, here, forbidding price discrimination when the airport and the airline not vertically integrated seems to distort the optimal regulation. So maybe we have to think again about what you should allow in the practice between airports and airlines. And I'm not talking about the last points that I did not cover here. Yes, so I think I'm done, thank you. Okay, thank you, Guillaume. Thank you, you were good on time. Okay, so I think I can collect the questions from the audience in three big questions because actually there is quite a lot of consensus on the questions and they mostly deal with the assumptions of the model. So the first one raised by Marc Ramon and Carlos is on the assumption on airline-airport integration, okay? And Marc was also asking why you don't consider airports as two-sided platforms, okay? So this is the first question about the integration and the assumption why you are assuming this and how can you just try that? Okay, sorry? You can answer if you want to. Yes, but sorry, but I didn't hear the last one. Okay, on airline-airport integration that you are making at the beginning and why you don't consider airports as two-sided platforms. So this is the question. Okay, I'm gonna make the three questions and then you can answer because otherwise it's gonna be... Okay, the second one has to do with a question raised by Tiziana is that in a way you are assuming that consumers are fully myopic in the sense that the facilities from retail are not actually conditioning their decisions, okay? So this is the second question. And the third question from Nicole has to do with the fact that consumers, they all buy in the terminals. So they are in a sense they have trapped, right? So these are the three questions that come from the audience. So if you can give some feedback, fast feedback about these three groups of questions, will be nice. Okay, perfect. So thanks for the question. So about the first one about two-sided platforms. So yes, obviously this is another way of considering this mobilization. So we haven't taken this route. And I actually, the route that we take is a little bit more on the side that airlines do, I mean, two-sided platform, airlines do profit more on the fact that there are more consumers and the airport could, sorry, and the customer services could benefit from the fact that there are more airlines and so on. But here with our assumption, what we would like to emphasize is actually the fact that there is some kind of capture of the airports within, of our consumers on the commercial activity side. So this is obviously different from a two-sided platform, but this is probably investigating another way of seeing these commercial activities as being maybe the most extreme form of consumer being captive of the airport. So I do not say that this is the ultimate way of doing this. And maybe, I mean, probably in two-sided ways are also valuable, but this is one take among the others. So yeah, I guess that's the best answer I can give for this question. Okay, so I don't know if... No, go ahead. If you have anything to say about the other points. So about the consumers that are fully myopic. So I guess that the question is related to the fact that consumers do not foresee that they are going to be exploited by the airport as they, this is a, sorry. In fact, that the retail business does not simulate their demand. So that's the point, I guess. Yes, so actually, so this is the way we model this. And I think that, okay. So this is a bit extreme because we assume that this absolutely does not affect consumers. So probably it affects some consumers, probably people who travel for business much more affected by this. But the idea was to make this very extreme and here form of saying that you are not going to enter the airport if you have lower valuation for integral services that the price is. And this cannot be compensated because you have an incredible, you have incredible restaurants or car parking inside the airport. So obviously we could relax this by saying that maybe there are some consumers that can be interested by the two sides. But I guess that this will just reduce a little bit the effects that we have that the airport would set the monopoly price in case where it's under-regulated. So I guess that we probably can add some parameter here and see what happens, but it's just going to affect the results marginally. The airport is just going to have less market power on consumers in the end. It's not going to change the qualitative results of the analysis. Okay, Kiyom, I think that's enough for the first presentation. Is there anything you wanna say very fast or still? But thanks again for the opportunity. I think we should move forward to the next. Okay, so. I'm going to, yeah, to kill my. Thank you so much, Kiyom. We appreciate your presentation, your insights, thank you. Thank you also for the question from the audience. And then we'll move to the second presenter of the session, which is Mod Janosz Burgost, okay, from University of Copenhagen. And the title of the paper is Commuting and the Gender Pay Gap. So, okay. So now we should see the screen. From multi, if I pronounce it correctly. Hello, do you hear me? Do you see my screen? Yes, now we see you. Okay, I don't know if I pronounce correctly your name, sorry. Yes, it's Malto. And I'm working at the University. Okay, sorry, sorry, sorry for that. Okay, so you know the rules. You can start whenever you want. Try to control your time. Otherwise, I will warn you when you have about five minutes left, okay? Okay, great, thank you. So, thank you for organizing this seminar. And thank you for the opportunity to present our joint work here. It's work together with Ismael Mullalic and Jospa Nomen. And we'll be talking about the commuting and the gender pay gap. So, our motivation is that over the recent decades we've seen a decrease in the gender pay gap, mainly due to increasing labor participation rates for women and reduced gaps in education skills. But in the last 10 years or so, we see that it has been stagnating in most of the OECD countries, whereas Denmark is around 50% depending on how you control for it, but yeah, and this puts Denmark at the lower end of the, compared to the OECD countries. We also see that the household burden still remains unequally shared. So, we are investigating how this unequally shared household burden is transmitted into the labor market. And we claim that the gender commuting gap that you can see on the right-hand side has something to do with it. So, there we see just a density distribution over the commuting distance for men and women separately. And we see that men commute longer distances on average. So, yeah, what is the commuting, the role of the commuting gender commuting gap in explaining the gender wage gap? We draw up on existing literature. We see there's a study from Cleveland and others. They found that children are at the core of gender inequalities and they claim it's mainly because of career interruption, more part-time jobs, reduced working hours and slower wage progression for women that come with a childbirth because of the increase in the household burden to the child. Women would work less hours or maybe have slower wage progression or take breaks and this has long-lasting impacts on the wages. But we claim that it's also, women also reconsider the time allocated to commuting. So, this event study on the birth of the first child will give us a good identification strategy to find more about the commuting distance and the impact on the gender pay gap. So, I will give you already an idea of what our findings are. We will show you that the cost of commuting after the childbirth are a lot larger than for the male counterparts. And we will also see that women receive less compensation for this increased cost compared to men. And we then will use this compensation for commuting to look at whether the impact on the gender pay gap is coming from the reduced distance or from the reduced compensation for it. And I can tell you now that it's the compensation. Yeah, this is my layout. Let's dive right in. This is our data. And we use full-time workers between 2003 and 30. We use data from the Danish administrative like then in a Denmark statistic, which is very detailed. And we modeled in a way that we observed the birth of the first child between 2000 and 2016. So, the timeframe is actually a little wider. So, we have less attrition. So, some of the parents already start with a child that is three years old. Commuting distance is calculated as the distance between workplace and residential address. And I come to the model. Let's see my internet connection. That's not very good. Is it okay? Okay. Okay, good. Okay, this is how the events that he looks like. On the x-axis, we have the event time with a vertical line indicating the birth of the first child. And we have the five years before and the seven years afterwards. And on the y-axis, on the graph to the left, we have the annual income. And this is percentage change to the reference year. So, what we do, it's just a set of dummies where we exclude the reference year before the birth of the first child. So, in the case of the annual income, it would be at minus two. And then we can interpret each of these estimates as, for instance, at time, event time two. Women, roughly 18% less compared to the year minus two before the child birth. And for the annual income, we see that there's quite a large gap of around 20%. So, this is normalized. It's not absolute values. It's compared to the year before. And on the right-hand side, we see that, so this, sorry, the annual income has been shown in the Cleveland paper that I mentioned earlier. And they even show that after 20 years, the gap still exists. So, it's quite a big impact. And we then look at the commuting distance, where again, on the x-axis, we have the time event at the event time from minus five until seven. And then on the y-axis, we have the effect of the event time effect on the compared to the reference year at minus one in this case. And we see that the male trend is pretty much uninterrupted. And they're both converging before and then afterwards women commute less and less. And we see a clear impact of the first child. So, in year two, women commute around 6% less. So, moving on, we look at where does the difference in the commuting distance come from? And there's two options, either you change your job or you move your residence. Here we have the percentage of our sample who moved jobs in the given event time and who moved residence also in the given event time. And what we can see is that the level of job moves is already higher than the level of residence moves. And moreover, we find that the relevance of residence moves after the childbirth, which is what we're interested in, is declining in importance. If we then split the sample into subgroups and carry out the event study again, we look at one sample where we keep the residence fixed and people only move the job in the time after the childbirth. And then the other sample to the right, we have people who move the residence but keep their job. So, we see the variation that comes from each step scenario. And what we can see is that the gap between the gender is larger for the labor market. So, this again, like the one at the slide that I just showed you, shows that the labor market is more important than the residential market. And it also means that women move not closer to their job but rather look for jobs that are closer to their residence location. And this also speaks for the Danish setting because the residential market is quite rigid where the labor market is relatively flexible. Yeah, moving on to the cost of commuting. Here, we estimate a job mobility model in the form of a linear probability model. We also test the hazard models and other specifications. But here, we can keep the full sample of around 3 million and include fixed effect as well. So, how it looks like is that we're on the left-hand side, we have an indicator for the job change in the following period. And then regress it on the log income and the distance in absolute values in kilometers, actually. And these coefficients can be interpreted as how much less is the person inclined to change job if the income is raised by 1% or how much more willing is the person to change job if they are commuting one kilometer longer. We also include a couple of controls which are time invariant. So, yeah, we have a family status, sector controls, firm size, region and job tenure. And some of them family status and sector, they are gender specific because we find strong evidence in the literature that this is the case. We also use individual and time fix as I mentioned. What we can do then is we can follow Mons Foskow and Jos van Nomen, who's my co-author, that we can, that a worker trade off between lower commuting distance and more income. So we can actually take the ratio and this will tell us something about the marginal willingness to pay for commuting. So, if we want to calculate the marginal willingness to commuting and express it in terms of Danish Kroner or monetary value per hour, we would first need to transform it into travel time instead of distance, which we do with the average speed. Then we have this ratio, which indicates the trade off. And then we multiply it by the hourly wage. So we take the annual income divided by the annual hours and multiply it by the hours of the day. So we get the value for a whole workday. And then to make it more comparable, because we have, this would be now a value of Danish Kroner per hour. We would compare it to the hourly wage which is the W bar for men and women before and after, just to give you a better feeling of how it relates to the hourly wage. If you compare it to travel time survey analysis, where often there's also a monetary value for travel time reduction, this result is expected to be higher because it reflects the long-term estimates and includes monetary and time costs, which are important to note. Okay, this is how it looks like. So as I mentioned, the coefficients can be interpreted as 1% increase in income for women before the child leads to 5% less likelihood to change jobs. And similarly, the distance is positive. So an increase of one kilometer in commuting distance or commuting distance that is one kilometer higher can be, it's leading to a higher probability to change job spots, which is quite low, but yeah. So when we look at the trade-off, we see the marginal willingness to pay per hourly wage. And it's expected to be roughly around two. So then in monetary value, it's probably something around 40 euros, if that helps. And we see for men, it's very close, but we see that before and after, but for women, we see that before it's two and then there's a three-fold increase in the marginal willingness to pay. For commuting, which means that women dislike commuting more after the child birth. They want to trade off more money for one hour less of commuting. We can also look at the distributions and look at their individual income. And then it is probably even more clear that we have the density plots for commuting. Commuting costs per average hourly wage again. Finishing about three, four minutes, okay? Yes. And we see women before, men before and men after, they're very close together. And then we see that women have very high costs of commuting. Okay, let's move on to the compensation. So are women compensated for this high cost? And we find, we use a wage regression. And we find that before, men and women have very similar estimates, roughly about 0.1%, which is around 120 kroner, so 15 euros per hour. And afterwards, it's increasing and the gap is increasing between the men and women as well. So we see that the gap goes from in the natural effect to rather like it's 10 fold. So one digit increase. If individually fixed effects are not included, we actually see an increase, which is in line with, for instance, Manning-Ellen Manning. And if we then look at how it relates to the gender pay gap, we find that the distance itself, so what if women would commute the same distance as men has a natural effect, but the compensation differential has actually a substantial effect. So if women would be compensated the same as men for commuting, the gender pay gap would actually be reduced. We find something in the area of three percentage points for a 30% gender pay gap at the means. So the commuting gender gap contributes to the gender income gap, not because of the shorter commutes, but due to less compensation for commuting. So to sum it up, we find that mothers commute less, that the cost of commuting increase. And we find that mothers are less compensated for commuting, although the cost increase. And it is mainly due to this compensation, rather than the shorter commuting distance that this is contributing to the gender income gap. Thank you very much. And I'm looking forward to questions. Hey, Malti, thank you very much for your nice presentation. And also thank you for speaking to your time constraint. Okay, so far there's just a single question from the audience coming from Mark Valdi. He's asking, is there any potential in the gender identity problem between the revenue and the decision to move? So the revenue, the wages, the income. So the question is about the gender identity problem between the revenue and the decision to move. Yes, there are some problems. However, we use, we're very careful with the sample selection first of all, and we also use fixed effects, which should capture a lot of the variation in the sample. I have seen some analysis who also do it the other way, but for our question at hand, I think it's the correct way to do it. And we follow the literature very closely. And Nicole is asking if you control for the fact that women are more likely to control, to work part-time. Yes, we use a full-time sample. Okay, so you just, for some of this. But those analysis, we use a full-time sample. We also check for the part-time, and the effects are even larger, as we would expect. But for some analysis, especially when we look at the cost of commuting, the amount of hours is very important. So we decided to stick with the full-time sample for now. Okay, I don't know if there's any other question from the audience, let me check. Yeah, it's another one. Okay, Mark, do you want to make your own question? Or just a question from, you mean the question from Maria. Are there any difference in the gender gap with respect to cost of commuting time, depending on the job sector or the income level? Yes, I can show you one more slide that I actually prepared. So we see, for instance, do you still see it? Yeah. So here we have one version of the mod. Do you also see my cursor? Yes. Okay, good. Here we see that for high incomes, it's actually larger, but we would also expect that high income earners have a higher willingness to pay. But the gap is larger, or just the willingness to pay? The willingness to pay. Yeah, but not the gender gap. No, we haven't looked at the gap yet. All right. But you might, because what I thought was maybe if you, like the gender gap, if it's like women have less to gain, it's become less compensated by coming along and maybe women with higher wages would be, so maybe then gender gaps would be reduced. I think if we look at the distribution, we see quite clearly that the costs are higher, and women on average have lower incomes. So if we multiply it with the lower income and we still get a higher margin of willingness to pay, I think it's very robust. Yes, on the average. So my question is only that different parts of the distribution, income distribution. Maybe another paper. How do you explain why the spread is much larger? So you have a big peak with not much variances, and then you have a small peak with large variances. Yes, I think this is mainly due to the fitting of the, if you're talking about the tails, I think it's about the fitting of this kernel density, but I see that the variance is larger. I guess if we're talking about mothers, there's still a lot of, they're in very different, so we need to keep in mind that we have, we're looking at parents, and I think that mothers are differently affected by this. So we would have, the share of single mothers is higher, for instance, and the share of, I mean, we don't have it here, but part-time employment is also an issue. I think the variation of people and life circumstances we're looking at is just bigger. I think that to women before, I mean. Okay, thank you. Mati, thank you for, I mean, also to the audience for all your questions, and then we should move to our next speaker. Okay, who is Juan Pablo Montero from Pontificia University of Catolica de Chile, presenting a paper entitled, A Practical Approach for Carving Congestion and Air Pollution, Driving Restrictions with Tall and Vintage Exceptions. Okay, so now we already have Juan Pablo here. Thank you, Ricardo. So whenever you want, you can share your screen and the floor is yours, and remember to stick to your time, which is 20 minutes. All right, thank you, Ricardo, very much, and also thank you, the organizers, for selecting the paper, which is, John worked with Leonardo Basso, who will be handling all the questions, and also Felipe Esipulvia. And so, let's see. Okay, so here you have some pictures of different cities in Latin America, but this is a problem that happens all over the world. Vehicle congestions, and also local air pollution. By local air pollution, I mean carbon dioxide, I'm sorry, monoxide, NOx, and hydrocarbons. So I'm not focusing here on global pollutants like CO2. Okay, so that's important to keep in mind. And so the question is, what do we do about this? And for congestion, of course, we economists understand that the best instrument is congestion pricing, but unfortunately, we don't see that much around the world. Only a few cities have implemented that. New York, apparently, is implementing that as well in New York City, I mean, in 2021. And it may be, I don't know, we'll see that after the COVID-19 pandemic, maybe more cities are willing to do that because they need to create more incentives for people to move from cars to eventually do public transport. I think we've already seen some of that in London. They have increased the congestion charts. But this is an interesting question. What is gonna happen to public transportation and to congestion around the world after the pandemic is over, which may take some time, as we all know. So congestion pricing, we have discussed that, for example, in my home city in Santiago, but currently there is no intention to implement anything like it. And I think in looking to the far future, eventually with the autonomous vehicles, I think there's a need for road pricing, urgent need for road pricing. Absolutely, in order to these vehicles to operate more freely. And so here is a political economy question. So if congestion pricing is so unpopular in many, many places, what are the alternatives? And here there's one, these are driving restrictions, which we've seen them in many places in Latin America, not only Latin America, but also in other places that I'll show you in a minute. So the question is, how can you use these policies that apparently have some support in order to implement some congestion alleviation and at the same time pollution alleviation? And typically for these driving restrictions come in the form of licensing plate bands, in which, for example, depending on the last digits of your license plate, you may not be able to use your car every day of the week, maybe only a few days. In Bogotá today, you can only use it only two and a half days of the week. And so, and these policies in different formats are becoming more and more popular over the years. So I think it was the first to implement something like that, then Santiago. But now you have cities in Europe implementing some formal driving restrictions. For example, the low emission zones in many cities in Germany, these are driving restrictions. I mean, you cannot enter the city center if your car is, if the emissions rates of your car are above certain level. And Madrid also, they have these driving restrictions, Madrid Central, in which if you have a hybrid or electric vehicle, you may enter the city center. If not, you cannot do that. So they come in different formats. And the idea we're proposing here in this paper is how you combine these some exemptions because the problem with these driving restrictions, this goes back, for example, to the Mexico famous policy, Oino Circula, or today you cannot use your car, in which creating incentive for people to buy a second car. This is the typical critique that these driving restrictions policies receive that you create incentives for buying a second car. So what we're proposing here is to create exemptions to these driving restrictions in order to avoid this problem. And at the same time to, of course, increase welfare. And you can start, for example, with a simple design of one or two days a week restrictions and eventually you can go higher with that. And so the idea is the following. The day in which your car is restricted, you will have the option to pay a daily pass or a toll to use your car that day. However, in order to also control for pollution, that exemption will be available only to some cars during the time in which pollution is a problem. When pollution is not a problem, that exemption is available to all cars regardless of their vintage or regardless of clean they are. But during the time of the year in which pollution is a problem, that exemption to pay a toll to use your car the day in which you have a restriction will be only available to relatively clean cars. I'll come back in a minute to what would mean by clean cars. Okay, so here you have the two, you know, the two exemptions working. One is the toll and the second one is a vintage of your car. Those two things together give you absolute exemption from the restriction. And the logic of this vintage thresholds comes from a paper in which we only focus on pollution. What we're adding here is that we are combining these two problems together, pollution and congestion. We're treating them both together. Although, with a short-term perspective, in that paper also we look at the long-term perspective because when you introduce this vintage, driving restrictions, also what happens is that you create incentives, you know, to accelerate the fleet turnover toward cleaner cars. In this paper we are not paying attention to that dynamic aspect of it. And only, you know, we're taking a short-term perspective, say, you know, during the first year when the fleet is constant and consumers don't update their cars. So one question you may have is, do we see these restrictions anywhere else anywhere in the world? And in terms of vintage exemptions, yes, we see them in many places. In Santiago, for example, in 1992, cars that were equipped with a catalytic converter were exempt from the restrictions. And also in Mexico City today, for example, if you have a new car for the first eight years, you are not facing restrictions. You have a electric car, the same thing, or a hybrid. In Paris, for example, you also have these vintage restrictions. Cars that are 98 or older cannot enter the city center during weekdays from, I think, 8 a.m. to 10 p.m., something like that. In Germany's low emission zones is the same. Madrid as well, you know, basically you are differentiating cars depending on their pollution rates. Regarding tall exemptions, basically you can pay a daily pass in order to use your car the day in which you face a restriction. There are only two places, two cities in the world in which, at least I'm aware of this, only these two places in which you can do that. One is in Cali that was implemented back in 2017 and the other one in Bogota that was implemented back in October last year. So these are the only two places in which you can pay a daily pass. It's like a congestion charge, basically. During the day in which you have restriction, you pay this congestion charge and you just can use your car. So I think in any public policy, it's important not only to pay attention to efficiency but also to distributional implications and more so now than ever. So let me tell you a little bit about this efficiency and distributional implication of thinking about this hybrid policy. I think the answer was the first, talking about this, driving restrictions with exemptions. And the idea, I mean, he was saying that maybe we can introduce this policy that with alleviate congestion by leaving everyone better off, rich and poor. And the idea is this, a low income individual probably cannot afford to pay it at all because it's too expensive but still that person's gonna be happy to with this policy because the loss from not using the car maybe one or two days a week is gonna be more than compensated with the fact that he's gonna be able or she's gonna be able to use the car the other three days of the week or four days of the week and commuting is gonna be faster. So everyone is gonna be worse off. I mean, the rich is gonna be using the car all the time. He's able to afford to pay for this toll and he's gonna be commuting faster and the poor will still benefit but perhaps not as much, but it will benefit. Well, it turns out at least for data coming from our home city of Santiago that that is not true. If you introduce a one or two day restrictions and you implement the optimal toll, the rich are better off but the poor are worse off. So it doesn't work. So what we are saying, what one important lesson from this paper is that all these distributional concerns must be handled separately. You should collect all the toll revenues and distributed back to individuals in a way in which you can leave everyone better off. So that's a very important part of our paper. So let me tell you a little bit about model and parameters. It's a very simple model which is based on Basso and Silva. You know, people that had to commute every day to work or to school during peak hours. You have two modes of transportation, cars and public transport. Some people own cars, some people don't. Cars are a three genius in their emissions rates and you have your surplus from commuting. You don't have three components, depending on your preferences for transportation modes. Some people like public transport better than the cars. Also, the cost in comparing using one or the other which depends on the toll that you have to pay on the public transport fare. And then the travel time, which is this combined. Perhaps one, you know, driving by car takes less time than commuting by public transport. And the regulator will have basically here three instruments. The number of data restrictions are, which would be zero, no restrictions. Basically it's a fare to five, which is a full restriction every day. Basically it's like you go to congestion chart because every time in which you want to use your car you have to pay at all. And then the level of this toll, this PC is the level of the toll. And then the emissions rate, this is when pollution is a problem, cars with emissions rate below this threshold are the only ones that are entitled to pay this toll. The other cars cannot do that. And finally also the regulator may want to reduce the bus fare or the public transport fare. Here I talk about buses, but also include Metro as well. And so a very important element of our analysis is the distributional implication of these policies. So we divide commuters in five income groups. What is important here, pay attention. This is the lowest income group and the highest group number five. Pay attention to the car ownership. The low income group only 16% of that group own a car and among the rich 95% own a car. And that is going to play an important role. And up here there are different parameters that enter into the model. I'm going to skip that. And this is also very important. You see, depending on your income group, the car you own is very different. So here you have a distribution of all the cars owned by the different income groups where they're subcompact, compact, mid-sized, SUV, and so on. And also some cars are run on gasoline, petrol, and the other run on diesel. And why this is important is because pollution, they emit different gases, as I can show you in the next slide. Here you have the different type of cars, compact the different age of the car and the different fuel. For example, here you have two pollutants, local pollutants, hydrocarbons and NOx and nitrogen oxides. And you can see, for example, older cars emit a lot more than new cars. This is very different from global pollutants like CO2 in which the emissions rate doesn't change much over the years, actually remains quite constant. But for local pollutants, it makes a huge difference whether your car is a new car or it's an old car. And that is very important to keep in mind. As well, there's an important difference between the type of cars, SUVs emit more than subcompact and compact, for example, and also depending on whether you run your car on diesel or gasoline, there's a difference. All that play a role in designing the optimal policy. And here you have a fit of our model. And let me show you some results. Let me go first to the case in which pollution is not a problem. Here you have only concerns about congestion. And depending on whether you implement a one-day restriction, this is this lower line here or a five-day restriction, depending on the toll that you establish, your surplus is going to change. For example, if you introduce a $13 toll daily pass that you have to pay to use your car and you introduce a five-day restriction, basically this is congestion charge. Every time that you want to use your car, you have to pay this daily toll. Well, you can increase welfare by a good amount, which is about 0.5% of our country's GDP. So we're talking about a big, big number here. So here is one question of whether you want to go. Eventually you want to go to a five-day restriction for political economy reasons. Maybe you want to start with one-day restrictions or two-day restrictions. And as you move on and you get more public support, then you may start moving to a higher level. Now, that was the overall surplus gained from introducing this policy. Now the question is, what happened to the distributional implication of this policy? And let me first look at the case in which all the toll revenues are sent back to the different income groups in the same numbers that were collected. So there's no transfer between income groups. If you do that, the poor, this is the lowest income group, they are really worse off. Why? Well, because mostly here, I will show you why is that. It's because car owners in that group, they cannot afford a toll. And they can have to leave the car at home and they're really worse off. The people that don't own a car also are worse off. But most of the damage is done by people that own a car. And why these people are worse off? For example, the ones that just were using public transfer before, well, because it's more crowded. That's the reason why they're also worse off. But most of the damage comes from the fact that many of these people were commuting by car and now they cannot afford to do that. And the more you move from one to five-day restrictions, here is the dark line, is the five-day restriction, of course, the damage is even higher. What happened to the highest income group? Group number five? Well, these people are very happy. These are the richest people and most of them own a car, they commute by car. Not all of them, but most of them commute by car. And they're very happy because they can commute much faster than before. And that's why they like, for example, a five-day restriction, basically congestion charge. Now, if we look at how the surplus from these people split between those that own a car and between those that do not, again, the people that do not own a car and that they still commute, there are very few, but still there are some. They're worse off because the public transfer system is more crowded. That's the reason why they're worse off. Now, let's think about more intelligent ways in which you can recycle this toll revenues back to the commuters. And one way to do it is to take all this money collected from the toll paid mostly by the rich and to reduce the public transportation fare. And if you do that and if you implement, for example, a five-day restriction, and the toll is set at the optimal level, about $12, $13, well, you can reduce the fare by 70%. Today is about $1 each right is about that. So we can reduce it in 70%. So it's a huge reduction in terms of prices. If you do that, now people in the lowest income group are better off, strictly better off because of this huge transfer. Now, interestingly, of course, the people that were using the subway before and the buses before are better off. This group here to the right, because they are paying less. Even it's more crowded. Still, for them, the fact that they're paying a lot less makes them worse, much better off. And more so, if you introduce a five-day restriction, this upper dark line here, then if you do, if you introduce only a one-day or a two-day restriction. Interestingly, here is a group of people that own a car. Not many, but still some. And for this guy, what would be optimal, if you're going to do this, is to introduce, for example, a five-day restriction, collect a lot of money and give that money back to them in the form of lower fares. So this is the way in which you can make everyone in this lowest income group better off. By implementing a five-day restriction, very aggressive, and the total set at the optimal level. Why? Very simple, because you can collect a lot of money, so you can come up with a big reduction in the public transport fare. Not news for the high income groups. They're still much better off because of the faster commute. And this guy, for example, is still worse off, even if you reduce the public transport fare, these people are still worse off, and the reason is because for these people it's more important the quality of the service than the price. And the fact that now it's more crowded really reduces their benefits. That's the reason why this guy is still worse off, it's prices, this big transfer in terms of lower prices. Now, you can also combine reductions in the public transport fare with improvements in the quality of the service. So you can use the total revenues to do both. And when you do it in the optimal way, at least using, for example, all this data is based on Santiago. In some places, maybe the fraction of money that you want to use to improve the quality of the service could be more than in others. So you should finish in about three minutes. Three minutes, OK. But I'm good. Then you are, for example, in that particular case, for example, here is the highest income group that were commuting by public transport before, and now they're better off because you're also increasing the quality of the transport service. Now, I have two more slides. When you add pollution, things change, but only one way. In the case of Santiago, for example, what happens in spring and summer is very different from what happened in fall and winter. In spring and summer, local pollution is not a problem. So what you have seen is the optimal design. So all car owners should be entitled to pay the toll if they want to. Pollution still goes down by 27% because many people decide not to pay the toll and leave the cars at home. And here, welfare comes mostly from less congestion, 86% of the welfare gain comes from less congestion, 14% from less pollution. In fall and winter, it's completely different. Because now pollution is a serious problem. So here, what you want to do is you want to maintain the toll exemptions but only make them available to gasoline cars that are 16 or less years old. Why? Because all the cars are very, very polluting. And also, you want to differentiate between gasoline and diesel cars. Diesel, they emit more. That's why the threshold is a little bit tighter for diesel cars. And when you do that, and you introduce, for example, a five day restrictions, pollution goes down by 71%. And now, the contributions to welfare between less pollution and less congestion completely reverse. Now pollution in fall and winter is much more important. They contribute with 59% to the welfare gains. And less congestion contributes with 41%. So my last slide, I guess I have one minute, regardless. And driving restrictions, here is a political economy. If you want motivation of working on this paper, we see these driving restrictions as increasingly popular in many places. So the question is how we can use them more wisely in terms to fight both congestion and local pollution. This has been our main motivation. And so following these two papers, you can combine both kind of exemptions, right? Total exemptions and also the intake exemptions to get the best out of these driving restrictions. Otherwise, if you don't introduce these exemptions, you are going to really do really, it's better to do nothing. I mean, we show that in Barona, there are many papers. If you don't introduce these exemptions, better do nothing. I said it again, why? Because if you don't, people are going to buy a second car. There is plenty of evidence documented that. We also do it in the Barona paper documenting that that is a really bad policy. Only you have to combine these restrictions with these exemptions. Otherwise, you're in really bad background. And the reason is because you eliminate this incentive to buy a second car. We think this, I think again, from a political economy perspective, you want to start perhaps with mild restrictions, maybe one day a week, two days a week, and eventually over time, as you get more public support for this policy, as people see less pollution and particularly less congestion, then you can start moving up higher until you reach five days of restrictions. And last point, distributional implications are crucial to be attended. The way to do it is to attend it separately, to take all the collection from the top and to use it wisely, mostly to reduce the public transport fare in order to leave all income groups better off. I finished with that. Okay, one more question. Looking forward to your question. Thank you. Thank you so much for your time. Thank you for the nice presentation. There's a question from Moise, but I don't know if you have already answered at the end of your presentation. He's wondering that the tolls may keep polluting Carson streets, but I think that you probably have already answered this by the end of your presentation. I don't know if you want to add something to that discussion. Marc is asking whether we should have taxes that depend on mileage. And then I have also some questions myself. First one, it comes to the beginning of your presentation. When you talk about driving restrictions, you talk about the restrictions that depend on the number of the plate. So the plate number and also the ones based on low emission zones. And I think these are two demand restrictions. So quantity restrictions that are not price based, but quantity based. And I think they are very, very different. So in the case of the plate number restrictions, you have these incentives to buy a second car. So you may end up with more pollution, more congestion, et cetera. But the fact is not the same with low emission zones. So I think it's interesting to differentiate between the two. So with low emission zones, it may be very regressive and so it may be about policy as well, but you don't have these incentives to buy another car. What did you have incentives to replace you the polluting cars by cleaner cars? Okay. And then what else? Okay, time up. Okay, let's go for the answers. Thank you, Ricardo. Let me start with your questions first. I think they're different, but if you both, we see both of these two restrictions can be understood as vintage restrictions. In which sense? For example, with a license plate, you can say, look, if you have an old car and if the last digit of your license plate ends, I don't know, in one, two, whatever, you cannot just, your car may be one, two days a week. But if you have a clean car, you face no restriction. And with that, you completely eliminate the incentive to buy, you know, a more polluting car to buy past restrictions. What you do, and this is exactly, we know the evidence from Santiago, what you do, you buy a more clean car. That's what you do. So this is the way in which you can kill that incentive, regardless of whether you're using, for example, a license plate ban or a low emission zone, because the low emission zones, you're right. It actually tells you no matter what car, I mean, the only cars that can enter the city are not this type of cars, clean cars. That's it, period. With a license plate, you can get the same idea, simply by providing, you know, the owners an alternative, which is to buy a cleaner car. And they're not going to buy, you know, a second old car to wipe out the restriction. It's more expensive. It's better to buy a cleaner car. And this is exactly the evidence that we found, for example, in Santiago and also in Mexico with the new programs. But at the end, both are vintage, specific restrictions. Now taxing by the use of the car is no so easy, because you can always manipulate, you know, the automator. And it has been proposed many, many times, but it's not so simple. And that's why, for example, in London, what you do is you just tax, you know, at all to enter the city. And so it's difficult to monitor that. I don't know of any system that use that, you know, taxing by the mileage. Just ask Google. Just ask Google. They have all your data. Well, eventually, maybe you can do it. But so far, I don't know of any place in which you do that. I think the place in which you have at all, I mean, you pay for using your car, basically is if you cross a portal and then you have to pay, for example, and this is the same way in which, for example, New York is planning to implement the toll. Right? If you enter, you know, the city, below, I don't know, the 29, I don't know, 59, I think, avenue, in the lower zone of Manhattan, if you enter that zone, then you pay the toll. But if I can, this is the way it's implemented. If I enter in London, with my small car and I keep running, my car, I'm going to pollute a lot compared to my very big car. And just, I will make just one kilometer with my big car. So it's very important. All the system where you have just the toll on access is probably less efficient. And it's for this reason that there is all this system of bypassing. I fully agree with you, Mark. Fully agree. I mean, these are second best or third best, you know, policies. And by this is the reality. Eventually, we would like to charge by the mile and also by the type of vehicle that you own. For example, you have a sport car that emits a lot, or you have a, you know, a smaller car, you should pay less. What London is doing, for example, is also adding to the congestion charge, for example, another charge depending on the type of vehicle that you run, which is, I think it's not the best, you know, it's not the ideal instrument, but it's going in the right direction, trying to, you know, create incentives for you to move to cleaner cars. These are all servers for second best policies, Mark. But I think we haven't yet reached the point in which you're using, you know, a first best instrument, like taxing by the mile at the time of the day in which there's congestion, because this is also, perhaps you have that also in mind, you know, congestion is created, you know, very localized and also depending on the time of the day. But we are not at that point yet. Eventually, I think we're going to move in that direction over time. We can see some of that in the future. But so far, we're talking about just trying to reduce congestion the best we can with, you know, simpler instruments, I think. And this paper is basically pointing in that direction. Using instruments that have some political support, they're not ideal, but at least, you know, provide you some congestion alleviation and also some pollution alleviation. And also there was a question about the toll at the beginning that, yes, you want to differentiate, for example, in London, the way you do it, as I just mentioned before, if you own an older car, you have to pay more. The congestion charge goes up by, I don't know, go up even by 50% or even more. I don't have the numbers on top of my head. But here we're trying to introduce something similar, not exactly equal. Instead of just charging more to the people that own an older car, we're just simply saying, look, because this is, you can just not buy this daily pass. So you cannot just use the toll. You have to leave your car at home. If your car is old enough, during the time in which pollution is a problem. So this is the way we do it here in this paper, which is more consistent with driving policies, which is whether, as you mentioned, regardless of these are quantity-based instruments. Okay, I think we should leave it here. I'm sorry, there are now several more questions in the chat, but we cannot cover all of them. There's a concern about the affordability to buy cars by poor people. But I guess that in your system, what you are doing is to provide subsidies to use public transportation. It's not to give incentives to buy new cars. Anyway, we can eventually keep the discussion later on, but now we need to move forward to the next speaker. Thank you. Thank you, Juan Pablo. Thank you very much. So then we move to our last speaker from this first idea webinar session. Initially, the presenter was Alex Lutman, but finally, the paper is going to be presented by his co-author, Daniel Lat, from GC Irvine. And the title you have already on your screens is Loyalty Rewards and Redemption Behavior, Stylized Facts for the US airline industry. So Daniel, Dan, if you are there, you can go ahead and remember to stick to your time. 20 minutes. Okay. Thank you for having me. This is my first ITA conference, so online, but hopefully look forward to getting to meet a lot of you guys in person next year in Rome. So like was mentioned, my name is Daniel Lat. I'm a graduate student at UC Irvine. This is joint work with Alexander Lutman, who's from the MITRE Corporation. What we're going to be presenting today is Loyalty Rewards and Redemption Behavior, Stylized Facts for the US airline industry. So to give you a sense of kind of where we're going to go for today's presentation, we're going to briefly define the size of the US frequent flyer market. We're going to describe our method that we have to identify frequent flyer tickets in one of the most used airline databases, which is the Department of Transportation's Airline Origin and Destination Survey, also known as the DB1B. We're going to describe how frequent flyer award tickets differ from paid tickets. And we're going to identify some of the characteristics of markets that have large shares of these frequent flyer passengers on it. To give you a preview of what these results are going to look like, we're going to show that frequent flyer tickets have more stops and are on longer and higher fare routes relative to paid tickets. However, it does appear that these differences between frequent flyer tickets and paid tickets is declining over time. That is, as we get further into our sample, we're finding that frequent flyer tickets and paid tickets are more similar. We're also going to show that frequent flyer tickets are more likely used to go to airports with really high seasonal variation in demand. This means airports that see lots of travel in certain times of the year and less travel otherwise. So this is often ski destinations or vacation spots. We're also going to show that markets that have large shares of passengers traveling on frequent flyer awards tend to have lower load factors and also lower fares if they're coming out of a hub airport, but higher fares if they're coming from non-hub airports. So like I said, a lot of this is going to revolve around the business sector, but this is probably the case with the ABB1B database. For those of you who aren't directly familiar with this, this is a 10% sample of all domestic passenger tickets on US carriers and this is reported quarterly. This is one of the main datasets that gets used in both reduced form and structural I.O. It's really common dataset that's been used for wide variety of papers to answer a lot of different questions data set. One of the things that is is a common assumption that most of the papers that use this database take is they assume and they remove tickets with fares below a certain cutoff, usually 20 or 25 dollars. And this is done because they are often assumed that these are either heavily discounted frequent flyer tickets, special sale tickets that may not be available, or various other non-revenue passengers. We wanted to explore how valid is this assumption. We want to know how large is the frequent flyer market and how is it kind of changing over time, and do we have a method to credibly identify frequent flyer award tickets in this data set. Right, so to get a sense of how big this market is, here's for example six large airlines in the United States. We have this for other carriers as well as going back slightly further, but for this slide just kind of shortening it down. Right, and so you can see that frequent flyer tickets are a non-insubstantial portion of the market. So this is reported by airline. It's the percent of revenue passenger miles that were frequent flyer award tickets. You can see there's a lot of variation among the different carriers. Southwest tends to have one of the higher rates of frequent flyer award tickets versus JetBlue is much smaller. Like I said, this is a pretty large segment of the market, right, somewhere around six or seven percent. To give you an idea of these particular six carriers in 2018, this represents roughly just shy of 50 million frequent flyer award tickets. So again, it's a reasonably large part of the market. So the question is how can we identify this? And so not all tickets below this $20 fair cutoff are actually frequent flyer awards, right? There could be zero fair non-revenue passengers. So this can be friends and family of airline employees that might be flying standby, actual airline employees themselves. You can also imagine airlines sometimes give away flights if things are canceled or rerouting and that kind of stuff. Some are actual paid fairs, right? There are discount carriers such as Frontier and Allegiant that actually sell tickets that would be less than $20, right? So these could also show up. To identify these frequent flyer award tickets, we're going to explore a federal regulation that established the TSA security fee also called the September 11 security fee. And crucially, so this is a security fee that's levied on all revenue tickets. So when you go buy a ticket, you'll often see a fee attached to it. Every ticket that is purchased in the United States among domestic destinations has to pay this fee. It helps subsidize the TSA, the security administration that oversees airports. Crucially, the security service fee must also be imposed on passengers who obtain the ticket for air transportation with a frequent flyer award, but you do not impose it on any other non-revenue passengers. And this is what's written in the federal register for how to apply this. Right? So what does this mean? This means that any ticket that you actually purchase, you have to pay the price of the actual ticket plus this fee. And any flight that you're getting for free, you don't end up paying this TSA security fee unless it's a frequent flyer ticket. And so this is really important because in the DB1B database, right, that 10% sample of tickets, it includes, it gives you the fare of the ticket inclusive of these fees. Right? So what this means is if we look at a distribution, well, sorry, before we get to that, there was also a law change that happened in 2014 that changed the structure of how this fare was applied. So the fare depends based on whether or not you have a round-trip ticket and how many segments you are. So in 2002, when this law was initially passed, right, the fee increased for the more segments you had on a one-way trip and then increased additionally for round trips. In 2014, they simplified that fee structure and now you just pay $5.60 for a one-way ticket, irrespective of the number of segments, and you pay $11.20 for any round-trip ticket. So when we go into the DB1B database, if we restrict fares just looking at fares below $20 and do a histogram, this is what it looks like per variety of years. Right? So on the left-hand side here, this is the share of tickets that are under $20. Across the bottom here is just the distribution of where those fares are. For simplicity, I've been to all the fares between $12 and $20 into that one category. So you can see for 2005 and 2010, these are under the old fare structure, right? You can see there are spikes at zero. These would be non-revenue passengers that are not frequent flyer tickets because if they were frequent flyer tickets, they would have to pay the TSA security fee. And then we see spikes at $2, $5, $7, and $10, which is exactly where we would expect to see spikes if these were frequent flyer tickets that were just paying the TSA security fee. We also see there are some tickets that are in this $12 to $20 category, which are likely these heavily discounted fare. So these are tickets that people are actually purchasing that they had to pay the cost of the ticket and the TSA security fee. Right? We can also see looking in 2015 and 2018, this is after that change. Now, again, it simplifies and now we only see a spike at $5 and a spike at $11, but we continue this spike at zero. Right? So again, these are likely these frequent flyer tickets. Now, the other thing we can do is we can exploit the fact that we know what the fare should be for each ticket type, right, depending on round trip and number of cycles. So we can actually divide that out for each carrier. So this is the distribution for Delta. These are flight fares under $20 in 2013, which is in red and 2015, which is in blue. It's arranged in four panels here by what the expected TSA security fare, expected TSA security fee would be in 2013. So this panel here, which is $2.50, you remember back to that previous table, that would be corresponding to a one-way nonstop ticket. And we can see here, right, that there is about 80% of those observations that are tickets under $20 that are one-way nonstop. Right? About 80% of those, we actually see the recorded fare of $2, right? So again, this should really only be frequent flyer tickets because if these were revenue tickets, it would be recording the price of the fare plus the TSA security fee. And if they were non-revenue that weren't frequent flyer tickets, then they should be showing up at this zero. We can also see in 2015, after the policy change, right, that now our spike is at five, and this directly reflects what happened with the TSA security fee change, where now any one-way ticket, irrespective of how many segments, is now $5 and 60 cents. Right? We see the same spikes going on here, where we expect $5. In 2013, this could be a round-trip nonstop, or it could be a one-way with multiple stops, which is why in 2015, you see a spike at both five and 11. And again, for $750 and $10, we see the spikes directly match what we would expect they would be given the TSA security fees. We can see this is relatively similar for other carriers. Here's the same distribution for American. And here's the distribution for Southwest. Right? And you can see Southwest is actually one of those carriers that doesn't do as many of these non-revenue passenger miles. Right? So now that we kind of have established how we can identify what are very likely frequent flyer tickets, we want to know, okay, what are the characteristics of these tickets, and how do they differ from paid tickets? And we're going to do that in this descriptive analysis by estimating the following equation. So on the left-hand side, we're going to have a series of different dependent variables that we want to test how they differ between frequent flyer and non-frequent flyer passengers. We're going to estimate this as an equation of whether or not it is a frequent flyer ticket, and then control for the roundtrip status. We're also going to include a quarter of year fixed effects and airline origin and airline destination fixed effects. The variable of interest that we're looking at here is Beta 1. And what I'm actually going to show you are a series of charts that are going to display Beta 1 and its 95% confidence interval. And we estimate this equation separately for every year, so allowing for a fully flexible model. We're going to do this with the airline destination and airline origin fixed effects, and without, so you can get a sense of how that changes. So see what that looks like. So here, this is different in number of segments. So our left-hand side variable here is the number of segments that each ticket has. Right, looking at the red line first, this is where we don't include the airline origin and airline destination fixed effects. This still has quarter of year fixed effects in it. So we can see here that this is the Beta 1 coefficient. So this is the coefficient on frequent flyer tickets. So we can see that frequent flyer tickets relative to paid tickets have about two tenths more segments on them, on average. Right, controlling for airline origin and airline destination fixed effects, that falls to around 0.1 cent. Right, so what we can say from this is it appears like on average frequent flyer award tickets have more stops. So each segment is separated by stop, right, so there's going to be more stops or more segments than paid tickets. The other interesting thing about this is we can see that this effect is declining over time. Right, when we look at 2015, 2016, 2017, 2018, right, frequent flyer award tickets don't look as different from paid tickets as they did back in 2005. So this is suggested potentially that individuals are now viewing frequent flyer awards as more, you know, direct cash. Right, it's more of a commodification of the frequent flyer. We don't know for sure, but it's at least suggesting that. We can also see how this looks for other variables of interest. So if we look at the difference in distance flown, right, again, you can see frequent flyer award tickets are likely to be longer distances. This can be a function of both that there are more stops, so they might not have as efficient of a route. But it's also an indication that the particular markets they're choosing to fly in might be longer distances. Right, so for instance, in the United States, if you're going to be using a frequent flyer award ticket disproportionately more often to go to someplace like Hawaii, that's going to be a longer distance flown. Again, we see the same relationship where this is declining over time. Right, we can also see this for difference in weighted average fare, frequent flyer award tickets tend to be on slightly more expensive routes. For HHI, it's very noisy, but maybe we can say it's slightly more competitive routes, but again, the same thing where it appears like this relationship deteriorating over time. Finally, the last one is difference in seasonal variation in demand. So this is looking just at the destination where individuals flying to. Higher seasonal variation in demand says that these these are airports that have more spiky demand, right? They're more likely to have more travel in certain parts of the year than others, right? As we can see that people are going to these. These are likely vacation destinations to give you a sense of why we're also thinking that I'm going to show you a map of one particular carrier just simplify. So this is Delta's route network limited to flight segments that had at least 15 percent frequent flyer share on them in quarter one of 2016. So quarter one, this is January, February and March. So this is winter time. You can see that this is Delta. So it has its, you know, normal hub structure where there's a hub of Atlanta, Minneapolis, Salt Lake City, LA, and New York City, right? And you've noticed that there are a whole bunch of flights coming in right here for those of you not familiar with the United States. These are the Rocky Mountains, but these are ski destinations, right? January, February, March. This is prime ski season. This is when you have holidays in the US such as Martin Luther King weekend and President's Day weekend, where people are going to go skiing a lot. You also see a lot of flights going to Hawaii as well, which is another common popular ski destination in the winter. Now we can look at how this changes in the summer, right? So looking at quarter three now, you can see that the flights kind of ship further north, right? Just flipping back and forth between these. You can see that there's a ship more northerly, right? It's warmer now. You can go more north. There is some overlap with, we're still in the Rocky Mountains, but now this is Yellowstone National Park right here. This is Glacier National Park. This is Olympic up here, and Orlando is where Disney World is. It also turns out Traverse City, Michigan has a really big cherry blossom festival that people get very excited about. So we think that's also what's driving there, right? So this is suggestive of the fact that again, these markets that these segments that consumers are choosing to use frequent fliers on are likely more vacation-oriented destinations as opposed to business destinations. Okay, so that was kind of the ticket-level descriptive analysis. We now kind of want to look at the market-level analysis and get a sense to determine how market and product quality characteristics can influence the share of passengers traveling on frequent flyer awards. This gives us some idea of potentially both the consumer's choice, right? Where consumers are choosing to use frequent fliers, what types of market, but also airline-choicing their ability to restrict what markets frequent fliers can use their tickets on. So in this case, we're going to be estimating estimating the equation of the following form. So on the left-hand side, we have the frequent flyer share of a particular carrier's product. And in this case, we're saying a product is a specific route in. So this is direct service or sequence of connections that are in a particular market. A market is an origin destination pair, and this is directional. So the idea is one market would be LAX to LaGuardia Airport, right? That would be separate from LaGuardia to LAX. That would be a market. It's that pair. The particular products that could exist would be airline-specific and would be potentially direct flights, direct round-trip flights, or a separate product that would connect through Chicago, right? We're going to estimate this product-level frequent flyer share as a function of fare of that airline's product, fare interacted with whether or not it left from a hub origin for that carrier, the HHI of the market, an indicator of whether or not this product was a nonstop product, whether it was a round-trip product, what the routing quality of that product is. In this case, routing quality is a ratio of the distance you had to travel compared to the straight-line distance and load factor. Load factor is going to be a measure of capacity constraint. So it's going to say on the product you're traveling for the carrier you're traveling, what's the segment that was most capacity constrained? How full was that segment, right? So what percent of seats were filled by passengers? We're also going to include market-level fixed effects, airline quarter year fixed effects, airline origin fixed effects, and airline destination fixed effects. Now before we can estimate this, right, fare is likely endogenous due to potential selection bias. So we're going to instrument for fare using two different instruments. One is the number of competing products offered by other carriers with an equivalent number of connections, right? So we're going to look within the same market, we're going to look at other products that don't necessarily follow the exact same route, but have the exact same number of stocks, right? This gives us a sense of what the competition is within that kind of market conditional on number of stocks. We're also going to use as an instrument the interaction of product distance with jet fuel price. We're just going these results and then we're done. So to give you just a sense of these summary statistics, we're limiting from years 2005 to 2018. We're going to restrict to airline markets with at least 500 passengers in the quarter and products with at least 50 passengers in the quarter. The key point here is that the frequent flyer share, the average frequent flyer share of these products is about 6.7%. And again to get the to understand what the interpretation of this routing quality is that 1.3 suggests that on average these products are 30% further than just straight line distance, right? So to then run this market level regression as the IV, what we can see is that for consumers leaving from non-hub airports, right? They're more likely to be on higher fair routes. However, when they're leaving from a hub airport, this is likely an airport with a higher level of dominance for that carrier, right? The sign now switches and it's larger in magnitude, right? This is suggested that the airlines actually have more control over what routes passengers are allowed to fly on going out of hubs, right? The other thing to note here is for non-stop routes, right? Again, this is negative, which suggests that there are fewer frequent flyer passengers on non-stop routes. Again, this is maybe that airlines are restricting access to these poor frequent flyers. Routing quality is negative. Remember the interpretation of that is somewhat opposite of what you would expect, right? What this is saying is that conditional on being a connecting route, passengers are frequent flyer passengers are more likely to be on more efficient, right? So that suggests that airlines might restrict their ability to go on non-stop routes, but when passengers are choosing between connecting routes, they're going to choose the more efficient route. Finally, a load factor is relatively large and suggests that on capacity constrained segments, airlines are less likely to allow frequent flyer passengers to use those, right? So hopefully this was kind of illuminating of what kind of the frequent flyer market looks like. And we kind of wanted to get this out as a potential jumping off point for other people to start looking at what the effects might be of various other exogenous shocks on frequent flyer awards. So the idea being you could look at the effective co-sharing of airline branded credit cards. You could look at changes to frequent flyer award programs over time. So we're really excited to kind of share this. We're going to open it up for questions now in the chat, but also if you guys have any questions, feel free to reach out to me or my co-author, Alexander. We're happy to talk about this more. Okay, Dan, thank you so much for your presentation. You also respect the time constrained. So thanks for that. I don't know if there's any question from the audience. So far there were no, there was no question in the chat. There's Mark who is asking why do airlines keep these programs? Airlines keep these programs because they can foster brand loyalty, right? So it gives them incentive for individuals to continue to fly a particular carrier repeatedly, which gives them some degree of market power. So it allows them to hopefully raise prices a little bit. So there's been some evidence of that. Mara Letterman has a paper that suggests that frequent flyer programs might actually be part of the cause of the hub premium effect that airlines can charge more at a hub because individuals are willing to fly that carrier more to get more frequent flyer miles. So it's a way to enhance brand loyalty, to some extent. Yeah, and which also makes this somewhat broadening to other cases where companies are trying to establish brand loyalty through rewards programs and stuff. There'll be this Starbucks has a rewards program. Lots of other online retailers and stuff allow the same sort of point generation, but then might restrict how you can redeem those points. So, well, I don't know if is there any other question so far? So you basically think that we should not neglect this part of the sample, right? Yeah, I don't think we should be throwing out this, you know, six to seven percent of the market just because, you know, we don't know as much about it. We're trying to illuminate here. And, you know, there could even be effects of particular routes, you know, might have higher or lower weighted average fare than what the airline actually cares about because they're subsidizing that because this is fair that lots of people want frequent flyer miles on versus one that not people do. So that's the thing that I find particularly interesting is that in this part of the sample that is often neglected, there are things that are interesting because precisely the behavior is different to the rest of the sample. Yeah, and an interesting question that we're hopefully going to try and dig into as well is kind of who's getting potentially bumped from these flights, right? Are these placements for low fare passengers or are these replacements for potentially higher fare passengers? And that's not abundantly clear. And it's something that people have been looking at. Absolutely. Ricardo may ask a question? Sure, because it's the end. Yeah, very quickly. So I noticed, or maybe there is a tendency downward trend, temporal trend. So it seems like they're being less used. Is it because, you know, the impact of let's say the low cost carriers model is kind of becoming more widespread also includes I would actually caution. It's not necessarily that they're becoming less used, right? If you look at the shares here, there's not necessarily a clear direction in terms of the presented passengers using them. What's happening in these charts where I'm comparing here is that the difference between frequent flyer tickets and paid tickets is changing, right? So they're becoming more similar, which I kind of referenced at the beginning, we think might be due to just the commodification, right? I think consumers used to view these as like special things that you'd save up for a particular trip. And I think now, especially with, you know, there are lots of websites and stuff that tell you how best to use them, consumers are more just using them as direct substitutes for cash, right? As soon as I have enough award miles to buy an efficient ticket that's going to give me a reasonably good reward. I'm just going to do that as opposed to saving it for some particular trip. Ricardo, I think you're muted. I think Ricardo, you don't have your mic. Sorry for that. I was saying that if there are no more questions, I will give the word to the chair of ITS scientific committee, Leo Basso, who had some connectivity problems at the beginning of the webinar, and he joined us later on, but now he's here. So it's my pleasure to give him the word to conclude this first webinar series. Thank you, Ricardo. It's good to see you, hear you, or at least feel that you're closing off all of you. Well, I give my excuses. There was a total collapse of connectivity here. It's an externality of the total love that we are living. Everyone is trying to be connected somehow. So just a few ideas that I was supposed to give at the beginning. We decided, first of all, to cancel. I guess that's pretty obvious. We also decided with the executive committee not to hold a virtual conference. It is already hard enough for people from all the world around the globe to be here. Having a full conference dealing with time zones was impossible. So we decided to hold these four sessions of seminars during June once a week to try to stay connected. And so we had to choose 12 papers. We actually chose 13. The scientific committee did the job twice because they first selected 175. I mean, we received 175 papers. Then we went down to 100 or 120. And then we had to select 13. I know it is disappointing. Many of you prepared these papers. We had to make a choice based on not only on quality, but on coverage of topics and other things. We do hope that the next scientific committee, because this scientific committee is finishing its two years term in a rather strange way, it's going to work well. Now, I do hope these four weeks will give us an efficient second best to the virtual conference. Third best, perhaps I should say. This was a good start. Good papers, good questions. And you can see that people are connected. So it's a good thing to see. And I want to remind you that the next webinar is going to be on Wednesday 10th, the same time. So next week, it's on Wednesday 10th, and not Thursday. We have staggered the date because people may have fixed meetings. It seems that we're meeting a lot. So we did Thursday this week. We're going to do Wednesday next week, same time. And to tease you a little, we're going to have three papers next week, the impact of access prices and train traffic. And our kinematic study, Carlos and Arte Bacares, will present this, the long unwinding roads, Guillermo Sinisterra, and then bike lanes, congestion and road safety, evidence from New York City with Joris Kringen. The chair is going to be Mohen Fosgerau, who's around there. So you can see three very different subjects. Luckily, they all have to do it with the transportation. So I guess that's all I have to say. Thanks, everyone. I want to thank the scientific committee for the effort. One last comment I'd like to make or to remind is that all these webinars will be recorded and uploaded to ITS webpage. And that's it. I don't know if Mark, you want to say some final words? Well, just thank you, everybody. I think I went very well for the first IT webinar. I think it seems to me that this technology will be helpful, not only in this period, but in the future. Thank you very much. And I'll see you next week. Bye.